EA043601B1 - METHODS AND COMPOSITIONS FOR TREATING INEFFECTIVE ERYTHROPOIESIS - Google Patents

Description

Cross reference to related applications

This application claims priority to US Provisional Application No. 61/547932, filed October 17, 2011. All information in the above application is incorporated herein by reference.

State of the art

The mature red blood cell, or red blood cell, is responsible for oxygen transport in the vertebrate circulatory system. Red blood cells contain high concentrations of hemoglobin, a protein that binds oxygen in the lungs at a relatively high partial pressure of oxygen ( _pO2 ) and delivers oxygen to areas of the body with a relatively low _pO2 .

Mature red blood cells are formed from pluripotent hematopoietic stem cells in a process called erythropoiesis. Postnatal erythropoiesis occurs primarily in the bone marrow and red pulp of the spleen. The coordinated action of various signal transduction pathways controls the balance of cell proliferation, differentiation, survival and death. Under normal conditions, red blood cells are produced at a rate that maintains a constant mass of red cells in the body, and production can increase or decrease in response to various stimuli, including increased or decreased oxygen pressure or tissue demand. The process of erythropoiesis begins with the formation of a lineage of competent progenitor cells and progresses through a series of different types of progenitor cells. At the final stages of erythropoiesis, reticulocytes are formed, which are released into the bloodstream and lose their mitochondria and ribosomes, acquiring the morphology of a mature red blood cell. An increased reticulocyte count or increased reticulocyte/red blood cell ratio in the blood indicates an increased rate of red blood cell production.

Erythropoietin (EPO) is widely recognized as the most significant positive regulator of postnatal erythropoiesis in vertebrates. EPO regulates the compensatory erythropoietic response to decreased tissue oxygen pressure (hypoxia) and low red blood cell levels or low hemoglobin levels. In humans, elevated levels of EPO promote the formation of red blood cells by stimulating the formation of erythroid progenitor cells in the bone marrow and spleen. In mice, EPO enhances erythropoiesis mainly in the spleen.

The effects of EPO are mediated through a cell surface receptor belonging to the cytokine receptor superfamily. The human EPO receptor gene encodes a 483 amino acid transmembrane protein, while the active EPO receptor is believed to exist as a multimeric complex, even in the absence of a ligand (See US Pat. No. 6,319,499). The cloned full-length EPO receptor, expressed in mammalian cells, binds EPO with an affinity similar to that of the native receptor on erythroid progenitor cells. Binding of EPO to its receptor causes a conformational change that leads to receptor activation and biological effects, including increased proliferation of immature erythroblasts, increased differentiation of immature erythroblasts, and decreased apoptosis of erythroid progenitor cells (Liboi et al., 1993, Proc Natl Acad Sci USA 90: 11351-11355; Koury et al., 1990, Science 248:378-381).

Doctors use various forms of recombinant EPO to increase red blood cell levels in a number of clinical conditions, and in particular to treat anemia. Anemia, in its broadest sense, is a condition characterized by lower than normal levels of hemoglobin or red blood cells in the blood. In some cases, anemia is caused by a primary problem in the production or lifespan of red blood cells. More often, anemia is secondary to diseases of other systems (Weatherall & Provan (2000) 355 Lancet 1169-1175). Anemia may result from a decreased rate of production or increased rate of destruction of red blood cells or from loss of red blood cells due to bleeding. Anemia can result from a number of disorders, which include, for example, chronic renal failure, treatment with chemotherapy drugs, myelodysplasia, rheumatoid arthritis and bone marrow transplantation.

Treatment with EPO typically causes an increase in hemoglobin of approximately 1 to 3 g/dL in healthy individuals over a week. When administered to individuals with anemia, this treatment regimen often provides significant increases in hemoglobin and red blood cell levels and results in improved quality of life and prolonged survival. EPO is not uniformly effective, and many individuals are refractory to even high doses (Horl et al. (2000) Nephrol Dial Transplant 15, 43-50). More than 50% of patients with cancer have an inadequate response to EPO, approximately 10% with end-stage renal disease have a poor response (Glaspy et al. (1997) J Clin 15 Oncol 1218-1234; Demetri et al. (1998) J Clin 16 Oncol 3412-3425), and less than 10% with myelodysplastic syndrome have a favorable response (Estey (2003) Curr Opin Hematol 10, 60-67). Several factors, including inflammation, iron and vitamin deficiencies, impaired dialysis, aluminum toxicity, and hyperparathyroidism, may predict poor therapeutic response. The molecular mechanisms of resistance to EPO are still unclear. Recent research suggests that higher doses of EPO may be associated with an increased risk of cardiovascular disease, tumor growth, and mortality in some patient populations (Krapf et al., 2009, Clin J Am Soc Nephrol 4:470-480; Glaspy 2009 Annu Rev Med 60: 181-192). Therefore, it is recommended that EPO-based therapeutic compounds (erythropoietin-stimulating agents,

ESA) was administered at the smallest dose sufficient to avoid the need for red blood cell transfusion (Jelkmann et al., 2008, Crit Rev Oncol. Hematol 67:39-61).

Ineffective erythropoiesis is a term used to describe a group of erythroid disorders in which the production of red blood cells is reduced despite an increase in the early stages of erythropoiesis (see, for example, Tanno, 2010, Adv Hematol 2010:358283). Ineffective erythropoiesis often results in anemia, elevated erythropoietin levels, excess red blood cell precursors, and iron overload. These symptoms, in turn, can lead to splenomegaly, liver and heart disease and bone damage, among other complications. Because endogenous erythropoietin levels are typically very high in patients with ineffective erythropoiesis, EPO-based therapeutics often do not treat anemia in these patients or may lead to an increase in other aspects of the disease, such as splenomegaly and iron overload.

Thus, it is an object of the present invention to provide alternative methods for increasing red blood cell levels and/or targeting other disorders associated with ineffective erythropoiesis.

The essence of the invention

In part, the invention demonstrates that GDF traps can be used to treat ineffective erythropoiesis and the disorders and symptoms that are associated with ineffective erythropoiesis, including, but not limited to, anemia, splenomegaly, iron overload, hypercellular bone marrow, elevated endogenous erythropoietin levels, and bone damage . In particular, the invention demonstrates that the GDF decoy, which is a soluble form of the ActRIIB polypeptide with an acidic residue at position 79 in SEQ ID NO: 1, when administered in vivo, increases red blood cell levels in normal healthy individuals as well as in animal models anemia and ineffective erythropoiesis. Surprisingly, in addition to directly increasing red blood cell levels, the molecules described alleviate other symptoms associated with ineffective erythropoiesis, including splenomegaly, iron overload, and bone damage. In some cases, these associated disorders are of equal or greater importance to the patient's health and quality of life than the anemic condition. Thus, in certain embodiments, the invention provides methods of using GDF traps to increase red blood cell and hemoglobin levels in patients and to treat disorders associated with low red blood cell levels in patients in need thereof. In particular, the invention provides methods for using GDF decoy polypeptides to treat complications arising from disorders of ineffective erythropoiesis, including such serious complications as anemia, tissue iron overload, extramedullary erythropoiesis, erythroblast-induced bone pathology, and inappropriately elevated erythropoietin levels. . In such disorders, GDF traps can be used to increase red blood cell levels while reducing the need for red blood cell transfusions and iron chelator therapy, and thus reducing the morbidity and mortality associated with iron accumulation in vulnerable tissues. In part, the GDF trap can be used in combination with existing supportive care for failed erythropoiesis, including red blood cell transfusion and iron chelator therapy. GDF traps can also be used in combination with a hepcidin agonist to treat ineffective erythropoiesis. As described in US Patent Application No. 12/012652, incorporated herein by reference, GDF traps can also be used to increase muscle mass and reduce fat mass.

In certain aspects, the present invention provides GDF traps that are variant ActRIIB polypeptides, including ActRIIB polypeptides with amino- and carboxy-terminal truncations and sequence changes. Optionally, the GDF traps of the invention can be designed to be preferential antagonists of one or more ActRIIB receptor ligands, such as GDF8 (also called myostatin), GDF11, Nodal and BMP7 (also called OP-1). Examples of GDF decoys include a set of ActRIIB-derived variants that have significantly reduced affinity for activin. These options demonstrate the desired effects on red blood cells while reducing effects on other tissues. Examples of such variants include variants that have an acidic amino acid (e.g., aspartic acid, D, or glutamic acid, E) at a position corresponding to position 79 in SEQ ID NO: 1. In certain embodiments, the GDF decoy polypeptide contains an amino acid sequence that includes, consists of, or consists essentially of the amino acid sequence of SEQ ID NO: 7, 26, 28, 29, 32, 37, or 38, and polypeptides that are at least 80, 85, 90, 95, 97, 98, or 99% identical to any of the above.

In certain aspects, the invention provides pharmaceutical preparations comprising a GDF decoy that binds to an ActRIIB ligand, such as GDF8, GDF11, activin (eg, activin B), BMP7 or nodal, and a pharmaceutically acceptable carrier. Optionally, the GDF trap binds to the ActRIIB ligand with a Kd of less than 10 μM, less than 1 μM, less than 100 nM, less than 10 nM, or less than 1 nM. Optionally, the GDF trap inhibits signaling through ActRIIB, such as intracellular signaling events that are triggered by ActRIIB ligand. The GDF trap for use in such a formulation may be any of those described herein, including, for example, GDF traps with an amino acid sequence selected from SEQ ID NO: 2, 3, 7, 11, 26, 28, 29, 32, 37, 38 or 40, or a GDF trap with an amino acid sequence that is at least 80, 85, 90, 95, 97 or 99% identical to an amino acid sequence selected from SEQ ID NO: 2, 3, 7, 11, 26, 28, 29, 32, 37, 38 or 40, or a GDF trap with an amino acid sequence that is at least 80, 85, 90, 95, 97 or 99% identical to an amino acid sequence selected from SEQ ID NO: 2, 3, 7, 11, 26, 28, 29, 32, 37, 38 or 40, wherein the position corresponding to L79 in SEQ ID NO: 1 is an acidic amino acid. Preferably, the GDF trap for use in such a preparation consists of or consists essentially of the amino acid sequence of SEQ ID NO: 26. The GDF trap may include a functional fragment of a naturally occurring ActRIIB polypeptide, such as a polypeptide containing at least 10, 20 or 30 amino acids of the sequence, selected from SEQ ID NO: 2, 3, 7, 11, 26, 28, 29, 32, 37, 38 or 40 or the sequence of SEQ ID NO: 2, without C-terminal 1, 2, 3, 4, 5 or from 10 to 15 amino acids and without 1, 2, 3, 4 or 5 amino acids at the N-terminus. Preferably, the polypeptide contains a truncation relative to SEQ ID NO: 2 or 40 of 2 to 5 amino acids at the N-terminus and no more than 3 amino acids at the C-terminus. The GDF trap may contain one or more changes in the amino acid sequence of the ActRIIB polypeptide (eg, in the ligand binding domain) relative to the native ActRIIB polypeptide. The change in amino acid sequence may, for example, involve glycosylation of the polypeptide when produced in a mammalian, insect or other eukaryotic cell, or involve proteolytic cleavage of the polypeptide relative to the native ActRIIB polypeptide.

A GDF trap may be a fusion protein that contains as one domain an ActRIIB polypeptide (e.g., an ActRIIB ligand binding domain with one or more sequence changes) and one or more additional domains that provide a desired property, such as improved pharmacokinetics, easier cleaning, targeting specific fabrics, etc. For example, the domain of a fusion protein may improve one or more of the following properties: in vivo stability, in vivo half-life, uptake/administration, tissue localization or distribution, protein complex formation, multimerization of the fusion protein, and/or purification. GDF decoy fusion proteins may contain an immunoglobulin Fc domain (wild type or mutant) or serum albumin. In certain embodiments, the GDF fusion trap comprises a relatively unstructured linker located between the Fc domain and the extracellular domain of ActRIIB. This unstructured linker may correspond broadly to the unstructured 15 amino acid region at the C-terminus of the ActRIIB extracellular domain (tail), or it may be an artificial sequence ranging from 3 to 5, 15, 20, 30, 50 or more amino acids that are relatively free in secondary structure. The linker may be rich in proline or glycine residues and may, for example, contain repeating sequences of threonine/serine and glycines (for example, TG ₄ (SEQ ID NO: 13) or SG ₄ (SEQ ID NO: 14) singlets or repeats) or series of three glycines. The fusion protein may include a purification subsequence such as an epitope tag, a FLAG tag, a polyhistidine sequence, and a GST fusion. In certain embodiments, the GDF decoy fusion protein comprises a leader sequence. The leader sequence may be a native ActRIIB leader sequence or a heterologous leader sequence. In certain embodiments, the leader sequence is a tissue plasminogen activator (TPA) leader sequence. In an embodiment, the GDF decoy fusion protein contains the above amino acid sequence of the formula A-B-C. Part B is an N- and C-terminally truncated ActRIIB polypeptide consisting of an amino acid sequence corresponding to amino acids 25131 of SEQ ID NO: 2 or 40. Parts A and C may independently be zero, one or more amino acids, and both parts A and C are heterologous to B. Part A and/or C may be attached to part B via a linker sequence.

Optionally, the GDF trap includes a variant ActRIIB polypeptide with one or more modified amino acid residues selected from the following: a glycosylated amino acid, a pegylated amino acid, a farnesylated amino acid, an acetylated amino acid, a biotinylated amino acid, an amino acid conjugated to a lipid moiety, and an amino acid conjugated to an organic derivatizing agent . The pharmaceutical preparation may also include one or more additional compounds, such as a compound that is used to treat disorders associated with ActRIIB. Preferably, the pharmaceutical preparation is substantially pyrogen-free. In general, it is preferable that the GDF trap is expressed in a mammalian cell line that mediates suitable natural glycosylation of the GDF trap, so as to reduce the likelihood of unwanted immune

- 3 043601 responses from the patient. Human and CHO cell lines have been used successfully, and it is suggested that other common mammalian expression vectors may be suitable.

In certain aspects, the invention provides a packaged pharmaceutical preparation comprising a pharmaceutical preparation described herein and instructions for use in increasing red blood cell levels in humans.

In certain aspects, the invention provides GDF decoys that are soluble ActRIIB polypeptides containing an altered ligand binding (eg, GDF8 binding) domain. GDF traps with altered ligand binding domains may contain, for example, one or more mutations in amino acid residues such as E37, E39, R40, K55, R56, Y60, A64, K74, W78, L79, D80, F82 and F101 of human ActRIIB (numbering according to SEQ ID NO: 1). Optionally, the altered ligand binding domain may have increased selectivity for a ligand, such as GDF8/GDF11, compared to the wild type ActRIIB receptor ligand binding domain. By way of illustration, these mutations are demonstrated herein to increase the selectivity of the altered ligand binding domain for GDF11 (and thus likely GDF8) over activin: K74Y, K74F, K74I, L79D, L79E, and D80I. The following mutations have the opposite effect, increasing the activin binding ratio compared to GDF11: D54A, K55A, L79A and F82A. The overall binding activity (of GDF11 and activin) can be increased by including a tail region or possibly an unstructured linker region, and also by using the K74A mutation. Other mutations that cause an overall decrease in ligand binding affinity include: R40A, E37A, R56A, W78A, D80K, D80R, D80A, D80G, D80F, D80M, and D80N. Mutations can be combined to achieve desired effects. For example, many of the mutations that affect the GDF11:Aktubuh binding ratio have an overall negative effect on ligand binding, and thus these mutations can be combined with mutations that generally increase ligand binding to produce an improved binding protein with selectivity for to ligands. In an illustrative embodiment, the GDF trap is an ActRIIB polypeptide that contains the L79D or L79E mutation, optionally in combination with additional amino acid substitutions, additions or deletions.

Optionally, a GDF trap that contains an altered ligand-binding domain has a ratio of Kd for activin binding to Kd for GDF8 binding that is at least 2-, 5-, 10-, or even 100-fold greater than the ratio for ligand-binding domain. wild type binding domain. Optionally, the GDF trap that contains an altered ligand binding domain has a ratio of IC ₅₀ for activin inhibition to IC ₅₀ for GDF8/GDF11 inhibition that is at least 2, 5, 10, or even 100 times greater than the ratio for the wild-type ActRIIB ligand-binding domain. Optionally, a GDF trap that contains an altered ligand binding domain inhibits GDF8/GDF11 with an IC ₅₀ of at least 2, 5, 10, or even 100 times less than the IC ₅₀ for activin inhibition. These GDF traps can be fusion proteins that include the Fc domain of an immunoglobulin (wild type or mutant). In specific cases, these soluble GDF traps are antagonists (inhibitors) of GDF8 and/or GDF11.

Other GDF traps such as the following are also covered. A GDF decoy fusion protein that contains a portion derived from the sequence ActRIIB of SEQ ID NO: 1 or 39, and a second polypeptide portion wherein the portion derived from ActRIIB corresponds to a sequence that begins with any of amino acids 21-29 of SEQ ID NO: 1 or 39 (optional, which begins with 22-25 of SEQ ID NO: 1 or 39) and ends with any of amino acids 109-134 of SEQ ID NO: 1 or 39, and wherein the GDF decoy fusion protein inhibits signal transduction activin, myostatin and/or GDF11 in cell assays. The above-described GDF decoy fusion protein, wherein the portion derived from ActRIIB corresponds to the sequence that begins with any of amino acids 20-29 of SEQ ID NO: 1 or 39 (optionally, which begins with 22-25 of SEQ ID NO: 1 or 39) and ends with any of amino acids 109-133 of SEQ ID NO: 1 or 39. The above-described GDF decoy fusion protein, wherein the portion derived from ActRIIB corresponds to the sequence that begins with any of amino acids 20-24 of SEQ ID NO: 1 or 39 (optionally beginning with 22-25 of SEQ ID NO: 1 or 39) and ending with any of amino acids 109-133 of SEQ ID NO: 1 or 39. The above-described GDF decoy fusion protein, wherein the portion derived from ActRIIB, corresponds to a sequence that begins with any of amino acids 21-24 of SEQ ID NO: 1 or 39 and ends with any of amino acids 109-134 of SEQ ID NO: 1 or 39. The above-described GDF decoy fusion protein, wherein the portion obtained of ActRIIB, corresponds to a sequence that begins with any of amino acids 20-24 of SEQ ID NO: 1 or 39 and ends with any of amino acids 118-133 of SEQ ID NO: 1 or 39. The above-described GDF decoy fusion protein, wherein part, derived from ActRIIB, corresponds to a sequence that begins with any of amino acids 21-24 of SEQ ID NO: 1 or 39 and ends with any of amino acids 118-134 of SEQ ID NO: 1 or 39. The above-described GDF decoy fusion protein, wherein part , derived from ActRIIB, corresponds to a sequence that begins with any of amino acids 20-24 of SEQ ID NO: 1 or 39 and ends with any of amino acids 128-133 of SEQ ID NO: 1 or 39. The above-described GDF decoy fusion protein, where the ActRIIB-derived portion corresponds to a sequence that begins with any of amino acids 20-24 of SEQ ID NO: 1 or 39 and ends with any of amino acids 128-133 of SEQ ID NO: 1 or 39. The above-described fusion protein -GDF trap, wherein the portion derived from ActRIIB corresponds to a sequence that begins with any of amino acids 21-29 of SEQ ID NO: 1 or 39 and ends with any of amino acids 118-134 of SEQ ID NO: 1 or 39. The above-described fusion a GDF decoy protein, wherein the ActRIIB-derived portion corresponds to a sequence that begins with any of amino acids 20-29 of SEQ ID NO: 1 or 39 and ends with any of amino acids 118-133 of SEQ ID NO: 1 or 39. The above a GDF proteintrap fusion, wherein the portion derived from ActRIIB corresponds to a sequence that begins with any of amino acids 21-29 of SEQ ID NO: 1 or 39 and ends with any of amino acids 128-134 of SEQ ID NO: 1 or 39. The above-described fusion a GDF decoy protein, wherein the ActRIIB-derived portion corresponds to a sequence that begins with any of amino acids 20-29 of SEQ ID NO: 1 and ends with any of amino acids 128-133 of SEQ ID NO: 1 or 39. Surprisingly, the constructs , which begin with 22-25 of SEQ ID NO: 1 or 39, had activity levels higher than proteins with the complete extracellular domain of human ActRIIB. In a preferred embodiment, the GDF fusion protein includes, contains essentially, or contains an amino acid sequence that begins at amino acid position 25 of SEQ ID NO: 1 or 39 and ends at amino acid position 131 of SEQ ID NO: 1 or 39. In other preferred embodiments The GDF decoy polypeptide comprises essentially or contains the amino acid sequence of SEQ ID NO: 7, 26, 28, 29, 32, 37 or 38. Any of the above-described GDF decoy fusion proteins can be prepared as a homodimer. Any of the above-described GDF decoy fusion proteins may have a heterologous portion that consists of an IgG heavy chain constant region, such as an Fc domain. Any of the above-described GDF decoy fusion proteins may contain an acidic amino acid at a position corresponding to position 79 of SEQ ID NO: 1, optionally in combination with one or more additional amino acid substitutions, deletions or insertions relative to SEQ ID NO: 1.

Other GDF traps such as the following are also covered. A GDF decoy protein comprising an amino acid sequence that is at least 80% identical to the amino acid sequence 29-109 of SEQ ID NO: 1 or 39, wherein the position corresponding to 64 in SEQ ID NO: 1 is R or K, and wherein GDF decoy protein inhibits activin, myostatin, and/or GDF11 signaling in a cell assay. The above-described GDF decoy protein, wherein at least one change relative to the sequence of SEQ ID NO: 1 or 39 is located outside the ligand binding pocket. The above-described GDF decoy protein, wherein at least one change relative to the sequence of SEQ ID NO: 1 or 39 is a conserved change located within the ligand binding pocket. The above-described GDF decoy protein, wherein at least one change relative to the sequence of SEQ ID NO: 1 or 39 is a change at one or more positions selected from the group consisting of K74, R40, Q53, K55, F82 and L79. The above-described GDF decoy protein, wherein the protein contains at least one N-X-S/T sequence at a position other than the native N-X-S/T sequence in ActRIIB and at a position outside the ligand-binding pocket.

Other GDF traps such as the following are also covered. A GDF decoy protein comprising an amino acid sequence that is at least 80% identical to amino acid sequence 29-109 of SEQ ID NO: 1 or 39, and wherein the protein contains at least one N-X-S/T sequence at a position other than the native sequence N-X-S/T in ActRIIB, and in a position outside the ligand-binding pocket. The above-described GDF trap protein, wherein the GDF trap protein contains an N at a position corresponding to position 24 of SEQ ID NO: 1 or 39, and an S or T at a position corresponding to position 26 of SEQ ID NO: 1 or 39, and wherein the GDF trap inhibits activin, myostatin and/or GDF11 signaling in cell-based assays. The above-described GDF trap, wherein the GDF trap protein contains R or K at a position corresponding to position 64 of SEQ ID NO: 1 or 39. The above-described GDF trap, wherein the ActRIIB protein contains D or E at a position corresponding to position 79 of SEQ ID NO: 1 or 39, and wherein the GDF trap inhibits activin, myostatin and/or GDF11 signaling in a cell assay. The above-described GDF trap, wherein at least one change relative to the sequence of SEQ ID NO: 1 or 39 is a conserved change located within the ligand binding pocket. The above-described GDF trap, wherein at least one change relative to the sequence of SEQ ID NO: 1 or 39 is a change at one or more positions selected from the group consisting of K74, R40, Q53, K55, F82 and L79. The above-described GDF trap, wherein the protein is a fusion protein further comprising a heterologous moiety. Any of the above-described GDF decoy fusion proteins can be prepared as a homodimer. Any of the above-described GDF decoy fusion proteins may have a heterologous portion that includes an IgG heavy chain constant region, such as an Fc domain.

In certain aspects, the invention provides nucleic acids encoding a GDF decoy polypeptide. The isolated polynucleotide may contain a coding sequence for races

- 5 043601 soluble GDF decoy polypeptide such as described above. For example, the isolated nucleic acid may include a sequence encoding a GDF trap that contains an extracellular domain (e.g., a ligand binding domain) of an ActRIIB polypeptide with one or more sequence changes, and a sequence that could encode part or all of the transmembrane domain and/or the cytoplasmic domain an ActRIIB polypeptide other than a stop codon located within the transmembrane domain or cytoplasmic domain, or located between the extracellular domain and the transmembrane domain or cytoplasmic domain. For example, an isolated polynucleotide encoding a GDF trap may include the full-length ActRIIB polynucleotide sequence, such as SEQ ID NO: 4, with one or more changes, or a partially truncated version of said isolated polynucleotide, further including a transcription termination codon at least six hundred nucleotides before 3' end or other position such that translation of the polynucleotide produces an extracellular domain, optionally fused to a truncated portion of full-length ActRIIB. The nucleic acids described herein can be operably linked to a promoter for expression, and the invention provides cells transformed with such recombinant polynucleotides. Preferably the cell is a mammalian cell, such as a CHO cell.

In certain aspects, the invention provides methods for producing a GDF decoy polypeptide. Such a method may involve expressing any of the nucleic acids (eg, SEQ ID NO: 5, 25, 27, 30, or 31) described herein in a suitable cell, such as a Chinese hamster ovary (CHO) cell. Such a method may include: a) culturing a cell under conditions suitable for expression of a GDF trap polypeptide, wherein said cell is transformed with a GDF trap expression construct; and b) recovering the GDF decoy polypeptide thus expressed. GDF decoy polypeptides can be recovered as crude, partially purified, or highly purified fractions using any well known methods for obtaining protein from cell cultures.

In certain aspects, a GDF decoy polypeptide described herein can be used in a method to facilitate the production of red blood cells or increase the levels of red blood cells in an individual. In certain embodiments, the invention provides methods for treating a disorder associated with low red blood cell counts or low hemoglobin levels (eg, anemia, myelodysplasia syndrome, etc.) or for facilitating red blood cell production in patients in need thereof. The method may include administering to an individual in need thereof an effective amount of a GDF decoy polypeptide. In certain aspects, the invention provides the use of GDF decoy polypeptides to produce a medicament for treating a disorder or condition described herein.

In part, the invention demonstrates that GDF scavengers can be used in combination (e.g., administered at the same time or at different time points, but typically in such a way as to achieve overlapping pharmacological effects) with EPO receptor activators to increase red blood cell levels (erythropoiesis) or treating anemia in patients who require it. In part, the invention demonstrates that a GDF trap can be administered in combination with an EPO receptor activator to synergistically increase red blood cell production in a patient. Thus, the effect of this combination treatment may be significantly greater than the sum of the effects of the GDF scavenger and the EPO receptor activator when administered separately at appropriate doses. In certain embodiments, this synergy may be beneficial in allowing target red blood cell levels to be achieved with lower doses of EPO receptor activator, thereby avoiding potential adverse effects or other problems associated with high levels of EPO receptor activation.

The EPO receptor activator can stimulate erythropoiesis by directly contacting and activating the EPO receptor. In certain embodiments, an EPO receptor activator is one of a class of compounds based on the 165 amino acid sequence of native EPO, generally known as erythropoiesis-stimulating agents (ESAs), examples of which are epoetin alfa, epoetin beta, epoetin delta, and epoetin omega. In other embodiments, ESAs include synthetic erythropoietin proteins (SEPs) and EPO derivatives with non-peptide modifications to impart desired pharmacokinetic properties (extended circulation half-life), examples of which are darbepoetin alfa and methoxy-polyethylene glycol epoetin beta. In certain embodiments, the EPO receptor activator may be an EPO receptor agonist that does not contain an EPO polypeptide backbone or is not generally classified as an ESA. Such EPO receptor agonists may include, by way of non-limiting examples, peptide and non-peptide EPO mimetics, agonist antibodies targeting the EPO receptor, EPO mimetic domain-containing fusion proteins, and extended-acting erythropoietin receptor limited agonists (EREDLA).

In certain embodiments, an EPO receptor activator can stimulate erythropoiesis indirectly, without contact with the EPO receptor, by promoting the production of endogenous EPO. For example,

- 6 043601 hypoxia-inducible transcription factors (HIFs) are endogenous stimulators of EPO gene expression that are suppressed (destabilized) under normal oxygen conditions due to cellular regulatory mechanisms. In part, the invention provides increased erythropoiesis in a patient through combination treatment with a GDF decoy and an indirect EPO receptor activator with HIF-stabilizing properties, such as a propyl hydroxylase inhibitor.

