EP0662125A1

EP0662125A1 - Delivery of proteins by intermembrane transfer for preaccommodation of xenogeneic organ transplants and other purposes

Info

Publication number: EP0662125A1
Application number: EP93922259A
Authority: EP
Inventors: Guerard W. Byrne; David Lee Kooyman; John Steele Logan
Original assignee: DNX Biotherapeutics Inc
Current assignee: Nextran Inc
Priority date: 1992-09-22
Filing date: 1993-09-22
Publication date: 1995-07-12
Also published as: FI951325A; WO1994006903A1; FI951325A0; NO951074D0; CA2144767A1; NO951074L; JPH08501451A; AU5132593A; AU671158B2

Abstract

GPI-linked proteins are transferred, especially in vivo, from vertebrate carrier cells to cells of a target vertebrate tissue. For example, for purposes of xenogeneic transplantation, cells bearing complement inhibition factor(s) specific for the recipient species, e.g., genetically engineered, red blood cells, are incubated with transplantable cells of a second, discordant donor species until transfer of the factor(s) occurs, The transfer may occur in vivo or in vitro, and the factor-bearing cells may be normal cells of the recipient species, or genetically engineered cells of the donor species which express a gene encoding a complement inhibition factor(s) of the recipient species. The method is particularly useful in modifying pig organs for xenotransplantation into humans.

Description

i DELIVERY OF PROTEINS BY INTERMEMBRANE TRANSFER FOR PREACCOMMODATION OF XENOGENEIC ORGAN TRANSPLANTS AND OTHER

PURPOSES

BACKGROUND OF THE INVENTION

This application is a continuation-in-part of Serial No. 07/948,521, filed September 22, 1992, hereby incorporated by reference. Field of the Invention

The invention relates to the field of delivery of proteins to a target vertebrate animal or to a vertebrate animal's isolated tissues or organs. In one embodiment, the proteins are complement inhibitors and are delivered to the vascular endothelium of a xenogeneic organ transplant. Description of the Background Art The Cell Membrane

The cell membrane is, simultaneously, both the shield by which the cell's internal processes are protected from the outside, and the interface at which it communicates with its environment. It is particularly important in complex organisms in which protein receptors within the membrane respond to stimuli, activating or deactivating cellular functions, and thereby define the predominant characteristics and functions of many cell types. The cell membrane is composed of lipids, proteins and carbohydrates (See Singer S.J. and Nicolson G.L. (1972) Science 175:720-731, for review of the fluid mosaic model of the membrane.) The lipids are polarized phospholipid molecules with hydrophilic heads containing glycerol, phosphate and other components, and a hydrophobic tail usually composed of two 14-18 carbon hydrocarbon chains. Phosphatidic acid, phosphatidylinositol, and phosphatydylcholine are three common phospholipids found in the cell membranes of most animal and plant cells. Due to the polarized nature of these molecules, in an aqueous environment they will spontaneously combine into a lipid bilayer with the hydrophobic tails on the inside of the bilayer and the hydrophilic heads arranged on the surface. This is the fundamental lipid structure observed in all plasma membranes. In addition to the lipid bilayer. cell membranes contain an array of carbohydrates and proteins producing a complex mosaic surface both outside and inside of the cell. The carbohydrate component is covalently attached both to the lipid heads to produce glycolipids and to membrane associated proteins called glycoproteins.

Intrinsic Membrane Proteins

Proteins which are integrated into cell membranes (intrinsic proteins) are held in place by hydrophobic interactions between the internal region of the bilayer and hydrophobic domains within the protein. The hydrophobic protein domain usually comprises an alpha helix with 19 or more hydrophobic amino acids. Intrinsic membrane proteins control a wide array of cellular functions, including ion and metabolite transport, cell - cell interaction, cell - substrate adhesion, hormone and cytokine reception, and interactions with internal cytoskeletal elements.

Proteins may be incorporated into cellular membranes through several strategies (see Capaldi, Roderick A. (1982) Structure of intrinsic membrane proteins) Trends in Biological Science 7:292- 295) . Globular membrane proteins, such as neuraminidase, often have a single long hydrophobic amino acid sequence which anchors them to the lipid bilayer leaving the hydrophilic globular catalytic domain exposed on the surface of the membrane. (See Fields, S., Winter, G., and Brownlee, G.G. (1981) Nature 290:213- 217.) Other transmembrane proteins may have cne (glycophorin) or multiple hydrophobic amino acid sequences (cytochrome c oxidase) embedded in the lipid bilayer with hydrophilic portions of the protein protruding through both sides of the membrane.

GPI -Anchored Membrane Proteins A more unusual method of anchoring proteins to the cell membrane is through the covalent post-translational addition of glycolipids to the newly formed polypeptide. The best characterized such lipid anchor is the addition of glycosyl- phosphatidyl inositol (GPI) . Addition of GPI occurs in the endoplasmic reticulum. Once added, the hydrocarbons of the phosphatidyl inositol moiety embed into the hydrophobic space of the bilayer. In this way, the GPI group functions to tether or anchor the protein to the cell membrane. (For a review of GPI synthesis and addition, see Ferguson (1991)).

Proteins with GPI anchors include hydrolytic enzymes, cell adhesion molecules, proteins involved in immune cell regulation or complement regulation, and many proteins expressed in parasites. For some of these proteins, particularly the hydrolytic enzymes and complement regulatory proteins, their cellular functions are believed to be well defined, whereas for many others their functions are not well understood although they are clearly important to a variety of cellular processes.

Although the function of these proteins is diverse, GPI- linked proteins do share some common characteristics. The presence of the GPI linkage provides unique chemical possibilities not associated with intrinsic membrane proteins. The GPI anchor is sensitive to deamination by nitrous acid, and to hydrolysis by phosphatidylinositol specific phospholipase C (PIPLC) phosphatidylinositol specific phospholipase D (PIPLD) and phospholipase A2 (PLA-2) . Bacterial PIPLC is used routinely in a test to detect the presence of a GPI linkage since this enzyme can remove at least a portion of the GPI-linked protein population from the surface of a cell. When the inositol ring is modified with an ester linked fatty acid, the GPI anchor is not sensitive to PIPLC, and this and other modifications may explain the presence of PIPLC resistant GPI proteins. See Walter, Elizabeth I., Roberts, William L., Rosenberry, Terrone L., Ratnoff, William D., and Medof, M. Edward, (1990), "Structural basis for variations in the sensitivity of human decay accelerating factor to phosphatidylinositol-specific phospholipase C cleavage," Journal of Immunology, 144:1030-1036. The sensitivity of the GPI linkage to PIPLC hydrolysis and the frequently observed presence of soluble forms of many GPI- linked proteins has led to the suggestion that the GPI anchor may provide a method for cells to control the release of GPI-linked surface proteins. For example lipoprotein lipase is released from 3T3-L1 adipocytes by both PIPLC, indicating its GPI anchor, and by stimulation with physiologically relevant levels of insulin. See Chan, Betty Liwah, Lisanti, Michael P., Rodriguez- Boulan, Enrique, and Saltiel, Alan R. (1988), "Insulin-stimulated release of lipoprotein lipase by metabolism of its phosphatidylinositol anchor," Science, 241:1670-1672. Similar observations have been made for other GPI-linked proteins such as 5' -nucleotidase, alkaline phosphatase and heparin sulfate proteoglycan. See review by Lisanti, Michae.'. P., Rodriguez- Boulan, Enrique, and Saltiel, Alan R., (1990), "Emerging functional roles for the glycosyl-phosphatidylinositol membrane protein anchor," Journal of Membrane Biology, 117:1-10. Insulin's effect on cells is thought to occur by activating specific phospholipases which hydrolyze molecules of glycosyl phosphatidylinositol in the inside of the plasma membrane, releasing diacyglycerol and inositol phosphate glycan which subsequently modulate other insulin sensitive enzymes. See Saltiel, A.R., Fox, J.A. , Sherline P., and Cuatrecasas, P., (1986), Schience, 233:967). Although the free glycosyl phosphatidylinositol is structurally similar to that in the GPI anchor, it must be kept in mind that the insulin dependent hydrolysis occurs within the cytoplasm and would be unlikely to affect extracellular membrane proteins. If the release of these surface proteins is occurring through hydrolysis of the GPI anchor, then it predicts the existence of additional insulin dependent phospholipase(s) which act on the surface or within the plane of the cell membrane. Such an enzyme has not yet been identified, although mammalian PIPLD is known to be an abundant serum protein. See Davitz, Michael A. , Horn, Judy, and Schenkman, Sergio, (1989) , "Purification of a glycosyl-phosphatidylinositol- specific phospholipase D from human plasma, " Journal of Biological Chemistry, 264:13760-13764. Since PIPLD is not capable of releasing GPI proteins from the cell surface it seems unlikely that PIPLD could be the suspected enzyme. This enzyme can, however, hydrolyze the GPI anchor from proteins in solution and therefore may be involved in removing GPI anchors from shed protein. In any case the soluble protein released from the cell via insulin or cytokine regulated shedding, if it does occur in vivo, is not expected to carry an intact GPI anchor.

As a result of the lipid anchor, GPI-linked proteins sometimes display enhanced mobility within the plane of the membrane. Thy-1, DAF, alkaline phosphatase and other GPI-linked proteins have been shown to posses diffusion coefficients which are lower than free lipids but higher than transmembrane proteins. This is not the case, however, for all GPI-linked proteins, and for a given protein only a portion of the population may show enhanced mobility. For some proteins, however, enhanced mobility within the plane of the membrane nicely fits their known or suspected functions. For example both CD59 and DAF regulate complement activation. CD59 stops the formation of membrane attack complexes by interfering with the polymerization of C9 and DAF functions by preventing the formation of C3 convertase and accelerating the decay of existing C3 convertase. Enhanced mobility within the plane of the membrane would increase the probability that these regulatory- proteins would encounter a site under active complement attack. GPI-linked proteins have also been observed to sort to the apical surface of polarized epithelial cells. This sorting behavior may be signaled at least in part by the GPI moiety. For example herpes simplex glycoprotein D is normally expressed on the basolateral surface of polarized MDCK cells. When the carboxy terminal 37 amino acids of DAF, which contains the signal for GPI addition are fused to the cytoplasmic domain of glycoprotein D the resulting GPI-linked fusion protein is transported to the apical surface, not the basolateral surface. Thus addition of the GPI linkage appears to act as a dominant apical sorting signal. See Lisanti, Michael P., Caras, Ingrid W. , Davitz, Michael, A., and Rodriguez-Boulan, Enrique, (1989), "A glycophospholipid membrane anchor acts as an apical targeting signal in polarized epithelial cells," Journal of Cell biology, 109:2145-2156; Lisanti, Michael P., Le Bivic, Andre, Saltiel, Alan R. , and Rodriguez-Boulan, Enrique, (1990), "Preferred apical distribution of glycosyl-phosphatidylinositol (GPI) anchored proteins: A highly conserved feature of the polarized epithelial cell phenotype," Journal Membrane Biology, 113:155-167.

A common feature of GPI-linked proteins appears to be their ability to spontaneously insert into cell membranes. When DAF, is isolated from cells such that the GPI anchor is intact, the purified protein can then be added to heterologous cells where they integrate into the plasma membrane of the cell in a functional way. See Medof, M. Edward, Kinosnita, Taroh, and Nussenzweig, Victor, (1984) , "Inhibition of complement activation on the surface of cells after incorporation of decay-accelerating factor (DAF) into their membranes," Journal of Experimental Medicine, 160:1558-1578. That this membrane integration is dependent on the intact GPI anchor is evident since recombinant soluble DAF can not insert itself into membranes. See Moran, Paul, Beasley, Hannah, Gorrell, Aldona, Martin, Evy, Gribling, Peter, Fuchs, Henry, Gillet, Nancy, Burton, Louis E., and Caras, Ingrid W. , (1992) , "Human recombinant soluble decay accelerating factor inhibits complement activation in vitro and in vivo, " Journal of Immunology 149:1736-1743. Similar studies have been done using CD59 and Thy-1. See Rollins, Scott A., and Sims, Peter J. (1990) , "The complement-inhibitory activity of CD59 resides in its capacity to block incorporarion of C9 into membrane C5c-9," Journal of Immunology, 144:3478-3483; Zhang, Fen, Schmidt, William G. , Hou, Yu, Williams, Alan F., Jacobson, Ken, (1992) , "Spontaneous incorporation of the glycosyl- phosphatidylinositol-linked protein Thy-1 into cell membranes, " Proceeding of the National Academy of Science, 89:5231-5235. Several GPI-linked proteins have soluble, non-GPI-linked analogues. In some cases, the soluble forms are derived from alternative processing of the mRNA resulting in an excreted protein. In other instances, the soluble forms are thought to be derived by PIPLC or protease hydrolysis of membrane bound protein. In the vast majority of cases these soluble proteins no longer contain a GPI anchor. The only exception to this which we are aware of is the presence of soluble GPI containing DAF and CD59 in certain tissues. See Rooney, I.A., and Morgan, B.P., (1992) , "Characterization of the membrane attach complex inhibitory protein CD59 antigen on human amniotic cells and in amniotic fluid," Immunology, 76:541-547. These soluble GPI- linked forms are typically found in regions with high potential for complement activity such as amniotic and seminal fluid and are associated with extracellular vesicles. See Rooney, Isabelle A., Atkinson, John P., Krul, Elaine S., Schonfeld, Gustav, Polakoski, Kenneth, Saffitz, Jeffrey E., and Morgan, B. Paul, (1993) "Physiologic relevance of the membrane attack complex inhibitory protein CD59 in human seminal plasma: CD59 is present on extracellular organelles (prostasomes) , binds cell membranes, and inhibits complement-mediated lysis," Journal of Experimental Medicine, 177:1409-1420. The origin of these GPI-linked proteins in lipid vesicles is not known. For example, it is not clear whether the GPI-linked proteins are shed from the cell surface, and then incorporate into the vesicles, or whether the vesicles shed from cells already contain the proteins.

