WO 99/47156 PCT/US99/05533 COMPOSrIONS FOR MODULATING CELL DIFFERENTIATION COMPRISING A LIPID AND A MORPHOGEN Field of the Invention This invention is generally in the field of cell differentiation. In particular, this invention provides compositions and methods of use thereof which modulate the process of cell differentiation. Background of the Invention 5 The process of cell differentiation is continuous process by which cells become more specialized as evidenced by a distinctive function, biochemical capability, and morphology. A collection of such differentiated cells result in a characteristic, functioning tissue. In the absence of disease or trauma, once a cell has acquired a particular differentiated state, its fate is typically irreversible, and the characteristic phenotype of a differentiated cell will be passed on to its 10 progeny in those cases where a particular differentiated cell retains the ability to divide. A well known example of cell differentiation is hematopoiesis in which precursor stem cells may be directed down one of many possible differentiation pathways to become a particular type of blood cell, e.g., erythrocyte, basophil, eosinophil, neutrophil, monocyte, platelet. Various hematopoietic growth factors have been identified which are capable of influencing cells at 15 particular steps in the process so that cells are committed to following the appropriate pathway that leads to a particular differentiated blood cell type. Morphogens are known to induce the proliferation and differentiation of progenitor cells in numerous soft and hard tissues. Morphogens include members of the family of bone morphogenetic proteins (BMPs) identified by their ability to induce ectopic, endochondral bone 20 morphogenesis. The morphogens, also referred to as, osteogenic proteins generally are classified as a subgroup of the TGF-P superfamily of growth factors (Hogan (1996) Genes & Development 10:1580-1594). Members of the morphogen family of proteins include the mammalian osteogenic protein-I (OP-1, also known as BMP-7, and the Drosophila homolog 60A), osteogenic protein-2 (OP-2, also known as BMP-8), osteogenic protein-3 (OP-3), BMP-2 (also known as BMP-2A or 25 CBMP-2A, and the Drosophila homolog DPP), BMP-3, BMP-4 (also known as BMP-2B or CBMP-2B), BMP-5, BMP-6 and its murine homolog Vgr-1, BMP-9, BMP-10, BMP-11, BMP-12, GDF3 (also known as Vgr2), GDF8, GDF9, GDF10, GDF11, GDF12, BMP-13, WO 99/47156 PCT/US99/05533 -2 BMP-14, BMP-15, GDF-5 (also known as CDMP-1 or MP52), GDF-6 (also known as CDMP-2), GDF-7 (also known as CDMP-3), the Xenopus homolog Vgl and NODAL, UNIVIN, SCREW, ADMP, and NEURAL. Mature morphogens results from processing through a "pro-form" to yield a mature 5 polypeptide chain competent to dimerize and containing a carboxy terminal active domain of approximately 97-106 amino acids. All members share a conserved pattern of cysteines in this domain and the active form of these proteins can be either a disulfide-bonded homodimer of a single family member or a heterodimer of two different members (see, e.g., Massague (1990) Annu. Rev. Cell Biol. 6:597; Sampath, et al. (1990) J. Biol. Chem. 265:13198). See also, U.S. 10 5,011,691; U.S. 5,266,683, Ozkaynak et al. (1990) EMBO J. 9: 2085-2093, Wharton et al. (1991) PNAS 88:9214-9218), (Ozkaynak (1992) J. Biol. Chem. 267:25220-25227 and U.S. 5,266,683); (Celeste et al. (1991) PNAS 87:9843-9847); (Lyons et al. (1989 ) PNAS 86:4554-4558). These disclosures describe the amino acid and DNA sequences, as well as the chemical and physical characteristics, of osteogenic proteins. See also, Wozney et al. (1988) 15 Science 242:1528-1534); BMP 9 (W093/00432, published January 7, 1993); DPP (Padgett et al. (1987) Nature 325:81-84; and Vg-1 (Weeks (1987) Cell 51:861-867). Proper tissue repair, preservation, and regeneration depends upon the ability of cells to enter and faithfully continue down specific pathways of cell differentiation. It is, therefore, be desirable to control the fate of cells to promote repair, regeneration, or maintenance of healthy 20 tissue. Summary of the Invention It has now been discovered that morphogenic proteins