In certain aspects, the invention provides methods for administering a GDF polypeptidal trap to a patient. In part, the invention demonstrates that GDF decoy polypeptides can be used to increase hemoglobin and red blood cell levels. In part, the invention demonstrates that GDF decoy polypeptides can be used to treat patients with ineffective erythropoiesis and to reduce one or more conditions associated with ineffective erythropoiesis, such as anemia, splenomegaly, iron overload, or bone disorders. In certain aspects, the invention provides methods for using GDF decoy polypeptides to treat ineffective erythropoiesis in a patient with thalassemia while reducing the need for red blood cell transfusions and reducing the morbidity and mortality associated with iron accumulation in vulnerable tissues. In particular, the GDF decoy polypeptide can be used in this manner for the treatment of [beta]-thalassemia syndromes, including [beta]-thalassemia intermediate. GDF decoy polypeptides can also be used to treat or prevent other therapeutic conditions, such as promoting muscle growth. Under certain circumstances, when administering a GDF trap polypeptide to promote muscle growth, it may be desirable to monitor the effect on red blood cells during administration of the GDF trap polypeptide, or to determine or adjust the dosage of the GDF trap polypeptide in order to reduce undesirable effects on red blood cells. For example, increases in red blood cell levels, hemoglobin levels, or hematocrit levels can lead to increased blood pressure.

Brief description of drawings

In fig. 1 shows an alignment of the extracellular domains of human ActRIIA (SEQ ID NO: 15) and human ActRIIB (SEQ ID NO: 2) with the residues identified herein based on composite analysis of several crystal structures of ActRIIB and ActRIIA for direct contact with the ligand (ligand -binding pocket) by the indicated frames.

In fig. 2 shows a multiple sequence alignment of various vertebrate ActRIIB and human ActRIIA proteins (SEQ ID NO: 16-23).

In fig. 3 shows the complete amino acid sequence for the GDF trap ActRIIB(L79D 20-134)-hPc (SEQ ID NO: 11), including the TPA leader sequence (double underline), the extracellular domain of ActRIIB (residues 20-134 in SEQ ID NO: 1; underlined ), and hFc domain. The aspartate substituted at position 79 in the native sequence is double underlined and highlighted in color, as is the glycine identified by sequencing as the N-terminal residue in the mature fusion protein.

In fig. Figure 4 shows the nucleotide sequence encoding ActRIIB(L79D 20-134)-hFc. SEQ ID NO: 25 corresponds to the sense strand, and SEQ ID NO: 33 corresponds to the antisense strand. The TPA leader sequence (nucleotides 1-66) is underlined by a double line, and the extracellular domain of ActRIIB (nucleotides 76-420) is underlined by a single line.

In fig. 5 shows the complete amino acid sequence for the truncated GDF trap ActRIIB(L79D 25-131)-hFc (SEQ ID NO: 26), including the TPA leader sequence (underlined by a double line), the truncated extracellular domain of ActRIIB (residues 25-131 in SEQ ID NO: 1; underlined with one line), and hFc domain. The aspartate substituted at position 79 in the native sequence is double underlined and highlighted in color, as is the glutamate identified by sequencing as the N-terminal residue in the mature fusion protein.

In fig. Figure 6 shows the nucleotide sequence encoding ActRIIB(L79D 25-131)-hPc. SEQ ID NO: 27 corresponds to the sense strand, and SEQ ID NO: 34 corresponds to the antisense strand. The TPA leader sequence (nucleotides 1-66) is underlined by a double line, and the truncated extracellular domain of ActRIIB (nucleotides 76-396) is underlined by a single line. Also provided is the amino acid sequence for the extracellular domain of ActRIIB (residues 25-131 in SEQ ID NO: 1).

In fig. 7 shows the amino acid sequence for the truncated GDF trap ActRIIB(L79D 25-131)-hPc without the leader sequence (SEQ ID NO: 28). The truncated extracellular domain of ActRIIB (residues 25-131 in SEQ ID NO: 1) is underlined. The aspartate substituted at position 79 in the native sequence is double underlined and highlighted in color, as is the glutamate identified by sequencing as the N-terminal residue in the mature fusion protein.

In fig. 8 shows the amino acid sequence for the truncated GDF trap ActRIIB(L79D 25-131) without the leader sequence, hFc domain and linker (SEQ ID NO: 29). The aspartate substituted at position 79 in the native sequence is double underlined and highlighted in color, as is the glutamate identified by sequencing as the N-terminal residue in the mature fusion protein.

In fig. 9 shows an alternative nucleotide sequence encoding

- 7 043601

ActRIIB(L79D 25-131)-hFc. SEQ ID NO: 30 corresponds to the sense strand, and SEQ ID NO: 35 corresponds to the antisense strand. The TPA leader sequence (nucleotides 1-66) is double underlined and the truncated ActRIIB extracellular domain (nucleotides 76-396) is single underlined, and substitutions in the wild-type extracellular domain nucleotide sequence are double underlined and highlighted (compare with SEQ ID NO: 27, Fig. 6). Also provided is the amino acid sequence for the extracellular domain of ActRIIB (residues 25-131 in SEQ ID NO: 1).

In fig. 10 shows nucleotides 76-396 (SEQ ID NO: 31) of the alternative nucleotide sequence shown in FIG. 9 (SEQ ID NO: 30). The same nucleotide substitutions indicated in Fig. 9, here also underlined with a double line and highlighted in color. SEQ ID NO: 31 encodes only the truncated extracellular domain of ActRIIB (corresponding to residues 25-131 in SEQ ID NO: 1) with the L79D substitution, for example, ActRIIB(L79D 25-131).

In fig. 11 shows the effect of ActRIIB(L79D 25-131)-hFc on hemoglobin concentration in a mouse model of chemotherapy-induced anemia. Data represent mean ±SEM. **, P < 0.01 compared with paclitaxel at the same time point. This GDF trap reverses the anemia induced by paclitaxel treatment.

In fig. 12 shows the effect of ActRIIB(L79D 25-131)-hFc on red blood cell (RBC) levels in the unilateral nephrectomy chronic kidney disease (NEPHX) mouse model. Data represent mean ±SEM. ***, p<0.001 compared to the original data. This GDF trap reverses the course of nephrectomy-induced anemia in control mice.

In fig. 13 shows the effect of ActRIIB(L79D 25-131)-hFc on red blood cell (RBC), hemoglobin (HGB) and hematocrit (HCT) levels in the unilateral nephrectomy chronic kidney disease (NEPHX) mouse model. Data represent mean changes from baseline at 4 weeks (±SEM). *, P<0.05; **, P<0.01; ***, P < 0.001 compared with NEPHX controls. This GDF trap prevents the decline in these erythrocyte parameters associated with nephrectomy, increasing each of them to values similar to those in sham mice.

In fig. 14 shows the effect of ActRIIB(L79D 25-131)-hFc on red blood cell (RBC) levels in a rat model of acute blood loss-induced anemia. Blood removal was performed on Day -1, with dosing on Days 0 and 3. Data are means ±SEM. **, P<0.01; ***, P < 0.001 compared to solvent at the same time point. This GDF trap improves the rate and extent of recovery from blood loss-induced anemia.

In fig. 15 shows the effect of treatment with ActRIIB(L79D 20-134)-hFc (gray) or ActRIIB(L79D 25-131)-hFc (black) on the absolute change in red blood cell concentration from baseline in cynomolgus monkeys. VEH = solvent. Data represent mean ±SEM. n=4-8 per group.

In fig. 16 shows the effect of treatment with ActRIIB(L79D 20-134)-hFc (gray) or ActRIIB(L79D 25-131)-hFc (black) on the absolute change in hematocrit from baseline in cynomolgus monkeys. VEH = solvent. Data represent mean ±SEM. n=4-8 per group.

In fig. 17 shows the effect of treatment with ActRIIB(L79D 20-134)-hFc (gray) or ActRIIB(L79D 25-131)-hFc (black) on the absolute change in hemoglobin concentration from baseline in cynomolgus monkeys. VEH = solvent. Data represent mean ±SEM. n=4-8 per group.

In fig. 18 shows the effect of treatment with ActRIIB(L79D 20-134)-hFc (gray) or ActRIIB(L79D 25-131)-hFc (black) on the absolute change in circulating reticulocyte concentration from baseline in cynomolgus monkeys. VEH = solvent. Data represent mean ±SEM. n=4-8 per group.

In fig. 19 shows the effect of combination treatment with erythropoietin (EPO) and ActRIIB(L79D 25-131)-hFc for 72 hours on hematocrit in mice. Data are means ±SEM (n=4 per group), and means that are significantly different from each other (p<0.05, unpaired t test) are indicated by different letters. Combination treatment increased hematocrit by 23% compared to vehicle, a synergistic increase greater than the sum of the individual effects of EPO and ActRIIB(L79D 25-131)-hFc.

In fig. 20 shows the effect of combined treatment with EPO and ActRIIB(L79D 25-131)-hFc for 72 hours on hemoglobin concentrations in mice. Data are means ±SEM (n=4 per group), and means that are significantly different from each other (p<0.05) are indicated by different letters. The combination treatment increases hemoglobin concentrations by 23% compared to the vehicle, which is also a synergistic effect.

In fig. 21 shows the effect of combined treatment with EPO and ActRIIB(L79D 25-131)-hFc for 72 hours on red blood cell concentrations in mice. Data are means ±SEM (n=4 per group), and means that are significantly different from each other (p<0.05) are indicated by different letters. The combination treatment increases concentrations by 20% compared to the solvent, which is also a synergistic effect.

- 8 043601

In fig. 22 shows the effect of combined treatment with EPO and ActRIIB(L79D 25-131)-hFc for 72 hours on the number of erythropoietic progenitor cells in the spleen of mice. Data are means ±SEM (n=4 per group), and means that are significantly different from each other (p<0.05) are indicated by different letters. While EPO alone dramatically increased the number of basophilic erythroblasts (BasoE) through late-stage progenitor maturation, the combination treatment increased BasoE numbers to a lesser but still significant extent while maintaining undiminished late-stage progenitor maturation.

Fig. 23 compares RBC parameters in the Hbb ^−/− mouse model of β-thalassemia with those in wild-type (WT) mice. Blood samples from untreated mice aged 2–6 months were analyzed to determine red blood cell count (RBC; A), hematocrit (HCT; B), and hemoglobin concentration (Hgb; C). Data represent mean ±SEM (n=4 per group), p<0.001. Hbb ^−/− mice were confirmed to have severe anemia.

In fig. 24 shows the effect of ActRIIB(L79D 25-131)-mFc on the number of RBCs in the Hbb ^-/- β-thalassemia mouse model. Blood samples were collected after four weeks of treatment. Data are means of 2 per group, with bars indicating range. ActRIIB(L79D 25-131)-mFc treatment halved RBC deficiency in Hbb ^−/− mice.

In fig. 25 shows the effect of ActRIIB(L79D 25-131)-mFc on RBC morphology in the Hbb ^- / ^- β-thalassemia mouse model.

Images of Giemsa-stained blood smears from mice treated for four weeks were taken at 100× magnification. Note hemolysis, cell debris, and many small or irregularly shaped RBCs in the blood of vehicle-treated Hbb ^−/− mice. In comparison, treatment with ActRIIB(L79D 25-131)-mFc significantly reduced hemolysis, cellular breakdown, and the appearance of irregularly shaped RBCs, while increasing the number of normal shaped RBCs.

In fig. 26 shows the effect of ActRIIB(L79D 25-131)-mFc treatment for two months on RBC numbers in the Hbb β-thalassemia mouse model, with data from vehicle-treated wild-type mice shown for comparison. Data represent mean ±SEM; n=7 per group. **, P < 0.01 compared with vehicle-treated Hbb ^- / ^- mice. Treatment with ActRIIB(L79D 25131)-mFc reduced the mean RBC deficiency in Hbb ^−/− mice by more than 50%.

In fig. 27 shows the effect of ActRIIB(L79D 25-131)-mFc treatment for two months on serum bilirubin levels in the Hbb ^-/- β-thalassemia mouse model, with data from vehicle-treated wild-type mice shown for comparison. Data represent mean ±SEM. ###, P < 0.001 compared with vehicle-treated wild-type mice; *, P < 0.05 compared with vehicle-treated Hbb ^−/− mice. ActRIIB(L79D 25-131)-mFc treatment significantly reduced total bilirubin levels in Hbb ^−/− mice.

In fig. 28 shows the effect of ActRIIB(L79D 25-131)-mFc treatment for two months on serum EPO levels in the Hbb ^-/- β-thalassemia mouse model, with data from vehicle-treated wild-type mice shown for comparison. Data represent mean ±SEM. ###, P < 0.001 compared with vehicle-treated wild-type mice; *, P < 0.05 compared with vehicle-treated Hbb ^−/− mice. Treatment with ActRIIB(L79D 25-131)-mFc reduced mean circulating EPO levels in Hbb ^−/− mice by more than 60%.

In fig. 29 shows the effect of ActRIIB(L79D 25-131)-mFc on splenomegaly in the Hbb ^-/- β-thalassemia mouse model, with data from vehicle-treated wild-type mice shown for comparison. A. Mean ±SEM from mice starting at three months of age following treatment with 1 mg/kg twice weekly for two months. ###, P < 0.001 compared with vehicle-treated wild-type mice; *, P < 0.05 compared with vehicle-treated Hbb ^−/− mice. B. Typical spleen sizes observed in a separate study in mice starting at 6-8 months of age following treatment with 1 mg/kg twice weekly for three months. Treatment with ActRIIB(L79D 25131)-mFc significantly reduced spleen weight in Hbb ^−/− mice.

In fig. 30 shows the effect of ActRIIB(L79D 25-131)-mFc treatment for two months on bone mineral density in the Hbb ^-/- β-thalassemia mouse model, with data from vehicle-treated wild-type mice shown for comparison. Mean ±SEM based on femur analysis. #, P < 0.05 compared with vehicle-treated wild-type mice; *, P < 0.05 compared with vehicle-treated Hbb ^−/− mice. Treatment with ActRIIB(L79D 25-131)mFc normalized bone mineral density in Hbb ^−/− mice.

In fig. 31 shows the effect of ActRIIB(L79D 25-131)-mFc treatment for two months on parameters of iron homeostasis in the Hbb ^-/- β-thalassemia mouse model. Mean ±SEM for serum iron (A), serum ferritin (B), and transferrin saturation (C). *, P<0.05; **, P < 0.01 compared with vehicle-treated Hbb ^−/− mice. Treatment with ActRIIB(L79D 25-131)-mFc significantly reduced each parameter of iron overload in Hbb ^−/− mice.

In fig. 32 shows the effect of treatment with ActRIIB(L79D 25-131)-mFc for two months on

- 9 043601 tissue iron overload in a mouse model of β-thalassemia Hbb ^-/- . Iron levels in tissue sections (200 µm) from spleen (A-C), liver (DF) and kidney (GI) were determined using Perls staining. Iron staining in wild-type spleen (A) was abundant in the red pulp (arrows) but absent in the white pulp (*). Increased levels of iron staining in the spleen of Hbb' ^/_ mice (B) reflect expansion of red pulp areas due to extramedullary erythropoiesis. ActRIIB(L79D 25-131)-mFc in Hbb' ^/_ mice reduced splenic erythropoiesis and restored the wild-type splenic iron staining pattern (C). In addition, abnormal iron staining in Kupffer cells in the liver (H, arrows) and renal cortex (E, arrows) was normalized in Hbb' ^/_ mice by ActRIIB(L79D 25-131)-mFc (F and I). Magnification, 200χ.

In fig. 33 shows the effect of ActRIIB(L79D 25-131)-mFc treatment for two months on hepcidin mRNA levels in the liver of a mouse model of Hbb β-thalassemia. Mean ±SEM; *, P < 0.05 compared with vehicle-treated Hbb ⁻ / ⁻ mice. ActRIIB(L79D 25-131)-mFc treatment significantly increased hepcidin mRNA expression in Hbb ⁻ / ⁻ mice.

In fig. 34 shows the effect of ActRIIB(L79D 25-131)-mFc on circulating reactive oxygen species (ROS) levels in the Hbb ^-/- β-thalassemia mouse model, with data from vehicle-treated wild-type mice shown for comparison. Data represent geometric mean ±SEM. ###, P < 0.001 compared with vehicle-treated wild-type mice; ***, P < 0.001 compared with vehicle-treated Hbb ^−/− mice. Treatment with ActRIIB(L79D 25-131)mFc significantly reduced ROS in Hbb ^−/− mice.

Detailed Description of the Invention

1. General overview.

The transforming growth factor beta (TGF-β) superfamily includes a number of growth factors that share common sequence elements and structural motifs. These proteins are known to have biological effects on a range of cell types in both vertebrates and invertebrates. Members of the superfamily perform important functions during embryonic development in the formation of structures and tissue specialization and can influence a number of differentiation processes, including adipogenesis, myogenesis, chondrogenesis, cardiogenesis, hematopoiesis, neurogenesis and epithelial cell differentiation. By manipulating the activity of a member of the TGF-β family, it is often possible to cause significant physiological changes in the body. For example, the Piedmontese and Belgian Blue cattle breeds carry a loss-of-function mutation in the GDF8 gene (also called myostatin) that causes a marked increase in muscle mass. Grobet et al., Nat Genet. 1997, 17(l):714. Additionally, in humans, inactive GDF8 alleles are associated with increased muscle mass and, reportedly, exceptional strength. Schuelke et al., N Engl J Med 2004, 350:2682-8.

TGF-β signaling is mediated by heteromeric type I and type II serine-threonine kinase receptor complexes that phosphorylate and activate downstream Smad proteins upon ligand stimulation (Massague, 2000, Nat. Rev. Mol. Cell Biol. 1: 169-178). These type I and type II receptors are transmembrane proteins consisting of a ligand-binding extracellular domain with a cysteine-rich region, a transmembrane domain, and a cytoplasmic domain with predicted serine/threonine specificity. Type I receptors are required for signal transduction. Type II receptors are required for ligand binding and expression of type I receptors. Type I and type II activin receptors form a stable complex upon ligand binding, resulting in phosphorylation of type I receptors by type II receptors.

Two related type II receptors (ActRII), ActRIIA and ActRIIB, have been identified as type II receptors for activins (Mathews and Vale, 1991, Cell 65:973-982; Attisano et al., 1992, Cell 68:97-108). In addition to activins, ActRIIA and ActRIIB can interact biochemically with several other TGF-β family proteins, including BMP7, Nodal, GDF8 and GDF11 (Yamashita et al., 1995, J. Cell Biol. 130:217-226; Lee and McPherron, 2001 , Proc. Natl. Acad. Sci. 98:9306-9311; Yeo and Whitman, 2001, Mol. Cell 7: 949-957; Oh et al., 2002, Genes Dev. 16:2749-54). ALK4 is a major type I receptor for activins, particularly activin A, and ALK-7 may also serve as a receptor for activins, particularly activin B. In certain embodiments, the present invention relates to antagonism of a ligand of ActRIIB receptors ( also referred to as ActRIIB ligand) using the subject matter of the GDF decoy polypeptide. Examples of ActRIIB receptor ligands include some members of the TGF-β family, such as activin A, activin B, Nodal, GDF8, GDF11 and BMP7.

Activins are dimeric polypeptide growth factors that belong to the TGF-β superfamily. There are three main forms of activin (A, B, and AB), which are homo/heterodimers of two closely related β subunits (β _A β _A , β β b b and β _A β b , respectively). The human genome also encodes activin C and activin E, which are mainly expressed in the liver, and heterodimeric forms including β _C or β _E are known. Within the TGF-β superfamily, activins are unique and multifunctional factors that can stimulate hormone production in ovarian and placental cells, support neural cell survival, positively or negatively influence cell cycle progression depending on cell type, and induce mesotherapy.

- 10 043601 dermal differentiation, at least in amphibian embryos (DePaolo et al., 1991, Proc Soc Ep Biol Med. 198:500-512; Dyson et al., 1997, Curr Biol. 7:81-84; Woodruff , 1998, Biochem Pharmacol 55:953963). In addition, erythroid differentiation factor (EDF) isolated from stimulated human monocytic leukocytes was found to be identical to activin A (Murata et al., 1988, PNAS, 85:2434). Activin A is believed to promote erythropoiesis in the bone marrow. In some tissues, activin signaling is antagonized by its related heterodimer, inhibin. For example, during the release of follicle-stimulating hormone (FSH) from the pituitary gland, activin promotes FSH secretion and synthesis, while inhibin prevents FSH secretion and synthesis. Other proteins that may regulate activin bioactivity and/or binding include follistatin (FS), follistatin-related protein (FSRP) and _α2 -macroglobulin.

Nodal proteins are involved in the induction and formation of mesoderm and endoderm, as well as the subsequent organization of axial structures such as the heart and stomach in early embryogenesis. It has been shown that dorsal tissue in the developing vertebrate embryo participates primarily in the axial structures of the notochord and prechordal plate, although it recruits surrounding cells to form non-axial embryonic structures. Nodal appears to signal through both type I and type II receptors and intracellular effectors known as Smad proteins. Recent studies support the idea that ActRIIA and ActRIIB serve as type II receptors for Nodal (Sakuma et al., Genes Cells. 2002, 7:401-12). Nodal ligands are proposed to interact with their cofactors (e.g., cripto) to activate activin receptors type I and II, which phosphorylate Smad2. Nodal proteins are involved in a variety of events important for early vertebrate embryogenesis, including mesoderm formation, anterior formation, and left-right axis determination.

Experimental data showed that the Nodal signal activates pAR3-Lux, a luciferase reporter gene that previously showed a response exclusively to activin and TGF-β. However, Nodal is unable to induce pTlx2-Lux, a reporter gene that responds exclusively to bone morphogenetic proteins. The latest results provide direct biochemical evidence that the Nodal signal is mediated by both Smad proteins of the activin-TGF-e pathway, Smad2 and Smad3. Additional research has shown that Nodal signaling requires the extracellular protein cripto, which distinguishes this signal from activin or TGF-β.

Growth and differentiation factor-8 (GDF8) is also known as myostatin. GDF8 is a negative regulator of skeletal muscle mass. GDF8 is expressed at high levels in developing and adult skeletal muscle. The GDF8 null mutation in transgenic mice is characterized by severe hypertrophy and hyperplasia of skeletal muscle (McPherron et al., Nature, 1997, 387:83-90). A similar increase in skeletal muscle mass occurs with naturally occurring GDF8 mutations in cattle (Ashmore et al., 1974, Growth, 38:501-507; Swatland and Kieffer, J. Anim. Sci., 1994, 38:752-757; McPherron and Lee, Proc. Natl. Acad. Sci. USA, 1997, 94:12457-12461; and Kambadur et al., Genome Res., 1997, 7:910-915) and, strikingly, in humans (Schuelke et al. , N Engl J Med 2004;350:2682-8). Studies have also shown that muscle wasting associated with HIV infection in humans is accompanied by increased expression of the GDF8 protein (Gonzalez-Cadavid et al., PNAS, 1998, 95: 14938-43). In addition, GDF8 can regulate the production of muscle-specific enzymes (eg creatine kinase) and regulate myoblast cell proliferation (WO 00/43781). The GDF8 propeptide can be non-covalently bound by the mature GDF8 domain dimer, inactivating its biological activity (Miyazono et al. (1988) J. Biol. Chem., 263: 6407-6415; Wakefield et al. (1988) J. Biol. Chem., 263 7646-7654 and Brown et al (1990) Growth Factors, 3: 35-43). Other proteins that bind to GDF8 or structurally related proteins and inhibit their biological activity include follistatin and possibly follistatin-related proteins (Gamer et al. (1999) Dev. Biol., 208: 222-232).

Growth and differentiation factor-11 (GDF11), also known as BMP11, is a secreted protein (McPherron et al., 1999, Nat. Genet. 22: 260-264). GDF11 is expressed in the tail bud, limb bud, maxillary and mandibular arches, and dorsal root ganglia during mouse development (Nakashima et al., 1999, Mech. Dev. 80: 185-189). GDF11 plays a unique role in patterning both mesodermal and neural tissue (Gamer et al., 1999, Dev Biol., 208:222-32). GDF11 has been shown to be a negative regulator of chondrogenesis and myogenesis in the developing chick limb (Gamer et al., 2001, Dev Biol. 229:407-20). Expression of GDF11 in muscle also suggests a role in regulating muscle growth, similar to GDF8. In addition, the expression of GDF11 in the brain suggests that GDF11 may also have activities that are related to the functioning of the nervous system. Notably, GDF11 was found to inhibit neurogenesis in the olfactory epithelium (Wu et al., 2003, Neuron. 37: 197-207).

Bone morphogenetic protein (BMP7), also called osteogenic protein-1 (OP-1), is a well-known inducer of cartilage and bone formation. In addition, BMP7 regulates a wide range of physiological processes. For example, BMP7 may be an osteoinductive factor that is responsible for the phenomenon of epithelial osteogenesis. It has also been established that BMP7 takes part in the regulation of calcium metabolism and bone homeostasis. Like activin, BMP7 binds to type II receptors,

- 11 043601

ActRIIA and ActRIIB. However, BMP7 and activin use different type I receptors in a heteromeric receptor complex. The main type I receptor observed for BMP7 was ALK2, while activin bound exclusively to ALK4 (ActRIIB). BMP7 and activin induce different biological responses and activate different Smad pathways (Macias-Silva et al., 1998, J Biol Chem. 273:2562836).

As shown herein, the GDF decoy polypeptide, which is a variant ActRIIB polypeptide (ActRIIB), is more effective in increasing red blood cell levels in vivo compared to soluble wild-type ActRIIB polypeptide and has beneficial effects in a number of animal models of anemia and ineffective erythropoiesis. Additionally, it has been shown that the use of a GDF decoy polypeptide in combination with an EPO receptor activator causes a significant increase in red blood cell production. It should be noted that hematopoiesis is a complex process that is regulated by a number of factors, including erythropoietin, G-CSF and iron homeostasis. The terms increases red blood cell levels and promotes red blood cell production refer to clinically observable measures such as hematocrit, red blood cell count, and hemoglobin values, and are neutral with respect to the mechanism by which such changes occur.

In addition to stimulating red blood cell levels, GDF decoy polypeptides are suitable for a number of therapeutic applications, including, for example, stimulation of muscle growth (see PCT Publication No. WO 2006/012627 and WO 2008/097541, which are thus incorporated by reference in in full). In some cases, when administering a GDF decoy polypeptide for the purpose of muscle augmentation, it may be desirable to reduce or minimize the effect on red blood cells. By monitoring various hematological parameters in patients being treated or those who are candidates for treatment, the appropriate dosage of the GDF decoy polypeptide (including the amount and frequency of administration) can be determined based on the individual needs of the patient, baseline hematological parameters and the goal of treatment. In addition, therapeutic progress and effects on one or more hematologic parameters over time may be useful in the management of patients receiving GDF decoy polypeptide by facilitating patient care, determining appropriate maintenance dosing (both amount and frequency), etc.

EPO is a glycoprotein hormone that is involved in the growth and maturation of erythroid progenitor cells into red blood cells. EPO is produced by the liver during fetal life and by the kidneys in adults. Decreased EPO production, which often occurs in adults as a consequence of kidney failure, leads to anemia. EPO is produced using genetic engineering techniques based on the expression and secretion of the protein by a host cell that has been transfected with the EPO gene. The administration of such recombinant EPO has proven effective in the treatment of anemia. For example, Eschbach et al. (1987, N Engl J Med 316:73) describes the use of EPO to correct anemia caused by chronic renal failure.

The effects of EPO are mediated by binding to and activating a cell surface receptor that belongs to the cytokine receptor superfamily, designated EPO receptor. Human and mouse EPO receptors have been cloned and expressed (D'Andrea et al., 1989, Cell 57:277; Jones et al., 1990, Blood 76:31; Winkelman et al., 1990, Blood 76:24; WO 90 /08822/US Patent No. 5278065). The human EPO receptor gene encodes a 483 amino acid transmembrane protein that contains an extracellular domain of approximately 224 amino acids and exhibits approximately 82% amino acid sequence identity with the murine EPO receptor (See US Pat. No. 6,319,499). The cloned full-length EPO receptor, which is expressed in mammalian cells (66-72 kDa), binds EPO with an affinity (K _D =100-300 nM) similar to that of the native receptor on erythroid progenitor cells. Thus, this form is believed to contain the major EPO binding determinant and is designated the EPO receptor. By analogy with other closely related cytokine receptors, the EPO receptor is believed to dimerize upon agonist binding. However, the detailed structure of the EPO receptor, which may be a multimeric complex, and its specific mechanism of activation have not been fully elucidated (US Patent No. 6,319,499).