Unlike transmembrane proteins, GPI-linked proteins do not have a cytoplasmic (intracellular) domain. The absence of a cytoplasmic domain might suggest that GPI-linked proteins would be incapable of transducing signals across the membrane. This does not appear to be the case however. Signal transduction as measured by cell proliferation or cytokine release has been demonstrated for many GPI-linked proteins on T or B cells (Thy-1,

Ly-6, RT-6, DAF, Qa-2, 5* -nucleotidase) , monocytes (CD14, LFA-3) , and thy ic epithelium (LFA-3). For a review, see Lublin, D.M.,

(1992) "Glycosyl-phosphatidylinositol anchoring of membrane proteins," Current Topics in Microbiology and Immunology, 178:141-1162. There is good evidence that the GPI moiety is required for this signal transduction. Crosslinking of Qa-2, a murine class I antigen, normally induces proliferation of T cells whereas antibody crosslinking of murine H-2 class I antigens is not mitogenic. Transgenic mice expressing normal Qa-2, or a transmembrane form produced by fusing the extracellular portion of Qa-2 to the transmembrane domain of H-2D^b or a GPI-linked form of H-2D^b made by fusing the extracellular domain of H-2D^b to the GPI attachment signal of Qa-2 have been produced. T cells from these animals demonstrate that antibody crosslinking of the normal Qa-2 and the GPI-linked H-2D^b-Qa-2 fusion elicits T cell proliferation. In contrast, crosslinking of the transmembrane Qa-2-H-2D^b fusion protein was not mitogenic. See Robinson, Peter J. , Millrain, Margaret, Antoniou, Jane, Simpson, Elizabeth, and Mellor, Andrew L., (1989), "A glycophospholipid anchor is required for Qa-2-mediated T cell activation," Nature, 342:85- 87. Several GPI anchored proteins have been reported to associate with p571ck, and intercellular protein kinase. See Sefanova, I., Horejsi, V., Ansotegui, I.J., Knapp, W. , and Stockinger, H., (1991), "GPI-anchored cell-surface molecules complexed to protein tyrosine kinases," Science, 254:1016-1019. This provides a possible route for GPI anchored signal transduction, although the mechanism and means of association with the kinase is far from understood.

Pathological Role of GPI -Linked Proteins

Paroxysmal nocturnal hemoglobinuria is a human disease in which GPI-linked proteins play an important role. Paroxysmal nocturnal hemoglobinuria (PNH) is an acquired hematopoietic disease affecting some hematopoietic stem cells. These affected stem cells produce a variety of blood cells including erythrocytes, monocytes granulocytes B cells and neutrophils which have reduced or are devoid of all GPI-linked proteins. PNH is caused by an acquired somatic mutation of the PIG-A gene which is involved in the early stages of GPI biosynthesis. See Takeda, Junji, Miyate, Toshio, Kawagoe, Kazuyoshi, Iida, Yoshiyasu, Endo, Yichi, Fujita, Teizo, Takahashi, Minour, Kitani, Teruo, and Kinoshita, Taroh, (1993), "Deficiency of the GPI anchor caused by a somatic mutation of the PIG-A gene in paroxysmal nocturnal hemoglobinuria," Cell, 73:703-711.

The erythrocytes and other blood cells of patients with PNH are unusually sensitive to complement mediated lysis. This symptom is consistent with the loss of the GPI-linked proteins DAF and CD59, both of which are expressed on erythrocytes. See Okuda, Keiko, Kanamaru, Akihisa, Ueda Etsuko, Kitani, Teruo, Okada, Noilo, Okada, Hidechika, Kakishita, Eizo, and Nagai, Kiyoyasu, (1990) , "Expression of decay-accelerating factor on hematopoietic progenitors and their progeny cells grown in cultures with fractionated bone marrow ceils from normal individuals and patients with paroxysmal nocturnal hemoglobinuria, " ExperimentalHematology, 13:1132-1136; Fletcher, A., Bryant, J.A. , Gardner, B., Judson, P.A., Spring, F.A., Parsons, S.F., Mallinson, G. , and Anstee, D.J., (1992), "New monoclonal antibodies in CD59: use for the analysis of peripheral blood cells from paroxysmal nocturnal hemoglobinuria (PNH) patient and for the quantitation of CD59 on normal and decay accelerating factor (DAF) -deficient erythrocytes," Immunology, 75:507-512. Indeed, the exogenous addition and incorporation of CD59 into these affected erythrocytes greatly improves their resistance to complement. See Okada, N. , Harada, R. , Taguchi, R. , and Okada, H. (1989), "Complete deficiency of 20kDa homologous restriction factor (HRF20) and restoration with purified HRF20." Biochemical, Biophysical Research Communication, 164:468-473.

In addition to sensitivity to complement lysis, patients with PNH are susceptible to bacterial infections. The FcRIII (CD16) is a GPI-linked immunoglobin receptor present on neutrophils, large granular lymphocytes, eosinophils, natural killer cells, macrophages and some T Cells. The GPI-linked form is present at 135,000 sites per cell on neutrophils, and represents the most prevalent form of Ig receptor in the blood. A transmembrane form, encoded by a separate gene, is expressed on macrophages and natural killer cells. The GPI-linked PcRIII appears to be the dominant receptor for neutrophil activation by immune complexes. See Hundt, Matthias and Schmidt, Reinhold E. (1992) , "The glycophosphatidylinositol-linked Fc receptor III represents the dominant receptor structure for immuno complex activation of neutrophils," European Journal of Immunology, 22:811-816. In PNH affected neutrophils, expression of this GPI anchored protein is absent. This may help explain the presence of circulating immune complexes and susceptibility to bacterial infections in these patients. See Selvaraj, Periasamy, Rosse, Wendal F., Silber, Robert, and Springer, Timothy M. (1988), "The major Fc receptor in blood has a phosphatidylinositol anchor and is deficient in paroxysmal nocturnal hemoglobinuria, " Nature, 333:565-567; Simmons, David, and Seed, Brian, (1988), "The Fc receptor of natural killer cells is a phcspholipid-linked membrane protein," Nature, 333:568-570).

Some pathological conditions are associated with altered patterns of GPI protein expression. In a variety of immunodeficiency diseases the level of 5' -nucleotidase expressed on B and T cells is reduced. This reduction appears to be a true decrease in the level of expression and not the result of a decrease in the number of B and T cells. It appears that 5' - nucleotidase may be a useful marker to measure the degree of maturation for these cells. See Thompson, Linda F. , Ruedi, Julie M. , and Low, Martin G. , (1987), "Purification of 5' -nucleotidase from human placenta after release from plasma membranes by phosphatidylinositol-specific phospholipase C. , " Biochemical and Biophysical Research Communications, 145:118-125; Thompson, Linda F., Ruedi, Julie M. , Low, Martin G. and Clement, Loran T. , (1987) Journal of Immunology, 139:4042-4048.

The protein CEA, a member of the Ig super family, is a tumor-associated antigen first identified in adenocarcinomas. The level of circulating CEA is widely used r.o monitor colon carcinomas and other malignancies. Circulating CEA does not appear to contain an intact GPI anchor. See Jean, Frederic,

Malapert, Pascale, Rougon, Genevieve, and Barbet, Jacques,

(1988) , "Cell membrane, but not circulating carcinoembryonic antigen is linked to a phosphatidylinositol-containing hydrophobic domain. , " Biochemical and Biophysical Research Communication 155:794-800; and Takami, Noboru, Misumi, Yoshio, Kuroki, Motomu, Matsuoka, Yuji, and Ikehara, Yukio (1988) , "Evidence for carboxy-terminal processing and glycolipid- anchoring of human carcinoembryonic antigen, " Journal of Biological Chemistry, 263:12716-12720).

GPI-linked variable surface glycoproteins are the major coat components of African Trypanosome parasites and are the primary defensive barrier against the host immune system. Trypanosomas contain a large number of VSG genes. During infection, one VSG is predominantly expressed as a very dense outer coat. When the host mounts an antibody response to this VSG the parasites are efficiently cleared, however, minor populations, expressing alternative VSGs, rapidly reproduce, yielding a new peak of parasites. This process will continue, producing a regular series of parasitic fluctuations. Similarly, many surface proteins in protozoa, including Plasmodium and Leishmania, are GPI-linked. See Braun-Breton, Catherine, Rosenberry, TerroneL., and Pereira da Silvia Luiz, (1988) , "Induction of the proteolytic activity of a membrane protein in Plasmodium falciparum by phosphatidyl incsitol-specific phospholipase C, " Nature 332:457- 459; Ferguson, Michael A.J. (1991), "Lipid anchors on membrane proteins, " Current Opinion in Structural Biology, 1:522-529. The frequency of GPI linkage for surface proteins in these parasites suggests an important functional role. African Trypanosome VSG proteins have been detected on erythrocytes after infection and may be responsible for the degree of anemia found in chronic infections.

Intermembrane Transfer of Proteins

Rifkin and Landsberger (1990) report intermembrane transfer of GPI-linked trypanosome variable surface glycoprotein (VSG) from African trypanosomas to sheep red blood cells in vitro. They suggest that this transfer is a two step process: (1) the trypanosome sheds the protein into the aqueous phase, and (2) the protein is taken up by the membrane of the red blood cell. Referring to intermembrane transfer, of, e.g., GPI-linked acetylcholinesterase (Cook, etal., 1980), they comment, "kinetic studies of these systems are consistent with models of protein transfer through an aqueous phase and rule out models that require membrane fusion or collision to effect transfer."

Ferguson (1991) has pointed out that the Rifkin study "did not show whether or not the transfer can occur in the presence of the serum GPI-specific phospholipase D that would be encountered in vivo. " Consequently, it is not clear whether this phenomenon occurs in vivo. If it does, the release may be attributable to an unusual feature of trypanosome membranes: they have a concentration of GPI-linked protein which exceeds, by several orders of magnitude, that found in vertebrate cell membranes. This unusual richness in GPI-linked proteins may faciliatate the release of these proteins from the trypanosomal membrane.

Human DAF has been detected on the surface of Schistosoma mansoni schistosomula and shown to transfer to the schistosomula upon co-cultivation with human red blood cells. See Fatima, M. , Horta, M. , and Ramalho-Pinto, F. Juarez, (1991), "Role of human decay-accelerating factor in the evasion of Schistosoma mansoni from the complement-mediated killing in vitro, " Journal of Experimental Medicine, 174:1399-1406. Fatima et al, (1991) comment that "preliminary unpublished data lead us to presume that DAF sheds from the N-HuE membrane, as shown in human polymorphonuclear cells, goes into solution and then binds to the surface of the parasite, possibly to a specific protein acceptor molecule." Schistosomes can't synthesis their own lipids, and therefore may have evolved several mechanisms to facilitate uptake of exogenous lipids and lipoproteins into their membrane. The existence of a special mechanism is implied by the schistosome's ability to take up non-GPI-linked membrane proteins. The schistosome may not so much "take up", as "tear off", lipoproteins from host cell membranes. It is clear from the foregoing that several members of the scientific community rejected the notion that intermembrane transfer of GPI-linked proteins occurred as a result of direct membrane-to-membrane contact, without the protein transiently entering the aqueous phase. Hourcade, et al., Adv. Immunol. 4j5:381 (1989) review the regulators of the complement activation gene cluster. They note that decay acceleration factor (DAF) , a GPI-linked protein, has been found on the membranes of erythrocytes, T-cells, B-cells, monocytes, granulocytes, platelet, endothelial cells, epithelial cells and fibroblasts. Hourcade, et al. suggest that the appearance of DAF on these different cell types, including endothelial cells, is the result of its tissue-specific expression in these cells. Thereby, Hourcade, et al. teach against the attribution of the appearance of DAF on endothelial cell membranes to a different mechanism, e.g., intermembrane transfer from red blood cells.

The ability of erythrocytes to transfer membrane-associated GPI-linked proteins to other vertebrate cells, in vivo, was not established prior to the present invention.

Pharmaceutical Use of GPI-linked Proteins

Many non-GPI-linked proteins have been purified to homogeneity and formulated as pharmaceuticals for oral or parenteral (intravenous, intramuscular, subcutaneous, etc.) administration. However, the GPI-modified proteins are not readily purified, as the GPI "tail" tends to stick nonspecifically to a multitude of surfaces.

Moran, et al. J. of Immunology 149:1736-1743 (1992) purified three forms of DAF: a GPI-linked membrane bound form and two non-GPI-linked soluble forms. Small scale purification of these molecules required both ion exchange and immuno-affinity chromatography. An organic solvent (7.5% n-butanol) was required for the final elution of GPI-linked membrane DAF, but not of the soluble forms. Thus, the GPI-linked form was more difficult to purify.

All three molecules were capable of inhibiting both the classical and alternative complement pathways, however, the GPI- linked membrane DAF, because it could integrate into cellular membranes, was over 50 times more effective in serum-free media than either soluble form. Unfortunately, the effectiveness of the membrane DAF was greatly diminished in the presence of serum. This report discourages the use of GPI-linked DAF, and presumably other GPI-linked proteins, as an injectable therapeutic, or in any other therapeutic modality which depends on its movement through the bloodstream.

Moreover, a more general problem with the conventional pharmaceutical approach to drug delivery is that if the patient requires continuing treatment, one must either repeatedly administer new doses of the drug, or adopt expedients such as slow-release implants. The conventional approach also falters when the need is for numerous small doses, not a few large ones.

Gene Therapy and Trangenic Animals An alternative approach is to produce the drug in situ, i.e., to provide a gene encoding the drug and cause the patient's cells to express it. In general, this technique, known as gene therapy, has taken two forms: one for the delivery of soluble proteins normally secreted in nature (e.g., hormones), the other for the delivery of membrane or cytoplasmic proteins. For soluble proteins, it is not critical which cells are engineered to contain and express the protein, as long as the protein reaches the bloodstream. Thus, for example, growth hormone, normally made by the pituitary gland, can be expressed by genetically engineered hematopoietic stem cells and provide the relevant physiological function. In contrast, for membrane proteins, gene therapy has been envisaged as requiring expression of the relevant genes in the actual cells which are the final target of the protein. Thus, for example, gene therapeutic delivery of cystic fibrosis transmembrane receptor (CFTR) protein, as a cure for cystic fibrosis, has been envisaged as requiring the direct introduction into the patient's airway of a viral vector (containing the CFTR genes) , such as adenovirus, which is capable of infecting the cells of the lung where it is desired that the CFTR protein be expressed.