can be combined with other factors to direct cell fate. This invention provides compositions and methods for influencing cell fate. In particular, the compositions provided herein modulate the process of cell differentiation to 25 promote regeneration, repair, or maintenance of healthy tissue. Compositions of this invention comprise a morphogen combined with a lipid. Such compositions influence the commitment of a cell down a specific pathway of differentiation. Lipid molecules useful in this invention are organic compounds which are extractable from biological material by nonpolar solvents, such as ether, chloroform, benzene, and not readily 30 extractable by aqueous solvents. Such lipids, include but are not limited to, neutral lipids, such as WO 99/47156 PCT/US99/05533 -3 steroids, including cholesterol and glucocorticoids, such as dexamethasone; charged and polar lipids, such as phospholipids, fatty acids, and sphingolipids; and eicosanoids, such as prostaglandins, thromboxanes, leukotrienes, and lipoxins. In a preferred embodiment, the lipid in the compositions of this invention is a glucocorticoid, such as dexamethasone. 5 Compositions of the invention may further comprise a biocompatible matrix, such as hydroxyapatite, collagen (particularly bovine bone collagen), or other biocompatible synthetic matrices. Compositions of the invention may also comprise a carrier such as carboxymethyl cellulose or fibrin glue. Morphogen/lipid compositions of the invention may be administered in a physiologically-acceptable buffer, such as an acetate or saline buffer. 10 This invention also provides methods of modulating cell differentiation comprising treating cells with an effective amount of a composition comprising a morphogen and a lipid in order to direct a population of cells down a differentiation pathway to regenerate, repair, or preserve tissue. According to the methods of this invention, a composition may be applied to cells in vitro ex vivo, or in vivo. The cells may be autogeneic, allogeneic, or xenogeneic. Methods of the 15 invention are useful to influence the differentiation of any cell type at any stage of developement. However, such methods are especially useful to alter the differentiation of multi-potent or pluripotent stem cells. Such methods are useful, for example, to cause differentiation or redifferentiation of cell types selected from mesenchymal cells, dedifferentiated cells, cancer cells, epithelial cells, hematopoietic cells, erythropoietic cells, and other stem cells. 20 Detailed Description Morphogens have previously been shown to induce tissue morphogenesis, including osteogenesis and adipogenesis (see, for example, Sampath et al., J. Biol. Chem., 267: 20352 20362 (1992); Asahina et al., Exp. Cell Res., 222: 38-47 (1996); Reddi, Nature Biochemistry, 1998). It has now been discovered that compositions containing a morphogen and a lipid are 25 capable of modulating cell differentiation. Cell differentiation refers to the process by which a multi- or pluripotent mesenchymal cell or other precursor (stem) cell changes to become a differentiated cell as identified by characteristic proteins, biochemical activities, and morphology. The compositions described herein are particularly useful for influencing the cell fate of multi- or pluripotent mesenchymal cells. Such mesenchymal cells have the potential to 30 differentiate into a particular cell type under the influence of differentiation factors present in a particular local environment. Compositions of this invention are capable of influencing the WO 99/47156 PCTIUS99/05533 -4 direction and extent of development of such pluripotent cells. The endpoint such influence may be a fully-differentiated cell, for example, as found in liver, heart, or bone. Alternitavely, the composition-induced differentiation may result a progenitor cell for a defined cell lineage. Such lineage precursor cells include muscle progenitor cells, adipocytes, and osteoprogenitor cells. 