Activation of the EPO receptor leads to several biological effects. These effects include increased proliferation of immature erythroblasts, increased differentiation of immature erythroblasts, and decreased apoptosis of erythroid progenitor cells (Liboi et al., 1993, Proc Natl Acad Sci USA 90: 11351-11355; Koury et al., 1990, Science 248:378- 381). The signal transduction pathways for proliferation and differentiation through the EPO receptor appear to be different (Noguchi et al., 1988, Mol Cell Biol 8:2604; Patel et al., 1992, J Biol Chem 1992, 267:21300; Liboi et al. ., ibid.) . Some results suggest that an accessory protein may be required to mediate the differentiation signal (Chiba et al., 1993, Nature 362:646; Chiba et al., 1993, Proc Natl Acad Sci USA 90: 11593); however, there is controversy regarding the role of accessory proteins in differentiation, since the constitutively activated form of the receptor can stimulate both proliferation and differentiation (Pharr et al., 1993, Proc Natl Acad Sci USA 90:938). EPO receptor activators include small molecule erythropoiesis-stimulating agents (ESAs) as well as EPO-based compounds. An example of the latter is a dimeric peptide-based agonist covalently linked to polyethylene glycol (proprietary name Hematid), which has demonstrated seropoiesis-stimulating properties in healthy volunteers and in patients with both chronic kidney disease and endogenous anti-EPO antibodies (Stead et al., 2006 , Blood 108: 1830-1834; Macdougall et al., 2009, N Engl J Med 361: 1848-1855). Other examples include non-peptide ESAs (Qureshi et al., 1999, Proc Natl Acad Sci USA 96: 12156-12161).

EPO receptor activators also include compounds that stimulate erythropoiesis indirectly, without contact with the EPO receptor, by increasing the production of endogenous EPO. For example, hypoxia-inducible factors (HIFs) are endogenous stimulators of EPO gene expression that are suppressed (destabilized) by cellular regulatory mechanisms under normal oxygen conditions. Thus, inhibitors of HIF prolyl hydroxylase enzymes are being investigated for EPO-inducing effects in vivo. Other indirect activators of the EPO receptor include inhibitors of the transcription factor GATA-2 (Nakano et al., 2004, Blood 104:4300-4307), which tonically inhibits EPO gene expression, and inhibitors of hematopoietic cell phosphatase (HCP or SHP-1), which functions as a negative regulator of signal transduction from the EPO receptor (Klingmuller et al., 1995, Cell 80:729-738).

The terms used in this specification generally have their ordinary meaning in the art, within the context of the present invention, and within the particular context in which each term is used. Certain terms are discussed below or elsewhere in the description in order to provide additional guidance to the practitioner in describing the compositions and methods of the invention, and how to make and use them. The scope or meaning of any application of a term will be clear from the particular context in which the term is applied.

Approximately and approximately should generally be understood as the permissible degree of error for a quantity measured, taking into account the basic properties or accuracy of measurements. Typically, approximate degrees of error are within 20 percent (%), preferably within 10%, and more preferably within 5% of a given value or range of values.

Alternatively, and specifically in biological systems, the terms about and about may mean values that are within an order of magnitude, preferably within 5 times the magnitude and more preferably within 2 times the value given. Numerical values given herein are approximate unless otherwise stated, meaning that the term approximate or approximately is implied unless expressly stated.

The methods of the invention may include the steps of comparing sequences to each other, including wild-type sequences with one or more mutant (variant sequences). Such comparisons typically involve polymer sequence alignments, for example, using sequence alignment programs and/or algorithms that are well known in the art (eg, BLAST, FASTA, and MEGALIGN, to name a few). Those skilled in the art will readily appreciate that, in such alignments where the mutation involves the insertion or deletion of a residue, the sequence alignment will introduce a gap (typically represented by a dash or A) in the polymer sequence that does not contain the insertion or deletion of the residue.

Homologous, in all its grammatical forms and spellings, refers to the relationship between two proteins that share a common evolutionary origin, including proteins from superfamilies of the same species of organisms, as well as homologous proteins from different species of organisms. Such proteins (and the nucleic acids encoding them) have sequence homology, which is reflected in their sequence similarity, either in percentage identity, or through the presence of specific residues or motifs and conserved positions.

The term sequence similarity, in all its grammatical forms, refers to the degree of identity or correspondence between nucleic acids or amino acid sequences that may or may not have a common evolutionary origin.

However, in common parlance and as used herein, the term homologous when modified by an adverb such as may highly refer to sequence similarity and may or may not be associated with a common evolutionary origin.

2. GDF decoy polypeptides.

In certain aspects, the invention relates to GDF decoy polypeptides, for example, soluble variant ActRIIB polypeptides, including, for example, fragments, functional variants and modified forms of ActRIIB polypeptides. In certain embodiments, the GDF decoy polypeptides have at least one similar or the same activity as the corresponding wild-type ActRIIB polypeptide. For example, a GDF decoy polypeptide of the invention can bind to an ActRIIB ligand (eg, Activin A, Activin AB, Activin B, Nodal, GDF8, GDF11 or BMP7) and inhibit its function. Optionally, the GDF decoy polypeptide by 13 043601 increases the level of red blood cells. Examples of GDF decoy polypeptides include human ActRIIB polypeptide precursors (SEQ ID NO: 1 or 39) with one or more sequence changes, and soluble human ActRIIB polypeptides (e.g., SEQ ID NO: 2, 3, 7, 11, 26, 28, 29,

32, 37, 38, 40 and 41) with one or more sequence changes.

As used herein, the term ActRIIB refers to the family of activin type IIb receptor proteins (ActRIIB) from any species and variants derived from such ActRIIB proteins by mutagenesis or other modification. Reference to ActRIIB in this document should be understood as a reference to any one of the forms identified to date. Members of the ActRIIB family are typically transmembrane proteins consisting of a ligand-binding extracellular domain with a cysteine-rich region, a transmembrane domain, and a cytoplasmic domain with predicted serine/threonine kinase activity.

The amino acid sequences of the human ActRIIA soluble extracellular domain (shown for comparison) and the ActRIIB soluble extracellular domain are shown in FIG. 1.

The term ActRIIB polypeptide includes polypeptides containing any naturally occurring polypeptide of a member of the ActRIIB family, as well as any variants thereof (including mutants, fragments, fusions and peptidomimetic forms) that retain beneficial activity. See, for example, WO 2006/012627. For example, ActRIIB polypeptides include polypeptides derived from the sequence of any known ActRIIB with at least about 80% sequence identity to an ActRIIB polypeptide, and optionally with at least 85, 90, 95, 97, 99% or greater identity. For example, an ActRIIB polypeptide can bind to the ActRIIB protein and/or activin and inhibit its function. The ActRIIB polypeptide, which is a GDF trap, can be selected to act to promote red blood cell formation in vivo. Examples of ActRIIB polypeptides include human ActRIIB polypeptide precursor (SEQ ID NOs: 1 and 39) and soluble human ActRIIB polypeptides (e.g., SEQ ID NOs: 2, 3, 7, 11, 26, 28, 29, 32, 37, 38, 40 and 41). Amino acid numbering for all ActRIIB polypeptides described herein is based on SEQ ID NO: 1 unless specifically indicated otherwise.

The sequence of the human ActRIIB precursor protein is shown below:

MTAPWVALALLWGSLWPGSGRGEAETRECIYYNANWELERT^QSGLERC EGEQDKRLHCYASWRgsSGTIELVKKGCWLDDFNCYDRQECVATEENPQ VYFCCCEGNFCNERFTHLPEAGGPEVTYEPPPTAPTLL TVLAY S LL ΡIG

GLSLIVLLAFWMYRHRKPPYGHVDIHEDPGPPPPSPLVGLKPLQLLEIK ARGRFGCVWKAQLMNDFVAVKIFPLQDKQSWQSEREIFS T PGMKHENLL QFIAAEKRGSNLEVELWLITAFHDKGSLTDYLKGNIITWNELCHVAETM SRGLSYLHEDVPWCRGEGHKPSIAHRDFKSKNVLLKSDLTAVLA DFGLA VRFEPGKPPGDTHGQVGTRRYMAPEVLEGAINFQRDAFLRIDMYAMGLV LWELVSRCKAADGPVDEYMLPFEEEIGQHPSLEELQEVVVHKKMRPTIK DHWLKHPGLAQLCVTIEECWDHDAEARLSAGCVEERVSLIRRSVNGTTS DCLVSLVTSVTNVDLPPKESSI (SEQ ID NO: 1)

The signal peptide is underlined with a single line; the extracellular domain is highlighted in bold, and potential N-linked glycosylation sites are boxed.

The form with alanine at position 64, also described in the literature, is presented below: MTAPWVALALLWGSLWPGSGRGEAETRECIYYNANWELERT^QSGLERC

EGEQDKRLHCYASWA§SSGTIELVKKGCWLDDFNCYDRQECVATEENPQ VYFCCCEGNFCNERFTHLPEAGGPEVTYEPPPTAPTLL TVLAY S LL ΡIG

GLSLIVLLAFWMYRHRKPPYGHVDIHEDPGPPPPSPLVGLKPLQLLEIK ARGRFGCVWKAQLMNDFVAVKIFPLQDKQSWQSEREIFS T PGMKHENLL QFIAAEKRGSNLEVELWLITAFHDKGSLTDYLKGNIITWNELCHVAETM SRGLSYLHEDVPWCRGEGHKPSIAHRDFKSKNVLLKSDLTAVLA DFGLA VRFEPGKPPGDTHGQVGTRRYMAPEVLEGAINFQRDAFLRIDMYAMGLV LWELVSRCKAADGPVDEYMLPFEEEIGQHPSLEELQEVVVHKKMRPTIK DHWLKHPGLAQLCVTIEECWDHDAEARLSAGCVEERVSLIRRSVNGTTS DCLVSLVTSVTNVDLPPKESSI (SEQ ID NO: 39)

The sequence of the soluble (extracellular) processed human ActRIIB polypeptide is given below:

- 14 043601

GRGEAETRECIYYNANWELERTNQSGLERCEGEQDKRLHCYASWRNSSG

TIELVKKGCWLDDFNCYDRQECVATEENPQVYFCCCEGNFCNERFTHLP

EAGGPEVTYEPPPTAPT (SEQ ID NO: 2)

An alternative form with A64 is presented below:

GRGEAETRECIYYNANWELERTNQSGLERCEGEQDKRLHCYASWANSSG

TIELVKKGCWLDDFNCYDRQECVATEENPQVYFCCCEGNFCNERFTHLP

EAGGPEVTYEPPPTAPT (SEQ ID NO: 40)

Under certain conditions, a protein can be produced with an SGR... sequence at the N-terminus. The terminal tail of the extracellular domain is underlined. The sequence with the tail removed (Δ15 sequence) is shown below:

GRGEAETRECIYYNANWELERTNQSGLERCEGEQDKRLHCYASWRNSSG

TIELVKKGCWLDDFNCYDRQECVATEENPQVYFCCCEGNFCNERFTHLP

EA (SEQ ID NO: 3)

An alternative form with A64 is presented below:

GRGEAETRECIYYNANWELERTNQSGLERCEGEQDKRLHCYASWANSSG

TIELVKKGCWLDDFNCYDRQECVATEENPQVYFCCCEGNFCNERFTHLP

EA (SEQ ID NO: 41)

Under certain conditions, a protein can be produced with an SGR... sequence at the N-terminus. The nucleic acid sequence encoding the human ActRIIB precursor protein is shown below: (nucleotides 5-1543 from Genbank entry NM 001106) (As shown, the sequence contains an alanine at position 64 and can be modified to contain arginine instead)

ATGACGGCGCCCTGGGTGGCCCTCGCCCTCCTCTGGGATCGCTGTGGC

CCGGCTCTGGGCGTGGGGAGGCTGAGACACGGGAGTGCATCTACTACAA

CGCCAACTGGGAGCTGGAGCGCACCAACCAGAGCGGCCTGGAGCGCTGC

GAAGGCGAGCAGGACAAGCGGCTGCACTGCTACGCCTCCTGGGCCAACA

GCTCTGGCACCATCGAGCTCGTGAAGAAGGGCTGCTGGCTAGATGACTT

CAACTGCTACGATAGGCAGGAGTGTGTGGCCACTGAGGAGAACCCCCAG

GTGTACTTCTGCTGCTGTGAAGGCAACTTCTGCAACGAGCGCTTCACTC

ATTTGCCAGAGGCTGGGGGCCCGGAAGTCACGTACGAGCCACCCCCGAC

AGCCCCCACCCTGCTCACGGTGCTGGCCTACTCACTGCTGCCCATCGGG

GGCCTTTCCCTCATCGTCCTGCTGGCCTTTTGGATGTACCGGCATCGCA

AGCCCCCCTACGGTCATGTGGACATCCATGAGGACCCTGGGCCTCCACC

ACCATCCCCTCTGGTGGGCCTGAAGCCACTGCAGCTGCCTGGAGATCAAG

GCTCGGGGGCGCTTTGGCTGTGTCTGGAAGGCCCAGCTCATGAATGACT

TTGTAGCTGTCAAGATCTTCCCACTCCAGGACAAGCAGTCGTGGCAGAG

- 15 043601

TGAACGGGAGATCTTCAGCACACCTGGCATGAAGCACGAGAACCTGCTA CAGTTCATTGCTGCCGAGAAGCGAGGCTCCAACCTCGAAGTAGAGCTGT GGCTCATCACGGCCTTCCATGACAAGGGCTCCCTCACGGATTACCTCAA GGGGAACATCATCACATGGAACGAACTGTGTCATGTAGCAGAGACGATG TCACGAGGCCTCTCATACCTGCATGAGGATGTGCCCTGGTGCC GTGGCG AGGGCCACAAGCCGTCTATTGCCCACAGGGACTTTAAAAGTAAGAATGT ATTGCTGAAGAGCGACCTCACAGCCGTGCTGGCTGACTTTGGCTTGGCT GTTCGATTTGAGCCAGGGAAACCTCCAGGGGACACCCACGGACAGGTAG GCACGAGACGGTACATGGCTCCTGAGGTGCTCGAGGGAGCCATCAACTT CCAGAGAGATGCCTTCCTGCGCATTGACATGTATGC CATGGGGTTGGTG CTGTGGGAGCTTGTGTCTCGCTGCAAGGCTGCAGACGGACCCGTGGATG AGTACATGCTGCCCTTTGAGGAAGAGATTGGCCAGCACCCTTCGTTGGA GGAGCTGCAGGAGGTGGTGGTGCACAAGAAGATGAGGCCCACCATTAAA GATCACTGGTTGAAACACCCGGGCCTGGCCCAGCTTTGTGTGACCATCG AGGAGTGCTGGGACCATGATGCAGAGGCT CGCTTGTCCGCGGGCTGTGT GGAGGAGCGGGTGTCCCTGATTCGGAGGTCGGTCAACGGCACTACCTCG GACTGTCTCGTTTCCCTGGTGACCTCTGTCACCAATGTGGACCTGCCCC CTAAAGAGTCAAGCATCTAA (SEQ ID NO: 4)

The nucleic acid sequence encoding the soluble (extracellular) human ActRIIB polypeptide is shown below (As shown, the sequence contains an alanine at position 64 and can be modified to contain arginine instead):

GGGCGTGGGGAGGCTGAGACACGGGAGTGCATCTACTACAACGCCAACT

GGGAGCTGGAGCGCACCAACCAGAGCGGCCTGGAGCGCTGCGAAGGCGA

GCAGGACAAGCGGCTGCACTGCTACGCCTCCTGGGCCAACAGCTCTGGC ACCATCGAGCTCGTGAAGAAGGGCTGCTGGCTAGATGACTTCAACTGCT ACGATAGGCAGGAGTGTGTGGCCACTGAGGAGAACCCCCAGGTGTACTT CTGCTGCTGTGAAGGCAACTTCTGCAACGAGCGCTTCACTCATTTGCCA GAGGCTGGGGGCCCGGAAGTCACGTACGAGCCACCCCCGACAGCC CCCA CC (SEQIDNO:5)

In a specific embodiment, the invention relates to GDF decoy polypeptides, which are variant forms of soluble ActRIIB polypeptides. As described herein, the term soluble ActRIIB polypeptide generally refers to polypeptides containing the extracellular domain of the ActRIIB protein. As used herein, the term soluble ActRIIB polypeptide includes any naturally occurring extracellular domain of the ActRIIB protein, as well as any variants thereof (including mutants, fragments, and peptidomimetic forms) that retain beneficial activity. For example, the extracellular domain of the ActRIIB protein binds to a ligand and is typically soluble. Examples of soluble ActRIIB polypeptides include soluble ActRIIB polypeptides (eg, SEQ ID NOs: 2, 3, 7, 11, 26, 28, 29, 32, 37, 38, 40 and 41). Other examples of soluble ActRIIB polypeptides contain a signal sequence in addition to the extracellular domain of the ActRIIB protein, see Example 1. The signal sequence may be a native ActRIIB signal sequence or a signal sequence from another protein, such as a tissue plasminogen activator (TPA) signal sequence or a signal sequence honeybee melittina (HBM).

The invention contemplates functionally active parts and variants of ActRIIB. The inventors have determined that an Fc fusion protein with the sequence described by Hilden et al. (Blood. 1994 Apr 15;83(8):2163-70), which has alanine at the position corresponding to amino acid 64 of SEQ ID NO: 1 (A64), has a relatively low affinity for activin and GDF-11. In contrast, the same Fc fusion protein with arginine at position 64 (R64) has an affinity for activin and GDF-11 ranging from low nanomolar to high picomolar. Thus, the sequence with R64 in this invention is used as a wild-type reference sequence for human ActRIIB.

Attisano et al. (Cell. 1992 Jan 10;68(1):97-108) showed that deletion of the proline knot at the C terminus of the extracellular domain of ActRIIB reduces the affinity of the receptor for activin. ActRIIB-Fc fusion protein containing amino acids 20-119 of SEQ ID NO: 1, ActRIIB(20-119)-Fc, has reduced binding

- 16 043601 with GDF-11 and activin compared to ActRIIB(20-134)-Fc, which contains a proline knot region and a complete juxtamembrane domain. However, the ActRIIB(20-129)-Fc protein retains similar but slightly reduced activity compared to the wild type, even despite disruption of the proline knot region. Thus, the extracellular domains of ActRIIB that end at amino acids 134, 133, 132, 131, 130, and 129 are expected to be active, and constructs that end at 134 or 133 may be the most active. Likewise, mutations at any of residues 129-134 are not expected to change the affinity for the ligand over a large range. In support of this, the P129 and P130 mutations do not substantially reduce ligand binding. Thus, a GDF decoy polypeptide that is an ActRIIB-Fc fusion protein can end as early as amino acid 109 (the last cysteine), however, forms that end at or between 109 and 119 are expected to have reduced binding ligand. Amino acid 119 is weakly conserved and can therefore be easily modified or deleted. Forms that end at 128 or further retain ligand binding activity. Forms that end at or between 119 and 127 will have intermediate binding activity. Any of these forms may be suitable for use, depending on the clinical or experimental setting.

At the N terminus of ActRIIB, a protein that begins at amino acid 29 or earlier is expected to retain ligand binding activity. Amino acid 29 is the initial cysteine. Mutation of alanine to asparagine at position 24 introduces an N-linked glycosylation sequence with essentially no change in ligand binding. This confirms that mutations in the region between the signal peptide cleavage and the cross-linked cysteine region corresponding to amino acids 20-29 are well tolerated. In particular, constructs that begin at positions 20, 21, 22, 23, and 24 will remain active, and constructs that begin at positions 25, 26, 27, 28, and 29 are expected to also retain activity. The data shown in the examples demonstrate that, surprisingly, a construct that begins with 22, 23, 24 or 25 will have the greatest activity.

Collectively, the active portion of ActRIIB contains amino acids 29-109 of SEQ ID NO: 1, and GDF trap constructs may, for example, contain a portion of ActRIIB that begins at the residue corresponding to amino acids 20-29 of SEQ ID NO: 1 or 39 and ends at position , corresponding to amino acids 109-134 SEQ ID NO: 1 or 39. Other examples include constructs that begin at position 20-29 or 21-29 and end at position 119-134, 119-133, 129-134, or 129- 133 SEQ ID NO: 1 or 39. Other examples include designs that begin at position 20-24 (or 21-24 or 22-25) and end at position 109-134 (or 109-133), 119-134 (or 119-133) or 129-134 (or 129-133) SEQ ID NO: 1 or 39. Variants within these ranges are also contemplated by the invention, in particular those having at least 80, 85, 90, 95 or 99% identity with the corresponding portion of SEQ ID NO: 1 or 39. In certain embodiments, a GDF decoy polypeptide includes, contains essentially, or contains a polypeptide with an amino acid sequence that is at least 80, 85, 90, 95, 96, 97, 98, 99, or 100% identical to amino acid residues 25-131 of SEQ ID NO: 1 or 39. In certain embodiments, a GDF decoy polypeptide includes, contains substantially, or contains a polypeptide with an amino acid sequence that is at least 80, 85 , 90, 95, 96, 97, 98, 99, or 100% identical to SEQ ID NO: 7, 26, 28, 29, 32, 37, or 38. In preferred embodiments, the GDF decoy polypeptide comprises or contains substantially the amino acid sequence of SEQ ID NO: 7, 26, 28, 29, 32, 37 or 38.

The invention includes the results of composite analysis of the ActRIIB structures shown in FIG. 1, demonstrating that the ligand binding pocket is defined by residues Y31, N33, N35, L38 to T41, E47, E50, Q53 to K55, L57, H58, Y60, S62, K74, W78 to N83, Y85, R87, A92 and E94 to F101. Conservative mutations are expected to be tolerated at these positions, although the K74A mutation is well tolerated, as are R40A, K55A, F82A, and mutations at position L79. R40 is K in the clawed frog (Xenopus), indicating that basic amino acids are tolerated at this position. Q53 is R in bovine ActRIIB and K in Xenopus ActRIIB, and thus amino acids including R, K, Q, N and H would be tolerated at this position. Thus, the general formula for a GDF decoy protein is one that contains amino acids 29-109 of SEQ ID NO: 1 or 39, but optionally starts at a position in the range 20-24 or 22-25 and ends at a position in the range 129 -134, and includes no more than 1, 2, 5, 10 or 15 conservative amino acid changes in the ligand binding pocket, and none, one or more non-conservative changes at positions 40, 53, 55, 74, 79 and/or 82 in the ligand binding pocket. Such a protein may retain greater than 80, 90, 95, or 99% sequence identity with amino acid sequence 29-109 of SEQ ID NO: 1 or 39. Regions outside the ligand-binding pocket in which variability may be particularly well tolerated include the amino and carboxy termini extracellular domain (as above), and positions 4246 and 65-73. Substitution of asparagine by alanine at position 65 (N65A) does improve ligand binding in the A64 background, and thus is not expected to have a deleterious effect on ligand binding in the R64 background. This substitution probably removes glycosylation at N65 in the A64 background, thus demonstrating that significant change in this region appears to be tolerated. Although replacement

R64A is poorly tolerated, R64K is well tolerated, and thus another basic residue such as H may be tolerated at position 64.

ActRIIB is highly conserved in almost all vertebrates, with large stretches of the extracellular domain that are completely conserved. Many of the ligands that bind to ActRIIB are also highly conserved. Thus, comparisons of ActRIIB sequences from different vertebrate organisms provide insight into which residues may be altered. Thus, an active variant human ActRIIB polypeptide suitable as a GDF trap may include one or more amino acids at corresponding positions from the ActRIIB sequence of another vertebrate, or may include a residue that is similar to a residue in a human or other vertebrate sequence. The following examples illustrate this approach for identifying the active ActRIIB variant. L46 is a valine in ActRIIB of Xenopus, and thus this position can be changed, and optionally can be changed to another hydrophobic residue such as V, I or F, or to a non-polar residue such as A. E52 is K in Xenopus , indicating that a wide range of changes are allowed in this region, including polar residues such as E, D, K, R, H, S, T, P, G, Y and probably A. T93 is K in Xenopus , indicating that extensive structural variation is tolerated at this position, with favored polar residues such as S, K, R, E, D, H, G, P, G, and Y. F108 is a Y in Xenopus, and thus thus, Y or another hydrophobic group such as I, V or L should be acceptable. E111 is a K in Xenopus, indicating that charged residues will be tolerated at this position, including D, R, K and H, as well as Q and N. R112 is a K in Xenopus, indicating that basic residues will be tolerated at at this position, including R and H. The A at position 119 is relatively weakly conserved, being P in rodents and V in Xenopus, so essentially any amino acid will be allowed at this position.

The invention demonstrates that the addition of an additional N-linked glycosylation site (N-X-S/T) increases the serum half-life of the ActRIIB-Fc fusion protein compared to the ActRIIB(R64)-Fc form. By introducing asparagine at position 24 (construct A24N), an NXT sequence is created that may have a longer half-life. Other NX(T/S) sequences are found at positions 42-44 (NQS) and 65-67 (NSS), although the latter may not be efficiently glycosylated with R at position 64. N-X-S/T sequences can generally be introduced into positions outside the ligand binding pocket defined in FIG. 1. Particularly suitable sites for introducing non-endogenous N-X-S/T sequences include amino acids 20-29, 20-24, 22-25, 109-134, 120-134 or 129-134. N-X-S/T sequences can also be introduced into the linker between the ActRIIB sequence and the Fc or other fusion component. Such a site can be introduced with minimal effort by introducing an N into the correct position in relation to a pre-existing S or T, or by introducing an S or T into a position corresponding to a pre-existing N. Thus, the desired changes that can create an N-linked glycosylation site , are: A24N, R64N, S67N (possibly in combination with the N65A replacement), E106N, R112N, G120N, E123N, P129N, A132N, R112S and R112T. Any S predicted to be glycosylated can be changed to T without creating an immunogenic site, due to the protection provided by glycosylation. Likewise, any T predicted to be glycosylated can be changed to S. Thus, the changes considered are S67T and S44T. Likewise, the A24N variant can use the S26T replacement. Thus, the GDF trap may be an ActRIIB variant with one or more additional, non-endogenous N-linked glycosylation consensus sequences.

The L79 position of ActRIIB can be altered to acquire altered activin-myostatin (GDF-11) binding properties. L79P reduces GDF-11 binding to a greater extent than activin binding. L79E or L79D retains GDF-11 binding. Notably, the L79E and L79D variants have significantly reduced activin binding. In vivo experiments indicate that these non-activin receptors retain significant red blood cell-enhancing ability but show reduced effects on other tissues. These data demonstrate the desirability and feasibility of obtaining polypeptides with reduced activity on activin. In illustrative embodiments, the methods described herein use a GDF decoy polypeptide that is a variant ActRIIB polypeptide containing an acidic amino acid (e.g., D or E) at a position corresponding to SEQ ID NO: 1 or 39, optionally in combination with one or more additional amino acid substitutions, additions or deletions.

The described variations can be combined in various ways. Additionally, the results of the mutagenesis programs described herein indicate that there are amino acid positions in ActRIIB that are often advantageous to maintain. These positions include position 64 (basic amino acid), position 80 (acidic or hydrophobic amino acid), position 78 (hydrophobic, and specifically tryptophan), position 37 (acidic, and specifically aspartic or glutamic acid), position 56 (basic amino acid), position 60 (hydrophobic amino acid, specifically phenylalanine or tyrosine). Thus, in each of the options described herein

- 18 043601 document, the invention relates to an amino acid framework that can be stored. Other positions that may be desirable to retain are: position 52 (acidic amino acid), position 55 (basic amino acid), position 81 (acidic), 98 (polar or charged, specifically E, D, R or K).