As membrane-bound proteins that are not normally secreted, GPI-linked proteins seemingly do not lend themselves to the first mode of gene tiierapy. They would seem to require expression within the final target cells. .And, given the present stage of molecular biology, precise control over the cells to which a foreign gene is delivered, or the type of cells in which will be expressed, is limited.

We find a similar situation in the field of transgenic animal development. If the protein to be expressed will be secreted into the bloodstream, it does not matter usually which cells produce it; it will reach its site of action eventually. For example, transgenic animals with increased growth have been developed using growth hormone genes with promoters that cause the hormone to be primarily expressed in an abnormal (PEPCK) site

(the liver) . For transgenic animals with transgenes encoding membrane proteins, the selection of promoters has previously been limited to those capable of expression in the final target.

Organ Transplantation

In 1989, a major human organ was transplanted into another human being every 40 minutes. The 16,000 organs transplanted that year included almost 9,000 kidneys, nearly 2,000 each of hearts and livers, and close to 3,000 bone marrow transplants. Such transplants, which are from an animal of one species to another of the same species, are termed allografts.

While allograft therapy is one of the most dramatic procedures of modern medicine, it is subject to certain practical limitations. Currently, the greatest limitation is the acute shortage of donor organs of all types. This general shortage is compounded by the need to match donors and recipients for tissue type. While, in 1989, almost 9,000 patients received kidneys, nearly 17,000 remained on the waiting list. They were joined by 1,500 prospective heart recipients and a 1,000 hopeful liver donoees. In the case of heart disease, it is more likely that the patient will die while waiting for a suitable donor organ to become available, than as a result of the complications of cardiac transplantation itself.

There are further disadvantages with current transplantation therapy. Many organ donors are themselves victims of some accident (for example, a road accident) or disease which may have caused some injury to their organ, rendering them less than ideal for transplantation.

Further, because of the unpredictable availability of organs, transplant surgery often cannot be scheduled as a routine operation. All too frequently, surgical teams and hospital administrators have to react the moment a donor organ is identified, thereby causing administrative difficulties. In the case of heart, liver and lung transplants, if rejection is encountered it will not usually be possible to retransplant unless by chance another suitable donor becomes available within a short space of time.

Xenograf ting

Attention therefore has turned to the possibility of using xenografts in transplantation. Xenografting is the generic term commonly used for the cross-species transfer of tissues.

There has been limited use of tissue xenografts in therapy. For example, recent years have witnessed the use of pig tissue for aortic valve replacement, pig skin to covϋr patients with severe burns, and cow umbilical vein as a replacement vein graft. However, more extensive xenotransplants, using entire (or large portions of) functional organs, have generally not been possible due to a rapid immune response by the recipient to the transplant. Within minutes to hours of transplant, hyperacute rejection (HAR) occurs, destroying the xenotransplant. HAR typically involves extensive interstitial edema and hemorrhage, tissue necrosis, and rapid loss of organ function. HAR is most pronounced when transplantation is between distantly related species, such as pig and man. Xenografts subject to HAR are known as discordant xenografts.

Xenotransplantation between closely related species, on the other hand, escapes this ferocious immune reaction (HAR) . Thus, tissue from the chimpanzee, which is a primate closely related to man, can survive without undergoing HAR, xenografts not subject to HAR are known as concordant xenografts. Longer term rejection can occur but is typically less rapid and less severe. While it may be thought that concordant xenografts might provide the solution to the problems with allografts, in practice this is usually not the case. As chimpanzees are much smaller than humans their organs are generally not big enough to provide adequate function in humans. Furthermore, chimpanzees breed slowly in nature and poorly in captivity, and the demand for chimpanzees as experimental animals (particularly in the current era of research into Acquired Immune Deficiency Syndrome (AIDS)) means that, yet again, demand is outstripping supply. Additionally, it may be difficult to obtain public acceptance of the use of non-human primates as xenograft donors. In contrast, the ability to use certain discordant species as sources of xenografts would overcome many if not all of these disadvantages. For example, the pig is potentially a good source of organs which are similar in size and physiology to human organs. Pigs can readily be raised and bred in a domesticated, agricultural setting. Over 50 million are raised in the United States annually. With a short gestation period (4 months) and time to sexual maturity (6 months) , this litter bearing species can be proliferated rapidly and raised cost-effectively. The domestication and use of pigs for human purposes (namely, for food products) is a practice established over centuries. Consequently, an extension of this practice for another human purpose (namely, as a source of xenografts) does not pose the same ethical obstacles as the use of non-domesticated species such as chimpanzees. Unfortunately, with respecr. to humans, pigs are a discordant species. Consequently, HAR is the major impediment to using pig organs in human therapy (see Platt et al., 1990). Complement Activation and Inhibition

The prevailing scientific view is that HAR is produced by the activation of the host's complement lysis defense system. Complement and its activation are now well known, and are described in Roitt, Essential Immunology (Fifth Edition, 1984) Blackwell Scientific Publications, Oxford. The activity ascribed to complement (C) depends upon the operation of nine protein components (Cl to C9) acting in concert, of which the first consists of three major sub-fractions termed Clq, Clr, and Cis. Complement can be activated by the classical or alternative pathway, both of which will now be briefly described.

In the classical pathway, Cl binds to antibody. The Cis subunit acquires esterase activity and brings about the activation and transfer to sites on the membrane or immune complex of first C4 and then C2. This complex has "C3- convertase" activity and splits C3 in solution to produce a small peptide fragment C3a and a residual molecule C3b, which have quite distinct functions. C3a has anaphylatoxin activity and plays no further part in the complement amplification cascade. C3b is membrane bound and can cause immune adherence of the antigen-antibody-C3b complex, so facilitating subsequent phagocytosis.

In the alternative pathway, the C3 convertase activity is performed by C3bB, whose activation can be triggered by extrinsic agents, in particular microbial polysaccharides such as endotoxin, acting independently of antibody. The convertase is formed by the action of Factor D on a complex of C3b and Factor B. This forms a positive feedback loop, in which the product of C3 breakdown (C3b) helps form more of the cleavage enzyme. In both the classical and alternative pathways, the C3b level is maintained by the action of a C3b inactivator (Factor I) . C3b readily combines with Factor H to form a complex which is broken down by Factor I and loses its he olytic and immune adherence properties. The classical and alternative pathways are common after the C3 stage. C5 is split to give C5a and C5b fragments. C5a has anaphylatoxin activity and gives rise to chemotaxis of polymorphs. C5b binds as a complex with C6 and C7 to form a thermostable site on the membrane which recruits the final components C8 and C9 to generate the membrane attack complex (MAC) . This is an annular structure inserted into the membrane and projecting from it, which forms a transmembrane channel fully permeable to electrolytes and water. Due to the high internal colloid osmotic pressure, there is a net influx of sodium ions and water, leading to cell lysis.

Complement inhibition (restriction) factors have been identified which interfere with the action of the complement cascade in such a way as to reduce or prevent its lytic activity; they are used by the host animal to label tissue as "self" to avoid an immune reaction. These factors may be cell membrane bound, or free in serum. Most often they intervene in one of the steps common to both complement activation pathways, however, some factors may be specific to either the classical or the alternative pathway.

Platt and Bach (1991) (and references therein) believe that a discordant xenograft triggers the host's complement (C) through the classical complement pathway. Preformed natural antibodies (PNA) circulating in the host's bloodstream recognize and bind to the donor organ, particularly on the luminal surface of the vascular endothelium. Binding of the PNAs serves to trigger the host's C -system. This attack leads to endothelial cell activation, adhesion of platelets and leukocytes, thrombosis and eventual necrosis of the xenograft organ within a few hours after transplantation. The capillary beds of the transplanted organ appear to be the most sensitive site for attack by the host's complement activity.

In contrast, White, et al., WO91/05855 urge that hyperacute xenograft rejection is not necessarily antibod -mediated, i.e., that it may also arise from the alternative pathway of complement activation. If White, et al are correct, then removal of PNAs will not always prevent hyperacute xenograft rejection.

It has long been known that complement activation plays a critical role in HAR (Schilling (1976), etc.). More recently, the involvement of complement in the hyperacute rejection has been dramatically demonstrated by exogenous inhibition of host complement activity prior to xenotransplantation. Several groups have developed experimental methods to inhibit complement by depleting the level of preformed natural antibodies in the host. The preformed natural antibodies are removed either by perfusing the host's blood through a donor organ such as a pig kidney, or by passing the blood over an immunoaffinity column which removes immunoglobin molecules (Moberg et al., 1971; Fischel et al., 1991; Gianello et al., 1991; Ye et al., 1991; Agishi et al., 1991). These methods are not directly transferrable to -a clinical transplantation setting. Alternatively, the administration of large amounts of cobra venom factor (McConnell and Lachman, 1976) or soluble complement receptor (Weisman et al., 1990; Hebell et al., 1991; Pruitt et al., 1991; Xia et al., 1991) has also been shown to be effective in reducing complement activity. Using these methods at least two independent groups have shown that inhibition of host complement prior to transplantation leads to prolonged xenograft survival (Platt et al., 1990; Fischel et al., 1991; Moberg et al., 1971; Lexer et al., 1987) . Xenografts which would normally be rejected in a few hours have been maintained for days and weeks if the host complement is continuously suppressed.

After approximately two weeks of such exogenous suppression, a fundamental change in the xenograft occurs. This change is referred to as accommodation. A xenograft which has been accommodated by the host, while still recognized by host immunoglobin molecules, is no longer subject to a hyperacute rejection and will survive even in the absence of exogenous complement inhibition. This state of accommodation occurs only when the organ has first gone through a period during which the host complement has been suppressed. For example, a host with one accommodated organ will, in the absence of further exogenous complement inhibition, immediately reject a second xenograft transplant but not the original accommodated graft even when the second graft is from the same donor as the first, e.g., two kidneys [Platt, personal communication] . While the mechanism which underlies accommodation is unknown

(Platt and Bach, 1991) , Platt has suggested three alternative explanations. First, the endothelial cells of the graft may be more sensitive to complement due to the trauma associated with transplantation. If complement is inhibited, then later the endothelial cells of the graft may recover and achieve a resistant condition. Second, as the level of antibodies in the host slowly returns to normal, the epitopes expressed on the endothelial cells of the graft may be altered so that complement is no longer effective. Finally, the host antibodies produced after depletion of preformed natural antibodies may be different in their specificity and/or affinity then the preformed natural antibodies which were originally present. The first two proposals both imply that the endothelial cells of the graft effect some change in their own cellular membrane. They are both rather vague explanations which fail to describe how such a dramatic change might be orchestrated. The final explanation, which may indeed be true, does not adequately explain the accommodation process since it ±r clear that the returning host antibodies are sufficient to trigger an hyperacute rejection of a second similar xenograft. Consistent with these theories, analysis of accommodated tissue has been directed at detecting host immunoglobins and components of the host complement system (Platt et al., 1990; Platt et al., 1991; Rose et al., 1991), not at determining the presence of or absence of host complement-inhibitors in the graft endothelium assayed. Thus, the possibility that C-inhibitors are being transferred to the graft from host cells has not been considered. Additionally, none of the previous theories readily suggest methods for overcoming the HAR problem.

Ability of Endothelial Cells to Take Up Complement Inhibitors From Culture Media

Platt, et al. (1990) , after "inserting human DAF... into pig endothelial cells in vitro". reported that "pig endothelial cells expressing human DAF in their membranes are very significantly protected from lysis by the human serum compared with pig endothelial cells not containing human DAF." It is unclear from this reference whether the DAF was expressed in the pig cells in the molecular biology sense, i.e., the pig cells were genetically engineered to express a human DAF gene, or in a cell biology sense, that is, the pig cells displayed exogenous human DAF on their cell membranes. It appears from Bach, et al. (1991) and Dalmasso et al. (1991) that the latter was the case.

In any event, Dalmasso, et al. (1991) demonstrated that porcine aortic endothelial cells cultured in a medium containing purified human DAF and 10% serum took up DAF and exhibited reduced sensitivity to the cytotoxic effect of human complement

(C) .

Since DAF is GPI-linked, Dalmasso, et al. suggest that analogues of CR1 or membrane cofactor protein (MCP) which are artificially linked with a glycosyl phosphatidyl inositol segment may be more useful in controlling HAR than their non-GPI-linked wild-type forms.

As pointed out previously, membrane DAF is less effective as a complement inhibitor when exposed to serum. There are several reasons this might be the case. First, in serum it will come into contact with soluble lipoproteins and other lipids, which may capture and thereby sequester it. Second, serum contains phospholipases which can cleave off the GPI tail of membrane DAF, so that the DAF molecule can no longer integrate into cell membranes.

It is therefore doubtful that it would be practical to administer human DAF to a pig, intravenously, as a means of delivering an adequate coating of human DAF to all the vascular endothelial cells of the intended xenograft. Perhaps for this reason, Dalmasso did not advocate this approach, instead calling for using a transgenic pig whose endothelial cells expressed DAF.

Significantly, Dalmasso et al. (1991) did not even mention the possibility of using red blood cells to deliver GPI-linked

C-inhibitors to endothelial cells by intermembrane protein transfer, even though he acknowledged that such inhibitors were present on erythrocytes from normal patients. He may well have expected that the degree of shedding of C-inhibitors by normal RBCs was very low, or invading cells would be able to take up those C-inhibitors and be gratuitously protected against host complement. Certainly, the concentration in the aqueous phase would be much less than that used in his studies. Moreover, in the body, the concentration of lipoproteins, lipids and phospholipases would be higher than in his culture medium, which was only 10% FCS. (Indeed, commercial FCS probably does not contain significant levels of phospholipase D.) Clearly, even if he thought that intermembrane transfer of DAF between RBCs and endothelial cells was possible, in view of the teachings in the art that intermembrane transfer of protein in vitro proceeds through the aqueous phase, he would have doubted that it was significant in vivo.