5 Such lineage precursor cells retain the potential to differentiate further, but can be identified by characteristic morphological and/or biochemical markers (see, e.g., Asahina et al., Exp. Cell Res., 222: 38-47 (1996)). Compositions of this invention contain a morphogen and a lipid. In its mature, native form, natural-sourced morphogen is a glycosylated dimer, typically having an apparent molecular 10 weight of about 30-36 kDa as determined by SDS-PAGE. When reduced, the 30 kDa protein gives rise to two glycosylated peptide subunits having apparent molecular weights of about 16 kDa and 18 kDa. In the reduced state, the protein has no detectable osteogenic activity. The unglycosylated protein, which also has osteogenic activity, has an apparent molecular weight of about 27 kDa. When reduced, the 27 kDa protein gives rise to two unglycosylated polypeptide 15 chains, having molecular weights of about 14 kDa to 16 kDa. Typically, the naturally occurring osteogenic proteins are translated as a precursor, having an N-terminal signal peptide sequence typically less than about 30 residues, followed by a "pro" domain that is cleaved to yield the mature C-terminal domain. The signal peptide is cleaved rapidly upon translation, at a cleavage site that can be predicted in a given sequence using the method of Von Heijne (1986) Nucleic 20 Acids Research 14:4683-4691. The pro domain typically is about three times larger than the fully processed mature C-terminal domain. Morphogens useful herein include any known naturally-occurring native proteins including allelic, phylogenetic counterpart and other variants thereof, whether naturally-occurring or biosynthetically produced (e.g., including "muteins" or "mutant proteins"), as well as new, 25 osteogenically active members of the general morphogenic family of proteins. Particularly useful sequences include those comprising the C-terminal 96 or 102 amino acid sequences of DPP (from Drosophila), Vgl (from Xenopus), Vgr-1 (from mouse), the OPI and OP2 proteins, proteins (see U.S. Pat. No. 5,011,691 and Oppermann et al., as well as the proteins referred to as BMP2, BMP3, BMP4 (see W088/00205, U. S. Patent No. 5,013,649 and 30 W091/18098), BMP5 and BMP6 (see W090/11366, PCT/US90/01630), BMP8 and BMP9.
WO 99/47156 PCT/US99/05533 -5 Other proteins useful in the practice of the invention include active forms of OP 1, OP2, OP3, BMP2, BMP3, BMP4, BMP5, BMP6, BMP9, GDF-5, GDF-6, GDF-7, DPP, Vgl, Vgr, 60A protein, GDF-1, GDF-3, GDF-5, GDF-6, GDF-7, BMP10, BMP1 1, BMIP13, BMP15, UNIVIN, NODAL, SCREW, ADMP or NURAL and amino acid sequence variants thereof In one 5 currently preferred embodiment, osteogenic protein include any one of: OPI, OP2, OP3, BMP2, BMP4, BMP5, BMP6, BMP9, and amino acid sequence variants and homologs thereof, including species homologs, thereof Publications disclosing OP-I and OP-2 sequences, as well as their chemical and physical properties, include U.S. 5,011,691, U.S. 5,266,683 incorporated by reference herein. 10 In preferred embodiments, morphogens for use in methods of the invention include proteins having at least 70% homology with the amino acid sequence of the C-terminal Seven cysteine domain of human OP-1, SEQ ID NO: 2, and having the ability to induce endochondral bone formation in the Reddi and Sampath assay described herein. Compounds that meet these requirements are considered functionally equivalent to a known response morphogen. To 15 determine whether a candidate amino acid sequence is functionally equivalent to a reference morphogen, the candidate sequence and the reference sequence are aligned. The first step for performing an alignment is to use an alignment tool, such as the dynamic programming algorithm described in Needleman et al., 48 J. Mol. Biol. 443 (1970), and the Align Program, a commercial software package produced by DNAstar, Inc. the teachings of which are incorporated by 20 reference herein. After the initial alignment is made, it is then refined by comparison to a multiple sequence alignment of a family of related proteins, such as those shown in figures 1 A through IL, which is a multiple sequence alignment of a family of known morphogens, including hOP-1. Once the alignment between the candidate and reference sequences is made and refined, a percent homology score is calculated. The