In certain embodiments, isolated fragments of ActRIIB polypeptides can be obtained by screening polypeptides obtained by recombinant methods from a corresponding nucleic acid fragment encoding an ActRIIB polypeptide (eg, SEQ ID NOs: 4 and 5). In addition, fragments can be synthesized chemically using methods known in the art, such as the conventional Merrifield solid-phase peptide synthesis f-Moc or t-Boc. The fragments can be produced (by recombinant methods or chemical synthesis) and tested to identify those peptidyl fragments that can act, for example, as antagonists (inhibitors) or agonists (activators) of the ActRIIB protein or ActRIIB ligand.

In certain embodiments, the GDF decoy polypeptide is a variant ActRIIB polypeptide with an amino acid sequence that is at least 75% identical to an amino acid sequence selected from SEQ ID NO: 2, 3, 7, 11, 26, 28, 29, 32, 37, 38, 40, or 41. In certain cases, the GDF trap has an amino acid sequence that is at least 80, 85, 90, 95, 97, 98, 99, or 100% identical to the amino acid sequence selected from SEQ ID NO: 2. 3, 7, 11, 26, 28, 29, 32, 37, 38, 40, or 41. In certain embodiments, the GDF trap includes, contains essentially, or contains an amino acid sequence of at least 80, 85, 90, 95, 97, 98, 99 or 100% identical to the amino acid sequence selected from SEQ ID NO: 2, 3, 7, 11, 26, 28, 29, 32, 37, 38, 40 or 41, where the position corresponding to L79 SEQ ID NO : 1, is an acidic amino acid (for example, amino acid residue D or E).

In certain embodiments, the present invention relates to the production of functional variants by modifying the structure of a GDF decoy polypeptide for purposes such as increasing therapeutic efficacy or stability (eg, ex vivo storage and resistance to in vivo proteolytic degradation). GDF decoy polypeptides can also be produced by substitution, deletion or addition of an amino acid. For example, it is logical to expect that individual substitution of isoleucine or valine for leucine, glutamate for aspartate, serine for threonine, or similar amino acid substitution for a structurally related amino acid (eg, conservative mutations) would not have much effect on the biological activity of the resulting molecule. Conservative substitutions are substitutions that occur within a family of amino acids that are related along their side chains. Which of the substitutions in the amino acid sequence of a GDF decoy polypeptide results in a functional variant can be readily determined by assessing the ability of the GDF decoy polypeptide to elicit a response in cells compared with an unmodified GDF decoy polypeptide or a wild-type ActRIIB polypeptide, or by binding to one or more ligands such as activin, GDF-11 or myostatin compared to unmodified GDF decoy polypeptide or wild-type ActRIIB polypeptide.

In certain specific embodiments, the present invention relates to creating mutations in the extracellular domain (also referred to as the ligand binding domain) of an ActRIIB polypeptide such that the ActRIIB polypeptide has altered ligand binding activity (eg, binding affinity or binding specificity). In certain cases, such GDF decoy polypeptides have altered (increased or decreased) binding affinity for a particular ligand. In other cases, GDF decoy polypeptides have altered binding specificity for ActRIIB ligands.

For example, the invention provides GDF decoy polypeptides that preferentially bind to GDF8/GDF11 over activins. The invention further establishes the desirability of such polypeptides for reducing off-target effects, although such selective variants may be less desirable for the treatment of severe diseases where significant improvements in red blood cell levels are required for therapeutic effect and where some level of off-target effects is acceptable. For example, amino acid residues of the ActRIIB protein such as E39, K55, Y60, K74, W78, D80 and F101 are located in the ligand-binding pocket and mediate binding to ligands such as activin and GDF8. Thus, the present invention provides a GDF trap containing an altered ligand binding domain (eg, GDF8 binding domain) of the ActRIIB receptor that contains one or more mutations at these amino acid residues. Optionally, the altered ligand binding domain may have increased selectivity for a ligand, such as GDF8, compared to the wild type ActRIIB receptor ligand binding domain. To illustrate, these mutations increase the selectivity of the altered ligand binding domain for GDF8 over activin. Optionally, the altered ligand binding domain has a ratio of Kd for activin binding to Kd for GDF8 binding that is at least 2, 5, 10, or even 100 times greater than the ratio for the wild type ligand binding domain. Optionally, the altered ligand binding domain has a ratio of IC ₅₀ for activin inhibition to IC ₅₀ for GDF8 inhibition that is at least 2, 5, 10, or even 100 times greater than the wild type ligand binding domain. Optionally, the altered ligand binding domain inhibits

GDF8 has an IC ₅₀ of at least 2, 5, 10, or even 100 times less than the IC ₅₀ for activin inhibition.

As a specific example, the positively charged amino acid residue Asp (D80) from the ActRIIB ligand binding domain can be mutated to another amino acid residue to produce a GDF decoy polypeptide that preferentially binds to GDF8 rather than activin. Preferably, the D80 residue is changed to an amino acid residue selected from the group consisting of: an uncharged amino acid residue, a negatively charged amino acid residue and a hydrophobic amino acid residue. As an additional specific example, the hydrophobic residue, L79, can be changed to the acidic amino acids aspartic acid or glutamic acid in order to significantly reduce activin binding while maintaining GDF11 binding. One skilled in the art will appreciate that most of the mutations, variants or modifications described can be produced at the nucleic acid level or, in some cases, by post-translational modification or chemical synthesis. Such methods are well known in the art.

In certain embodiments, the present invention provides GDF decoy polypeptides that have specific mutations in ActRIIB to alter the glycosylation of the ActRIIB polypeptide. Examples of glycosylation sites in GDF decoy polypeptides are shown in FIG. 1 (e.g. NX(S/T) sites underlined). Such mutations may be selected to introduce or remove one or more glycosylation sites, such as O-linked or N-linked glycosylation sites. Recognition sites for asparagine-linked glycosylation generally contain a tripeptide sequence, asparagine-X-threonine (where X is any amino acid), which is specifically recognized by the corresponding cellular glycosylation enzymes. The change can also be made by adding or substituting one or more serine or threonine residues to the wild-type ActRIIB polypeptide sequence (for O-linked glycosylation sites). A number of amino acid substitutions or deletions at one or both the first and third positions of the glycosylation recognition site (and/or an amino acid deletion at the second position) do not result in glycosylation in the tripeptide-modified sequence. Another way to increase the number of carbohydrate moieties on a GDF decoy polypeptide is to chemically or enzymatically couple glycosides to the GDF decoy polypeptide. Depending on the joining method used, the sugar(s) may be attached to (a) arginine and histidine; (b) free carboxyl groups; (c) free sulfhydryl groups such as cysteine groups; (d) free hydroxyl groups such as serine, threonine or hydroxyproline groups; (e) aromatic residues such as phenylalanine, tyrosine or tryptophan residues; or (f) a glutamine amide group. These methods are described in WO 87/05330 and Aplin and Wriston (1981) CRC Crit. Rev. Biochem., pp. 259-306, incorporated by reference herein. Removal of one or more carbohydrate moieties present on a GDF decoy polypeptide can be accomplished chemically and/or enzymatically. Chemical deglycosylation includes, for example, exposing the GDF decoy polypeptide to a trifluoromethanesulfonic acid compound or an equivalent compound. This treatment results in the cleavage of most or all sugars except the binding sugar (N-acetylglucosamine or N-acetylgalactosamine), leaving the amino acid sequence intact. Chemical deglycosylation is further described by Hakimuddin et al. (1987) Arch. Biochem. Biophys. 259:52 and Edge et al. (1981) Anal. Biochem. 118: 131. Enzymatic cleavage of carbohydrate moieties on GDF decoy polypeptides can be achieved using a variety of endo- and exoglycosidases described by Thotakura et al. (1987) Meth. Enzymol. 138:350.

The sequence of the GDF decoy polypeptide can be adjusted accordingly depending on the type of expression system used, since mammalian, yeast, insect and plant cells can introduce different glycosylation profiles, which can be influenced by the amino acid sequence of the peptide. In general, GDF decoy polypeptides for use in humans can be expressed in mammalian cell lines that ensure proper glycosylation, such as the HEK293 or CHO cell lines, although other mammalian expression cell lines are expected to be suitable as well.

This specification further provides a method for generating variants, particularly sets of combinatorial variants, of a GDF decoy polypeptide, including, optionally, truncated variants; combinatorial mutant pools are particularly suitable for identifying GDF trap sequences. The goal of screening such combinatorial libraries may be to obtain, for example, GDF decoy polypeptide variants that have altered properties, such as altered pharmacokinetics or altered ligand binding. A number of screening studies are presented below, and such studies can be used to evaluate options. For example, a variant of the GDF trap polypeptide may be screened for the ability to bind to the ActRIIB polypeptide to

- 20 043601 to prevent the binding of an ActRIIB ligand to an ActRIIB polypeptide or to disrupt signal transduction caused by an ActRIIB ligand.

The activity of the GDF decoy polypeptide or variants thereof can also be tested in a cell-based or in vivo assay. For example, the impact of a GDF polypeptide trap variant on the expression of genes involved in hematopoiesis can be assessed. If desired, this can be accomplished in the presence of one or more recombinant ActRIIB ligand proteins (eg, activin), and cells can be transfected to produce a GDF decoy polypeptide and/or variants thereof and, optionally, an ActRIIB ligand. Likewise, the GDF decoy polypeptide can be administered to a mouse or other animal and can be assessed using methods well known in the art for one or more blood parameters such as red blood cell count, hemoglobin level, iron accumulation, or reticulocyte count.

Variants of combinatorial origin can be generated that have selective potency over the reference GDF decoy polypeptide. Such variant proteins, when expressed from recombinant DNA constructs, can be used in gene therapy protocols. Likewise, mutagenesis can give rise to variants that have intracellular half-lives that differ sharply from the corresponding unmodified GDF decoy polypeptide. For example, the modified protein may become either more stable or less stable with respect to proteolytic cleavage or other cellular processes that result in the destruction or other inactivation of the unmodified GDF decoy polypeptide. Such variants and the genes that encode them can be used to alter GDF decoy polypeptide levels by modulating the half-life of GDF decoy polypeptides. For example, a short half-life may result in more erratic biological effects and, in terms of the inducible expression system, may allow greater control over the levels of the recombinant GDF decoy polypeptide in the cell. In an Fc fusion protein, mutations can be made in the linker (if present) and/or in the Fc portion to change the half-life of the protein.

In certain embodiments, the GDF decoy polypeptides of the invention may further contain post-translational modifications beyond any naturally present in the ActRIIB polypeptides. Such modifications include, but are not limited to, acetylation, carboxylation, glycosylation, phosphorylation, lipidation and acylation. As a result, GDF decoy polypeptides may contain non-amino acid elements such as polyethylene glycols, lipids, poly- or monosaccharides and phosphates. The effect of such non-amino acid elements on the functional activity of the GDF trap polypeptide can be tested as described herein for other variants of the GDF trap polypeptide. In cases where the GDF decoy polypeptide is produced in cells by cleavage of the nascent form of the GDF decoy polypeptide, post-translational processing may also be important for proper folding and/or function of the protein. Various cells (such as CHO, HeLa, MDCK, 293, WI38, NIH-3T3 or HEK293) have specific cellular machinery and characteristic mechanisms for such post-translational activities and can be selected to ensure proper modification and processing of GDF decoy polypeptides.

In certain aspects, GDF trap polypeptides include fusion proteins having at least a portion of an ActRIIB polypeptide and one or more fusion domains. Well-known examples of such fusion domains include, but are not limited to, polyhistidine, Glu-Glu, glutathione S transferase (GST), thioredoxin, protein A, protein G, immunoglobulin heavy chain constant region (Fc), maltose binding protein (MBP) or human serum albumin. The fusion domain can be selected to impart the desired properties. For example, some fusion domains are particularly suitable for isolating fusion proteins using affinity chromatography. For affinity purification purposes, appropriate affinity chromatography matrices such as glutathione-, amylase-, and nickel- or cobalt-conjugated resins are used. Many of these matrices are available in kit form, for example the Pharmacia GST purification system and the QIAexpress™ system (Qiagen) are used with fusion partners (HIS6). As another example, a fusion domain may be selected to facilitate detection of GDF decoy polypeptides. Examples of such detection domains include various fluorescent proteins (eg, GFP), as well as epitope tags, which are usually short peptide sequences for which a specific antibody is available. Well-known epitope tags for which specific monoclonal antibodies are readily available include FLAG, influenza virus hemagglutinin (HA) and c-myc tags. In some cases, the fusion domains have a protease cleavage site, for example for factor Xa or thrombin, which allows the corresponding proteases to partially cleave the fusion proteins and thus release recombinant proteins from them. The released proteins can then be isolated from the fusion domain by subsequent chromatographic separation. In certain preferred embodiments, the GDF decoy polypeptide is fused to a domain that stabilizes the GDF decoy polypeptide in vivo (stabilizing domain). By stabilizing is meant an increase in serum half-life, regardless of whether this is due to decreased degradation, decreased renal excretion, or another pharmacokinetic effect. Fusions with the Fc portion of immunoglobulin are known to impart desired pharmacokinetic properties to a wide range of proteins. Likewise, fusions with human serum albumin can impart desired properties. Other fusion domains that may be selected include multimerization domains (eg, dimerization, tetramerization) and functional domains (which confer additional biological function, such as further increasing red blood cell levels).

As a specific example, the present invention relates to a GDF trap that is an ActRIIB-Fc fusion protein comprising an extracellular (eg, ligand binding) domain of an ActRIIB polypeptide fused to an Fc domain. The sequence of an example Fc domain is shown below (SEQ ID NO: 6).

THTCPPCPAPELLGGPSVFLFPPKPKDTLMISRTPEVTCVVVD(A)VSHEDPEVKFNWYVDG IVEVHNAKTKPREEQYNSTYRVVSVLTVLHQDWLNGKEYKCK(A)VSNKALPVPIEKTISKAK GQPREPQVYTLPPSREEMTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPV LDSDG PFFLYSKLTVDKSRWQQGNVFSCSVMHEALHN(A)HYTQKSLSLSPGK*

Optionally, the Fc domain has one or more mutations at residues such as Asp-265, lysine 322 and Asn-434. In some cases, a mutant Fc domain having one or more of these mutations (eg, the Asp-265 mutation) has a reduced ability to bind to the Fcy receptor compared to the wild-type Fc domain. In other cases, a mutant Fc domain having one or more such mutations (eg, the Asn-434 mutation) has an increased ability to bind to the MHC class I Fc receptor (FcRN) compared to the wild-type Fc domain.

It is understood that the various elements of the fusion proteins can be arranged in any manner that is compatible with the desired functional activity. For example, a GDF decoy polypeptide may be C-terminal to a heterologous domain, or, alternatively, a heterologous domain may be C-terminal to a GDF decoy polypeptide. The GDF decoy polypeptide domain and the heterologous domain do not need to be contiguous in the fusion protein, and additional domains or amino acid sequences may be inserted C- or N-terminally into any domain or between domains.

In certain embodiments, the GDF decoy fusion protein comprises an amino acid sequence represented by the formula A-B-C. Part B is an ActRIIB polypeptide truncated at the N- and C-termini, which includes the amino acid sequence corresponding to amino acids 26-132 of SEQ ID NO: 26. Parts A and C may independently be zero, one or more amino acids, and both parts A and C, if present, are heterologous to part B. Parts A and/or C may be attached to part B via a linker sequence. Examples of linkers include short polypeptide linkers such as 2-10, 2-5, 2-4, 2-3 glycine residues, such as, for example, the Gly-Gly-Gly linker. Other suitable linkers are described above herein. In certain embodiments, the GDF decoy fusion protein comprises an amino acid sequence represented by the formula A-B-C, wherein A is a leader sequence, B consists of amino acids 26-132 SEQ ID NO: 26, and C is a polypeptide moiety that improves one or more of the following: in vivo stability, in vivo half-life, uptake/administration, tissue localization or distribution, protein complexation, and/or purification. In certain embodiments, the GDF decoy fusion protein comprises an amino acid sequence represented by the formula A-B-C, wherein A is a TPA leader sequence, B consists of amino acids 26-132 of SEQ ID NO: 26, and C is an immunoglobulin Fc domain . A preferred GDF decoy fusion protein contains the amino acid sequence set forth in SEQ ID NO: 26.

In certain embodiments, the GDF decoy polypeptides of the present invention contain one or more modifications that are capable of stabilizing the GDF decoy polypeptides. For example, such modifications increase the in vitro half-life of GDF decoy polypeptides, increase the half-life of GDF decoy polypeptides in circulation, or reduce the proteolytic degradation of GDF decoy polypeptides. Such stabilizing modifications include, but are not limited to, fusion proteins (including, for example, fusion proteins containing a GDF decoy polypeptide and a stabilizing domain), glycosylation site modifications (including, for example, adding a glycosylation site to a GDF decoy polypeptide ) and modification of the carbohydrate moiety (including, for example, removal of carbohydrate moieties from the GDF polypeptidal trap). In the case of fusion proteins, the GDF decoy polypeptide is fused to a stabilizing domain, such as an IgG molecule (eg, an Fc domain). As used herein, the term stabilizing domain not only refers to a fusion domain (eg, Fc) as in the case of fusion proteins, but also includes non-protein modifications, such as a carbohydrate moiety or a non-protein polymer, such as polyethylene glycol.

- 22 043601

In certain embodiments, the present invention makes it possible to obtain isolated and/or purified forms of GDF decoy polypeptides that are isolated from, or substantially free from, other proteins.

In certain embodiments, the GDF decoy polypeptides of the invention (unmodified or modified) can be produced using a variety of known methods. For example, such GDF decoy polypeptides can be synthesized using standard protein chemistry methods, such as those described in Bodansky, M. Principles of Peptide Synthesis, Springer Verlag, Berlin (1993) and Grant G.A. (ed.), Synthetic Peptides: A User's Guide, W.H. Freeman and Company, New York (1992). In addition, automated peptide synthesizers are commercially available (eg, Advanced ChemTech Model 396; Milligen/Biosearch 9600). Alternatively, GDF decoy polypeptides, fragments or variants thereof can be produced recombinantly using various expression systems (eg, E. coli, Chinese hamster cells (CHO), COS cells, baculovirus) that are well known in the art. In a further embodiment, modified or unmodified GDF decoy polypeptides can be produced by digesting recombinant full-length GDF decoy polypeptides using, for example, a protease such as trypsin, thermolysin, chymotrypsin, pepsin, or a paired amino acid cleaving enzyme (PACE). . Computer analysis (using commercially available software, e.g., MacVector, Omega, PCGene, Molecular Simulation, Inc.) can be used to identify proteolytic cleavage sites. Alternatively, such GDF decoy polypeptides can be prepared from recombinant full-length GDF decoy polypeptides using standard methods known in the art, such as chemical digestion (eg, cyanogen bromide, hydroxylamine).

3. Nucleic acids encoding GDF decoy polypeptides.

In certain aspects, the invention relates to isolated and/or recombinant nucleic acids encoding any of the GDF decoy polypeptides described herein. SEQ ID NO: 4 encodes the natural ActRIIB precursor polypeptide, while SEQ ID NO: 5 encodes the soluble ActRIIB polypeptide, and SEQ ID NOs: 25, 27, 30 and 31 encode soluble GDF decoys. Said nucleic acids may be single-stranded or double-stranded. Such nucleic acids may be DNA or RNA molecules. These nucleic acids can be used, for example, in methods for producing GDF decoy polypeptides or as direct therapeutic agents (eg, gene therapy).

In certain aspects, said nucleic acids encoding GDF decoy polypeptides should be further understood to include nucleic acids that are variants of SEQ ID NOs: 5, 25, 27, 30 and 31. Variant nucleotide sequences include sequences that differ in one or more nucleotide substitutions, additions or deletions, such as allelic variants; and will thus include coding sequences that differ from the nucleotide sequence of the coding sequence designated in SEQ ID NOs: 5, 25, 27, 30 and 31.

In certain embodiments, the invention provides isolated or recombinant nucleic acid sequences that are at least 80%, 85%, 90%, 95%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 5, 25, 27, 30, or 31. One skilled in the art will appreciate that nucleic acid sequences complementary to SEQ ID NO: 5, 25, 27, 30 or 31, and variants of SEQ ID NO: 5, 25, 27, 30 or 31 are also within the scope of the present invention. In additional embodiments, the nucleic acid sequences of the invention may be isolated, recombinant and/or fused to a heterologous nucleotide sequence, or may be contained in a DNA library.

In other embodiments, the nucleic acids of the invention also include nucleotide sequences that hybridize under very stringent conditions to the nucleotide sequence designated in SEQ ID NO: 5, 25, 27, 30 or 31, the complementary sequence of SEQ ID NO: 5, 25, 27 , 30 or 31, or fragments thereof. As stated above, one skilled in the art will readily understand that the appropriate stringent conditions that promote DNA hybridization can be varied. For example, hybridization can be performed at 6.0x sodium chloride/sodium citrate (SSC) at approximately 45°C, followed by a wash with 2.0x SSC at 50°C. For example, the salt concentration in the wash step can be selected from a low hardness of approximately 2.0x SSC at 50°C to a high hardness of approximately 0.2x SSC at 50°C. In addition, the temperature of the wash step can be increased from low severity conditions at room temperature, approximately 22°C, to high severity conditions at approximately 65°C. Both temperature and salt can change, or temperature or salt concentration can be held constant while other variables change. In one embodiment, the invention relates to nucleic acids that hybridize under low stringency conditions with 6x SSC at room temperature followed by a wash with 2x SSC at room temperature.

- 23 043601

Isolated nucleic acids that differ from the nucleic acids specified in SEQ ID NO: 5, 25, 27, 30 or 31 due to degeneracy of the genetic code are also within the scope of the invention. For example, a number of amino acids are defined by more than one triplet. Codons that designate the same amino acid, or synonyms (for example, CAU and CAC are synonyms for histidine) can result in silent mutations that do not affect the amino acid sequence of the protein. In certain embodiments, the GDF decoy polypeptide is encoded by an alternative nucleotide sequence. The alternative nucleotide sequences are degenerate to the native GDF trap nucleic acid sequence but still encode the same fusion protein. In certain embodiments, the GDF trap of SEQ ID NO: 26 is encoded by an alternative nucleic acid sequence comprising SEQ ID NO: 30. However, it is believed that DNA sequence polymorphisms that still result in substitutions in the amino acid sequences of these proteins will occur in mammalian cells . One skilled in the art will appreciate that these variations in one or more nucleotides (up to about 3-5% of the nucleotides) of the nucleic acids encoding a particular protein may occur among individuals of a given species due to natural allelic variability. Any and all of such nucleotide variations, and the resulting amino acid polymorphisms, are within the scope of the present invention.

In certain embodiments, the recombinant nucleic acids of the invention may be operably linked to one or more regulatory nucleotide sequences in an expression construct.

Regulatory nucleotide sequences will generally correspond to the host cell being used for expression. There are many types of suitable expression vectors and suitable regulatory sequences known in the art for a variety of host cells. Typically, the one or more regulatory nucleotide sequences may include, but are not limited to, promoter sequences, leader or signal sequences, ribosome binding sites, transcription start and termination sequences, translation start and termination sequences, and enhancer or activator sequences. Constitutive or inducible promoters known in the art are contemplated by this invention. Promoters can be either natural promoters or hybrid promoters that combine elements of more than one promoter. The expression construct may be present in the cell in an episome, such as a plasmid, or the expression construct may be inserted into a chromosome. In a preferred embodiment, the expression vector contains a selectable marker gene to enable selection of transformed host cells. Selectable marker genes are well known in the art and vary depending on the host cell used.

In certain aspects of the invention, the nucleic acid is provided in an expression vector that contains a nucleotide sequence encoding a GDF decoy polypeptide and is operably linked to at least one regulatory sequence. The regulatory sequences are generally accepted in the art and are chosen for direct expression of the GDF decoy polypeptide. Thus, the term regulatory sequence includes promoters, enhancers and other expression control elements. Examples of regulatory sequences are described in Goeddel; Gene Expression Technology: Methods in Enzymology, Academic Press, San Diego, CA (1990). For example, to express DNA sequences encoding a GDF decoy polypeptide in these vectors, any of a wide variety of expression control sequences that direct expression of the DNA sequence when operably linked thereto can be used. Such suitable expression control sequences include, for example, the SV40 early and late promoters, the tet promoter, the adenovirus or cytomegalovirus immediate early promoter, the RSV promoters, the lac system, the trp system, the TAC or TRC system, the T7 promoter, the expression of which is driven by T7 RNA polymerase , basic operator and promoter regions of phage lambda, control regions for the fd coat protein, promoter of 3-phosphoglycerate kinase or other glycolytic enzymes, acid phosphatase promoters, e.g. Pho5, yeast α-mating factor promoters, polyhedron promoter of the baculovirus system, and other sequences known for control of gene expression of prokaryotic or eukaryotic cells or their viruses, and their various combinations. It should be understood that the design of the expression vector may depend on factors such as the choice of host cell for transformation and/or the type of protein that is desired to be expressed. In addition, the copy number of the vector, the ability to control that copy number, and the expression of any other protein encoded by the vector, such as antibiotic markers, should also be considered.

The recombinant nucleic acid of the invention can be produced by ligating a cloned gene or part thereof into a vector suitable for expression in either prokaryotic cells or eukaryotic cells (yeast, birds, insects or mammals), or both. Expression vehicles for producing recombinant GDF decoy polypeptide include

- 24 043601 plasmids and other vectors. For example, suitable vectors include plasmid types: pBR322-derived plasmids, pEMBL-derived plasmids, pEX-derived plasmids, pBT-derived plasmids and pUC-derived plasmids for expression in prokaryotic cells such as E. coli.

Some mammalian expression vectors contain both prokaryotic sequences to facilitate propagation of the vector in bacteria and one or more eukaryotic transcription units that are expressed in eukaryotic cells. Vectors derived from pcDNAI/amp, pcDNAI/neo, pRc/CMV, pSV2gpt, pSV2neo, pSV2-dhfr, pTk2, pRSVneo, pMSG, pSVT7, pkoneo and pHyg are examples of mammalian expression vectors suitable for transfection of eukaryotic cells. Some of these vectors are modified with bacterial plasmid sequences, such as pBR322, to facilitate replication and selection for drug resistance in both prokaryotic and eukaryotic cells.

Alternatively, derivatives of viruses such as bovine papillomavirus (BPV1) or Epstein-Barr virus (pHEBo, pREP-derivatives and p205) can be used to transiently express proteins in eukaryotic cells. Examples of other viral (including retroviral) expression systems can be found below in the description of delivery systems for gene therapy. Various methods used in the production of plasmids and in the transformation of host organisms are well known in the art. For other suitable expression systems for both prokaryotic and eukaryotic cells, as well as general recombinant methods, see Molecular Cloning A Laboratory Manual, 3rd Ed., ed. by Sambrook, Fritsch and Maniatis (Cold Spring Harbor Laboratory Press, 2001). In some cases, it may be desirable to express recombinant polypeptides using a baculovirus expression system. Examples of such baculovirus expression systems include pVL-derived vectors (such as pVL1392, pVL1393 and pVL941), pAcUW-derived vectors (such as pAcUW1) and pBlueBac-derived vectors (such as pBlueBac III containing β-gal).

In a preferred embodiment, a vector is constructed to produce the subject GDF decoy polypeptides in CHO cells, for example, Pcmv-Script vector (Stratagene, La Jolla, Calif.), pcDNA4 vectors (Invitrogen, Carlsbad, Calif.), and pCI-neo vectors (Promega , Madison, Wise). As will be appreciated, the subject gene constructs can be used to stimulate expression of the subject GDF decoy polypeptides in cells expanded in culture, for example, to produce proteins, including fusion proteins or protein variants for purification.

The present invention also provides host cells that are transfected with a recombinant gene containing a coding sequence (eg, SEQ ID NO: 4, 5, 25, 27, 30, or 31) for one or more of the subject GDF decoy polypeptides. The host cell can be any prokaryotic or eukaryotic cell. For example, the GDF decoy polypeptide of the invention can be expressed in bacterial cells such as E. coli, insect cells (eg, using a baculovirus expression system), yeast or mammalian cells. Other suitable host cells are known to those skilled in the art.

Thus, the present invention further relates to methods for producing the subject GDF decoy polypeptides. For example, a host cell that has been transfected with an expression vector encoding a GDF decoy polypeptide can be cultured under appropriate conditions that allow expression of the GDF decoy polypeptide.