Transgenic Animals Expressing Human Complement Inhibitors in Vascular Endothelial Cells As a Source of Xenografts Platt, et al. (1990) speculated that it might be possible to produce a transgenic pig expressing human DAF, and perhaps other human membrane-associated inhibitors of complement, as a potential donor animal. This suggestion was further detailed in Platt, et al. (1991), which called for directing expression of human inhibitors of complement to the endothelial cell membranes of transgenic pigs. Dalmasso, et al. likewise suggested engineering a transgenic donor animal, such as the pig, with human membrane-associated C-inhibitor genes to achieve a high level of expression of the corresponding proteins in the endothelial cells of the xenograft. Again, the concept is to express the C-inhibitor directly in endothelial cells; no other mechanism of delivering them to endothelial cell membranes is contemplated.

Consistent with the foregoing strategy, White et al, WO 91/05855, prepared transgenic mice bearing a transgene encoding human membrane cofactor protein (MCP) (also known as CD46) (Ex.11) or human decay accelerating factor (DAF) (Ex.12). However, they did not determine whether these genes were expressed, and, if so, in which tissues, or whether a graft from the transgenic animal would elicit HAR in a discordant animal. Similarly, Yannoutsos et al (1991) describe the development of transgenic mice expressing human DAF or MCP. In this study a series of broad spectrum promoters were used so that it is possible that some of the total complement-inhibitor expression would take place in endothelial cells. Most of their animals appeared to have very low levels of complement-inhibitor expression. Furthermore, they did not confirm that expression in endothelial cells had been achieved, or complement-resistance obtained.

Animals transgenic for human DAF have been produced using a partial genomic DNA fragment. See Cary, N. , Moody, J. Yannoutsos, N. , Wallwork, J. , and White, D., (1993), "Tissue expression of human decay accelerating factor, a regulator of complement activation expressed in mice: A potential approach to inhibition of hyperacute rejection." Transplant. Proc, 25:400-401 (Feb. 1993) These animals allegedly exhibited widespread expression of DAF although there appeared to be little expression in hematopoietic tissues. The most consistent expression was observed in vascular smooth muscle, with variable expression in endothelium.

The transgenic animals of the cited references share in common what, over a decade of transgenic anir-ial studies with multiple gene-promoter combinations, has evolved as a standard experimental design and approach. It may be termed a direct approach: if the goal is the presence of a transgene encoded protein on the surface of a particular group or type of cells, couple the transgene to a promoter which can be reasonably expected to express the transgene directly in (at least) those very same cells.

This direct approach may not succeed in producing organs suitable for transplantation. It is doubtful that this direct approach will ensure a uniform distribution of complement- inhibitors throughout the vasculature, particularly the capillary beds, as broad spectrum promoters may not be adequately expressed in all target cells, e.g., endothelial cells. If some of the cells do not express sufficient complement inhibitors, this could result in hyperacute rejection of at least a part of the organ.

There is no suggestion in the multitude of previous transgenic animal studies, including but not limited to White, et al. and Yannoutsos, et al., that a non-standard, indirect approach may be possible and prove perhaps preferable for achieving the presence of transgene encoded protein on a particular group or type of cells. In such an indirect approach, promoters would be selected for expression of the transgene in a group or type of cells different than those cells on whose surface it is desired that the transgene-encoded protein will come to reside. The possibility of such indirect expression, with the transgene-encoded protein being transferred in vivo to distinct target cells, has not been demonstrated in transgenic animals prior to the present invention. Nor would it have been considered rational to seek to express the complement inhibitor in cells other than those of the organ to be transplanted, prior to Applicants' discoveries. No admission is made that any reference cited herein is prior art or pertinent prior art, and Applicants reserve the right to challenge the accuracy of the contents or nominal publication date of any reference.

SUMMARY OF THE INVENTION The present invention overcomes the deficiencies of the background art. More particularly, it contemplates delivering a protein of interest to a target vertebrate animal by providing a mobile cell to whose membrane the protein is attached by a GPI anchor. The GPI anchor. The GPI-linked protein is then delivered to other cells by intermembrane transfer.

Suitable mobile cells include red blood cells, macrophages, and fibroblasts. A protein of interest which is not natively GPI-linked may be attached to a mobile cell by expressing a non- naturally occurring precursor protein which provides a GPI attachment signal recognized by the cell.

In a preferred embodiment, the present invention relates to the intermembrane transfer of complement restriction factors from mobile cells, such as red blood cells, to discordant xenograft tissue, in vivo or in vitro, to diminish the risk of hyperacute rejection. These complement restriction factors bear natural or

"artificial" GPI anchors which facilitate intermembrane transfer.

We believe that accommodation of a xenograft results from the transfer of one or more species-specific complement inhibition factors from the recipient animal's red blood cells (RBCs) to the vascular endothelium of the xenograft. Until this transfer is completed, the xenograft must be protected from HAR by cell-free administration of exogenous inhibitors or removal of the host's antibodies. Once a sufficient number of inhibitors are transferred, the xenograft is resistant to host complement and the xenograft is accommodated. In human recipients, this threshold level is apparently reached "'-10 days past transplantation. When the host antibody levels return to normal, the graft remains protected because of the continual addition of new C-inhibitors from the host RBCs. This compensates for the loss of C-inhibitors through normal turnover of the graft's endothelial cell membranes. Others, misled by the conventional theories discussed above, have sought to achieve accommodation either through administration of complement inhibitor per se. with consequent delivery problems, or through expression of genes encoding the inhibitors directly in the endothelial cells of the xenograft itself, but reportedly with only poor expression of the inhibitor protein on the surfaces of those same cells.

Applicants have discovered that transfer of C-inhibitors can occur, in vitro, from the membranes of red blood cells to the membranes of target endothelial cells, and th-sy have evidence indicating that this phenomenon occurs in vivo as well. The present invention overcomes the aforementioned problems associated with discordant xenograft transplantation by contacting the discordant xenograft with cells bearing suitable membrane-associated complement inhibition factors functional in the intended recipient, under conditions conducive to the cell- to-cell transfer of complement resistance to the xenograft. Preferably, when such contact occurs in vivo, the cells in question are mobile cells, such as red blood cells, so as to achieve a wide distribution of complement resistance. It is relatively difficult to obtain foreign gene expression in endothelial cells, and the difficulties of obtaining complete coverage of the vascular endothelium of the transgenic animal by direct expression of the foreign C-inhibitor transgene in endothelial cells are compounded by the variations among this heterogeneous group of cells; i.e., a promoter effective in aortic endothelial cells may be mediocre in capillary endothelial cells, and vice versa.

With the indirect method described in the instant application, the C-inhibitor is expressed in a red blood cell (or other mobile cell) , and then transferred to the endothelial cells by membrane-to-membrane transfer. This mechanism will be particularly efficient in capillaries, arterioles and venuoles, where cells are crowded together. These are the very areas first affected by HAR.

In an especially preferred embodiment, a transgenic nonhuman animal, such as a pig, of a species normally discordant to humans is prepared, whose erythrocytes have been genetically engineered to express one or more human-specific complement inhibition factors. After a short period, human complement resistance is conferred on the animal's vascularized tissues and organs. These tissues and organs may then be transplanted to humans, with a substantially reduced risk of hyperacute rejection of the xenograft. Alternatively, a normal, non-transgenic donor animals' tissues and organs may be exposed, in vivo or in vitro, to exogenous red blood cells which bear on their surface recipient-specific complement inhibition factors, thereby conferring resistance to the recipient's complement system. The invention relates to methods of transferring complement resistance to a discordant xenograft, to the xenografts thereby rendered tolerable, to the use of the modified xenograft in xenotransplantation, and to assays for identifying complement inhibition factors and for determining when an organ has become sufficiently mo-dified to avoid hyperacute rejection.

In other preferred embodiments, other GPI-linked proteins are provided on mobile cells and transferred to target cells.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a flow chart illustrating the combination of the human CD59 gene, the human alpha globin 5' and 3' flanking sequences, and the human globin LCR to form LCR0.CD59.

Figure 2 similarly illustrates the construction of LCRα-DAF.

Figure 3 shows a map of the plasmid LCRαDAF, and of the linearized Scal-Kpnl fragment. Figure 4 shows a map of the plasmid LCRαCD59, and of the linearized SacII fragment.

Figure 5 sets forth the DNA and translated amino acid sequence for the MCP:DAF Fusion. The DAF sequence begins at base 826.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

The present invention relates to the inter-.embrane transfer of GPI-linked proteins. The term "intermembrane transfer", while common in the art, is potentially ambiguous. First of all, cells have several membranes, and therefore the term could be misinterpreted as including the transfer of a protein from one membrane to another within the same cell. However, in the art, and in this specification, it refers to what is more aptly called "intercellular transfer", i.e., transfer from one cell to another, whether direct or indirect. For the purpose of the appended claims, we will use "intermembrane transfer" to refer to transfer of membrane-associated proteins between cells.

Intermembrane transfer may be direct, as a result of cell membrane-to-cell membrane contact, or it may be indirect. I f indirect, the protein may move through the aqueous phase (e.g., the blood) separating the cells, or it may be carried by a lipid vesicle released by one cell and adsorbed by t.ie other. While the point is not yet fully established, and applicants do not wish to be bound by this theory, applicants believe that in their system, the intermembrane transfer is direct. The evidence is mainly circumstantial: if the transfer were indirect, the protein would be exposed to serum lipids and lipoproteins, as well as, lipases, and our success is more likely to be due to these obstacles being avoided as a result of direct intermembrane transfer, rather than that our system shed so much GPI-linked protein that enough made it to the target cells despite capture and degradation by serum molecules. However, applicants' Example VIII describes several protocols for determining the mechanism of intermembrane transfer. Target Animal

The target animal for the protein transfer method of the present invention is a vertebrate animal, i.e., a mammal, bird, reptile, fish or amphibian. Among mammals, the target animal preferably belongs to the order Primata (humans, apes and monkeys), Artiodactyla (e.g., cows, pigs, sheep, goats, horses), Rodenta (e.g., rabbits, mice, rats) or Carnivora (e.g., cats, dogs). Among birds, the target animals are preferably of the orders Anseriformes (e.g., ducks, geese, swans) or Galliformes (e.g., quails, grouse, pheasants, turkeys and chickens). Among fish, the target animal is preferably of the order Clupeiformes (e.g., sardines, shad, anchovies, whitefish, salmon and trout). The "target tissues" are any tissues or organs of the target animal which may come into contact with the aforementioned GPI- linked protein bearing carrier cells. (The terms "target tissue" and "target organ" are hereafter used interchangeably.) An animal may be considered a "target animal" even though only selected tissues and organs, already explanted, are exposed to the carrier cells. The vascular endothelial cells are preferred target tissues.

In one embodiment, the target animal is a nonhuman mammal, from which organs or tissues are to be transplanted to a human subject. The GPI-linked protein is transferred, by intermembrane transfer, from mobile cells as hereafter discussed to the intended organ or tissue transplant (collectively, the "target tissue"), which subsequently is transplanted to the subject. Most commonly, this GPI-linked protein will be a complement inhibitor.

Raising non-human mammals specifically for use as donors of organ transplants has many advantages provided that the recipient will accept the organ or tissue. First, a carefully raised non- human animal is less likely to be damaged or to carry a pathogenic virus or neoplasm than a human donor. Note that human donors are frequently deceased, with the death resulting from a cause which may make the donated organ less than ideal. Secondly, the size and age may be carefully controlled with xenotransplants compared to the random choice of human organ donors. Thirdly, potential recipients frequently cannot wait for long periods of time until a donated organ becomes available. Organs from non-human mammals are likely to be available in greater quantities, and on a more consistent basis, than cognate human organs. If a graft fails, a backup organ should also be readily available. Finally, organs and recipients are presently matched primarily for MHC compatibility. If rejection is less of a concern, more attention may be given to other considerations, such as size matching.

In accordance with the present invention, a variety of discordant animals may be used as donors for organ and tissue transplantation into humans and other mammals. The choice of animal will depend on the particular organ or tissue desired, the size and sex of the recipient. For human recipients, a pig is generally preferred due to its size, similar physiology, ease of genetic manipulation, reproduction rate and convenience. Other mammals, such as sheep, goats, cattle, etc., rray also be used. Should the recipient not be human, the choice of animal may vary but by using the same selection criteria, one skilled in the art can choose an appropriate donor animal.

The list of organs which may be transplanted is very long and is primarily limited by known surgical techniques. Certain tissues or parts of organs also may be transplanted. For the purpose of the appended claims, the term "organ" includes whole organs, parts of organs, and miscellaneous tissues. Examples include kidney, eye, heart, heart valve, skin, liver, bone marrow, intestine, blood vessels, joints or parts thereof, pancreas or portions containing the islets, lung, bronchi, brain tissue, muscle and any other vascularized tissue. However, the transfusion of blood or components thereof is not to be construed as being an organ transplantation.

Transplantation may be performed to correct an organ which is improperly functioning as a result of injury, genetic defect, disease, toxic reaction etc. The recipient may receive the transplanted organ to supplement existing tissue such as using skin tissue for treating burns, pancreatic islets for diabetes or brain tissue for treating Parkinson's disease. Alternatively, the recipient's defective organ may be completely removed and replaced with the xenograft such as in kidney, heart, liver, lung or joint transplants. For some organ transplants, it is possible for either technique to be used such as with liver transplants for treating cirrhosis and hepatitic neoplasms and infections.

Carrier Cells

"Carrier Cells" are vertebrate cells which bear a GPI-linked protein of interest in their membrane, and are capable, under the desired transfer conditions, of transferring that protein via intermembrane protein transfer to a target tissue of a vertebrate animal.

The term "mobile cells" is intended to encompass both red blood cells, which are passively circulated, and true migratory cells, which can move of their own volition. When the intermembrane protein transfer is to occur in vivo, the carrier cells will be mobile cells. One may use any mobile cell as the primary site of expression of complement inhibition factors, and rely on it to distribute these factors to other cells as it moves through the body.

The term "red blood cells" is intended to encompass both erythrocytes and reticulocytes, as both are found in the red blood cell fraction of blood. Red blood cells are circulated in the bloodstream of the animal and can reach any vascularized tissue or organ. Other cells, such as macrophages and T-cells are not only capable of such movement, but may also migrate from the bloodstream into solid tissues, and therefore are termed "migratory cells." This term also includes fibroblasts, which are capable of movement within extracellular spaces.