individual amino acids of each sequence are compared 25 sequentially according to their similarity to each other. Similarity factors include similar size, shape and electrical charge. One particularly preferred method of determining amino acid similarities is the PAM250 matrix described in Dayhoff et al., 5 Atlas of Protein Sequence and Structure 345-52 (1978 & Supp.) incorporated by reference herein. A similarity score is first calculated as the sum of the aligned pairwise amino acid similarity scores. Insertions and 30 deletions are ignored for the purposes of percent homology and identity. Accordingly, gap penalties are not used in this calculation. The raw score is then normalized by dividing it by the WO 99/47156 PCTIUS99/05533 -6 geometric mean of the scores of the candidate compound and the seven cysteine domain of hOP 1. The geometric mean is the square root of the product of these scores. The normalized raw score is the percent homology. In a preferred embodiment, a functionally-equivalent morphogen sequence shares at least 5 60% amino acid homology with a reference sequence. That is, any 60% of the aligned amino acids are either identical to, or are conservative substitutions of, the corresponding amino acids in the reference sequence. Any one or more of the naturally-occurring or biosynthetic morphogens disclosed herein may be used as a reference sequence to determine whether a candidate sequence falls within the morphogen family. In a preferred embodiment, the reference sequence is the C 10 terminal seven-cysteine skeleton sequence of human OP-I as shown in SEQ ID NO: 2. Examples of conservative substitutions for use in the above calculations include the substitution of one amino acid for another with similar characteristics, e.g., substitutions within the following groups are well-known: (a) valine, glycine; (b) glycine, alanine; (c) valine, isoleucine, leucine; (d) aspartic acid, glutamic acid; (e) asparagine, glutamine; (f) seine, threonine; (g) lysine, arginine, 15 methionine; and (h) phenylalanine, tyrosine. The term "conservative variant" or "conservative variation" also includes the use of a substituted amino acid in place of an unsubstituted parent amino acid in a given polypeptide chain, provided that antibodies having binding specificity for the resulting substituted polypeptide chain also have binding specificity (i.e., "crossreact" or "immunoreact" with) the unsubstituted or parent polypeptide. 20 In a preferred embodiment, morphogens useful in the present invention are defined by a generic amino acid sequence that represents variations in known morphogens. For example, SEQ ID NOS: 4 and 5 encompass observed variations between preferred morphogens, including OP-1, OP-2, OP-3, CBMP-2A, CBMP-2B, BMP-3, 60A, DPP, Vgl, BMP-5, BMP-6, Vgr-1, and GDF- 1. SEQ ID NO: 5 includes all of SEQ ID NO: 4, and also includes at its N-terminus the five 25 amino acid sequence of SEQ ID NO: 8. The generic sequences include both the amino acid identity shared by these sequences in the C-terminal domain, defined by the six- and seven cysteine skeletons (SEQ ID NOS: 4 and 5, respectively), and alternative amino acids for variable positions within the sequence. Positions that allow for alternative amino acids are represented by "Xaa". Figure 9 shows the alternative amino acids for each "Xaa" position in SEQ ID NOS: 4, 5 30 and 8. For example, referring to SEQ ID NO: 5 and Figure 9, the "Xaa" at position 2 may be a tyrosine or a lysine. The generic sequences provide an appropriate cysteine skeleton for inter- or WO 99/47156 PCTIUS99/05533 -7 intramolecular disulfide bonding, and contain certain critical amino acids likely to influence the tertiary structure of the proteins. In addition, the "Xaa" at position 36 in SEQ ID NO: 4, or at position 41 in SEQ ID NO: 5, may be an additional cysteine, thereby encompassing the morphogenically-active sequences of OP-2 and OP-3. 