The GDF decoy polypeptide can be secreted and isolated from a mixture of cells and from a medium containing the GDF decoy polypeptide. Alternatively, the GDF decoy polypeptide may be retained in the cytoplasmic or membrane fraction, and the cells are harvested, lysed, and the protein is recovered. A cell culture contains host cells, media, and other intermediates. Suitable cell culture media are known in the art. Subject GDF decoy polypeptides can be isolated from cell culture medium, host cells, or both by methods known in the art for protein purification, including ion exchange chromatography, gel filtration, ultrafiltration, electrophoresis, immunoaffinity purification with using antibodies specific to specific epitopes of GDF trap polypeptides. In a preferred embodiment, the GDF decoy polypeptide is a fusion protein containing a domain that facilitates its purification.

In another embodiment, a fusion gene encoding a purification leader sequence, for example, a poly(His)/enterokinase cleavage site sequence at the N-terminus of a desired portion of a recombinant GDF decoy polypeptide, may allow the expressed fusion protein to be purified by affinity chromatography using Ni ²⁺ metal resin. The purification leader sequence can then be removed by treatment with enterokinase to produce a purified GDF decoy polypeptide (eg, see Hochuli et al., (1987) J. Chromatography 411:177 and Janknecht et al., PNAS USA 88:8972).

Methods for producing fusion genes are well known. Mainly, the joining of different DNA fragments encoding different polypeptide sequences is carried out in accordance with

- 25 043601 with conventional techniques using blind or dangling ends for ligation, restriction enzyme digestion to obtain the corresponding ends, filling of sticky ends if necessary, treatment with alkaline phosphatase to avoid unwanted connection and enzymatic ligation. In another embodiment, the fusion gene can be synthesized using conventional techniques including automated DNA synthesizers. Alternatively, PCR amplification of gene fragments can be accomplished using anchor primers that form complementary overhangs between two successive gene fragments, which can subsequently be annealed to produce a chimeric gene sequence (see, for example, Current Protocols in Molecular Biology, eds. Ausubel et al., John Wiley & Sons: 1992).

4. Screening tests.

In certain aspects, the present invention relates to the use of these GDF decoy polypeptides (eg, soluble variant ActRIIB polypeptides) to detect compounds that are agonists or antagonists of ActRIIB polypeptides. Compounds identified by this screen can be tested to assess their ability to modulate red blood cell, hemoglobin and/or reticulocyte levels in vivo or in vitro. These compounds can be tested, for example, in animal models.

There are a number of approaches to screen for therapeutics that increase red blood cell or hemoglobin levels by targeting the ActRIIB signaling pathway. In some embodiments, high throughput screening of compounds can be conducted to identify agents that specifically inhibit or reduce the binding of an ActRIIB polypeptide to its binding partner, such as an ActRIIB ligand. Alternatively, the assay can be used to identify compounds that enhance the binding of an ActRIIB polypeptide to its binding partner, such as an ActRIIB ligand. In a further embodiment, compounds can be identified by their ability to interact with an ActRIIB polypeptide.

A variety of studies of varying formats are suitable and, in light of the present disclosure, those studies that are not specifically presented herein will nonetheless be understandable to one skilled in the art. The test compounds (substances) of the invention described in the present invention can be obtained by any combinatorial chemical method.

Alternatively, the subject compounds may be natural biomolecules synthesized in vivo or in vitro. Compounds tested for their ability to act as tissue growth modulators may be produced, for example, by bacteria, yeast, plants or other organisms (e.g., natural products), chemically produced (e.g., small molecules, including peptidomimetics) or obtained recombinantly. Test compounds contemplated by the present invention include non-peptidyl organic molecules, peptides, polypeptides, peptidomimetics, sugars, hormones and nucleic acid molecules. In a specific embodiment, the test substance is a small organic molecule with a molecular weight of less than about 2000 Da.

The test compounds of the invention may be presented as individual discrete units or presented in libraries of greater complexity, such as those obtained using combinatorial chemistry. These libraries may contain, for example, alcohols, alkyl halides, amines, amides, esters, aldehydes, ethers and other classes of organic compounds. The introduction of test compounds into the test system can be either in isolated form or as a mixture of compounds, especially in the initial stages of screening.

Optionally, compounds can be derived from other compounds and have derivative groups that facilitate the isolation of these compounds. Non-limiting examples of derived groups include biotin, fluorescein, digoxigenin, green fluorescent protein, isotopes, polyhistidine, magnetic beads, glutathione S transferase (GST), photoactivatable cross-linkers, or combinations thereof.

In many drug screening programs that test libraries of compounds and natural extracts, high-throughput assays are desired in order to maximize the number of compounds analyzed in a given time period. Assays that are performed in cell-free systems, those that can be prepared with purified or semi-purified proteins, are often preferred as primary screens in which they can be designed for rapid development and relative ease in detecting a change in the molecular target that is mediated by the test compound . Moreover, the effects of cellular toxicity or bioavailability of the test compound may largely go undetected in an in vitro system; the assay instead focuses primarily on the effect of the drug on the molecular target, which may be manifested by a change in binding affinity between the ActRIIB polypeptide and its binding partner (eg ActRIIB ligand).

Just by way of illustration, in an exemplary screening assay of the present invention, a compound of interest is contacted with an isolated and purified ActRIIB polypeptide, which

- 26 043601 which is generally capable of binding to an ActRIIB ligand suitable for the purpose of the assay. A composition containing an ActRIIB ligand is then added to the mixture of compound and ActRIIB polypeptide. Detection and quantitation of ActRIIB/ActRIIB ligand complexes provides methods for determining the effectiveness of a compound in inhibiting (or potentiating) the formation of a complex between an ActRIIB polypeptide and its binding protein. The effectiveness of a compound can be assessed by constructing dose-response curves from the data obtained using different concentrations of the test compound. Moreover, a control study can also be performed to obtain a background value for comparison. For example, in a control assay, isolated and purified ActRIIB ligand is added to a composition containing an ActRIIB polypeptide and the formation of an ActRIIB/ActRIIB ligand complex is quantified in the absence of the test compound. It should be understood that, in general, the order in which the reagents can be mixed may vary, and they can be mixed simultaneously. Moreover, instead of purified proteins, cell extracts and lysates can be used to replicate a suitable cell-free assay system.

The formation of a complex between an ActRIIB polypeptide and its binding protein can be detected using a number of methods. For example, modulation of complex formation can be analyzed quantitatively, for example, using detectable labeled proteins, such as radiolabeled (eg, ³² P, ³⁵ S, ¹⁴ C or ³ H), fluorescently labeled (eg, FITC) or enzymatically labeled ActRIIB polypeptides or its binding protein, by immunoassay or by chromatographic determination.

In certain embodiments, the present invention relates to the use of fluorescence polarization assays and fluorescence resonance energy transfer (FRET) assays in measuring either directly or indirectly the extent of interaction between an ActRIIB polypeptide and its binding protein. In addition, other detection methods, such as those based on optical waveguides (PCT publication WO 96/26432 and US patent No. 5677196), surface plasmon resonance (SPR), surface charge detectors and surface tension force detectors, are compatible with many embodiments of the invention.

In addition, the present invention relates to the use of an interaction capture assay, also known as a two-hybrid assay, to identify compounds that disrupt or potentiate the interaction between an ActRIIB polypeptide and its binding partner; see, for example, US Pat. No. 5,283,317; Zervos et al. (1993) Cell 72:223-232; Madura et al. (1993) J. Biol. Chem. 268:12046-12054; Bartel et al. (1993) Biotechniques14:920-924 and Iwabuchi et al. (1993) Oncogene 8:16931696). In a specific embodiment, the present invention relates to the use of reverse two-hybrid systems to identify compounds (eg, small molecules or peptides) that uncouple interactions between an ActRIIB polypeptide and its binding protein; see, for example, Vidal and Legrain, (1999) Nucleic Acids Res 27:919-29; Vidal and Legrain, (1999) Trends Biotechnol 17:374-81; US Patent Nos. 5525490, 5955280 and 5965368.

In certain embodiments, the compounds of interest are identified by their ability to interact with the ActRIIB polypeptide. The interaction of the compound and the ActRIIB polypeptide may be covalent or non-covalent. For example, such interactions can be identified at the protein level by biochemical methods in vitro, including photocross-linking, radiolabeled ligand binding, and affinity chromatography (Jakoby W.B. et al., 1974, Methods in Enzymology 46:1). In certain cases, compounds may be screened in an assay based on a specific mechanism, for example, in an assay to identify compounds that bind to the ActRIIB polypeptide. This study may involve solid phase binding or liquid phase binding. Alternatively, the gene encoding the ActRIIB polypeptide can be transfected along with a reporter system (eg, β-galactosidase, luciferase, or green fluorescent protein) into a cell and screened against a library, preferably using a high-throughput screen, or using individual members from libraries. Other binding assays based on a specific mechanism may be used, such as binding assays that detect free energy changes. Binding assays can be performed using a target fixed in a well, on a bead or chip, or captured by an immobilized antibody, or separated by capillary electrophoresis. Bound compounds can be detected typically by colorimetric, fluorescence, or surface plasmon resonance.

5. Examples of therapeutic applications.

In certain embodiments, the GDF decoy polypeptides of the present invention can be used to increase red blood cell levels in mammals, such as rodents and primates, and specifically in human patients. GDF decoy polypeptides, optionally in combination with an EPO receptor activator, may be useful in the treatment of ineffective erythropoiesis. Initially distinguished from aplastic anemia, hemorrhage, or peripheral hemolysis based on ferrokinetic studies (Ricketts et al., 1978, Clin Nucl Med 3: 159-164), ineffective erythropoiesis describes a heterogeneous group of anemias in which the production of mature RBC is less than,

- 27 043601 than would be expected given the number of erythroid precursors (erythroblasts) present in the bone marrow (Tanno et al., 2010, Adv Hematol 2010:358283). In such anemias, tissue hypoxia persists despite elevated erythropoietin levels due to ineffective production of mature RBCs. Ultimately, a vicious cycle develops in which elevated erythropoietin levels drive a massive increase in the number of erythroblasts, potentially leading to splenomegaly (enlarged spleen) due to extramedullary erythropoiesis (Aizawa et al., 2003, Am J Hematol 74:68-72), a pathology erythroblast-induced bone (Di Matteo et al., 2008, J Biol Regul Homeost Agents 22:211-216), and tissue iron overload, even in the absence of therapeutic RBC transfusions (Pippard et al., 1979, Lancet 2:819-821 ). Thus, by increasing the efficiency of erythropoiesis, the GDF decoy polypeptide may break the above cycle and may alleviate not only the underlying anemia, but also complications associated with elevated erythropoietin levels, splenomegaly, bone pathology, and tissue iron overload. GDF decoy polypeptides can treat ineffective erythropoiesis, including anemia and elevated EPO levels, as well as complications such as splenomegaly, erythroblast-induced bone pathology and iron overload, and their associated pathologies. With splenomegaly, such pathologies include chest or abdominal pain and reticuloendothelial hyperplasia. Extramedullary hematopoiesis can occur not only in the spleen, but potentially in other tissues in the form of extramedullary hematopoietic pseudotumors (Musallam et al., 2012, Cold Spring Harb Perspect Med 2:a013482). With erythroblast-induced bone pathology, associated pathologies include low bone mineral density, osteoporosis and bone pain (Haidar et al., 2011, Bone 48:425-432). With iron overload, associated pathologies include hepcidin suppression and dietary iron hyperabsorption (Musallam et al., 2012, Blood Rev 26(Suppl 1):S16-S19), multiple endocrinopathies and liver fibrosis/cirrhosis (Galanello et al., 2010, Orphanet J Rare Dis 5:11), and iron overload cardiomyopathy (Lekawanvijit et al. 2009, Can J Cardiol 25:213–218).

The most common causes of ineffective erythropoiesis are thalassemic syndromes, hereditary hemoglobinopathies in which an imbalance in the production of intact alpha and beta chains of hemoglobin leads to increased apoptosis during erythroblast maturation (Schrier, 2002, Curr Opin Hematol 9:123-126). The thalassemias are collectively among the most common genetic disorders in the world, with changing epidemiological patterns predicted to contribute to a growing public health problem both in the United States and worldwide (Vichinsky, 2005, Ann NY Acad Sci 1054: 18- 24). Thalassemic syndromes are named according to their severity. Thus, α-thalassemias include α-thalassemia minor (also known as α-thalassemia minor; two α-globin genes affected), hemoglobin H disease (three α-globin genes affected), and α-thalassemia major (also known as hydrops fetalis ; four α-globin genes affected), β-thalassemias include β-thalassemia minor (also known as β-thalassemia minor; one β-globin gene affected), β-thalassemia intermedia (two β-globin genes affected), thalassemia with hemoglobin E (two affected β-globin genes) and β-thalassemia major (also known as Cooley's anemia; two affected β-globin genes resulting in complete absence of the β-globin protein). βthalassemia affects multiple organs, is associated with significant morbidity and mortality, and currently requires lifelong treatment. Although the average life expectancy of patients with β-thalassemia has increased in recent years due to the use of regular blood transfusions in combination with iron chelator therapy, iron overload from transfusions and from excess gastrointestinal iron absorption can cause serious complications such as heart disease , thrombosis, hypogonadism, hypothyroidism, diabetes, osteoporosis and osteopenia (Rund et al., 2005, N Engl J Med 353: 1135-1146). As demonstrated herein in a mouse model of β-thalassemia, a GDF decoy polypeptide, optionally in combination with an EPO receptor activator, can be used to treat thalassemic syndromes. GDF decoy polypeptides, optionally in combination with an EPO receptor activator, can be used to treat disorders of ineffective erythropoiesis, in addition to thalassemic syndromes. Such disorders include sideroblastic anemia (hereditary or acquired); dyserythropoietic anemia (Types I and II); sickle cell anemia; hereditary spherocytosis; pyruvate kinase deficiency; megaloblastic anemias, potentially caused by conditions such as folate deficiency (due to congenital diseases, decreased intake, or increased requirements), cobalamin deficiency (due to congenital diseases, pernicious anemia, impaired absorption, pancreatic insufficiency, or decreased intake), certain medications remedies, or unexplained causes (hereditary dyserythropoietic anemia, refractory megaloblastic anemia, or erythroleukemia); myelophthisis anemia, including myelofibrosis (myeloid metaplasia) and myelophthisis; hereditary erythropoietic porphyria; and lead poisoning.

As used herein, a therapeutic agent that prevents a disorder or condition refers to a compound that, in a statistical sample, reduces the incidence of the disorder or condition in a treated sample compared to an untreated sample.

- 28 043601 troll sample, or slows the manifestation or reduces the severity of one or more symptoms of a disorder or condition compared to an untreated control sample. As used herein, the term treatment includes improvement or elimination of a condition once it has been established. In any case, prevention or treatment can be distinguished by the diagnosis provided by the physician or other health care professional and the expected result of the administration of the therapeutic agent.

As shown herein, GDF decoy polypeptides, optionally in combination with an EPO receptor activator, can be used to increase red blood cell, hemoglobin, or reticulocyte levels in healthy individuals, and such GDF decoy polypeptides can be used in a selected patient population. Examples of suitable patient populations include populations with undesirably low red blood cell or hemoglobin levels, such as patients with anemia, and populations that are at risk for developing undesirably low red blood cell or hemoglobin levels, such as patients undergoing major surgery or other procedures that may lead to significant blood loss. In one embodiment, a patient with adequate red blood cell levels is treated with a GDF decoy polypeptide to increase red blood cell levels, and then blood is collected and stored for later use in transfusions.

GDF decoy polypeptides, optionally in combination with an EPO receptor activator, described herein can be used to increase red blood cell levels in patients with anemia. When observing hemoglobin levels in people, a level less than normal for the appropriate age and gender category may indicate anemia, although individual variations are taken into account. For example, a hemoglobin level of 12 g/dL is generally considered the lower limit of normal in the general adult population. Possible causes include blood loss, nutritional deficiency, drug reaction, various bone marrow problems, and a variety of medical conditions. More specifically, anemia is associated with a number of disorders that include, for example, chronic renal failure, myelodysplastic syndrome, rheumatoid arthritis, bone marrow transplantation. Anemia may also be associated with the following conditions: solid tumors (eg, breast cancer, lung cancer, colon cancer); tumors of the lymphatic system (eg, chronic lymphocytic leukemia, non-Hodgkin's lymphomas and Hodgkin's lymphoma); tumors of the hematopoietic system (for example, leukemia, myelodysplastic syndrome, multiple myeloma); radiation therapy; chemotherapy (eg, treatment regimens that include platinum drugs); inflammatory and autoimmune diseases, including, but not limited to, rheumatoid arthritis, other inflammatory arthritis, systemic lupus erythematosus (SLE), acute or chronic skin diseases (eg, psoriasis), inflammatory bowel disease (eg, Crohn's disease and ulcerative colitis); acute or chronic kidney disease or failure, including idiopathic or congenital conditions; acute or chronic liver disease; acute or chronic bleeding; situations where red blood cell transfusion is not possible due to the patient's allo- or autoantibodies and/or for religious reasons (for example, some Jehovah's Witnesses); infections (eg, malaria, osteomyelitis); hemoglobinopathies, including, for example, sickle cell disease, thalassemia; use or abuse of drugs, such as alcohol abuse; pediatric patients with anemia from any cause to avoid transfusion; and elderly patients or patients with underlying cardiopulmonary disease with anemia who cannot receive transfusions due to concerns about circulatory overload.

Myelodysplastic syndrome (MDS) is a heterogeneous collection of hematological conditions that are characterized by ineffective production of myeloid blood cells and the risk of transformation into acute myelogenous leukemia. In patients with MDS, blood stem cells do not mature into healthy red blood cells, white blood cells, or platelets. Myelodysplastic disorders include, for example, refractory anemia, refractory anemia with ring sideroblasts, refractory anemia with excess blasts, refractory anemia with excess blasts in transformation, refractory cytopenia with multilineage dysplasia, and myelodysplastic syndrome associated with an isolated abnormality of chromosome 5q. Because these disorders manifest as irreversible defects in both the quantity and quality of hematopoietic cells, most patients with MDS suffer from chronic anemia. Thus, patients with MDS eventually require blood transfusions and/or treatment with growth factors (eg, erythropoietin or G-CSF) to increase red blood cell levels. However, many patients with MDS develop side effects due to the frequency of this treatment. For example, patients who have received frequent red blood cell transfusions may have organ and tissue damage due to the accumulation of extra iron. As shown in the examples below, GDF decoy polypeptides have been used to treat anemia in a mouse model of MDS. Thus, the GDF decoy polypeptides described herein can be used to treat patients with MDS. In certain embodiments, patients suffering from MDS can be treated using a combination of a GDF decoy polypeptide in combination with an EPO receptor activator. In other embodiments, a patient suffering from MDS can be treated using a combination of a GDF decoy polypeptide and one or more additional therapeutic agents

- 29 043601 for the treatment of MDS, including, for example, thalidomide, lenalidomide, azacitadine, decitabine, erythropoietins, deferoxamine, antithymocyte globulin, filgrastim (G-CSF) and an erythropoietin signaling pathway agonist.

GDF decoy polypeptides, optionally in combination with an EPO receptor activator, are useful for the treatment of hypoproliferative bone marrow anemias, which are typically associated with subtle changes in red blood cell (RBC) morphology. Hypoproliferative anemias include: 1) anemia of chronic disease, 2) anemia of kidney disease, and 3) anemia associated with hypometabolic conditions. In each of these types, endogenous erythropoietin levels are excessively low for the degree of anemia observed. Other hypoproliferative anemias include: 4) early stage iron deficiency anemia, and 5) anemia caused by bone marrow damage. In these types, endogenous erythropoietin levels are increased according to the observed degree of anemia.

The most common type is anemia of chronic disease, which includes inflammation, infection, tissue damage, and conditions such as cancer, and is characterized by both low levels of erythropoietin and an inadequate bone marrow response to erythropoietin (Adamson, 2008, Harrison's Principles of Internal Medicine, 17th ed.; McGraw Hill, New York, pp 628-634). Many factors may contribute to cancer-related anemia. Some of them are themselves associated with the pathological process and the production of pro-inflammatory cytokines such as interleukin-1, interferon-gamma and tumor necrosis factor (Bron et al., 2001, Semin Oncol 28 (Suppl 8): 1-6). Among its effects, inflammation induces the key iron regulatory peptide, hepcidin, thereby inhibiting iron export from macrophages and generally limiting the availability of iron for erythropoiesis (Ganz, 2007, J Am Soc Nephrol 18:394-400). Blood loss through various routes may also contribute to cancer-associated anemia. The incidence of anemia associated with cancer progression varies depending on the type of cancer, ranging from 5% for prostate cancer to 90% for multiple myeloma. Cancer-related anemia has profound consequences for patients, including fatigue and decreased quality of life, decreased treatment response, and increased mortality.

Chronic kidney disease is associated with hypoproliferative anemia, the severity of which varies with the degree of renal failure. This anemia is predominantly associated with inadequate production of erythropoietin and decreased red blood cell survival. Chronic kidney disease usually progresses gradually over several years or decades to end-stage disease (stage-5), at which dialysis or a kidney transplant is necessary for the patient's survival. Anemia often develops early in this process and worsens as the disease progresses. The clinical consequences of anemia in kidney disease have been documented and include the development of left ventricular hypertrophy, decreased cognitive function, decreased quality of life, and altered immune function (Levin et al., 1999, Am J Kidney Dis 27:347-354; Nissenson, 1992, Am J Kidney Dis 20 (Suppl l):21-24; Revicki et al., 1995, Am J Kidney Dis 25:548-554; Gafter et al., 1994, Kidney Int 45:224-231). As demonstrated by the inventors in a mouse model of chronic kidney disease (see example below), a GDF decoy polypeptide, optionally in combination with an EPO receptor activator, can be used to treat anemia in kidney disease.

A variety of conditions that result in a low metabolic rate can cause mild to moderate hypoproliferative anemia. Among these conditions are conditions of endocrine insufficiency. For example, anemia can occur in Addison's disease, hypothyroidism, hyperparathyroidism, or in men who are castrated or taking estrogen. Mild to moderate anemia may also occur with low dietary protein intake, a condition that is especially common in older adults. Finally, anemia can develop in patients with acute liver disease from almost any cause (Adamson, 2008, Harrison's Principles of Internal Medicine, 17th ed.; McGraw Hill, New York, pp 628-634).

Anemia resulting from acute loss of sufficient volume of blood, such as from trauma or postpartum hemorrhage, is known as acute posthemorrhagic anemia. Acute blood loss initially causes hypovolemia without anemia because there is a proportional depletion of red blood cells along with other blood components. However, hypovolemia quickly triggers physiological mechanisms that move fluid from the extravascular to the vascular space, resulting in blood dilution and anemia. When chronic, blood loss gradually depletes the body's iron stores and eventually leads to iron deficiency. As the authors have shown in a mouse model (see example below), the GDF decoy polypeptide, optionally in combination with an EPO receptor activator, can be used to rapidly recover from anemia during acute blood loss.

Iron deficiency anemia is the final stage in a gradual progression of increasing iron deficiency, which includes negative iron balance and iron deficiency erythropoiesis as intermediate stages. Iron deficiency may result

- 30 043601 increased iron requirement, reduced dietary iron intake or increased iron loss, for example in conditions such as pregnancy, inadequate diet, small intestinal malabsorption, acute or chronic inflammation and acute or chronic blood loss. In this type of anemia of mild to moderate severity, the bone marrow remains hypoproliferative and red blood cell morphology remains essentially normal; however, even mild anemia can result in some microcytic hypochromic RBCs, and the progression to severe iron deficiency anemia is accompanied by hyperproliferation in the bone marrow and increased prevalence of microcytic and hypochromic RBCs (Adamson, 2008, Harrison's Principles of Internal Medicine, 17th ed.; McGraw Hill, 2008). New York, pp. 628-634). Appropriate therapy for iron deficiency anemia depends on its cause and severity, with oral iron supplements, parenteral iron formulations, and red blood cell transfusions as the main commonly accepted options. The GDF decoy polypeptide, optionally in combination with an EPO receptor activator, can be used to treat chronic iron deficiency anemias alone or in combination with conventional therapeutic approaches, in particular for the treatment of anemias of multifactorial origin.

Hypoproliferative anemias may result from a primary bone marrow dysfunction or failure rather than a dysfunction secondary to inflammation, infection, or progression of malignancy. A well-known example would be myelosuppression caused by chemotherapy for cancer or radiation therapy for cancer. A broad review of clinical trials found that mild anemia may occur in up to 100% of patients following chemotherapy, while more severe anemia may occur in up to 80% of such patients (Groopman et al., 1999, J Natl Cancer Inst 91: 1616- 1634). Myelosuppressive drugs include: 1) alkylating agents such as nitrogen mustards (eg, melphalan) and nitrosoureas (eg, streptozocin); 2) antimetabolites such as folate antagonists (eg, methotrexate), purine analogs (eg, thioguanine), and pyrimidine analogs (eg, gemcitabine); 3) cytotoxic antibiotics such as anthracyclines (for example, doxorubicin); 4) kinase inhibitors (for example, gefitinib); 5) mitosis inhibitors such as taxanes (eg, paclitaxel) and vinca alkaloids (eg, vinorelbine); 6) monoclonal antibodies (for example, rituximab); and 7) topoisomerase inhibitors (eg, topotecan and etoposide). As demonstrated in a mouse model of chemotherapy-induced anemia (see example below), a GDF decoy polypeptide, optionally in combination with an EPO receptor activator, can be used to treat anemia caused by chemotherapeutic agents and/or radiation therapy.

GDF decoy polypeptides, optionally in combination with an EPO receptor activator, may also be suitable for the treatment of anemias of impaired RBC maturation, which is characterized in part by smaller size (microcytes), larger size (macrocytes), deformation or abnormal coloration (hypochromia) of RBCs.

In certain embodiments, GDF decoy polypeptides can be used in combination (eg, administered at the same time or at a different time, but generally in such a way as to achieve overlapping pharmacological effects) with maintenance therapy for ineffective erythropoiesis. Such therapy involves transfusion of either red blood cells or whole blood to treat anemia. In chronic or hereditary anemias, the normal mechanisms for iron homeostasis are overwhelmed by frequent transfusions, eventually leading to toxic and potentially lethal accumulation of iron in vital tissues such as the heart, liver, and endocrine glands. Thus, maintenance therapy for patients chronically suffering from ineffective erythropoiesis also includes treatment with one or more iron chelating molecules to promote iron excretion in urine and/or feces and therefore prevent or reverse tissue iron overload (Hershko, 2006, Haematologica 91: 1307-1312; Cao et al., 2011, Pediatr Rep 3(2):el7). Effective iron chelators must be able to selectively bind and neutralize ferric iron, the oxidized form of iron not bound to transferrin, which likely accounts for much of the iron toxicity due to the catalytic production of hydroxyl radicals and oxidation products (Esposito et al., 2003 , Blood 102:2670–2677). These agents may be structurally different, but they all have donor oxygen or nitrogen atoms capable of forming neutralizing octahedral coordination complexes with individual iron atoms in a stoichiometric ratio of 1:1 (hexadentate substances), 2:1 (tridentate substances), or 3:1 ( bidentate) (Kalinowski et al., 2005, Pharmacol Rev 57:547-583). Effective iron chelators also have a relatively low molecular weight (less than 700 Daltons) with water and lipid solubility to allow access to affected tissues. Specific examples of iron chelator molecules are deferoxamine, a hexadentate substance of bacterial origin that requires daily parenteral administration, and the orally active synthetic substances deferiprone (bidentate) and deferasirox (tridentate). Combination treatment, including administration of two iron chelators on the same day, appears promising in patients not responding to chelator monotherapy and in overcoming problems with poor patient compliance with deferoxamine alone treatment regimens (Cao et al., 2011, Pediatr Rep.

- 31 043601

3(2):el7; Galanello et al., 2010, Ann NY Acad Sci 1202:79-86).

In certain embodiments, GDF decoy polypeptides can be used in combination with hepcidin agonists to ineffective erythropoiesis. A circulating polypeptide primarily produced in the liver, hepcidin is considered a major regulator of iron metabolism due to its ability to induce degradation of ferroportin, an iron export protein localized to absorptive enterocytes, hepatocytes, and macrophages. In general terms, hepcidin reduces the availability of extracellular iron, so hepcidin agonists may be useful for treating ineffective erythropoiesis (Nemeth 2010, Adv Hematol 2010:750643). This view is supported by the beneficial effects of overexpression of hepcidin in a mouse model of β-thalassemia (Gardenghi et al., 2010, J Clin Invest 120:4466-4477).