There are greater than 10⁹ RBC/ml of blood. This is several orders of magnitude greater than any other blood cell. Although other blood cells have higher levels of C-inhibitors on their surface, the number of RBCs greatly favors these cells as the primary vehicle of GPI-linked protein (e.g., C-inhibitor) transfer to vascular endothelial cells. For transfer of gene products, such as complement inhibitors, to tissues other than vascular endothelial cells, the preferred vehicles would be macrophages, T-cells, fibroblasts and other "migratory cells". The protein of interest may be delivered to the target, organism by mobile cells, compatible with the target animal, by providing the protein with a GPI anchor and permitting the GPI- linked protein to attach itself through the GPI anchor to the membrane of a suitable mobile cell.

Methods of Delivery. Generally

In one embodiment, genetically engineered mobile cells are administered to the target animal. The protein of interest is expressed within mobile cells, which post-translationally modify the protein, in response to a signal sequence fused to the protein of interest and expressed therewith, by attaching the desired GPI anchor. The GPI-linked protein is then exported to the membrane of the mobile cell from which it may be transferred, via membrane-to-membrane contact, to target tissues.

In a second embodiment, the target animal is a nonhuman chimeric or transgenic animal genetically engineered so that at least some of its mobile cells produce the desired GPI-linked protein and move it to their membranes. Such animals contain at least one foreign gene, called a transgene, in the genetic material of cells endogenous to the animal.

In a third embodiment, the protein of interest is produced outside the mobile cell, with a GPI anchor already attached. It is then incubated with a carrier cell under conditions conducive to the incorporation of the protein into the membrane of the cell. The cell is then incubated with the target tissue.

In a fourth embodiment, the GPI-linked protein is produced outside the target animal and administered to the target animal in such manner that it can be taken up by cells of the target animal, including possibly mobile cells thereof.

The first and second embodiments have the advantage that new GPI-linked protein continues to be produced by mobile cells within the target animal.

Delivery of GPI-Linked C-Inhibiters to Xenografts for Pre- Accommodation Thereof

In accordance with the present invention, the xenografts are "disguised" so that the recipient does not immediately recognize the transplant as foreign and initiate HAR, by contacting the organ or tissue to be transplanted (the graft) with an effective amount of mobile or other carrier cells which bear complement inhibitor factors preventing hyperacute rejection. The contacting is performed for a sufficient period of time for the xenograft to acquire sufficient inhibitors to prevent hyperacute rejection.

A xenograft may be given the complement inhibitors needed to prevent hyperacute rejection by a number of methods, which usually are specific adaptations of the generic method described in the last section. In a preferred embodiment, hyperacute rejection is avoided by engineering the donor animal to express the recipient's species-specific complement inhibition factor(s) in mobile cells, which then transfer the complement inhibition factor(s) to the cells to be grafted. Since discordant animals by definition lack this ability, the present invention overcomes this limitation by preparing transgenic animals where the transgene(s) encode the factor(s) believed necessary for preventing hyperacute rejection. Thus, one embodiment of the present invention involves a transgene which can be expressed in a primary tissue, but, whose protein product can be further distributed to secondary sites by way of mobile cells. The cells at these secondary sites do not need to synthesize the transgene product, but acquire it, through intermembrane transfer. Since red blood cells travel throughout the body this technique is expected to distribute hyperacute rejection inhibition factors widely.

In another embodiment, a tissue or organ is preaccommodated by contacting it with cells, such as red blood cells, from the proposed recipient. The red blood cells do not actually need to be from the individual recipient provided that they are at least concordant with the recipient. The closer the match of blood type, etc., the better. Since red blood cells are readily available in large quantities, the examples use this source.

The contacting step may be performed in vitro or in vivo. When contact is in vitro, use of mobile cells is not required, as adequate distribution may be obtained by assuring that all surfaces of the organ or tissue are in contact with the resistant cells often enough for effective transfer of resistance. However, use of red blood cells is still preferred. When contact is in vivo, use of mobile cells is necessary.

The organ or tissue may be either first removed from the donor before contacting or the contacting step may be performed in the donor before transplantation. In vitro contacting may be enhanced by perfusing the tissue or organ with suitable red blood cells to ensure adequate exposure throughout the tissue. In vivo treatment may be performed by transfusing suitable red blood cells to the donor to obtain donor organ-to-target RBC contact inside the donor animal.

For superior effect, the donor should be immunosuppressed so that it will not reject the transfused red blood cells. Immunosuppression may be achieved in a number of ways including removal or destruction of organs involved in the immune system such as the spleen, providing anti-lymphocyte antibody, plasmapheresis to remove antibody and/or complement and administering immunosuppressive drugs such as a cyclosporin, a steroid, thalidomide or succinylacetone.

A combination of transgenic animal and blood replacement techniques may also be used. In this situation, red blood cells from one or more transgenic animals expressing one or more different recipient-specific C-inhibitors may oe withdrawn and transfused into a recipient animal of the same species which eventually will become an organ donor. Immunosuppression is not needed, and particularly if multiple C-inhibitors are desired, this may prove easier than engineering of the desired inhibitors into a single line of transgenic animals. Where transfer of complement resistance occurs within a transgenic animal, the organs and tissues will have had sufficient contact with the engineered mobile cells to have received hyperacute rejection inhibitory factors. In other situations, one will need to allow contact to continue for a period of time for transfer of complement inhibitor to the xenograft. The period of time is dependant on the quantity of red blood cells or other mobile cells, the organ or tissue being treated and perhaps other factors as well. In preliminary experiments on certain endothelial tissues, it appears two days are not acceptable and six days provides adequate time.

In almost all non-emergency situations, it is preferable to test a sample of the organ or tissue being transplanted, or a related tissue, to determine whether or not the organ or tissue has become sufficiently accommodated. The sample is incubated with the serum from the prospective organ or tissue recipient and one measures the presence of any reaction. If complement mediated lysis or cell death occurs, accommodation obviously is not complete. Complement fixation tests are also useful indicators. Lesser reactions of antibody binding to sample may be observed by immunohistochemical staining of the sample with a labeled anti- (human immunoglobulin) antibody. A number of other suitable immunoassays are available and may be used for determining whether the organ or tissue has become sufficiently accommodated to the recipient before transplantation.

Proteins of Interest. Generally

The protein to be transferred may be a naturally occurring GPI-linked protein, or a functional fragment or homologue of such a protein. Naturally GPI-linked proteins are listed in Table 5.

The protein of interest may also be one whose closest naturally occurring cognate is not GPI-linked, but which has been modified so that it will be processed to acquire a GPI-anchor.

Such modifications are discussed in the next section. All proteins expressed on the cell surface, or excreted from the cell, also must have an amino terminal leader sequence. This is a hydrophobic sequence which directs the nascent polypeptide chain into the endoplasmic reticulum. Any protein which we engineer for GPI linkage must also have such a sequence if it is not already present. For example intracellular proteins would require addition of a signal sequence, but secreted proteins such as growth hormone already possess a signal sequence.

The amino acid sequence of a protein of interest may be modified, e.g., by site-specific or semirandom mutagenesis of the corresponding gene to obtain a mutant protein with a "substantially" corresponding" amino acid sequence.

In determining whether sequences should be deemed to "substantially correspond", one should consider the following issues: the degree of sequence similarity when the sequences are aligned for best fit according to standard algorithms, the similarity in the connectivity patterns of any crosslinks (e.g., disulfide bonds) , the degree to which the proteins have similar three-dimensional structures, as indicated by, e.g., X-ray diffraction analysis or NMR, and the degree to which the sequenced proteins have similar biological activity. In this context, it should be noted that among the serine protease inhibitors, there are families of proteins recognized to be homologous in which there are pairs of members with as little as 30% sequence homology.

Preferably, the sequence of the mature protein is at least 50% identical, more preferably at least 80% identical, with the sequence of its naturally occurring cognate.

The 3D-structure can be used to identify interior and surface residues; generally speaking, proteins mutated at surface residues (other than the receptor binding site) are more likely to remain functional. However, Creighton and Chothia, Nature, 339:14 (1989) discuss the toleration of mutations at buried residues. The structure may also be used to determine flexible surface "loops" and interdomain boundaries; proteins are more tolerant of deletions and insertions in such regions. In general, segments of the protein which are more difficult to resolve by NMR are likely to be segments which are freer to move, and hence more tolerant of mutation.

Insertions and deletions are preferably at the amino or carboxy termini, at loops (sequences joining helices to helices, helices to sheets, and sheets to sheets, and at interdomain boundaries) . At termini, internal insertions or deletions are preferably of no more than three consecutive amino acids, more preferably only of a single amino acid.

The mutations are preferably substitutions. In terms of the kinds of substitutions which may be made, one may look to analyses of the frequencies of amino acid changes between homologouys proteins of different organisms. Based on such analyses, we define conservative substitutions as exchanges within the groups set forth below:

I small aliphatic, nonpolar or slightly polar residues - Ala, Ser, Thr (Pro, Gly)

II negatively charged residues and their amides - Asn Asp Glu Gin

III positively charged residues - His Arg Lys IV large aliphatic nonpolar residues - Met Leu He Val (Cys) V large aromatic residues - Phe Tyr Trp Three residues are parenthesized because of their special roles in protein architecture. Gly is the only residue without a side chain and therefore imparts flexibility to the chain. Pro has an unusual geometry which tightly constrain--: the chain. Cys can participate in disulfide bonds which hold proteins into a particular folding; the four cysteines of bGH are highly conserved. Some authorities would merge I and II above. Note also that Tyr, because of its hydrogen bonding potential, has some kinship with Ser, Thr, etc. Acceptable substitutions also include substitutions already known, as a result of their appearance in proteins similar in biological activity and sequence with a protein of interest, to be likely to be tolerated. For example, if the protein of interest were to have superoxide dismutase activity, instead of using a protein identical with a naturally occurring SOD, it could be a chimera of several naturally occurring SODs.

Mutations which could prevent attachment of the GPI anchor, as well as mutations which reduce the desired activity, should be avoided. The active site residues may be determined, if not already known, by methodically testing fragments for activity, as was done for C4bp by Chung and Reid (1985) , or by systematic testing of mutants.

The nucleotide sequence which encodes the protein of interest may be, but need not be, identical to the naturally occurring sequence. "Silent" mutations may be made to improve transcriptional or translational efficiency, introduce or eliminate restriction sites, or reduce the probability of recombination. In addition, mutations which result in a change in the encoded amino acid sequence may be made as previously discussed.

Complement Inhibitors

For xenograft transplantation, the proteins of greatest interest are "complement inhibitors". In all species with a complement lysis system there are a variety of molecules which normally function to inhibit complement, as shown in Table 1. These complement inhibitors appear necessary to limit autologous cell lysis within the host. Some of the complement inhibitors are species specific. That is, a human complement inhibitor such as decay accelerating factor (DAF) will inhibit human complement, and complement of some other closely related primate species, but is ineffective against complement of more distant species such as the mouse or pig. (Atkinson, personal communication) . Indeed, this form of species-specific complement inhibition is thought to be one of the major contributory factors in determining whether or not a xenograft is concordant or discordant. At least two of the human complement inhibitors, DAF, and CD59, and possibly a third, homologous restriction factor (HRF) , are of particular interest because they are all thought to function solely to protect host cells from complement-mediated cell lysis. (Kinoshita, 1991) . These molecules inhibit complement by interfering with C3 and C5 convertase (DAF) or by preventing formation of the terminal membrane attack complex (CD59 and HRF) . However, the present invention is not limited to DAF, CD59 and HRF.

All of these molecules are found on red blood cells and endothelial cells and are anchored to the cellular surface through a glycosylphosphatidylinosital (GPI) linkage (See Lisanti, et al.). At the protein level all three complement- inhibitors are broadly distributed, as can be seen in Table l, and generally are found in any region which is in contact with complement activity.

Complementary DNAs corresponding to CR1, CR2, DAF, MCP, C4bp and H have been cloned and sequenced, as a result of which it is known that these proteins are composed mainly of a tandemly repeated motif of about 60-70 amino acids in length (see Hourcade, et al.), called "short consensus repeats" (SCRs) . Several pathogens carry genes encoding structurally related proteins. The 35k vaccinia protein has a signal sequence followed by four SCRs, and is 38% homologous at the amino acid level to the amino terminal half of C4 bp. The glycoprotein C-1 of HSV lacks a typical SCR structure, but nonetheless contains short stretches substantially homologous to various C-inhibitors and exhibits DAF-like activity. Anti -Oxidants During organ removal, perfusion, storage and reperfusion prior to transplant the vascular endothelium is exposed to dramatic environmental and metabolic changes. These changes lead to the production of reactive oxygen intermediates which cause some endothelial cell activation and tissue damage. This reperfusion injury renders the transplanted organ susceptible to further injury through neutrophil infiltration after transplant, particularly in the case of xenotransplants. One method of controlling this injury is to express super oxide dismutase (SOD) or catalase on the surface of the endothelial calls, with a GPI anchor, since both of these proteins scavenge oxygen radicals (Erzurum Serpil C, Lermarchand Patricia, Rosenfeld Melissa A., Yoo Jee-Hong, and Crystal Ronald G. (1993) , "Protection of human endothelial cells from oxidant injury by adenovirus-mediated transfer of the human catalase cDNA, " Nucleic Acids Research 21:1607-1612) . The preferred method of accomplishing this is likely to be through GPI transfer, since the long term constitutive expression of these proteins is likely to be detrimental to the health of the transplant recipient and efficient expression of these proteins throughout the vascular endothelial surface would be difficult to achieve with viral vector systems.

For catalase, an intracellular enzyme, it would be necessary to add a hydrophobic leader sequence, to target expression to the cell surface, as well as a GPI anchor sequence for the addition of the GPI tail. This protein could then be expressed in erythrocytes, and subsequently transferred to the vascular cell surface. Adhesion Molecules One of the most perplexing characteristics of GPI-linked proteins is that several of them are thought to function as adhesion molecules. For example N-CAM, a neural cell adhesion molecule, exists in both GPI-linked and transmembrane forms due to alternative processing of RNA. When neurons are cultured on top of non-neural cells, both forms of N-CAM displayed by the non-neuronal cells can promote neurite outgrowth. This response by the neurons is mediated by transmembrane N-CAM in the neurons, which stimulates a classical secondary message pathway. In this instance, it appears that the GPI-linked N-CAM may function to provide recognition or positional information whereas the transmembrane form may be required for cells to actively respond to that information. Consistent with this interpretation is the finding that substrate cells (cells that the neurons crawl over) and target cells (cells to which the neurons branch and establish synapses with) often express GPI-linked N-CAM. For example Schwann cells, over which neurons grow, and skeletal muscle, to which neurons establish synapses, preferentially express GPI- linked N-CAM (Schwann cells) or switch to GPI-linked form

(skeletal muscle) prior to synapse formation (Walsh F.S. and

Doherty P. (1991) , "Glycosylphosphatidylinositol anchored recognition molecules that function in axonal fasciculation, growth and guidance in the nervous system, " Cell Biology International Reports, 15.:1151-1166) .