5 In another embodiment, useful morphogens include those defined by SEQ ID NOS: 6 or 7, which are composite amino acid sequences of the following morphogens: human OP-1, human OP-2, human OP-3, human BMP-2, human BMP-3, human BMP-4, human BMP-5, human BMP-6, human BMP-8, human BMP-9, human BMP-10, human BMP- 11, Drosophila 60A, Xenopus Vg-1, sea urchin UNIVIN, human CDMP-1 (mouse GDF-5), human CDMP-2 (mouse 10 GDF-6, human BMP-13), human CDMP-3 (mouse GDF-7, human BMP-12), mouse GDF-3, human GDF- 1, mouse GDF- 1, chicken DORSALIN, Drosophila dpp, Drosophila SCREW, mouse NODAL, mouse GDF-8, human GDF-8, mouse GDF-9, mouse GDF-10, human GDF-11, mouse GDF-11, human BMP-15, and rat BMP-3b. SEQ ID NO: 7 includes all of SEQ ID NO: 6 and also includes at its N-terminus the five amino acid sequence of SEQ ID NO: 9. SEQ ID NO: 15 6 accommodates the C-terminal six-cysteine skeleton, and SEQ ID NO: 7 accommodates the seven-cysteine skeleton. Positions that allow for alternative amino acids are represented by "Xaa". Figure 10 shows the alternative amino acids for each "Xaa" position in SEQ ID NOS: 6, 7 and 9. As noted above, certain preferred morphogen sequences useful in this invention have 20 greater than 60% identity, preferably greater than 65% identity, with the amino acid sequence defining the preferred reference sequence of hOP-1. These particularly preferred sequences include allelic and phylogenetic variants of the OP-I and OP-2 proteins, including the Drosophila 60A protein, as well as the closely related proteins BMP-5, BMP-6 and Vgr-1. Accordingly, in certain particularly preferred embodiments, useful morphogens include proteins comprising the 25 generic amino acid sequence SEQ ID NO: 3 (referred to herein as "OPX"), which defines the seven-cysteine skeleton and accommodates the homologies between several identified variants of OP-I and OP-2. Positions that allow for alternative amino acids are represented by "Xaa". Figure 11 shows the alternative amino acids for each "Xaa" position in SEQ ID NO: 3. In still another preferred embodiment, useful morphogens include those having an amino 30 acid sequence encoded by a polynucleotide that hybridizes under high stringency conditions with DNA or RNA encoding a reference morphogen. Standard stringency conditions are well WO 99/47156 PCT/US99/05533 -8 characterized in standard molecular biology texts. See generally MOLECULAR CLONING A LABORATORY MANUAL, (Sambrook et al., eds., 1989); DNA CLONING, Vol. I & II (D.N. Glover ed., 1985); OLIGONUCLEOTIDE SYNTHESIS (M.J. Gait ed., 1984); NUCLEIC ACID HYBRIDIZATION (B. D. Hames & S.J. Higgins eds. 1984); B. Perbal, A PRACTICAL GUIDE To MOLECULAR 5 CLONING (1984). Lipids useful in the compositions and methods of this invention are organic compounds which are extractable from biological material by nonpolar solvents, such as ether, chloroform, benzene, but not readily extractable by aqueous solvents. Such lipids may be isolated from natural sources or synthesized, and may be neutral, polar, or charged molecules. Lipids useful in the 10 compositions and methods of this invention include, but are not limited to, steroids, such as cholesterol and glucocorticoids; fatty acids; phospholipids; sphingolipids; and eicosanoids, such as prostaglandins, thromboxanes, leukotrienes, and lipoxins. In an embodiment of this invention, the lipid component of the composition is a lipid characteristic of the cell of the differentiated state sought to be attained. In another embodiment, the lipid component is present in an amount 15 sufficient to result in the desired differentiated state. Compositions provided herein may also contain various other components. For example, in addition to a morphogen and a lipid, a composition of this invention may include a biocompatible, biodegradable matrix such as collagen or hydroxyapatite. Such matrices permit cells to infiltrate and grow into the matrix material during differentiation. As an alternative in 20 applications in which structural integrity of the resulting tissue is not of primary importance, compositions of the invention may be prepared in a non-matrix formulations, including carboxymethyl cellulose , fibrin glue, or a physiologically-acceptable buffer, such as an acetate or saline buffer. Cells may be exposed to a composition of this invention either in vitro or in vivo, or in ex 25 vivo applications. The extent