Additionally, as shown herein, GDF decoy polypeptides can be used in combination with EPO receptor activators to achieve an increase in red blood cells at a lower dose range. This may be beneficial in reducing the known off-target effects and risks associated with high doses of EPO receptor activators. In certain embodiments, the present invention provides methods for treating or preventing anemia in an individual in need thereof by administering to the individual a therapeutically effective amount of a GDF decoy polypeptide or a combination (or concomitant therapy) of a GDF decoy polypeptide and an EPO receptor activator. These methods can be used for therapeutic and prophylactic treatment of mammals, and in particular, humans.

GDF decoy polypeptides can be used in combination with EPO receptor activators to reduce the required dose of these activators in patients who are susceptible to the adverse effects of EPO. The primary adverse effects of EPO are excessive increases in hematocrit or hemoglobin levels and polycythemia. Elevated hematocrit levels can lead to hypertension (more specifically, worsening hypertension) and vascular thrombosis. Other adverse effects of EPO that have been reported, some of which are associated with hypertension, are headaches, influenza-like syndrome, shunt occlusion, myocardial infarction and cerebral seizures due to thrombosis, hypertensive encephalopathy and red blood cell aplasia (Singibarti, (1994) J. Clin Investig 72 (suppl 6), S36-S43; Horl et al. (2000) Nephrol Dial Transplant 15 (suppl 4), 51-56; Delanty et al. (1997) Neurology 49, 686-689; Bunn ( 2002) N Engl J Med 346(7), 522-523).

The rapid effects of the GDF decoy polypeptides described herein on red blood cell levels indicate that these substances act by a different mechanism than EPO. Thus, these antagonists may be useful in increasing red blood cell and hemoglobin levels in patients who do not respond to EPO. For example, a GDF decoy polypeptide may be beneficial for a patient in whom normal to increased dosage of EPO (>300 IU/kg/week) does not increase hemoglobin levels to target. Patients with an inadequate response to EPO occur in all types of anemia, but the greatest number of patients who do not respond to treatment was observed especially often among patients with malignancies and patients with end-stage renal disease. An inadequate response to EPO can be either constitutive (ie, observed after the first EPO treatment) or acquired (eg, observed after repeated EPO treatment).

Patients can be treated with a dosage regimen designed to restore the patient to a target hemoglobin level, typically between about 10 g/dL and about 12.5 g/dL, and typically around 11.0 g/dL (see also Jacobs et al (2000) Nephrol Dial Transplant 15, 15-19), although lower target levels may cause fewer cardiovascular side effects. Alternatively, hematocrit levels (the percentage of the volume of a blood sample occupied by cells) can be used as a parameter of red blood cell status. Hematocrit levels for healthy individuals range from 41 to 51% for adult men and 35 to 45% for adult women. Target hematocrit levels are typically approximately 30-33%. Additionally, hemoglobin/hematocrit levels vary from person to person. Thus, it is optimal if target hemoglobin/hematocrit levels can be set individually for each patient.

In certain embodiments, the present invention provides methods for managing a patient being treated with a GDF trap polypeptide or who is a candidate for treatment with a GDF trap polypeptide by measuring one or more hematologic parameters in the patient. Hematologic parameters can be used to evaluate appropriate dosing for a patient who is a candidate for treatment with a GDF decoy polypeptide, to monitor hematologic parameters during treatment with a GDF decoy polypeptide, to assess whether dosage adjustments need to be made during treatment with a GDF decoy polypeptide, and /or to evaluate an appropriate maintenance dose of the GDF decoy polypeptide. If one or more hematologic parameters are outside the normal range, GDF decoy polypeptide dosing may be reduced, delayed, or interrupted.

Hematologic parameters that can be measured in accordance with the methods provided herein include, for example, red blood cell levels, blood pressure, iron accumulation, and other substances that are found in body fluids and that correlate with elevated red blood cell levels, with using methods known in the art.

Such parameters can be determined using a blood sample from a patient. Elevated red blood cell, hemoglobin, and/or hematocrit levels may cause increased blood pressure.

In one embodiment, if one or more hematologic parameters are outside the normal range, or at the upper limit of normal, in a patient who is a candidate for treatment with a GDF decoy polypeptide, then the initiation of administration of the GDF decoy polypeptide may be delayed until until hematological parameters return to normal or acceptable levels, either naturally or through therapeutic intervention. For example, if a candidate patient has elevated or near-high blood pressure, the patient may be treated with a blood pressure lowering agent to reduce the patient's blood pressure. Any blood pressure lowering agent that is appropriate for the individual patient's condition may be used, including, for example, diuretics, adrenergic receptor inhibitors (including alpha blockers and beta blockers), vasodilators, calcium channel blockers, angiotensin converting enzyme (ACE) inhibitors. ) or angiotensin II receptor blockers. Blood pressure can alternatively be treated using a diet and exercise system. Likewise, if a candidate patient has iron stores below normal or at the lower limit of normal, then the patient can be treated with an appropriate diet regimen and/or iron supplements until the patient's iron stores return to normal or acceptable levels. For patients who have red blood cell and/or hemoglobin levels above normal, administration of GDF decoy polypeptide may be delayed until levels return to normal or acceptable levels.

In certain embodiments, if one or more hematological parameters are outside the normal range, or at the upper limit of normal, in a patient who is a candidate for treatment with a GDF decoy polypeptide, then the start of administration may be delayed. However, the dosage amount or frequency of dosing of the GDF decoy polypeptide may be set in an amount that would reduce the risk of unacceptable increases in hematological parameters occurring after administration of the GDF decoy polypeptide. Alternatively, a treatment regimen may be developed for a patient that combines a GDF decoy polypeptide with a therapeutic agent targeting the undesired level of a hematologic parameter. For example, if a patient has high blood pressure, a treatment regimen may be developed that includes administration of a GDF polypeptide trap and a blood pressure lowering agent. For a patient with less than desirable iron stores, a treatment regimen with a GDF decoy polypeptide and an iron supplement may be developed.

In one embodiment, a baseline value(s) for one or more hematologic parameters may be established for a patient who is a candidate for treatment with a GDF decoy polypeptide, and an appropriate dosing regimen for that patient based on the baseline value(s) may be established. Alternatively, established baseline parameters based on the patient's medical history could be used to appropriately schedule the dosage of the GDF decoy polypeptide for the patient. For example, if a healthy patient has an established baseline blood pressure value that is above a certain normal range, it may not be necessary to bring the patient's blood pressure within the range considered normal in the general population before treatment with a GDF decoy polypeptide. The baseline values of one or more hematologic parameters of a patient prior to treatment with a GDF decoy polypeptide may also be used as appropriate reference values to monitor any changes in hematologic parameters prior to treatment with a GDF decoy polypeptide.

In certain embodiments, one or more hematological parameters are measured in patients who will be treated with a GDF decoy polypeptide. Hematologic parameters can be used to monitor the patient during treatment and to allow adjustment or discontinuation of GDF decoy polypeptide dosing or additional dosing of another therapeutic agent. For example, if administration of a GDF decoy polypeptide results in an increase in blood pressure, red blood cell count, or hemoglobin level, or a decrease in iron stores, then dosing of the GDF decoy polypeptide may be reduced in amount or frequency in order to reduce the effect of the GDF decoy polypeptide on one or more hematological parameters. If administration of a GDF decoy polypeptide results in a change in one or more hematologic parameters that is unfavorable to the patient, then dosing of the GDF decoy polypeptide may be discontinued either temporarily until the hematologic parameter(s) return to an acceptable level, or constantly. Likewise, if one or more hematological parameters do not return to an acceptable range after reducing the dose or frequency of administration of the GDF decoy polypeptide, then dosing may be discontinued. As an alternative or in addition to reducing or stopping the dosage of GDF decoy polypeptide, the patient may

- 33 043601 receive doses of an additional therapeutic agent that targets an undesirable level of hematological parameter(s), such as, for example, a blood pressure lowering agent or an iron supplement. For example, if a patient being treated with a GDF decoy polypeptide has elevated blood pressure, then the dosing of the GDF decoy polypeptide may be continued at the same level and a blood pressure lowering agent is added to the treatment regimen, the dosage of the GDF decoy polypeptide may be reduced ( e.g., by amount and/or frequency), and a blood pressure lowering agent is added to the treatment regimen, or dosing of the GDF polypeptide trap may be discontinued and the patient may be treated with a blood pressure lowering agent.

6. Pharmaceutical compositions.

In certain embodiments, the compounds (eg, GDF decoy polypeptides) of the present invention are formulated with a pharmaceutically acceptable carrier. For example, the GDF decoy polypeptide can be administered alone or as a component of a pharmaceutical composition (therapeutic composition). The compounds in question can be formulated into a composition intended for administration by any convenient route and for use in humans or veterinary medicine.

In certain embodiments, a therapeutic method of the invention comprises administering the composition systemically or locally in the form of an implant or device. When administered, the therapeutic composition used in accordance with the invention is, of course, in a pyrogen-free, physiologically acceptable form. Therapeutically suitable agents other than the GDF decoy polypeptide, which may also optionally be included in the composition described above, can be administered simultaneously or sequentially with the subject compounds (eg, GDF decoy polypeptides) in the methods of the invention.

Typically the compounds are administered parenterally. Pharmaceutical compositions suitable for parenteral administration may contain one or more GDF decoy polypeptides in combination with one or more pharmaceutically acceptable sterile isotonic aqueous or non-aqueous solutions, dispersions, suspensions or emulsions or sterile powders that can be reconstituted into sterile injectable solutions or dispersions immediately before use, which may contain antioxidants, buffers, bacteriostatic agents, solutes that render the compositions isotonic with the blood of the intended recipient, or suspending agents or thickeners. Examples of suitable aqueous and non-aqueous carriers that can be used in the pharmaceutical compositions of the invention include water, ethanol, polyols (such as glycerin, propylene glycol, polyethylene glycol and the like) and suitable mixtures thereof, vegetable oils such as olive oil and injectable organic esters such as ethyl oleate. Appropriate fluidity can be maintained, for example, by using coatings such as lecithin, by maintaining the required particle size in the case of dispersions, and by using surfactants.

Additionally, the composition may be encapsulated or injected into a form for delivery to a target tissue site (eg, bone marrow). In some embodiments, the compositions of the present invention may include a matrix capable of delivering one or more therapeutic compounds (e.g., GDF decoy polypeptides) to a target tissue site (e.g., bone marrow), providing a structure for tissue development and optimally capable of resorption into body. For example, the matrix may provide a slow release of GDF decoy polypeptides. Such matrices can be formed from materials currently used for other implantable medical applications.

The choice of matrix material is based on biocompatibility, biodegradability, mechanical properties, cosmetic appearance and interface properties. The specific use of the compositions in question will depend on the respective formulation. Possible matrices for the compositions may be biodegradable and chemically defined calcium sulfate, tricalcium phosphate, hydroxyapatite, polylactic acid and polyanhydrides. Other possible materials are biodegradable and biologically well characterized, such as bone or dermal collagen. Additionally, the matrices are composed of pure proteins or extracellular matrix components. Other possible matrices are non-biodegradable and chemically characterized, such as synthesized hydroxyapatite, biological glass, aluminates or other ceramic materials. The matrices may consist of combinations of any of the types of material mentioned above, for example polylactic acid and hydroxyapatite or collagen and tricalcium phosphate. Bioceramics can be modified in composition, such as calcium aluminate phosphate, and processed to change pore size, particle size, particle shape, and biodegradability.

In certain embodiments, the methods of the invention may involve oral administration, for example, in the form of capsules, sachets, pills, tablets, lozenges (using a flavor base, typically sucrose and acacia or tragacanth), powders, granules, or as a solution or suspension in aqueous or non-aqueous liquid, or as an oil-in-water emulsion liquid

- 34 043601 or water-in-oil, or in the form of an elixir or syrup, or in the form of lozenges (using an inert base such as gelatin and glycerin or sucrose and acacia) and/or in the form of mouth rinses and the like, each contains a predetermined amount of the product as an active ingredient. The drug can also be administered as a bolus, electuary or paste.

In solid oral dosage forms (capsules, tablets, pills, dragees, powders, granules and the like), one or more therapeutic compounds of the present invention may be admixed with one or more pharmaceutically acceptable carriers, such as sodium citrate or dicalcium phosphate, and /or any of the following: (1) fillers or dry diluents such as starches, lactose, sucrose, glucose, mannitol and/or silicic acid; (2) binders such as, for example, carboxymethylcellulose, alginates, gelatin, polyvinylpyrrolidone, sucrose and/or gum arabic; (3) humectants such as glycerin; (4) disintegrants such as agar-agar, calcium carbonate, potato starch or tapioca, alginic acid, some silicates and sodium carbonate; (5) a solution of retarding agents such as paraffin; (6) absorption enhancers such as quaternary ammonium compounds; (7) humectants, such as cetyl alcohol and glycerol monostearate; (8) absorbents such as kaolin and bentonite clay; (9) lubricants such as talc, calcium stearate, magnesium stearate, solid polyethylene glycols, sodium lauryl sulfate and mixtures thereof; and (10) dyes. In the case of capsules, tablets and pills, the pharmaceutical compositions may also contain buffering agents. Solid compositions of a similar type can also be used as fillers in soft and hard gelatin capsules using excipients such as lactose or milk sugars, as well as high molecular weight polyethylene glycols and the like.

Liquid dosage forms for oral administration include pharmaceutically acceptable emulsions, microemulsions, solutions, suspensions, syrups and elixirs. In addition to the active ingredient, liquid dosage forms may contain inert diluents commonly used in this field, such as water or other solvents, solubilizing agents and emulsifiers, such as ethyl alcohol, isopropyl alcohol, ethyl carbonate, ethyl acetate, benzyl alcohol, benzyl benzoate, propylene glycol, 1. 3-butylene glycol, oils (especially cottonseed, peanut, corn, germ, olive, castor and sesame oils), glycerin, tetrahydrofuryl alcohol, polyethylene glycols and sorbitan fatty acid esters and mixtures thereof. In addition to inert diluents, oral compositions may also contain adjuvants such as humectants, emulsifiers and suspending agents, sweeteners, flavoring agents, colors, flavorings and preservatives.

Suspensions, in addition to the active compounds, may contain suspending agents such as ethoxylated isostearyl alcohols, polyoxyethylene sorbitol and sorbitan esters, microcrystalline cellulose, aluminum metahydroxide, bentonite, agar-agar and tragacanth, and mixtures thereof.

The compositions of the invention may also contain adjuvants such as preservatives, humectants, emulsifiers and dispersing agents. Prevention of the action of microorganisms can be ensured by the inclusion of various antibacterial and antifungal agents, for example, paraben, chlorobutanol, phenolsorbic acid and the like. It may also be desirable to include isotonic agents such as sugars, sodium chloride and the like in the compositions. In addition, prolonged absorption of the injectable pharmaceutical form may be caused by the inclusion of agents that delay absorption, such as aluminum monostearate and gelatin.

It is understood that the dosage regimen will be determined by the attending physician, taking into account various factors that modify the effect of the subject compounds of the invention (eg, GDF decoy polypeptides). Various factors include, but are not limited to, the patient's red blood cell count, hemoglobin level or other diagnostic indicators, the desired target red blood cell count, the patient's age, sex and diet, the severity of any disease that may contribute to the low red blood cell count, timing of administration, and other clinical factors. The addition of other known growth factors to the final composition may also affect the dose. Progress can be monitored by periodic assessment of red blood cell and hemoglobin levels, as well as by assessment of reticulocyte levels and other indicators of the hematopoietic process.

In certain embodiments, the present invention also provides gene therapy for the production of GDF decoy polypeptides in vivo. The therapeutic effect of such treatment can be achieved by introducing GDF trap polynucleotide sequences into cells or tissues having the above disorders. Delivery of GDF decoy polypeptide polynucleotide sequences can be achieved using a recombinant expression vector such as a chimeric virus or colloidal dispersion system. Preferred for therapeutic delivery of polynucleotide sequences of GDF decoy polypeptides is the use of targeted liposomes.

Various viral vectors that can be used for gene therapy as reported in the present invention include adenoviruses, herpes virus, smallpox virus or RNA virus such as a retrovirus. The retroviral vector may be a derivative of a murine or avian retrovirus. Examples of retroviral vectors into which a single foreign gene can be inserted include, but are not limited to, Moloney murine leukemia virus (MoMuLV), Harvey murine sarcoma virus (HaMuSV), murine mammary tumor virus (MuMTV), and Rous sarcoma virus (RSV). A number of additional retroviral vectors may contain multiple genes. All of these vectors can carry or contain a gene for a selectable marker, so that transduced cells can be identified and generated. Retroviral vectors can be made target specific by attaching, for example, a sugar, glycolipid or protein. Preferred targeting is accomplished by using an antibody. Those skilled in the art will appreciate that specific polynucleotide sequences can be inserted into the retroviral genome or attached to the viral envelope to allow specific targeting of a retroviral vector containing a GDF decoy polynucleotide.

Alternatively, tissue culture cells can be directly transfected with plasmids encoding the retroviral structural genes gag, pol and env by conventional calcium phosphate transfection. These cells are then transfected with a vector plasmid containing the genes of interest. The resulting cells release the retroviral vector into the culture medium.

Another targeted delivery system for GDF decoy polynucleotides is a colloidal dispersion system. Colloidal dispersion systems include macromolecular complexes, nanocapsules, microspheres, beads, and lipid-based systems, including oil-in-water emulsions, micelles, mixed micelles, and liposomes. The preferred colloidal system of the present invention is a liposome. Liposomes are artificial membrane vesicles that are used as carriers in vitro and in vivo. RNA, DNA and intact virions can be encapsulated in an aqueous interior and can be delivered to cells in a biologically active form (see, for example, Fraley, et al., Trends Biochem. Sci., 6:77, 1981). Methods for efficient gene transfer using liposome carriers are known in the art, see, for example, Mannino, et al., Biotechniques, 6:682, 1988. The composition of liposomes is usually a combination of phospholipids, usually in combination with steroids, especially cholesterol. Other phospholipids or other lipids may also be used. The physical characteristics of liposomes depend on pH, ionic strength, and the presence of divalent cations.

Examples of lipids used in the preparation of liposomes include phosphatidyl compounds such as phosphatidylglycerol, phosphatidylcholine, phosphatidylserine, phosphatidylethanolamine, sphingolipids, cerebrosides and gangliosides. Exemplary phospholipids include egg phosphatidylcholine, dipalmitoylphosphatidylcholine, and distearoylphosphatidylcholine.

Targeting of liposomes is also possible, for example based on organ specificity, cell specificity and organelle specificity, and is known in the art.

Examples

The invention, having been described generally, will be better understood with reference to the following examples, which are given only for the purpose of illustrating certain embodiments and embodiments of the present invention and are not intended to limit the invention.

Example 1: Creating a GDF trap.

Applicants have constructed a GDF trap as set forth below. A polypeptide with a modified ActRIIB extracellular domain with significantly reduced activin A binding compared to GDF11 and/or myostatin (due to the substitution of leucine for aspartate at position 79 of SEQ ID NO: 1) was fused to a human or mouse Fc domain with a minimal linker (three amino acids glycine) between them. The constructs are designated ActRIIB(L79D 20-134)-hFc and ActRIIB(L79D 20-134)-mFc, respectively. Alternative forms with glutamate instead of aspartate at position 79 were created similarly (L79E). Alternative forms with alanine instead of valine at position 226 SEQ ID NO: 7 below were also created and produced equivalently in all ratios tested. Aspartate at position 79 (relative to SEQ ID NO: 1, or position 60 relative to SEQ ID NO: 7) is highlighted in gray below. Valine at position 226 relative to SEQ ID NO: 7 is also highlighted in gray below.

The GDF trap ActRIIB(L79D 20-134)-hFc is shown below, purified from CHO cell lines (SEQ ID NO: 7).

GRGEAETRECIYYNANWELERTNQSGLERCEGEQDKRLHCYASWRNSSGTIELVKK gcw||ddfncydrqecvateenpqvyfcccegnfcnerfthlpeaggpevtyepppt APTGGGTHTCPPCPAPELLGGPSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVK FNWYVDGVEVHNAKTKPREEQYNSTY RVVSVLTVLHQDWLNGKEYKCKVSNKAL PVPIEKTISKAKGQPREPQVYTLPPSREEMTKNQVSLTCLVKGFYPSDIAVEWESNGQ PENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQQGNVFSCSVMHEALHNHYTQKSLS LSPGK

The ActRIIB-derived portion of the GDF trap has the amino acid sequence shown below (SEQ ID NO: 32), and this portion can be used as a monomer or as a separate

- 36 043601 that with Fc protein in the form of a monomer, dimer or higher order complex.

GRGEAETRECIYYNANWELERTNQSGLERCEGEQDKRLHCYASWRNSSGTIE

LVKKGCWDDDFNCYDRQECVATEENPQVYFCCCEGNFCNERFTHLPEAGGPEVTYE

PPPTART (SEQ ID NO: 32)

The GDF decoy protein was expressed in CHO cell lines. Three different leader sequences were considered:

(i) honey bee mellitin (HBML): MKFLVNVALVFMVVYISYIYA (SEQ ID NO: 8) (ii) tissue plasminogen activator (TPA): MDAMKRGLCCVLLLCGAVFVSP (SEQ ID NO: 9) (iii) native:

MTAPWVALALLWGSLCAGS (SEQ ID NO: 10).

The selected form used the TPA leader sequence and had the following unprocessed amino acid sequence:

MDAMKRGLCCVLLLCGAVFVSPGASGRGEAETRECIYYNANWELERTNQSGLERCE GEQDKRLHCYASWRNSSGTIELVKKGCWDDDFNCYDRQECVATEENPQVYFCCCE GNFCNERFTHLPEAGGPEVTYEPPPTAPTGGGTHTCPPCPAPELLGGPSVFLFPPKPKD TLMISRTPEVTCVVVDVSHEDPEVKFNWYV DGVEVHNAKTKPREEQYNSTYRVVSV LTVLHQDWLNGKEYKCKVSNKALPVPIEKTISKAKGOPREPOVYTLPPSREEMTKNO VSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQO GNVFSCSVMHEALHNHYTQKSLSLSPGK (SEQ ID NO: 11)

This polypeptide is encoded by the following nucleic acid sequence (SEQ ID NO:

12):

A TGGATGCAAT GAAGAGAGGG CTCTGCTGTG TGCTGCTGCT GTGTGGAGCA GTCTTCGTTT CGCCCGGCGC CTCTGGGCGT GGGGAGGCTG AGACACGGGA GTGCATCTAC TACAACGCCA ACTGGGAGCT GGAGCGCACC AACCAGAGCG GCCTGGAGCG CTGCGAAGGC GAGCAGGACA AGCGGCTGCA CTGCTACGCC TCCT GGCGCA ACAGCTCTGG CACCATCGAG CTCGTGAAGA AGGGCTGCTG GGACGATGAC TTCAACTGCT ACGATAGGCA GGAGTGTGTG GCCACTGAGG AGAACCCCCA GGTGTACTTC TGCTGCTGTG AAGGCAACTT CTGCAACGAG CGCTTCACTC ATTTGCCAGA GGCTGGGGGC CCGGAAGTCA CGTACGAGCC ACCCCCGACA GCCCCCACCG GTGGTGGAAC TCACACATGC CCACCGTGCC CAGCACCTGA ACTCCTGGGG GGACCGTCAG TTTTSSTT SSSSSSSAAA CCCAAGGACA CCCTCATGAT CTCCCGGACC CCTGAGGTCA CATGCGTGGT GGTGGACGTG AGCCACGAAG ACCCTGAGGT CAAGTTCAAC TGGTACGTGG ACGGCGTGGA GGTGCATAAT GCCAAGACAA AGCCGCGGGA GGAGCAGTAC AACAGCACGT ACCGTGTGGT CAGCGTCCTC ACCGTCCTGC ACCAGGACTG GCT GAATGGC AAGGAGTACA AGTGCAAGGT STSSAASAA GCCCTCCCAG TCCCCATCGA GAAAACCATC TCCAAAGCCA AAGGGCAGCC CCGAGAACCA CAGGTGTACA CCCTGCCCCC ATCCCGGGAG GAGATGACCA AGAACCAGGT CAGCCTGACC TGCCTGGTCA AAGGCTTCTA TCCCAGCGAC ATCGCCGTGG AGTGGGAGAG CAATGGGC AG CCGGAGAACA ACTACAAGAC CACGCCTCCC GTGCTGGACT CCGACGGCTC STTTTTSTS TATAGCAAGC TCACCGTGGA CAAGAGCAGG TGGCAGCAGG GGAACGTCTT CTCATGCTCC GTGATGCATG AGGCTCTGCA SAASSASTAS ACGCAGAAGA GCCTCTCCCT GTCTCCGGGT AAATGA

Purification can be achieved using a series of column chromatography steps including, for example, three or more of the following in any order: Protein A chromatography, Sepharose Q chromatography, phenyl Sepharose chromatography, size exclusion chromatography, and cation exchange chromatography. Purification may be completed by virus filtration and buffer exchange. In an example purification scheme, cell culture medium is passed through a Protein A column, washed with 150 mM Tris/NaCl (pH 8.0), then washed with 50 mM Tris/NaCl (pH 8.0) and eluted with 0.1 M glycine, pH 3.0. The low pH eluate is stored at room temperature for 30 min as a viral purification step. The eluate is then neutralized and passed through a Sepharose Q ion exchange column and washed with 50 mM Tris pH 8.0, 50 mM NaCl, and eluted in 50 mM Tris pH 8.0, with a NaCl concentration ranging between 150 mM and 300 mM. The eluate is then placed in 50 mM Tris pH 8.0, 1.1 M ammonium sulfate and passed through a phenyl Sepharose column, washed, and eluted in 50 mM Tris pH 8.0 with an ammonium sulfate concentration ranging between 150 and 300 mM. The eluate is dialyzed and filtered for use.

Additional GDF decoys (ActRIIB-Fc fusion proteins modified to reduce the binding ratio of activin A to myostatin or GDF11) are described in PCT/US2008/001506 and WO2006/012627, incorporated by reference herein.

Example 2: Bioassay to evaluate GDF-11 and activin mediated signal transduction.

The A-204 reporter gene assay was used to evaluate the effects of ActRIIB-Fc proteins and GDF decoys on signaling through GDF11 and activin A.

Cell line.

Human rhabdomyosarcoma (derived from muscle tissue). Reporter vector: pGL3(CAGA)12 (Described in Dennler et al., 1998, EMBO 17: 3091-3100). The CAGA12 motif is present in TGF-β response genes (PAI-1 gene), so this vector is widely used for Smad2 and 3 signaling factors.

- 37 043601

Day 1. Seed A-204 cells in a 48-well plate.

Day 2. A-204 cells were transfected with 10 μg of pGL3 (CAGA)12 or pGL3 (CAGA) 12 (10 μg) + pRLCMV (1 μg) and Fugene.

Day 3. Add factors (diluted in medium + 0.1% BSA). Inhibitors must be preincubated with factors for 1 hour before adding to cells. After six hours, the cells were washed with PBS and the cells were lysed.

Luciferase assay was then performed. In the absence of any inhibitors, activin A showed a 10-fold stimulation of reporter gene expression and an ED50 of ~2 ng/ml. GDF-11: 16-fold stimulation, ED50: ~1.5 ng/ml.

ActRIIB (20-134) is a potent inhibitor of the actions of activin, GDF-8 and GDF-11 in this assay. Variants were also tested in this analysis.

Example 3 Inhibition of GDF-11 by N-terminal and C-terminal truncations.

N-terminal and/or C-terminal truncation variants of ActRIIB(20-134)-hFc were prepared and tested for their activity as GDF-11 and activin inhibitors. Activities are shown below (measured in conditioned media).

C-terminal truncations of ActRIIB-hFc:

IC50 (ng/ml) GDF-11 Activin ActRIIB(20-134)-hFc 45 22 ActRIIB(20-132)-hFc 87 32 ActRIIB(20-131)-hFc 120 44 ActRIIB(20-128)-hFc 130 158

As can be seen, truncations of three (end with ...PPT), six (end with ...YEP) or more amino acids from the C-terminus cause a decrease in the activity of the molecule by a factor of three or more. Truncation of the last fifteen amino acids of the ActRIIB portion causes a significant loss of activity (see WO2006/012627).