As we have shown, the GPI linkage of a protein to the cell surface is in some sense tenuous. It may therefore be possible to take advantage of this condition to inhibit cell adhesion through the expression of GPI-linked adhesion molecules. Such molecules on the surface of a cell might simply release from the membrane when bound by an adhesion molecule on another cell. In this way GPI-linked adhesion proteins might function as a "slippery rock". For example, adhesion of lymphocytes to endothelial cells at the site of inflammation is mediated predominantly by VCAM-1. This adhesion molecule is transmembrane protein expressed on endothelial cells within 2 hours of treatment with IL-2 or TNF-alpha and this expression is maintained for at least 72 hours. The expression of this molecule permits lymphocyte adhesion and infiltration. It may be possible to delay or block lymphocyte infiltration of the endothelium by αisplaying a GPI-linked form of VCAM-l. This GPI- linked isoform could be expressed in erythrocytes of transgenic mice or pigs, where it would transferred to the vascular endothelial cells. Organs from such pigs could be transplanted to humans (assuming they also display human complement regulatory proteins) where they might have enhanced resistance to lymphocyte infiltration. Similar inhibition of endothelial binding to polymorphonuclear leukocytes could be achieved using a GPI-linked isoform of ELAM-1. UP A Receptor

An additional utility of GPI transfer which is applicable to a gene therapy approach is the introduction of hematopoietic stem cells and overexpression of the normally GPI-linked urokinase plasminogen activator receptor. This receptor and its ligand, urokinase plasminogen activator are involved in regulating the proenzyme plasminogen and are thus involved in many biological processes which involve tissue remodeling, including thrombosis. (McNeill Helen and Jensen Pamela J. (1990) , "A high affinity receptor for urokinase plasminogen activator on human keratinocytes: characterization and potential modulation during migration," Cell Regulation. 1:843-852). The overexpression of this protein, and its transfer to endothelial cells, might then be usefully applied to controlling thrombotic sites which accumulate in the extremities of patients suffering from phlebitis. While this approach could not restore blood flow to occluded vessels, since it requires blood flow for transfer, it would be expected to improve the flow through partially occluded vessels. Additionally this approach could represent a one time treatment that provides long term relief from recurrence of the disease. Similar gene therapy approaches to the treatment of phlebitis could be achieved using thrombomodulin and tissue plasminogen activator receptor, both intrinsic membrane proteins which would require modifications to incorporate GPI anchors (Nachman Ralph L. (1992) , " Thrombosis and atherogenesis: Molecular connections, " Blood.29.:1897-1906) .

GPI Attachment Signal Sequence

It is within the scope of the invention to incorporate a known GPI-attachment signal, naturally occurring or modified, into a protein (e.g., CR1, CR2, MCP) which is not normally GPI- linked in order to obtain the benefits of a GPI anchor. For studies of GPI-linked proteins, see Cross, et al. , 1990; Lisanti, et al., 1990; Ferguson, 1991; Ferguson and Williams, 1988. In order to cause a cell to attach a GPI anchor to a protein which is not normally GPI-linked, it is necessary to modify its sequence to provide a suitable signal sequence at the carboxy terminus of the protein (or a biologically active fragment of the protein of interest) .

Attachment of the GPI moiety is a post-translational modification which presumably occurs in the endoplasmic reticulum. As a result of GPI addition a hydrophobic sequence from the carboxy terminum of the nascent protein is removed. This hydrophobic sequence is a necessary but not fully sufficient portion of the signal for GPI addition. For example, the carboxy terminal 17 amino acid hydrophobic domain of DAF when deleted disrupts the addition of the GPI residue yet when this same sequence is fused to a heterologous protein, human growth hormone, it does not create a GPI anchored fusion protein. Caras, et al., "Analysis of the signal for attachment of a glycophospholipid membrane anchor", Journal of Cell Biology, 108:1387-1396, (1989).

In known GPI-linked proteins the last amino acid after cleavage and attachment is typically Gly, Ser, Ala, Asp, Asn, and (perhaps less preferably) Cys. (There is reason to believe , see Ferguson (1991) , that Glu, Gin, Pro, Trp, Leu, Val, Phe, Thr, Met and Tyr cannot serve in this role.) These are all small amino acids having at most a two carbon side chain. The small attachment amino acid is normally followed by a second such amino acid, which may be the same or different, which is cleaved off. (Together, these two amino acids are referred to herein as the cleavage/attachment site, or CAS, doublet.) This doublet, in conjunction with a hydrophobic carboxy terminus, thus seem to form the minimal sequence necessary for GPI addition.

There are, however, many subtle variations which affect the efficiency of GPI addition. The position of the CAS doublet relative to the hydrophobic domain is significant. A spacer of 5-20 amino acids, more preferably, 7-14, stil more preferably 8- 12 amino acids, is desirable. The sequence of this spacer does not appear to substantially affect the efficiency of GPI addition, though Ferguson (1991) notes that polar amino acids are common and this teaching may be followed. CAS doublets of Ser- Ser, Ser-Gly, Ser-Ala appear to be most effective. Finally the length of the hydrophobic domain can also affect the efficiency of GPI addition. The hydrophobic domain is thought to function the endoplasmic reticulum to slow or temporarily stop the transit of the nascent protein through the membrane of the ER so that attachment of the GPI moiety can occur. Using homopolymeric leucine as a hydrophobic domain, a tail of 14 residues was maximally efficient, an 11 residue tail was only slightly less efficient, but an 8 residue one was not sufficient for GPI attachment. See Coyne, et al., "Construction of synthetic signals for glycosyl-phosphatidylenositol anchor attachment", Journal of Biological Chemistry, 268:6689-6693 (1993). This indicates that even with constant hydrophobicity, the length of the hydrophobic domain is important. Analysis of hydrophobicity of naturally occurring GPI anchor signals (see TABLE 4) would indicate that the hydrophobic domain should comprise a minimum of 8 amino acids, and more preferably a longer sequence, with a minimum average hydrophobicity of -1.3 (typically higher for shorter sequences) . Preferably, the hydrophobic domain does not contain any charged amino acids, as these are the most hydrophilic.

It is possible that only a portion of the signal sequences recited in TABLE 4 are required for GPI attachment. Due to the degeneracy of the hydrophobic region, and the many possible combinations of small amino acid attachment sites, it is expected that conservative substitutions will be tolerated.

Addition of GPI linkages occurs broadly in most if not all types of cells. GPI addition is thought to be produced through a series of common, ubiquitous cell functions. Therefore we expect no species or cellular specificity to substantially affect GPI addition. One particular GPI signal sequence, should be effective in all cell types in all species. Nonetheless, the efficiency of GPI-linked protein synthesis depands not only on the signal sequence but also on the presence of other enzymes required for GPI synthesis (see pages 17-22 of Cross review) . For example, Amthauer, et al. (1993) have shown that the immunoglobin heavy chain binding protein (BiP) binds to metabolic intermediates involved in GPI synthesis. Additionally, these authors have provided evidence that the transamidase (not yet identified) is present in the endoplasmic reticulum.

In a preferred embodiment, the protein of interest is expressed as a fusion of at least that portion required for biological activity with the decay accelerating factor (DAF) - derived attachment signal (DAF29)

SGTTRLLSGHTCFTLTGLLGTLVTMGLLT (SEQ ID NO: 13) G wherein the C-terminal hydrophobic domain is underlined and the cleavage/attachment site is bolded as indicated, the first amino acid of the CAS doublet may be S or G. Lisanti, et al., J. Cell.

Sci., 99:637-640 (1991) reported that an hGH-DAF29 (Ser, Gly) fusion was expressed, GPI-linked, and properly relocated by transfected MDCK cells.

A large number of candidate attachment signals may be simultaneously produced and screened for effectiveness by an adaption of the method of Ladner, USP 5,223,409. A gene encoding a normally secreted and readily assayable protein, such as a growth hormone, is fused to "variegated" (semirandom) DNA encoding a large family of possible attachment signals. Some codon positions will encode the same amino acid in each version of the gene, and others will encode different amino acids depending on which DNA molecule is examined. Ladner refers to the corresponding amino acid positions as "constant" and "variable" residues, respectively. Constant and variable residues may be interspersed as desired. It is also possible for the variation to affect the number of residues. For example, this DNA may have the form

^α2 08-12 7s-₃o

Where the subscript indicates the number of codons of each of classes a, β , and γ, a codons encode the amino acids of the GAS doublet, β codons encode those of the spacer, and γ codons encode those of the hydrophobic tail.

Theoretically, the gene may be synthesized so that the "substitution set" at a given variable residue includes all twenty genetically encodable amino acids. More typically, the variation will be resticted, e.g, one or more of the α codons might each be randomized independently to encode any of Gly, Ser, Ala, Asp, Asn and Cys and one or more of the y codons might each be randomized independently to encode any of the more hydrophobic amino acids, e.g., He, Val, Leu, Phe, Cys, Met, Ala, Gly, Thr, Trp, Ser, and Tyr. (Other substitution sets are possible.) The β codons are of less importance. They could be selected to encode in all of the proteins, a single sequence modelled after the space of a known GPI-linked protein precursor, or the variation in the β codons could be in the number of codons rather than in the amino acid encoded, or one or more of the β codons could each be selected independent from codons encoding amino acids found frequently in the naturally occurring species.

The fused gene is expressed in cultivatable cells which are capable of adding a GPI anchor to a protein. The cells are first subjected to a negative screen. This looks (e.g., with a labeled antibody) for the parental protein (e.g., growtr. hormone) in the supernate. If found, then the protein is being secreted, rather than equipped with a GPI anchor and held in the membrane, and the candidate signal expressed in those cells is thus shown to be ineffectual.

The "non-secreting" cells, which passed the first screen, are then treated with phospholipase D, or some other enzyme that cleaves off the GPI-linkage. The supernate is then screened for the presence of the GH. If present, it implies that the cells produced GPI-linked GH, but that the phospholipase treatment removed the GPI, allowing the GH to escape nto the culture medium.

GH is a preferred "assay target" since GPI-linked forms of the GH have been produced by expressing a GH/DAF29 attachment signal chimera in cells. Therefore, it is known that the attachment of such a signal does not interfere w th the detection of GH.

Transgenic Animals

In a transgenic animal, the transgene is contained in eventually all of the animal's cells, including germ cells, such that it can be transmitted to the animal's offspring. In a chimeric animal, at least some cells endogenous to the animal bear the transgene, but germ line transmission is not necessarily possible. The term "genetically engineered animals" includes both transgenic and chimeric animals. However, since germ live transmission of transgenes is usually advantageous, production of trangenic animals is usually preferred. The discussion below therefore refers to "transgenic" animals, however, such references apply, mutatis mutandis, to other chimeric animals. A number of techniques may be used to introduce the transgene into an animal's genetic material, including, but not limited to, retroviral infection, electroporation, microinjection of the transgene into pronuclei of fertilized eggs, and manipulation of embryonic stem cells (U.S. Patent No. 4,873,191 by Wagner and Hoppe; Palmiter and Brinster, 1986, Ann. Rev. Genet. 20.:465-499; French Patent Application 2593827 published August 7, 1987) .

The overwhelming majority of transgenic animals (and all transgenic livestock) produced to date have resulted from pronuclear microinjection of DNA. The technique involves the delivery of DNA in solution to one of the pronuclei of a one cell fertilized ova. The pronuclei of the fertilized ova may be observed at 200X under Nomarski optics. In the case of ova from pigs, cattle and sheep, the ova should first La centrifuged to sediment cytoplasmic lipids which make visualization of the pronuclei difficult. (Swanson et al., 1992.)

Holding and microinjection pipettes utilized in the microinjection process are manufactured from (1mm O.D., 0.78 mm I.D.) borosilicate glass capillaries. The capillaries are heated in a microforge and pulled to make the microinjection or holding ends. After microinjection, the surviving ova are transferred back into a recipient female in the appropriate stage of estrus.

There are alternatives to pronuclear microinjection, for instance, the infection of preimplantation embryos (one cell to eight cell) with genetically engineered retroviruses. See Jaenisch, R. (1976) , "Germ line integration and Mendelian transmission of the exogenous Moloney leukemia virus"; Proc. Natl. Acad. Sci. USA 73:1260-1264. In this case, the zona pellucida is removed and the embryos are o-cultured with fibroblasts during the infection process. Infected embryos may then be placed back in a zona pellucida and transferred to an appropriate recipient. Another approach to producing transgenic animals involves electroporation. In this technique, the one cell ova is placed in a electroporation chamber with a solution of DNA. A pulsating electric field is generated through the chamber to drive the DNA into the ova.

Embryonic stem (ES) cell lines are derived from the cells of the inner cell mass of mammalian (e.g., mouse and hamster) blastocysts. ES cells are maintained in the stem cell state by growth on a feeder layer, e.g., of primary embryonic fibroblasts or of the embryonic fibroblastic cell line STO. The ES cells may be genetically modified by any technique suitable to in vitro mammalian cell culture, and then injected into the blastocyst. They then differentiate and colonize most if not all tissues of the animal, including the germline. See Dυetschman, "Gene Targeting in Embryonic Stem Cells", Chap. 4, pp. 89-100, of First and Haseltine, Transgenic .Animals (Butterworth-Heinemann: 1988) .

Transgenic animals may carry the transgene in all their cells or may be genetically mosaic. Although a number of studies have involved transgenic mice, other species of transgenic animals have also been produced, such as rabbits, sheep and pigs

(Hammer et al., 1985, Nature 315:680-683) chickens (Salter et al., 1987, Virology 117:236-240), rats (Mullins, et al, 1990,

Nature, 144.:541-544) , goats (Denman, et al., 1991),

Bio/Technology, 9_:839-843), and cows (Krimpenfort, et al. , 1991, Bio/Technology, 9_:844-847).