of resulting cell differentiation may be monitored by assaying for characteristic phenotypic markers, such as morphological or biochemical activities. The cells to which compositions of the invention are exposed may be autogeneic, allogeneic, or xenogeneic. The compositions of this invention are useful in regenerating, repairing, and maintaining tissue affected by disease, trauma or aging. In addition, compositions and methods described 30 herein are useful in tissue reconstructive procedures and in maintaining the integrity or viability of transplanted tissue. For example, a composition of this invention may be applied locally to WO 99/47156 PCT/US99/05533 -9 modulate differentiation of cells in a particular area of tissue that is in need of maintenance, restoration, or repair. Alternatively, mesenchymal cells may be cultured in vitro or ex vivo in the presence of a composition of this invention until the desired differentiated state is attained, and then transplanted into an individual in need of such differentiated cells. 5 Compositions of the invention are useful for maintaining the integrity of transplanted tissue or cells. For example, compositions of the invention are administered to a patient prior to, subsequent to, or concurrent with a transplant procedure in order to maintain the differentiated state of the transplanted organ, tissue, or cells consistent with the recipient environment. Compositions of the invention may be administered directly to cells in culture, or in vivo 10 by any of a variety of known modes of administration including intraperitoneally, intravenously, intramuscularly, subcutaneously, and sublingually. Example Morphogen and lipid containing composition for modulating differentiation of pluripotent mesenchymal cells 15 The following experiment examined the ability of a composition of OP-1 and dexamethasone to modulate osteogenesis and adipogenesis of the murine D1 mesenchymal cell line Materials and Methods Multipotential mesenchymal cell, D1, was cloned from mouse bone marrow stroma. D1 20 cells were plated in 35 mm culture wells and were maintained in DMIEM with 10% fetal bovine serum. After the cells reached confluence, they were treated with either OP-I at increasing concentrations of 10, 50, 100 ng/ml or concomitantly treated with dexamethasone (10-7 M). Cells were stained either with von Kossa stain to demonstrate matrix mineralization, or Sudan IV to identify lipid vesicles. Northern blots of total RNA were hybridized with osteocalcin cDNA, 25 collagen type 1 cDNA, and 422(aF2) cDNA, a fat specific gene. In addition, cells were treated with OP-1 (50 ptg/ml), antibody IB-12 against OP-1 or with 10- 7 M dexamethasone and IB-12 to determine if steroid-induced adipogenesis is mediated by OP-1. Results OP-1 induced both osteogenic and adipogenic changes in D1 cells. Accumulation of lipid 30 vesicles within the cells started at 4 days after treatment with OP-1. The number of adipocytes increased with greater concentrations of OP-I and with the time of treatment. The expression of WO 99/47156 PCT/US99/05533 - 10 422(aP2) was dose-dependent and increased with time of treatment, reaching a maximum at 10 days. In the control cultures that were not treated with OP-1, neither adipocytes were observed nor fat-specific gene expression was detected. OP-I enhanced the osteoblastic properties of D1 cells by increasing the number of mineralized nodules and osteocalcin gene expression, but the 5 effect was not dose-dependent between 10 and 100 ng/ml. However, OP-1 inhibited expression of collagen type I mRNA. Dexamethasone produced adipogenesis by stimulating 422(aF2) gene expression in D1 cells while it decreased the osteoblastic gene, osteocalcin and type I collagen expression. Simultaneous treatment with OP-I and dexamethasone showed a synergistic effect in 10 stimulating adipogenesis and on inhibiting type I collagen gene expression. Dexamethasone also counteracted the stimulatory effect of OP-I on osteocalcin gene expression. These results indicate that a composition comprising OP-I and the lipid, dexamethasone, have a synergistic effect in stimulating cell differentiation.