Amino-terminal truncations were made based on the ActRIIB (20-131)-hFc protein. Activities are shown below (measured in conditioned media).

N-terminal truncations of ActRIIB-hFc:

IC50 (ng/ml) GDF-11 Activin ActRIIB(20-131)-hFc (GRG...) 183 201 ActRIIB(21-131)-hFc (RGE...) 121 325 ActRIIB(22-131)-hFc (GEA...) 71 100 ActRIIB(23-131)-hFc (EAE...) 60 43 ActRIIB(24-131)-hFc (AET...) 69 105

Thus, truncations of two, three, or four amino acids from the N-terminus result in the production of a more active protein than versions with the full-length extracellular domain. Additional experiments show that a five amino acid truncation, ActRIIB(25-131)-hFc, has activity equivalent to the non-truncated form, and additional deletions at the N terminus continue to impair the activity of the protein. Thus, optimal designs will have a C-terminus that ends between amino acids 133-134 of SEQ ID NO: 1, and an N-terminus that begins with amino acids 22-24 of SEQ ID NO: 1. An N-terminus corresponding to amino acids 21 or 25 will give activity similar to that of the ActRIIB (20-134)-hFc construct. These truncations can also be used against GDF traps such as the L79D or L79E variant.

Example 4. ActRIIB-Fc variants, cellular activity.

The activity of ActRIIB-Fc and GDF trap proteins was tested in a cell assay as described above. The results are summarized in the table below. Some variants were tested on various C-terminally truncated constructs. As stated above, truncation of five or fifteen amino acids results in decreased activity. GDF traps (variants L79D and L79E) showed a significant reduction

- 38 043601 activin binding while maintaining wild-type GDF11 inhibition at near wild-type levels.

Binding of soluble ActRIIB-Fc to GDF 11 and activin A:

Variations ActRIIBFc ActRIIB-Fc portion (corresponding to amino acids SEQ ID NO: 1) GDF11 inhibition activity Activin inhibition activity R64 20-134 +++ (approximately 10~ ⁸ M Κι) +++ (approximately 10~ ⁸ M Kt) A64 20-134 + (approximately 10~ ⁶ M Kt) + (approximately 10~ ⁶ M Kt) R64 20-129 +++ +++ R64 K74A 20-134 ++++ +++ + R64A24N 20-134 +++ +++ R64A24N 20-119 ++ ++ R64 A24N K74A 20-119 + + R64 L79P 20-134 + + R64 L79P K74A 20-134 + + R64 L79D 20-134 +++ + R64 L79E 20-134 +++ + R64K 20-134 +++ +++ R64K 20-129 +++ +++ R64 P129S P130A 20-134 +++ +++ R64N 20-134 + +

+ Weak activity (approximately 1x10 ^-6 M K _I ).

++ Moderate activity (approximately 1x10 ^-7 M K _I ).

+++ Good activity (wild type) (approximately 1x10 ^-8 M K _I ).

++++ Higher than wild type activity.

For some variants, half-life in rat serum was assessed. ActRIIB (20-134)-Fc had a serum half-life of approximately 70 hours. ActRIIB (A24N 20-134)-Fc had a serum half-life of approximately 100-150 hours. The A24N variant in the cell assay (above) and in the assays in vivo (below) had activity equivalent to the wild type molecule. Combined with its longer half-life, this means that over time the A24N variant will have a greater effect per unit of protein than the wild-type molecule. The A24N variant, and any of the other variants examined above, can be combined with GDF decoy molecules such as the L79D or L79E variants.

Example 5: Binding of GDF-11 and Activin A.

Binding of specific ActRIIB-Fc proteins and GDF decoys to ligands was tested in the BiaCore™ assay.

ActRIIB-Fc variants or wild-type protein were captured by the system using an anti-hFc antibody. Ligands were injected and flowed through the captured receptor proteins. The results are summarized in the tables below.

Ligand binding specificity for IIB variants:

- 39 043601

GDF11 Protein Ko _P (1/ms) K _off (1/c) KD(M) ActRIIB(20-134)-hFc 1.34e-6 1.13e-4 8.42e-ll ActRIIB(A24N 20-134)-hFc 1.21e-6 6.35e-5 5.19e-ll ActRIIB(L79D 20-134)-hFc 6.7e-5 4.39e-4 6.55e-10 ActRIIB(L79E 20-134)-hFc 3.8e-5 2.74e-4 7.16e-10 ActRIIB(R64K 20-134)-hFc 6.77e-5 2.41e-5 3.56e-ll GDF8 Protein Ko _P (1/ms) K _off (1/c) KD(M) ActRIIB(20-134)-hFc 3.69e-5 3.45e-5 9.35e-ll ActRIIB(A24N 20-134)-hFc ActRIIB(L79D 20-134)-hFc 3.85e-5 8.3e-4 2.15e-9 ActRIIB(L79E 20-134)-hFc 3.74e-5 9e-4 2.41e-9 ActRIIB(R64K 20-134)-hFc 2.25e-5 4.71e-5 2,le-10 ActRIIB(R64K 20-129)-hFc 9.74e-4 2.09e-4 2.15e-9 ActRIIB(P129S, P130R 20134)-hFc 1.08e-5 1.8e-4 1.67e-9 ActRIIB(K74A 20-134)-hFc 2.8e-5 2.03e-5 7.18e-ll Activin A Protein K _op (1/ms) Koff (1/c) KD(M) ActRIIB(20-134)-hFc 5.94e6 1.59e-4 2.68e-ll ActRIIB(A24N 20-134)-hFc 3.34e6 3.46e-4 1.04e-10 ActRIIB(L79D 20-134)-hFc Weak connection ActRIIB(L79E 20-134)-hFc Weak connection ActRIIB(R64K 20-134)-hFc 6, 8 2 e 6 3.25e-4 4.76e-11 ActRIIB(R64K 20-129)-hFc 7.4 6e6 6.28e-4 8.41e-11 ActRIIB(P129S, P130R 20134)-hFc 5.02e6 4.17e-4 8.31e-11

These data support the cellular assays by demonstrating that the A24N variant retains ligand binding activity that is similar to that of the ActRIIB(20-134)hFc molecule, and that L79D or L79E retains myostatin GDF11 binding but shows a marked decrease (not quantifiable). measurement) binding to activin A.

Other variations have been created and tested, as described in WO2006/012627 (incorporated herein by reference in its entirety), see for example pages 59-60, using ligands attached to the device and flow of the receptor through the attached ligands. Notably, K74Y, K74F, K74I (and presumably other hydrophobic substitutions at position K74, such as K74L), and D80I cause a decrease in the ratio of activin A binding to GDF11 binding compared with the wild-type K74 molecule. A table with data related to these options is presented below.

Binding of soluble ActRIIB-Fc variants to GDF11 and activin A (BiaCore assay):

- 40 043601

ActRIIB ActA GDF11 WT (64A) KD=l,8e-7M (+) KD=2.6e-7M (+) WT (64R) pa KD=8.6е-8М ( + + +) +15 tail KD ~2.6е-8М ( + + +) KD=1.9е-8М (++++) E37A * 4g R4 0A - - D54A - 4g K55A ++ 4g R5 6A * 4g K7 4A KD=4.35е-9М + + + + + KD=5,Ze-9M + + + + + K74Y * — K74F * -- K74I 4g — W7 8A * * L7 9A + 4g D80K 4g 4g D80R 4g 4g D8 0A 4g 4g D80F 4g 4g D80G 4g 4g D8 0M 4g 4g D80N 4g 4g D80I 4g -- F82A ++ -

*No binding observed.

- <1/5 of wild type binding.

- ~1/2 of wild type + wild type binding.

++ <2x increase in binding +++ ~5x increase in binding ++++ ~10x increase in binding +++++ ~ 40x increase in binding.

Example 6 ActRIIB-hFc stimulates erythropoiesis in non-human primates.

ActRIIB (20-134)-hFc (IgG1) was administered once a week for 1 month to male and female cynomolgus monkeys by subcutaneous injection. Forty-eight cynomolgus monkeys (24/sex) were assigned to one of four treatment groups (6 animals/sex/group) and received subcutaneous injections of either vehicle or ActRIIB-hFc at concentrations of 3, 10, or 30 mg/kg q.i. week for four weeks (total 5 doses). Parameters including general clinical pathology (hematology, clinical chemistry, coagulation, and urinalysis) were assessed. ActRIIB-hFc caused a statistically significant increase in mean absolute reticulocyte values by day 15 in treated animals. By day 36, ActRIIB-hFc caused significant hematologic changes, including an increase in mean absolute reticulocyte values and red cell volume distribution values and a decrease in mean corpuscular hemoglobin concentration. Both sexes and all treated groups were affected. These effects are consistent with the positive effects of ActRIIB-hFc on the release of immature reticulocytes from the bone marrow. This effect was reversible after drug washout from treated animals (by study day 56). Thus, it is concluded that ActRIIB-hFc stimulates erythropoiesis.

Example 7 ActRIIB-mFc promotes aspects of erythropoiesis in mice by stimulating spleen erythropoietic activity.

This study analyzed the effects of in vivo administration of ActRIIB(20-134)-mFc on the frequency of hematopoietic progenitors in the bone marrow and spleen. One group of C57BL/6 mice was injected with PBS as a control, and a second group of mice was injected with two doses of ActRIIB-mFc at a concentration of 10 mg/kg, and both groups were sacrificed after 8 days. To perform a complete blood count, we used

- 41 043601 peripheral blood, and femurs and spleens were used to perform in vitro clonogenic assays to assess the content of lymphoid, erythroid and myeloid progenitor cells in each organ. During the short period of this study, no significant changes in red blood cell, hemoglobin, or white blood cell levels were observed in the treated mice. In the femur, there were no differences in the number of nucleated cells or progenitor content between the control and treated groups. In spleens, the compound-treated group showed a statistically significant increase in the number of colonies of mature erythroid progenitors (CFU-E) per plate, frequency and total number of progenitors per spleen. In addition, an increase in the number of myeloid (CFU-GM), immature erythroid (BFU-E) progenitors, and total progenitors per spleen was shown.

Animals.

Sixteen female C57BL/6 mice, 6–8 weeks of age, were used in the study. Eight mice were injected subcutaneously with test compound ActRIIB-mFc on days 1 and 3 at a dose of 10 mg/kg, and eight mice were injected subcutaneously with the vehicle control, phosphate-buffered saline (PBS), at a volume of 100 per mouse. All mice were sacrificed 8 days after the first injection according to the appropriate Animal Handling Guidelines. Peripheral blood (PB) samples were collected from individual animals by cardiac puncture and used for complete blood count and leukocyte count (CBC/Diff). Femurs and spleens were collected from each mouse.

Performed tests.

CBC/Diff analyses.

PB was collected from each mouse by cardiac puncture and placed into appropriate Microtamer tubes. Samples were sent to CLV for analysis on a CellDyn 3500 counter.

Clonogenic assays.

Clonogenic progenitors of the myeloid, erythroid and lymphoid lineages were assessed in vitro using the methylcellulose-based media systems described below.

Mature erythroid precursors.

Clonogenic progenitors of mature erythroid (CFU-E) lines were cultured in MethoCult™ 3334, a methylcellulose-based medium containing recombinant human (rh) erythropoietin (3 U/ml).

Lymphoid precursors.

Clonogenic precursors of the lymphoid (CFU-pre-B) lineage were cultured in MethoCult® 3630, a methylcellulose-based medium containing rh Interleukin 7 (10 ng/ml).

Myeloid and immature erythroid precursors.

Clonogenic progenitors of the granulocyte-monocyte (CFU-GM), erythroid (BFU-E) and multipotent (CFU-GEMM) lineages were cultured in MethoCult™ 3434, a methylcellulose-based medium containing recombinant murine (rm) stem cell factor (50 ng/ml ), rh Interleukin 6 (10 ng/ml), rm Interleukin 3 (10 ng/ml) and rh Erythropoietin (3 U/ml).

Ways.

Mouse femurs and spleens were processed according to standard protocols. Briefly, bone marrow was obtained by flushing the femoral cavity with Iscove's modified Dulbecco's medium containing 2% fetal bovine serum (IMDM 2% FBS) using a 21-gauge needle and ¹ cc syringe. Spleen cells were obtained by crushing the spleen through a 70 μm filter and washing the IMDM filter with 2% FBS. Counts of nucleated cells in 3% glacial acetic acid were then performed on single cell suspensions using a Neubauer counting chamber so that the total number of cells per organ could be counted. To remove contaminating erythrocytes, all spleen cells were then diluted with three times the volume of ammonium chloride lysis buffer and incubated on ice for 10 min. The cells were then washed and resuspended in IMDM 2% FBS, and a second cell count was performed to determine the cell concentration after lysis.

Cell stocks were prepared and a methylcellulose-based media formulation was added to each to obtain optimal seeding concentrations for each tissue in each media formulation. Bone marrow cells were plated at a concentration of 1x105 cells per plate in MethoCult™ 3334 to assess mature erythroid progenitors, 2x105 cells per plate in MethoCult™ 3630 to assess lymphoid progenitors, and 3x10 ⁴ cells per plate in MethoCult™ 3434 to assess immature erythroid and myeloid progenitors. Spleen cells were plated at a concentration of 4x105 cells per plate in MethoCult™ 3334 to evaluate mature erythroid progenitors, 4x105 cells per plate in MethoCult™ 3630 to evaluate lymphoid progenitors, and 2x105 cells per plate in MethoCult™ 3434 to evaluate immature erythroid and myeloid progenitors. Cultures were plated in triplicate and incubated at 37°C, 5% CO ₂ until colonies were counted and scored by trained personnel. Mature erythroid progenitors were cultured for 2 days, lymphoid progenitors were cultured for 7 days, and immature erythroid and myeloid progenitors

- 42 043601 were cultivated for 12 days.

Analysis.

Mean ±1 standard deviation was calculated for triplicate cultures from clonogenic assays and for control and treatment groups for all data sets.

The frequency of colony-forming cells (CFC) in each tissue was calculated as follows: cells plated per plate;

average CFC calculated per plate.

Total CFC per femur or spleen was calculated as follows:

total counted CFC x number of nucleated cells per femur or spleen (after erythrocyte lysis);

number of cultured nucleated cells.

Standard t tests were performed to evaluate whether there were differences in the mean number of cells or hematopoietic progenitors between PBS control and compound-treated mice. Due to the potential subjectivity of counting colonies, a p value less than 0.01 was considered significant. Mean values (±SD) for each group are shown in the tables below.

Hematological parameters

Treatment group White blood cells (x10 ⁹ /l) Red blood cells (x10 ⁹ /l) Hemoglobin (g/l) Hematocrit (l/l) PBS (n=8) ActRIIB-mFc (n=8) 9.53+1.44 9.77+1.19 CFC out of trouble 10.5+1.1 10.8+0.3 I renal bone ι 160.9+13.3 162.1+4.1 ι spleen 0.552+0.057 0.567+0.019 Treatment group Total CFC per femur General CFC to the spleen Total CFU-E per femur Total CFU-E to spleen PBS (n=8) 88+10 54+14 156+27 131+71 ActRIIB-mFc (n=8) 8 5+9 7 9+6* 164+23 436+86*

*preliminary analysis shows p<0.05.

Treatment of mice with ActRIIB(20-134)-mFc during the short period of this study did not result in a significant increase in red blood cell or hemoglobin levels. However, the effect on progenitor cell content was noticeable. In the femurs, there were no differences in the number of nucleated cells or progenitor content between the control and treated groups. In spleens, the compound-treated group showed a statistically significant increase in the number of nucleated cells before erythrocyte lysis and the number of colonies of mature erythroid progenitors (CFU-E) per plate, frequency and total number of progenitors per spleen. In addition, an increase in the number of myeloid (CFU-GM), immature erythroid (BFU-E) progenitors, and total progenitors per spleen was observed. Thus, it is expected that after a longer period of time, treatment with ActRIIB(20-134)-mFc may lead to increased red blood cell and hemoglobin levels.

Example 8 GDF Trap Increases RBC Levels in Vivo.

Twelve-week-old male C57BL/6NTac mice were assigned to one of two treatment groups (N=10). Mice were dosed with either vehicle or ActRIIB variant polypeptide (GDF Trap) [ActRIIB(L79D 20-134)-hFc] by subcutaneous injection (SC) at a concentration of 10 mg/kg twice weekly for four weeks. At the end of the study, whole blood was collected by cardiac puncture into tubes containing EDTA, and cell distribution was analyzed using an HM2 hematology analyzer (Abaxis, Inc).

- 43 043601

Group designations.

Group N Mice Injection Dose (mg/kg) Route of administration Frequency 1 10 C57BL/6 BBS 0 PC Twice a week 2 10 C57BL/6 Trap gdf [ActRIIB(L79D 20-134)-hFc] 10 PC Twice a week

GDF trap treatment had no statistically significant effect on white blood cell (WBC) counts compared with vehicle controls. Red blood cell (RBC) counts were increased in the treatment group compared to controls (see table below). Both hemoglobin (HGB) and hematocrit (HCT) were also elevated due to the extra red blood cells. Mean erythrocyte width (RDWc) was higher in treated animals, indicating an increased pool of immature erythrocytes. Thus, GDF trap treatment results in an increase in red blood cell counts, without any discernible effects on leukocyte populations.

Hematological results.

RBC (10 ¹² /l) HGB (g/dl) NST (%) RDWc (%) PBS 10.7+0.1 14.8+0.6 44.8+0.4 17.0+0.1 GDF trap 12.4+0.4** 17.0±0.7* 48.8+1.8* 18.4+0.2**

*=p<0.05, **=p<0.01.

Example 9: GDF Trap is superior to ActRIIB-Fc in increasing red blood cell levels in vivo.

Nineteen-week-old male C57BL/6NTac mice were randomly assigned to one of those groups. Mice were dosed with vehicle (10 mM Tris-buffered saline, TBS), wild-type ActRIIB(20-134)-mFc, or GDF trap ActRIIB(L79D 20-134)-hFc by subcutaneous injection twice weekly for three weeks. Blood was collected from the cheek at baseline and after three weeks of dosing and cell distribution using a hematology analyzer (HM2, Abaxis, Inc.).

Treatment with ActRIIB-Fc or GDF trap had no significant effect on white blood cell (WBC) counts compared with vehicle controls. Red blood cell count (RBC), hematocrit (HCT), and hemoglobin levels were increased in GDF trap-treated mice compared with either controls or the wild-type construct (see table below). Thus, in a direct comparison, the GDF trap promoted red blood cell production to a significantly greater extent than the wild-type ActRIIB-Fc protein. In fact, in this experiment, wild-type ActRIIB-Fc protein did not cause a statistically significant increase in red blood cells, suggesting that longer dosing or higher doses would be necessary to detect this effect.

Hematologic results after three weeks of dosing.

RBC (10 ¹² /ml) NST (%) HGB (g/dl) TBS 11.06+0.46 46, 78+1.9 15.7+0.7 ActRIIB-mFc 11.64+0.09 49.03+0.3 16.5+1.5 GDF trap 13.19+0.2** 53.04+0.8** 18.4+0.3**

**=p<0.01.

Example 10: Construction of a GDF Trap with a Truncated ActRIIB Extracellular Domain.

As described in Example 1, a GDF trap, designated ActRIIB(L79D 20-134)-hFc, was generated by N-terminal fusion of the TPA leader sequence to the extracellular domain of ActRIIB (residues 20-134 in SEQ ID NO: 1) containing a leucine substitution for aspartate (at residue 79 in SEQ ID NO: 1) and a C-terminal fusion of the human Fc domain with a minimal linker (three glycine residues) (Fig. 3). The nucleotide sequence corresponding to this fusion protein is shown in FIG. 4.

A GDF trap with a truncated extracellular domain of ActRIIB, designated ActRIIB(L79D 25131)-hFc, by N-terminal fusion of the TPA leader sequence with a truncated extracellular domain (residues 25-131 in SEQ ID NO: 1) containing a leucine to aspartate substitution (in balance 79 in

- 44 043601

SEQ ID NO: 1) and a C-terminal fusion of the human Fc domain with a minimal linker (three glycine residues) (Fig. 5).

The nucleotide sequence corresponding to this fusion protein is shown in FIG. 6.

Example 11: Selective Ligand Binding by GDF Trap Double Truncation of ActRIIB Extracellular Domain.

The affinities of GDF decoys and other ActRIIB-hFc proteins for several ligands were assessed in vitro using the Biacore™ instrument. The results are summarized in the table below. Kd values were obtained by equilibrium affinity fitting due to the very rapid association and dissociation of the complex, which prevents the accurate determination of k _on and kof.

Ligand selectivity for ActRIIB-hFc variants.

Fused design Activin A (Kd e-11) Activin B (Kd e-11) GDF 11 (Kd e-11) ActRIIB(L79 20-134)-hFc 16 1.2 3, 6 ActRIIB(L79D 20-134)-hFc 1350.0 78, 8 12.3 ActRIIB(L79 25-131)-hFc 18 1.2 3, 1 ActRIIB(L79D 25-131)-hFc 2290.0 62.1 7.4

A GDF trap with a truncated extracellular domain, ActRIIB(L79D 25-131)-hFc, had equal or greater ligand selectivity shown for the longer variant, ActRIIB(L79D 20134)-hFc, with a pronounced loss of activin A and activin B binding and almost complete preservation of GDF11 binding compared to ActRIIB-hFc partners without the L79D substitution. Note that truncation alone (without the L79D substitution) does not change selectivity for the ligands shown here [compare ActRIIB(L79 25-131)-hFc with ActRIIB(L79 20-134)-hFc].

Example 12. Construction of ActRIIB(L79D 25-131)-hFc with alternative nucleotide sequences.

To create ActRIIB(L79D 25-131)-hFc, the extracellular domain of human ActRIIB with an aspartate substitution at the native position position 79 (SEQ ID NO: 1) and with N-terminal and C-terminal truncations (residues 25-131 in SEQ ID NO : 1) fused from the N-terminus to the TPA leader sequence instead of the native ActRIIB leader sequence and from the C-terminus to the human Fc domain with a minimal linker (three glycine residues) (Fig. 5). One nucleotide sequence encoding this fusion protein is shown in FIG. 6 (SEQ ID NO: 27), and an alternative nucleotide sequence encoding the exact same fusion protein is shown in FIG. 9 (SEQ ID NO: 30). This protein was expressed and purified using the methods described in Example 1.

Example 13: ActRIIB Extracellular Domain Truncated GDF Trap Enhances Proliferation of Erythroid Progenitors in Mice.

ActRIIB(L79D 25-131)-hFc was evaluated to determine its effect on erythroid progenitor proliferation. Male C57BL/6 mice (8 weeks old) were treated with ActRIIB(L79D 25-131)-hFc (10 mg/kg, s.c.; n=6) or vehicle (TBS; n=6) on days 1 and 4. then euthanized on day 8 to collect spleens, tibias, femurs, and blood. Spleen and bone marrow cells were isolated, diluted in Iscove's modified Dulbecco's medium containing 5% fetal bovine serum, suspended in a special methylcellulose-based medium, and cultured for either 2 or 12 days to assess the levels of clonogenic precursors at the colony-forming unit stage of the erythroid series ( CFU-E) and at the burst-forming unit erythroid stage (BFU-E), respectively.

Methylcellulose BFU-E medium (MethoCult M3434, Stem Cell Technologies) included murine recombinant stem cell factor, interleukin-3, and interleukin-6, which were not present in methylcellulose CFU-E medium (MethoCult M3334, Stem Cell Technologies ), while both media contained erythropoietin, among other components. For BFU-E and CFU-E, the number of colonies in culture dishes was determined in duplicate from each tissue sample, and statistical analysis of the results was based on the number of mice per treatment group.

Cultures obtained from the spleens of ActRIIB(L79D 25-131)-hFc-treated mice had twice as many CFU-E colonies as corresponding cultures from control mice (P<0.05), while the number of BFU-E colonies did not differ significant when processed in vivo. The number of CFU-E or BFU-E colonies from bone marrow cultures also did not differ significantly between treatments. As expected, increased CFU-E colony counts in spleen-derived cultures were accompanied by highly significant (P < 0.001) changes in red blood cell levels (11.6% increase), hemoglobin concentration (12% increase), and hematocrit level (increase). by 11.6%) upon euthanasia in mice treated with ActRIIB(L79D 25-131)-hFc compared to controls. These results indicate that

- 45 043601 that administration of a GDF trap with a truncated extracellular domain of ActRIIB in vivo can stimulate the proliferation of erythroid progenitors as part of its overall effect to increase red blood cell levels.

Example 14: ActRIIB Extracellular Domain Truncated GDF Trap Compensates for Chemotherapy-Induced Anemia in Mice.

The authors of the application investigated the effect of ActRIIB(L79D 25-131)-hFc on erythropoiesis parameters in a mouse model of anemia induced by paclitaxel-based chemotherapy, which inhibits cell division by blocking microtubule polymerization. Male C57BL/6 mice (8 weeks old) were assigned to one of four treatment groups:

1) paclitaxel (25 mg/kg, i./p.);

2) ActRIIB(L79D 25-131)-hFc (10 mg/kg, i/p);

3) paclitaxel + ActRIIB(L79D 25-131)-hFc;

4) solvent (TBS).

Paclitaxel was administered on day 0, while ActRIIB(L79D 25-131)-hFc or vehicle was administered on days 0 and 3. Blood samples for CBC analysis were collected from separate groups on days 1, 3 and 5, and results for treatment groups 1-3 (above) were expressed as the percentage difference from the solvent at a given time. Tooth wear due to paclitaxel toxicity was a problem in the paclitaxel-only group on day 3 (where n=1); otherwise, n=3-5 per treatment group per time point. Compared with vehicle, paclitaxel alone reduced hemoglobin concentration by almost 13% by day 5, while addition of ActRIIB(L79D 25-131)-hFc prevented this paclitaxel-induced decrease (Fig. 11). Similar effects were observed for hematocrit and RBC levels. In the absence of paclitaxel, ActRIIB(L79D 25-131)-hFc increased hemoglobin concentration by 10% compared to vehicle on days 3 and 5 (Fig. 11). Thus, a GDF trap with a truncated extracellular domain of ActRIIB may increase red blood cell levels sufficiently to compensate for chemotherapy-induced anemia.

Example 15 GDF trap with truncated extracellular domain ActRIIB promotes regression of nephrectomy-induced anemia in mice.

We investigated the effect of ActRIIB(L79D 25-131)-hFc on anemia in a nephrectomy mouse model of chronic kidney disease. Male C57BL/6 mice (11 weeks old) underwent either sham surgery or unilateral nephrectomy to reduce the ability to produce erythropoietin. Mice were allowed to recover for a week postoperatively and were then treated twice weekly with ActRIIB(L79D 25-131)-hFc (10 mg/kg, i.p.; n=15 per condition) or vehicle (TBS; n=15 per condition). condition) in just 4 weeks. Blood samples were collected before dosing and after 4 weeks of treatment. While vehicle-treated nephrectomy mice showed a significant reduction in red blood cell counts after a 4-week treatment period, treatment with ActRIIB(L79D 25-131)-hFc not only prevented the decline but increased red blood cell levels by 17% (P<0.001). ) from the original ones (Fig. 12), despite the reduced ability of the kidneys to produce erythropoietin. In nephrectomy mice, ActRIIB(L79D 25-131)-hFc also caused a significant increase in baseline hemoglobin and hematocrit levels, and in particular stimulated each of these erythropoietic parameters to approximately the same extent in the nephrectomy condition as in the sham-operated condition. (Fig. 13). Thus, a GDF trap with a truncated extracellular domain of ActRIIB can increase red blood cell levels sufficiently to reverse anemia in a model of chronic kidney disease.

Example 16 ActRIIB extracellular domain truncated GDF trap improves recovery from blood loss anemia in rats.