Transgenic pigs may be produced by adaptation of the methods and materials described in Swanson, et al. (1992), with a suitably prepared insert encoding the protein of interest, replacing the alpha and/or beta globin gene(s).

Accommodation Assay

One aspect of the present invention is an in vitro assay which mimics the accommodation process. In this assay, endothelial cells are co-cultivated with a 1000 fold excess of human red blood cells to mimic the ratios present during an organ transplant. To avoid lysis of the red blood cells, the cells are washed and fresh red blood cells added every three days. After a period of cultivation, the endothelial cells are tested for complement-resistance by the addition of human serum. This assay has demonstrated three major results. First, over the course of six days, sufficient complement-inhibitory functions are transferred from the human red blood cells to the endothelium to provide nearly complete complement protection. Secondly, the complement-inhibitory functions are species specific. That is, co-cultivation with human red blood cells protects the endothelial cells from human serum, but not porcine or murine serum. Finally, the assay demonstrates that at least two known human complement-inhibitors, DAF and CD-59 are transferred after six days of co-cultivation to the membrane of the endothelial cells.

Complement-inhibitory functions apparently can be transferred from human cells to heterologous cells merely by co- cultivation. Furthermore the acquisition of species specific complement-resistance preferably includes intermembrane transfer of human DAF and CD-59 from the red blood cells to the endothelial cells. The time required to establish complement- resistance in vitro is similar to the time required for xenograft accommodation. Thus this assay mimics the naturally occurring process.

Production of GPI-linked Proteins (e.g., C-Inhibitors) in Mobile Cells

Preferred promoters for expression in various mobile cells, whether or not those cells are in transgenic animals, are set forth in Table 3. When the GPI-linked Protein (e.g., C- inhibitor) is to be expressed in a red blood cell (prior to its loss of its nucleus during maturation) , the transgene should include a promoter which is active in red blood cells of the animal which is to express the transgene. Preferably, the promoter then is one which is expressed primarily in erythroid cells. An example of such a promoter is a globin promoter, such as the human alpha, beta, delta, epsilon or zeta promoters, or their counterparts in other species. For the sequence of the alpha globin pr-o oter, see Liebhaber, et al., Proc. Nat. Acad. Sci. USA, 77:7054-56 (1980) . If the gene is to be expressed in a transgenic pig RBC, a pig globin promoter is especially preferred. However, use of an endogenous promoter is not required. Nor is it required that the promoter employed having a naturally occurring sequence. The sequence may be mutated to enhance the strength or tissue specificity of the promoter. Hybrid promoters may also be constructed, e.g., one which comprises a regulatory element of the globin promoter which confers erythroid cell specificity, with another promoter that has a higher level of transcriptional activity.

Besides globin promoters, other promoters functional in red blood cells may also be used to drive expression of the GPI- linked protein (e.g., C-inhibitor) in red blood cells, though their transcriptional efficiency and specificity may be inferior to that of the globin promoters.

Besides the globin promoters, other promoters suitable for expression of GPI-linked proteins (e.g., C-inhibitors) in RBCs include the promoters of genes encoding protein components of the RBC cytoskeleton. These include alpha-spectrin [Sahr, et al., "The Complete cDNA and Polypeptide Sequences of Human Erythroid Alpha-Spectrin , " J. Biol. Chem., 265:4434-43 (1990)]; beta- spectrin [Winkelmann, et al., "Molecular Cloning of the cDNA for Human Erythrocyte Beta-Spectrin, : Blood, 72:328-34 (1988)]; ankyrin [Lambert, et al., "cDNA Sequence for Human Erythrocyte Ankyrin, " Proc. Nat. Acad. Sci. USA, 87:1730-34 (1990)]; and the human erythrocyte anion exchange protein [Lux, et al., "Cloning and Characterization of Band 3, the Human Erythrocyte Anio -

Exchange Protein (AE1) , " Proc. Nat. Acad. Sci. USA, 86:9089-93

(1989)]. The CR1 promoter is also of interest, as CR1 is a C- inhibitor expressed only in RBCs. CR1 is a non-GPI-linked protein. Other possibilities include promoters of enzymes manufactured in red blood cells, such as superoxide dismutase and carbonic anhydrase.

For expression in macrophages, a preferred promoter is chick lysozyme. The CD2 and TCR promoters are favored for expression in T-cells. And for fibroblasts, transcription may be controlled with the promoter of the interferon beta (fibroblast interferon) gene or other genes especially active in fibroblasts.

The foregoing examples of promoters are not intended to be limiting. The basic requirement is that, at the appropriate time, the gene be expressed at sufficiently high levels so that an adequate supply of GPI-linked proteins (e.g., C-inhibitors) is directed to the surface of the mobile cells and transferred to the target tissue, e.g., the vascular endothelium of the xenograft. However, it is preferable, when the production is within a transgenic animal, that the expression be specific to the carrier mobile cells of choice so as to limit possible disruption of the life processes of the transgenic animal.

An advantage of expressing the C-inhibitor in the red blood cell (or other mobile cell) , but essentially not in the endothelial cell, is that the resulting accommodated organ or tissue will not express any foreign gene. Concerns have been voiced regarding the transplantation of organs and tissues which express foreign genes. Other regulatory elements, both upstream and downstream of a structural gene, or within an intron of a structural gene, may also be incorporated into a transgene. Preferably, for expression in RBCs, the transgene comprises a dominant activator (locus control region, LCR) sequence as described by Grosveld, WO 89/01517. There are several DNasel super hypersensitive sites, or LCRs, located about 50 kb 5' of the human adult beta globin gene and 5 kb 5' of the human epsilon globin gene. These sites are identified by controlled DNase I digestion as described in Grosveld, WO 89/01517. Grosveld describes construction of a beta globin "mini-locus" vector bearing some or all of the beta globin LCRs. EXAMPLES

REFERENCE EXAMPLE A: Protein Localization: Is it In the Membrane? A standard assay to determine if a protein is displayed on the surface of a cell would be to label the intact, non- permeabilized cells with an antibody specific to the protein of interest. The protein specific antibody is then detected by a second fluorescence conjugated antibody specific to the first antibody. Alternatively the primary antibody, the one specific to the protein of interest, may be directly conjugated to a fluorochrome such as FITC or Rhodamine. The cells with the antibody bound can then be visualized by fluorescence microscopy, or by fluorescence activated cell sorting. Fluorescence microscopy is the method used by Moran, et al., (1991) and fluorescence cell sorting is the method used by Coyne, et al., (1992) . Since the size of immunoglobin molecules prohibits their entry into intact cells, the protein can only be detected by the antibody if it is displayed on the outer cell surface. REFERENCE EXAMPLE B: Detecting GPI Anchors on Proteins

There are at least three methods for determining if the protein of interest is GPI-linked. No single method is considered definitive, but a positive result using all three methods currently constitutes proof of GPI linkage. The first method takes advantage of the sensitivity of the GPI-linkage to hydrolysis by Phospholipase C (PIPLC) . Cells expressing the protein of interest are incubated in a small volume of phosphate buffered saline with 2% heat inactivated fetal bovine serum and 4μg/ml PIPLC at 37° for 30-60 minutes. After incubation the cells are removed by centrifugation and the supernatant is assayed for the presence of the protein of interest, usually by an ELISA assay. This method was used by Moran and Cara (1991) and by Caras, Weddell and Williams (1989) .

The second method takes advantage of the unusual structure of the GPI linkage to specifically label GPI anchored protein. In general proteins do not contain ethanolamine, however this compound is a critical component of the GPI linkage. Cells can be metabolically labeled using tritiated ethanolamine and this radioactive tracer will be incorporated into the GPI anchor. Once labeled, the protein can be extracted from the membrane using detergents, or released from the surface by PIPLC. The resulting solubilized protein is then immunoprecipitated with an antibody specific to the protein of interest and then analyzed by polyacrylamide gel electrophoresis. Generally, this procedure is done in parallel with a standard protein labeling protocol using ³⁵S labeled methionine to place a radioactive tracer in the polypeptide. The methionine labeled protein then acts as a control for the immunoprecipitation. This method was used by Caras, Weddell and Williams (1989) .

A third method of demonstrating GPI linkage takes advantage of the ability of GPI-linked proteins to spontaneously integrate into cellular membranes in an active state. For this method the protein of interest with a GPI anchor is purified. The precise method of purification will depend on the particular protein, but usually consists of detergent solubilization and purification on ion-exchange and or immuno-affinity columns. Occasionally crude extracts made with organic solvents can be used. The purified protein is then incubated with a cell that does not express the protein of interest. Because of the GPI anchor, the protein will integrate into the membrane of the heterologous cell in a biochemically active state. The activity of the protein is then measured by some suitable assay. For example, Medof, et al.,

(1984) purified DAF from human erythrocytes using a combination of butanol extraction, DEAE-Sephacel and high pressure liquid chromatography. The purified protein was then added to sheep erythrocytes where it incorporated and provideα protection from human complement. Similar experiments using porcine endothelial cells were performed by Dalmasso, et al., (1991) and these experiments also demonstrated that DAF provides protection from human complement.

In general the first two methods are commonly used because they are relatively easy, and when done together provide essentially definitive proof that the protein of interest is GPI anchored. The last method is seldom done since it usually requires rigorous purification of the protein (if the protein is not highly purified this method can be criticized on the basis that a contaminant is responsible for the observed activity) . Although rarely done, this assay provides especially in conjunction with the other two, an extremely high level of assurance of the presence of a GPI linkage.

To demonstrate that a GPI-linked protein is capable of intercellular transfer, one may co-incubate a cell expressing the protein with a heterologous cell that does not express the protein. Assays for GPI anchor are available in kit form from Oxford GlycoSystems Cat #K-200, for "IDENTIFICATION of PIRLC sensitive proteins". EXAMPLE I: IN VITRO ASSAY FOR ACCOMMODATION CELL CULTURE Bovine aortic endothelial cells (#CCL 209 CPAE) were obtained from the American Type Culture Collection. After thawing, cells were diluted to 10 ml in RPMI 1640 supplemented with 10% calf serum, 300 mg/ml L-glutamine and 50 μg/ml gentamicin before plating in a 75 mm tissue culture flask. Cells were maintained in a 37°C, 5% C0₂ humidified incubator. After six days growth, cells were split into two 75 mm flasks following trypsinization. Type A+ blood was collected from a healthy adult male in a collection tube (VenoJect, Terumo Medical Elkton MD.) containing heparin using a PrecisionGlide Vacutainer (Becton, Dickinson and Company Rutherford NJ.) needle and holder. Blood was transferred to a 15 ml Falcon 2097 conical centrifugation tube (Becton, Dickinson and Company Rutherford NJ.) and centrifuged at approximately 21.00 rpm in an IEC model HN-SII (Damon/IEC Needham MA) table top centrifuge. The serum fraction was removed and stored at -80°C in 100 μl aliquots. The red blood cells were washed three times in phosphate buffered saline (PBS) by centrifugation to remove serum and serum components such as complement. The white cell enriched buffy layer was also removed. Following washing, the red blood cells were resuspended in supplemented RPMI 1640 and counted using a hemacytometer. Approximately 1X10⁵ red blood cells were suspended in 2 ml supplemented RPMI 1640 and were cultured in 10 wells of a 24 well tissue culture plate. Supernatant was removed daily and spectroscopically assayed at 415 nm for hemoglobin. Very little change in optical density readings occurred until after 96 hours of culturing. Blood was collected by jugular venipuncture from adult pigs and heparinized immediately. Approximately 24 hours after collection, the pig blood was washed as previously described for human blood and serum saved as previously stated. Mouse blood was collected from anesthetized mice by ocular collection into a 1.5 ml tube containing 20 μl heparin. Mouse blood was washed in a similar manner to that previously described. Mouse serum was obtained from Oncogene Science (Manhasset NY.) and stored as previously described. COMPLEMENT ASSAY The initial experiment involved the co-culture of human red blood cells with bovine endothelial cells treated with mitomycin C (0.02 μg/ml overnight incubation). Subsequent experiments involved untreated bovine endothelial cells. The culture experiments were performed in either two chamber tissue culture slides (#177380 Nunc Naperville IL) or 24 well plates. The tissue culture chambers were pre-coated at room temperature for 10 minutes with 1% gelatin in a laminar flow tissue culture hood. The gelatin solution was aspirated off and the slides were allowed to air dry before use. Approximately 1X10⁴ bovine endothelial cells were plated in each chamber or well and incubated for 24 hours in the supplemented RPMI 1640 to become established. Approximately a 1000 fold excess of freshly collected and prepared human red blood cells were added to each chamber and the cells were co-cultured. Every 72 hours the red blood cells were aspirated and the endothelial cells were rinsed free of remaining red blood cells with phosphate buffered saline prior to the addition of freshly collected and prepared red blood cells.

To assess sensitivity of endothelial cells to complement, the red blood cells were aspirated from a chamber/well and the endothelial cells were rinsed free of remaining red blood cells by three washings with phosphate buffered saline. Chambers/wells containing endothelial cells not cultured with red blood cells were washed in a similar manner. Both chambers/wells of cells were subsequently cultured in supplemented RPMI 1640 containing 10% normal human serum for four hours.