The authors of the application investigated the effect of ActRIIB(L79D 25-131)-hFc on erythropoiesis parameters in a rat model of anemia caused by acute blood loss (acute posthemorrhagic anemia). Male Sprague-Dawley rats (weighing approximately 300 g) were regularly injected with a jugular vein catheter (Harlan) from the manufacturer. On day -1, 20% of the total blood volume was collected from each rat over a 5-minute period through an isoflurane anesthetized catheter. Blood volume was collected based on total blood volume calculated according to the following relationship obtained by Lee et al (J Nucl Med 25:72-76, 1985) for rats with body weight greater than 120 g.

Total blood volume (ml) = 0.062 x body weight (g) + 0.0012.

Volume was replaced with an equal volume of phosphate-buffered saline through the catheter at the time of blood collection. Rats were treated with ActRIIB(L79D 25-131)-hFc (10 mg/kg, SC; n=5) or vehicle (TBS; n=5) on days 0 and 3. Blood samples for SBC analysis were collected via catheter on days -1 (original), 0, 2, 4 and 6.

Control rats responded to a 20% blood loss by dropping red blood cell levels by almost 15% on day 0. These levels remained significantly lower than baseline on days 2 and 4 and had not fully recovered by day 6 (FIG. 14). Although rats treated with ActRIIB(L79D 25-131)-hFc showed an almost identical drop in red blood cell levels after 20% blood loss, these rats then showed a complete recovery of these levels on day 2, followed by an increase on days 4 and 6, which - 46 043601 represents a very significant improvement compared to control levels at the corresponding time points (Fig. 14). Similar results were obtained for hemoglobin concentration.

These findings demonstrate that a GDF trap with a truncated extracellular domain of ActRIIB can create a more rapid recovery of red blood cell levels in anemia caused by acute bleeding.

Example 17: ActRIIB Extracellular Domain Truncated GDF Trap Increases RBC Levels in Non-Human Primates.

Two GDF traps, ActRIIB(L79D 20-134)-hFc and ActRIIB(L79D 25-131)-hFc, were evaluated for their ability to stimulate red blood cell production in cynomolgus monkeys. Monkeys were treated subcutaneously with GDF trap (10 mg/kg; n=4 males/4 females), or vehicle (n=2 males/2 females) on days 1 and 8. Blood samples were collected on day 1 (baseline before treatment), 3 8, 15, 29, and 44, and red blood cell levels (FIG. 15), hematocrit (FIG. 16), hemoglobin levels (FIG. 17), and reticulocyte levels (FIG. 18) were analyzed. Vehicle-treated monkeys exhibited decreased red blood cell, hematocrit, and hemoglobin levels at all time points after treatment, an expected effect of repeated blood draws. In contrast, treatment with ActRIIB(L79D 20-134)-hFc or ActRIIB(L79D 25-131)-hFc increased these parameters at the first time point after treatment (day 3) and maintained them at significantly elevated levels throughout the study (Fig. 15- 17). Importantly, reticulocyte levels in monkeys treated with ActRIIB(L79D 20-134)-hFc or ActRIIB(L79D 25-131)-hFc were significantly increased on days 8, 15 and 29 compared to vehicle (FIG. 18). This result demonstrates that GDF trap treatment increases the production of red blood cell precursors, leading to increased levels of red blood cells.

Taken together, these data demonstrate that truncated GDF traps, as well as full-length variants, can be used as selective antagonists of GDF11 and potentially related ligands to enhance red blood cell production in vivo.

Example 18 GDF Trap Derived from ActRIIB5.

Others have reported an alternative soluble form of ActRIIB (designated ActRIIB5) in which exon 4, including the transmembrane domain of ActRIIB, is replaced by a different C-terminal sequence (WO2007/053775).

The sequence of native human ActRIIB5 without the leader sequence is presented below:

GRGEAETRECIYYNANWELERTNQSGLERCEGEQDKRLHCYASWRNSSGTIELVK kgcw|ddfncydrqecvateenpqvyfcccegnfcnerfthlpeaggpegpwast TIPSGGPEATAAAGDQGSGALWLCLEGPAHE (SEQ ID NO: 36)

Substitution of leucine by aspartate or other acidic residue substitution can be made at the native position 79 (underlined and color coded) as described to create the ActRIIB5(L79D) variant, which has the following sequence:

GRGEAETRECIYYNANWELERTNQSGLERCEGEQDKRLHCYASWRNSSGTIELVK kgcw|ddfncydrqecvateenpqvyfcccegnfcnerfthlpeaggpegpwast TIPSGGPEATAAAGDQGSGALWLCLEGPAHE (SEQ ID NO: 37)

This variant can be linked to a human Fc using the TGGG linker to create a human ActRIIB5(L79D)-hFc fusion protein with the following sequence:

GRGEAETRECIYYNANWELERTNQSGLERCEGEQDKRLHCYASWRNSSGTIELVK kgcw|ddfncydrqecvateenpqvyfcccegnfcnerfthlpeaggpegpwast TIPSGGPEATAAAGDQGSGALWLCLEGPAHEiiliTHTCPPCPAPELLGGPSVFL FPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYV DGVEVHNAKTKPREEQYN STYRVVSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREPQVYTLP PSREEMTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLY SKLTVDKSRWQQGNVFSCSVMHEALHNHYTQKSLSLSPGK (SEQ ID NO: 38) This construct can be expressed in CHO cells.

Example 19: Effects in mice with combination treatment of EPO and ActRIIB extracellular domain truncated GDF trap.

EPO induces red blood cell formation by increasing the proliferation of erythroid precursors, while GDF decoys could potentially influence red blood cell formation in a manner that complements or enhances the effects of EPO. Thus, the authors of the application studied the effect of combined treatment with EPO and ActRIIB(L79D 25-131)-hFc on erythropoiesis parameters. Male C57BL/6 mice (9 weeks old) received a single i/p injection of recombinant human EPO alone (epoetin alfa, 1800 units/kg), ActRIIB(L79D 25-131)-hFc alone (10 mg/kg), EPO and ActRIIB(L79D 25-131)-hFc together, or vehicle (Tris-buffered saline). Mice were euthanized 72 h after dosing and blood, spleens, and femurs were collected.

Spleens and femurs were processed to obtain erythroid progenitor cells for flow cytometric analysis. After removal, the spleen was ground in Iscove's modified Dulbecco's medium containing 5% fetal bovine serum and mechanically separated by pushing through a 70-μm cell filter with a plunger from a sterile 1-ml syringe. The femurs were cleared of any remaining muscle or connective tissue and the ends were cut off to allow collection of the marrow by washing the remaining bone body with Iscove's Modified Dulbecco's Medium containing 5% Fetal Bovine Serum using a 21-gauge needle connected to a 3-mL syringe. . Cell suspensions were centrifuged (2000 rpm for 10 min), and cell pellets were resuspended in PBS containing 5% fetal bovine serum. Cells (106) from each tissue were incubated with anti-mouse IgG to block nonspecific binding, then incubated with fluorescently labeled antibodies against the mouse cell surface markers CD71 (transferrin receptor) and Ter119 (cell surface glycophorin A-associated antigen), washed and analyzed by flow cytometry. Dead cells in the samples were excluded from analysis by counterstaining with propidium iodide. Erythroid differentiation in the spleen or bone marrow was assessed by CD71 labeling, which decreases with differentiation, and Ter119 labeling, which increases during terminal erythroid differentiation, starting at the proerythroblast stage (Socolovsky et al., 2001, Blood 98:3261-3273; Ying et al., 2006, Blood 108: 123-133). Thus, flow cytometry was used to determine the number of proerythroblasts (CD71 ^high Ter119 ^low ), basophilic erythroblasts (CD71 ^high Ter119“ ^gh ), polychromatophil + orthochromatophilic erythroblasts (CD71 ^med Ter119 ^high ), and late orthochromatophilic erythroblasts and reticulocytes (CD71 ^low Ter119 ^high ) as described.

Combined treatment with EPO and ActRIIB(L79D 25-131)-hFc resulted in surprisingly vigorous growth of red blood cells. In the 72-hour time period of this experiment, neither EPO nor ActRIIB(L79D 25131)-hFc alone increased hematocrit significantly compared to vehicle, while combination treatment of the two agents resulted in an almost 25% increase in hematocrit, which was unexpected synergistically, i.e., greater than the sum of their individual effects (Fig. 19). Synergy of this type is generally taken as evidence that the individual agents act through different cellular mechanisms. Similar results were also observed for hemoglobin concentrations (Fig. 20) and red blood cell concentrations (Fig. 21), each of which was also increased synergistically by the combination treatment.

Analysis of erythroid precursor levels revealed a more complex pattern. In mice, the spleen is considered the primary organ responsible for inducible (stress) erythropoiesis. Flow cytometric analysis of spleen tissue after 72 hours revealed that EPO markedly changed the profile of erythropoietic progenitors compared to vehicle, increasing the number of basophilic erythroblasts by more than 170% at the expense of late progenitors (late orthochromatophilic erythroblasts and reticulocytes), which decreased by more than a third ( Fig. 22). Importantly, the combination treatment significantly increased basophilic erythroblasts compared to vehicle, but to a lesser extent than EPO alone, while maintaining unabated maturation of late-stage progenitors (Figure 22). Thus, combined treatment with EPO and ActRIIB(L79D 25-131)-hFc increased erythropoiesis based on a balanced increase in progenitor proliferation and maturation. In contrast to the spleen, the profile of progenitor cells in the bone marrow after combination treatment was not significantly different from the profile after EPO alone. The authors of the application predict, based on the profile of progenitors in the spleen, that the combination treatment could lead to increased levels of reticulocytes and could be accompanied by a sustained increase in levels of mature red blood cells if the experiment was continued longer than 72 hours.

Taken together, these results indicate that the GDF trap with a truncated extracellular domain of ActRIIB can be administered in combination with EPO to synergistically enhance red blood cell formation in vivo. Acting through an additional but undefined mechanism, the GDF Trap may mitigate the potent proliferative effects of pure EPO receptor activator while allowing target red blood cell levels to be achieved with lower doses of EPO receptor activator, thereby avoiding the potential adverse effects or other problems associated with high levels. activation of the EPO receptor.

Example 20 GDF trap with truncated extracellular domain ActRIIB increases red blood cell levels in a mouse model of myelodysplastic syndrome.

- 48 043601

Myelodysplastic syndromes (MDS) are heterogeneous diseases associated with bone marrow failure that are clinically characterized by peripheral cytopenia, refractory anemia, and risk of progression to acute myeloid leukemia. Red blood cell transfusion is a key supportive therapy in MDS to alleviate fatigue, improve quality of life, and prolong survival; however, chronic transfusions typically result in iron overload in these patients with adverse effects on morbidity and mortality, which may lead to the use of medications such as iron chelator therapy (Dreyfus, 2008, Blood Rev 22 Suppl 2:S29-34 ; Jabbour et al., 2009, Oncologist 14:489-496). Although recombinant erythropoietin (EPO) and its derivatives are an alternative therapeutic approach for a small percentage of patients with MDS (Estey, 2003, Curr Opin Hematol 10:60-67), recent studies suggest that this class of agents is associated with an increased risk of morbidity and mortality at some doses due to thromboembolic events and tumor growth (Krapf et al., 2009, Clin J Am Soc Nephrol 4:470-480; Glaspy, 2009, Annu Rev Med 60: 181-192). Thus, there is a need for an alternative therapy for MDS that increases red blood cell levels without the iron overload that accompanies chronic transfusions or the risks inherent with exogenous EPO and its derivatives.

Therefore, we investigated the effect of ActRIIB(L79D 25-131)-hFc on RBC levels in a transgenic mouse model that recapitulates the core features of MDS, including transformation to acute leukemia (Lin et al., 2005, Blood 106:287- 295; Beachy et al., 2010, Hematol Oncol Clin North Am 24:361-375). Beginning at three months of age, male and female NUP98-HOXD13 mice were treated twice weekly with ActRIIB(L79D 25-131)-hFc (10 mg/kg, SC) or vehicle (TBS). Wild-type litter was dosed with ActRIIB(L79D 25-131)-hFc or vehicle and used as a control. Blood samples were collected before dosing and at monthly intervals thereafter for CVS measurements.

Some differences were noted early in the study between NUP98-HOXD13 mice and wild-type controls. Specifically, male NUP98-HOXD13 mice showed significantly reduced RBC concentrations (-8.8%, p<0.05) and hematocrit (-8.4%, p<0.05) compared to wild-type mice and female mice NUP98-HOXD13 showed similar trends. Results after three months of dosing (mean ±SD) are shown in the following table.

NUP98-HOXD13 mice RBC concentration (10 ¹² cells/l) Hemoglobin concentration (g/dL) Hematocrit (%) Males Solvent (n=6) 6.5 6±0.51 10.68±0.68 31.83±2.13 ActRIIB(L79D 25131)-hFc (n=8) 8.65±0.54*** 13.54±0.91*** 39.20±2.82*** Females Solvent (n=5) 6.38±1.61 10.30±2.58 31.96±8.73 ActRIIB(L79D 25131)-hFc (n=6) 8.52±0.70* 13.52±0.56* 42.23±2.53t

*** p<0.001 compared to males + solvent * p<0.05 compared to females + solvent f p=0.056 compared to females + solvent

Compared with vehicle, treatment with ActRIIB(L79D 25-131)-hFc for three months significantly increased RBC and hemoglobin concentrations (by approximately 30%) in both male and female NUP98-HOXD13 mice. Hematocrit also increased significantly in these male mice and increased with a trend towards significance in female mice. Thus, a GDF trap with a truncated extracellular domain of ActRIIB can significantly increase red blood cell levels in a mouse model of MDS. While transfusions inherently provide a source of exogenous iron, the GDF trap increases RBC levels, promoting the utilization of endogenous iron stores through erythropoiesis, thereby avoiding iron overload and its negative consequences. In addition, as noted in Example 19, the GDF trap acts through a different (albeit complementary) cellular mechanism than that used by EPO receptor activators to stimulate erythropoiesis, and thereby bypasses the potential adverse effects associated with EPO receptor activation.

Example 21: Effects of ActRIIB extracellular domain truncated GDF trap on RBC levels and morphology in a mouse model of β-thalassemia.

In thalassemic syndromes, which represent the most common causes of ineffective erythropoiesis, an imbalance in the expression of α- and β-globin chains leads to anemia due to increased apoptosis during erythroblast maturation. Red blood cell transfusion is currently a key supportive therapy for thalassemia, but over time causes potentially fatal iron accumulation in certain tissues (Tanno et al., 2010, Adv Hematol

- 49 043601

2010:358283). For example, heart disease associated with iron overload may account for 50% of the mortality in patients with thalassemia major (Borgna-Pignatti et al., 2005, Ann NY Acad Sci 1054:40-47). It is important to note that endogenous EPO levels are typically elevated and contribute to the etiology of disease in thalassemic syndromes as well as other disorders of ineffective erythropoiesis; thus, therapeutic use of recombinant EPO may be inappropriate. Thus, there is a need for an alternative therapy for thalassemia and other disorders of ineffective erythropoiesis that increases red blood cell levels without the iron overload that accompanies chronic transfusions.

We investigated the effect of ActRIIB(L79D 25-131)-mFc on RBC formation in a mouse model of β-thalassemia intermedia in which the entire coding region of the major β-globin coding gene is deleted. Mice homozygous for this Hbb ^th 1 allele exhibit a hypochromic microcytic anemia with inclusion bodies in a high proportion of circulating RBCs (Skow et al., 1983, Cell 1043: 1043-1052). In a preliminary experiment, Hbb--β-thalassemic mice (C57BL/6J-Hbb ^d3th /J) aged 2-5 months were randomly assigned to receive ActRIIB(L79D 25-131)-mFc (10 mg/kg) or vehicle ( Tris-buffered saline) by subcutaneous injection twice weekly. Diluent-dosed wild-type litter served as an additional control. Blood samples (100 μl) were collected from the cheek before dosing and at regular intervals after dosing for SBC analysis. Characterization of hematological parameters at the start of the study confirmed that these Hbb' ^/_ βthalassemic mice had severe anemia (Fig. 23), and treatment of Hbb' ^/_ mice with ActRIIB(L79D 25-131)mFc for 4 weeks markedly increased the number of red blood cells compared with Hbb' ^/_ mice treated with vehicle, thereby halving the anemia observed in this model (Fig. 24). Treatment-associated increases in hematocrit and hemoglobin concentration were also observed. It is important to note that treatment of Hbb' ^/_ mice with ActRIIB(L79D 25-131)-mFc also resulted in improved RBC morphology and decreased hemolysis and erythrocyte debris compared to vehicle-treated Hbb' ^/_ mice (Figure 25). thus indicating a dramatic improvement in erythropoiesis. Thus, a GDF decoy polypeptide with a truncated extracellular domain of ActRIIB may provide a therapeutic benefit for anemia in a mouse model of β-thalassemia by both increasing RBC number and morphology. By promoting erythroblast maturation while reducing anemia, GDF decoy polypeptides may treat ineffective erythropoiesis. Unlike transfusions, which by nature provide a source of exogenous iron, GDF trap polypeptide can increase RBC levels, promoting the utilization of endogenous iron stores through erythropoiesis, thereby avoiding iron overload and its negative consequences.

Example 22: Effect of ActRIIB Extracellular Domain Truncated GDF Trap on EPO Levels, Splenomegaly, Bone Density, and Iron Overload in a Murine Model of β-Thalassemia.

Hypoxia associated with ineffective erythropoiesis results in elevated EPO levels, which can drive massive proliferation of erythroblasts within and outside the bone marrow, leading to splenomegaly (enlarged spleen), erythroblast-induced bone pathology, and tissue iron overload, even in the absence of therapeutic red blood cell transfusions . Untreated iron overload leads to iron accumulation in tissues, multiple organ dysfunction and premature death (Borgna-Pignatti et al., 2005, Ann NY Acad Sci 1054:40-47; Borgna-Pignatti et al., 2011, Expert Rev Hematol 4:353- 366), most often due to cardiomyopathy in severe forms of thalassemia (Lekawanvijit et al., 2009, Can J Cardiol 25:213-218). By increasing the efficiency of erythropoiesis, the GDF decoy polypeptide may not only alleviate the underlying anemia and elevated EPO levels, but also the associated complications of splenomegaly, bone pathology, and iron overload.

Applicants examined the effect of the GDF decoy polypeptide on these parameters in the same β-thalassemia intermedia mouse model studied in Example 21. Hbb' ^/_ β-thalassemic mice (C57BL/6J-Hbb ^d3th /J) at three months of age at random dispensed to receive ActRIIB(L79D 25-131)-mFc (1 mg/kg, n=7) or vehicle (Tris-buffered saline, n=7) by subcutaneous injection twice weekly for 2 months. Wild-type litter (n=13) dosed with vehicle served as an additional control. Blood samples (100 μl) were collected at the end of the study for CVS analysis. At the end of the study, bone mineral density was determined using dual-energy x-ray absorptiometry (DEXA), serum EPO levels were determined by enzyme-linked immunosorbent assay (ELISA), reactive oxygen species (ROS) were quantified using 2',7'-dichlorodihydrofluorescein diacetate and flow cytometry ( Suragani et al., 2012, Blood 119:5276-5284), and hepcidin mRNA levels were determined using quantitative polymerase chain reaction.

This GDF decoy polypeptide had multiple hematological effects indicating relief of ineffective erythropoiesis. Treatment of Hbb' ^/_ mice with ActRIIB(L79D 25-131)-mFc for 2 months increased the number of RBCs by 25% compared to vehicle-treated Hbb' ^/_ mice (Fig. 26). In Hbb' ^/_ mice, treatment with ActRIIB(L79D 25-131)-mFc also significantly increased the concentration

- 50 043601 hemoglobin rate and hematocrit after 2 months compared to vehicle controls. These changes were accompanied by a decrease in the level of circulating reticulocytes (31.3±2.3% compared with 44.8±5.0% for Hbb'/'mice treated with ActRIIB(L79D 25-131)-mFc or vehicle, respectively) , which corresponds to a decrease in anemia. As in Example 21, treatment of Hbb' ^/_ mice with ActRIIB(L79D 25-131)-mFc resulted in improved RBC morphology and reduced erythrocyte debris compared to vehicle-dosed Hbb ^-/ ' mice. Compared with healthy individuals, patients with thalassemia exhibit an increased rate of red blood cell destruction and elevated serum levels of bilirubin, which is a product of heme catabolism and a marker of hemolysis (Orten, 1971, Ann Clin Lab Sci 1:113-124). In Hbb' ^/_ mice, treatment with ActRIIB(L79D 25-131)-mFc reduced serum bilirubin levels at 2 months by almost half compared with vehicle (Fig. 27), thereby providing evidence that ActRIIB(L79D 25-131 )-mFc may unexpectedly improve the structural/functional integrity of mature red blood cells because it promotes red blood cell formation. Importantly, treatment of Hbb' ^/_ mice with ActRIIB(L79D 25-131)mFc reduced serum EPO levels at 2 months by more than 60% compared to vehicle in the same model (Figure 28). Since elevated EPO levels are the hallmark of ineffective erythropoiesis in β-thalassemia, the reduction in these levels provides strong evidence that ActRIIB(L79D 25-131)-mFc reduces ineffective erythropoiesis itself, and not just the anemia it causes, in this mouse models of thalassemia.

This GDF decoy polypeptide also produces positive changes in endpoints that represent major complications of ineffective erythropoiesis. In patients with thalassemia, both splenomegaly and bone damage are caused by EPO-stimulated erythroid hyperplasia and extramedullary erythropoiesis. In Hbb' ^/_ mice, treatment with ActRIIB(L79D 25-131)-mFc for 2 months significantly reduced spleen weight compared to vehicle (Figure 29) and completely restored bone mineral density to wild-type values (Figure 30) . Iron homeostasis was also significantly improved by treatment with this GDF decoy polypeptide. Serum iron consists of unbound (free) iron and iron bound to apotransferin (forming transferrin), a specialized protein for transporting elemental iron in the circulation. Serum iron represents a relatively small and labile component of total body iron, whereas serum levels of ferritin, another storage form of iron detected primarily intracellularly, represents a larger and less labile component. The third parameter of iron loading is transferrin saturation, the degree to which the iron-binding capacity of transferrin is occupied. In Hbb' ^/_ mice, treatment with ActRIIB(L79D 25-131)-mFc for 2 months significantly reduced each of these indicators of iron overload compared to vehicle (FIG. 31). In addition to its effect on these various parameters of iron homeostasis, ActRIIB(L79D 25131)-mFc normalizes tissue iron overload in Hbb' ^/_ mice, as determined by histochemical analysis of the spleen, liver and kidney (Fig. 32). In addition, this GDF decoy polypeptide had a positive effect on the expression of hepcidin, a liver protein that is considered a master regulator of iron homeostasis (Gantz, 2011, Blood 117:4425-4433), and whose levels vary inversely with dietary iron uptake. Treatment with ActRIIB(L79D 25-131)-mFc restored the abnormally low expression of hepcidin in the liver of Hbb' ^/_ mice (Fig. 33). Finally, another study with a similar design was conducted to determine the effects of this GDF trap on reactive oxygen species (ROS), which are believed to mediate many of the toxic effects of iron overload (Rund et al., 2005, N Engl J Med 353: 1135-1146 ). In 3-month-old Hbb'/'mice, treatment with ActRIIB(L79D 25131)-mFc at 1 mg/kg twice weekly for 2 months virtually normalized ROS levels (Figure 34) and would therefore be predicted to significantly will reduce ROS-mediated tissue damage in thalassemia and other diseases characterized by ineffective erythropoiesis.

Taken together, the above results demonstrate that GDF decoy polypeptides can treat ineffective erythropoiesis, including anemia and elevated EPO levels, as well as complications such as splenomegaly, erythroblast-induced bone pathology and iron overload, and their associated pathologies. In splenomegaly, such pathologies include chest or abdominal pain and reticuloendothelial hyperplasia. Extramedullary hematopoiesis can occur not only in the spleen, but potentially in other tissues in the form of extramedullary hematopoietic pseudotumors (Musallam et al., 2012, Cold Spring Harb Perspect Med 2:a013482). In erythroblast-induced bone pathology, associated pathologies include low bone mineral density, osteoporosis, and bone pain (Haidar et al., 2011, Bone 48:425-432). Associated pathologies with iron overload include hepcidin suppression and increased absorption of dietary iron (Musallam et al., 2012, Blood Rev 26(Suppl 1):S16-S19), multiple endocrinopathies and liver fibrosis/cirrhosis (Galanello et al., 2010, Orphanet J Rare Dis 5:11), and iron overload cardiomyopathy (Lekawanvijit et al. 2009, Can J Cardiol 25:213–218). In contrast to existing therapies for ineffective erythropoiesis, GDF decoy polypeptides such as ActRIIB(L79D 25-131)-mFc are capable of

-

Claims (20)

reduce iron overload in mouse models while increasing red blood cell levels. This new ability distinguishes GDF decoy polypeptides from blood transfusions, which inherently load the body with exogenous iron in the process of treating anemia and do so without alleviating the underlying condition of ineffective erythropoiesis. Incorporation by reference All publications and patents mentioned herein are hereby incorporated by reference in their entirety as if each individual publication or patent were specifically stated to be incorporated by reference. Although specific embodiments of the subject matter of the invention have been discussed, the above description is illustrative and non-limiting. Many variations will be apparent to those skilled in the art upon reading the present specification and the following claims. The entire scope of the invention is to be determined by reference to the claims, along with the full scope of their equivalents, and the description, along with variations. CLAIM 1. The use of a pharmaceutical composition for the treatment of myelofibrosis in a patient, wherein the composition contains a polypeptide containing an amino acid sequence that is at least 90% identical to the amino acid sequence 29-109 of SEQ ID NO: 1, wherein the polypeptide contains an acidic amino acid at a position corresponding to the position 79 SEQ ID NO: 1. 2. Use according to claim 1, wherein the polypeptide contains an amino acid sequence that is at least 95% identical to the sequence of amino acids 29-109 SEQ ID NO: 1. 3. Use according to claim 1, wherein the polypeptide contains an amino acid sequence that is at least 98% identical to the sequence of amino acids 29-109 SEQ ID NO: 1. 4. Use according to claim 1, wherein the polypeptide contains an amino acid sequence that is identical to the sequence of amino acids 29-109 of SEQ ID NO: 1, but the polypeptide contains an acidic amino acid at a position corresponding to position 79 of SEQ ID NO: 1. 5. Use according to claim 1, wherein the polypeptide contains an amino acid sequence that is at least 90% identical to the sequence of amino acids 25-131 of SEQ ID NO: 1. 6. Use according to claim 1, wherein the polypeptide contains an amino acid sequence that is at least 95% identical to the sequence of amino acids 25-131 of SEQ ID NO: 1. 7. Use according to claim 1, wherein the polypeptide contains the sequence of amino acids 25-131 of SEQ ID NO: 1, but the polypeptide contains an acidic amino acid at a position corresponding to position 79 of SEQ ID NO: 1. 8. Use according to claim 1, wherein the polypeptide contains an amino acid sequence that is at least 90% identical to the amino acid sequence of SEQ ID NO: 28. 9. Use according to claim 1, wherein the polypeptide contains an amino acid sequence that is at least 95% identical to the amino acid sequence of SEQ ID NO: 28. 10. Use according to claim 1, wherein the polypeptide contains the amino acid sequence of SEQ ID NO: 28. 11. Use according to claim 1, wherein the polypeptide consists of the amino acid sequence SEQ ID NO: 28. 12. Use according to any one of claims 1 to 11, wherein the polypeptide binds to GDF11. 13. Use according to any one of claims 1 to 12, wherein the polypeptide binds to GDF8. 14. Use according to any one of claims 1 to 13, wherein the polypeptide binds to GDF11 and GDF8. 15. Use according to any one of claims 1 to 9, wherein the acidic amino acid is aspartic acid (D). 16. Use according to any one of claims 1 to 9, wherein the acidic amino acid is glutamic acid (E). 17. Use according to any one of claims 1 to 16, wherein the polypeptide is in a composition for subcutaneous administration. 18. Use according to any one of claims 1 to 17, wherein the patient has splenomegaly, and wherein the pharmaceutical composition treats splenomegaly in the patient. 19. Use according to any one of claims 1 to 18, wherein the patient has extramedullary erythropoiesis, and wherein the pharmaceutical composition reduces extramedullary erythropoiesis in the patient. 20. Use according to any one of claims 1 to 19, wherein the patient has anemia, and wherein the pharmaceutical composition treats the anemia in the patient. -