ENDOTHELIAL CELL ACQUISITION OF COMPLEMENT RESISTANCE Endothelial cell acquisition of complement resistance was achieved by the co-culture of human red blood cells with bovine aortic endothelial cells (ATCC # CCL-209 CPAE) . After a co- culture incubation period of one, three, six, seven, eight, and nine days of culture in a chamber of endothelial cells with and without human red blood cells, the effect of human serum on the bovine endothelial cells was measured. Cells were observed under phase contrast microscopy for visual signs of cell damage which may have been mediated by complement components in the human serum. On days one and three there was a dramatic decrease in the number of endothelial cells, whether or not they had been cultured in the presence of human red blood cells, after exposure to human serum. Virtually all of the cells remaining in both groups were balled up and rounded in contrast to the normal adherent morphology. These observations suggested that complement components in the human sera had attacked the cells resulting in their abnormal morphology and lysis. The results following six days of co-culture with human red blood cells produced dramatically different results. After incubation with human serum as previously described, the control group appeared similar to the cells examined on day three. There was an overall large decrease in the number of intact cells, and the remaining cells were rounded and non-adherent. In contrast endothelial cells cultured with human red blood cells appeared normal after incubation with human serum with little or no anomalies observed. Similar results were observed at days seven, eight and nine. This indicates that at days one and three, a sub-optimal level of complement regulatory proteins had transferred from the red blood cells to endothelial cells. However, by day six, enough regulatory proteins (probably at least DAF and CD-59) had incorporated into the endothelial cell membranes to protect them from human complement. An initial concern was that the transfer of GPI-linked proteins may be a very slow and inefficient process. If the endothelial cells grew at a rate faster than proteins are transferred then they may not incorporate enough human DAF and/or human CD-59 to be protected from human complement. Therefore, the endothelial cells were treated with mitomycin C which acts as a mitotic arrestor by crosslinking DNA.

EXAMPLE II. IN VITRO ASSAY FOR TRANSFER OF MEMBRANE PROTEINS FROM RED BLOOD CELLS TO OTHER CELLS

Bovine endothelial cells were co-cultured with human red blood cells as previously described. Following three, six, seven and eight days of culture, red blood cells were removed as previously described. An anti-human DAF mouse IgG monoclonal antibody (Waco Pure Chemical Industries, Ltd., Richmond VA) and anti-human CD-59 rat IgG monoclonal antibody (Gift from Dr. Herman Waldman) were used in conjunction with a goat anti-mouse IgG conjugated with Texas Red to assess transfer of human DAF to bovine endothelial cells. A mouse anti-blood group A IgM (Pharmingen San Diego CA) was used in conjunction with a rat anti-mouse IgM conjugated with FITC to assess transfer of other red blood cell membrane proteins.

Endothelial cells were prepared for immunofluorescent staining by fixing with acetone for 10 minutes at room temperature. All incubations involving antibodies were performed at 4° C. Endothelial cells were incubated with antibodies against human DAF (5 μg/ml) simultaneously for 20 minutes in phosphate buffered saline supplemented with 5% fetal calf serum and 0.1% NaN₃. The cells were washed three times with phosphate buffered saline supplemented with 0.1% NaN₃. Cells were incubated with the anti-blood group A antibody in a similar fashion. Second antibody incubations were performed in the dark utilizing conjugated antibodies previously described at a concentration of 5 μg/ml. Following second antibody incubations cells were washed in the manner previously described and the fluorescence quantified.

EXAMPLE III. SPECIES SPECIFICITY OF COMPLEMENT RESISTANCE

To ascertain whether or not the transfer of complement resistance form red blood cells to bovine endothelial cells was a general phenomena or restricted to human red blood cells, another experiment was performed. In this example, bovine endothelial cells were cultured in 24 well plates at an initial concentration of 1 X 10⁴ cells per well. Then either human, pig, or mouse red blood cells were cultured with the cells. Observations were made on days two and nine. Four wells were allocated for each species of red blood cells on each observation day. Two of the four wells were exposed to 10% normal or heat inactivated serum from the same species as the red blood cells in that treatment group. Each of the two remaining wells were exposed to 10% serum from one of the remaining two species whose red blood cells were not used in that treatment group. Therefore, each treatment consisted of assessing the effects of serum from the same species as the red blood cells cultured with the endothelial cells as well as assessing the effect of serum from a species different than that cultured with the red blood cells.

As observed in the previous examples, the endothelial cells were not protected from complement attack whether or not it came from cross species after two days of incubation with red blood cells. This indicates once again that an insufficient level of complement regulatory proteins apparently had transferred from the red blood cells to the endothelial cells.

After nine days of co-culture with red blood cells, endothelial cells exposed to sera (normal or heat inactivated) from the same species as the red blood cells cultured with the cells were protected from complement lysis. This indicates once again that complement regulatory proteins had apparently transferred from the red blood cells to the endothelial cells. However, serum from species different from that which was a source of red blood cells lysed the endothelia¹. cells. Therefore, in the discordant xenograft, complement regulatory proteins of one species apparently do not function in the presence of discordant complement. Further, this indicates that red blood cell to endothelial cell transfer of complement inhibiting protein(s) is a general process not restricted to the human red blood cells-bovine endothelial cells discussed earlier.

EXAMPLE IV: GENE CLONING AND TRANSGENIC ANIMAL PREPARATION

4.1 Construction of CD-59 Transgenic Erythroid-Specific

Expression System

In vivo accommodation based on the transfer of certain GPI- linked proteins from the host to donor tissue may be performed by expressing the protein in transformed RBCs which then transfer the proteins to donor tissue.

A human CD-59 cDNA was kindly provided by Dr. Herman

Waldman. PCR primers CD-59 5 and CDN-59 3, shown in Table 2, were designed to include only the coding region with Ncol sites engineered into the ends for ease of cloning. The human alpha one globin gene was obtained from Dr. Frank Grosveld and cloned into cloning vector pSELECT-1 (Promega Biotechnology Madison, WI) generating pSELECT alpha. Many alpha globin gene clones are publicly available and any other source of the alpha globin promoter and terminator is acceptable. Since the sequence of the regulatory sequences has been published, they may also be synthesized if desired.

The PCR cloned cDNA was digested with Nco I prior to ligation into the alpha globin gene at the globin Nco I site. In this manner, the globin promoter could direct expression of CD-59 to red blood cells and the globin gene could provide elements (i.e., 3'flanking DNA and a polyadenylation signal) necessary for mRNA stability. The alpha globin/CD-59 chimeric gene was then removed from pSELECT alpha with a Cla I, Kpn I digest and inserted into a plasmid (pLCR) containing the human hemoglobin locus control region (LCR) (see Grosveld) . LCR- bearing plasmids are publicly available from Dr. Grosveld and the Medical Research Council of Great Britain; LCRs may also be recovered in the manner taught by Grosveld W) 89/01517, by suitable screening of human genomic DNA.

The LCR alpha CD-59 plasmid was digested with Sac II and the LCR alpha globin:CD-59 linear fragment was then removed and agarose gel purified prior to microinjection.

4.2 Production of CD-59 Transgenic Animal A linearized fragment bearing the C-inhibitor CD59 gene and globin regulatory sequences, but essentially free of the plasmid- specific sequences, was microinjected into fertilized mouse embryos using methods essentially described by Wagner (Wagner, et al., 1981). Founder mice (F₀) were identified by slot blot hybridization and bred to non-transgenic mice whenever possible to propagate F, transgenic mice. To begin a line of transgenic mice, F, X F_ matings were performed. The resulting offspring were designated F₂ and so on. Protein and RNA expression patterns were characterized using either F,, F₂ or F₃ transgenic mice.

4.3 Tissue Distribution of CD-59 in the Transgenic Animal

Protein expression of human CD59 on erythrocytes and lymphocytes in transgenic mice was assessed by FACS analysis while immunohistochemistry was utilized to characterize protein expression in tissue. .An F_j transgenic animal from each of 10 lines was sacrificed by cervical dislocation. A piece of tissue approximately 2mm cube was removed from the heart, spleen, lung, kidney, and liver. The tissue sample was placed on a piece of cork approximately 2 mm square, covered with Tissue-TEK O.C.T. (Miles Inc. Elkart IN) and frozen in liquid nitrogen cooled isopentane. Following freezing, 4 micron thick sections were cut with a cryostat and mounted on slides for immunohistochemistry analysis. Mounted sections were maintained at -80°C until analysis. Slides were thawed at room temperature for 10 minutes prior to immersion in room temperature acetone for 10 minutes to fix the tissue. After fixing, the slides were removed from acetone and allowed to air dry. The slide was then immersed in PBS for 12 minutes with the PBS changed every 3 minutes to prewet the tissue. Approximately 25 μl of primary antibody at a dilution of 1:250 in PBS was pipetted onto the tissue. The primary antibody was a rat IgG2b monoclonal antibody (YTH 53.1) obtained from Dr. Herman Waldmann (Meri, et al., 1990). Goat anti-rat IgG2b affinity purified antibody conjugated with FITC was used as a second antibody. After incubation with each antibody the slide was immersed in PBS for a total of 9 minutes with the PBS changed every 3 minutes. Immunohistochemistry staining results for endothelial cells (EC) in the tissues were: heart 2+, kidney 3+, spleen 3+, liver trace to 2+, and lung trace to 24-, on average, in the mice examined, depending on the level of CD59 expression on RBCs.

Whole blood was analyzed by FACS to examine erythroid specific protein expression of transgene encoded human CD59. Whole blood was collected by periorbital bleeding into a tube containing EDTA. The RBCs were washed in PBS and counted by hemocytometer. Approximately 1 x 10⁶ RBCs were resuspended in 10 μl PBS and 5 μg of either anti-CD59 (YTH53.1) or control non¬ specific rat IgG2b were added and incubated on ice for one hour. The RBCs were washed three times with PBS and resuspended in 100 μl PBS. One μg of goat anti-rat IgG2b conjugated with phycoerythrin was added and incubated for one hour. The RBCs were subsequently washed three times with PBS and preserved in 500 μl of PBS supplemented with 1% paraformaldehyde. Analysis was done in a Coulter Elite flow cytometer. All F, transgenic mice express human CD59 on RBCs at a level, on average, of twice as high as that detected on human RBCs. 4.4 Tissue Distribution of CD-59 Messenger RNA in the Transgenic Animal

The hematopoietic tissue specific expression of human o. or β globin genes in conjunction with the human LCR has been previously characterized (Behringer, et al., 1989). The presence of human CD59 antigen on EC must therefore be due to transfer of protein. To confirm this, RNA was examined from various tissues. One Fj transgenic mouse was anesthetized and restrained on its back. The chest cavity was opened by a mid-ventral incision to expose the heart. After severing the vena cava, a blunt end 23 gauge needle was inserted into the left ventricle and the body was perfused with PBS containing Na Heparin until the liver and kidney appeared blanched (about 30ml) . Kidney, heart, and liver samples were removed and RNA was isolated according to the procedure of Chomczynski, et al., (1987). A Northern analysis of 3 μg total cellular RNA per well was conducted. Control non- transgenic mouse RNA did not hybridize to a human CD59 probe. Size appropriate hybridization was detected only from transgenic mouse blood RNA but not in kidney, heart or liver samples. We conclude that the human CD59 antigen detected on transgenic mice ECs of vascularized grafts was not synthesized by the EC. The human CD59 antigen must have transferred from erythrocytes to the ECs. 4.5 Transfer of CD59 to Endothelial Cells

In an effort to demonstrate the transfer of CD59 from RBCs to EC in vivo, a kidney was transplanted from a non-transgenic mouse into a transgenic mouse. To accomplish the transplant, the renal artery was attached to the dorsal aorta and the renal vein to the vena cava. The ureter was attached to the bladder. In this manner, the transplanted kidney performed as a functional kidney. Five days after transplant, the transplanted normal mouse kidney was biopsied and a piece of tissue was assayed by immunohistochemistry as previously described. Human CD59 was detected on ECs at an intensity of 3+ in the non-transgenic kidney which had been exposed to transgenic mouse blood for 5 days. The human CD59 antigen detected on the ECs of the normal mouse kidney could only have come from the transgenic mouse RBCs. This experiment clearly demonstrated that human CD59 antigen can transfer from RBCs to ECs.

In order to provide further evidence of CD59 transfer from erythrocytes to EC, a bone marrow transplant experiment based on the Leong method was performed (Leong, et al., 1992). Twelve CD59 transgenic mice were used as bone marrow donors to twenty non-transgenic mice. On days 6 and 5 pre-transplant, transgenic donor mice received 0.25 mg I.V. injections of monoclonal antibodies YTS 169 and YTS 191 (Leong, et al., 1592) . Transgenic donor mice were sacrificed by cervical dislocation prior to harvesting cells for transplant. The femur and tibia were removed from each mouse and flushed to harvest bone marrow stem cells. The spleen was removed and a single cell suspension prepared. Stem cells and spleen cells from all donor mice were pooled into respective suspensions.

Fourteen days pre-transplant, recipient mice received I.V. injections of 0.25 mg monoclonal antibodies YTS 169 and YTS 191. Subsequently, recipient mice received 0.25 mg of YTS 169 and 191 by I.P. injection on days 11, 8, 6, 4, 10 days pre-transplant. In addition, the mice were administered 500 mg/1 dimetridazole in drinking water from days 10 through 0 pre-transplant and 50 mg/1 oxytetracycline in drinking water days 0 through 21 post- transplant. At 20 hours pre-transplant, recipient mice received I.P. injections of dimethylmyleran (8mg/kg) . Dimethylmyleran (DMM) acts in mice as a myeloablative agent (Leong, et al., 1992) . The mice received lxlO⁷ bone marrow cells and lxlO⁷ spleen cells from transgenic donor mice by I.V. injection.

One hundred days post-transplant, recipient mice were screened by FACS analysis for the presence of CD59 positive erythrocytes. Nephrectomies were performed on three mice identified to have approximately 30% of the erythrocytes positive for CD59. Immunohistochemistry analysis detected 3+ EC staining in the kidneys assayed. One mouse was sacrificed by cervical dislocation and biopsy of heart tissue also demonstrated EC specific stainiag for human CD59. As in the case of the kidney transplant previously described, this experiment clearly demonstrates that human CD59 antigen transfers from RBCs to ECs in a variety of vascularized organs.

4.6 Functional Integrity of CD-59 After Transfer The functional integrity of human CD59 once it had transferred from the RBCs to the EC of a heart was assessed by ex vivo organ perfusion essentially done by the Langendorf technique. Transgenic and non-transgenic mice hearts were perfused with oxygenated University of Wisconsin (UW) media initially until a stable basal heart (HR) was achieved

(approximately 20 minutes on average) . After HR stabilization, oxygenated UW media supplemented with 50% humar plasma or serum was used to perfuse the organ until 25% of initial basal HR was attained (approximately 30 minutes on average) . Upon reaching 25% of initial basal HR, the perfusion was stopped and a biopsy of the heart was taken for immunohistochemistry analysis as previously described. Biopsy samples were assayed for the presence of human membrane attack complex (MAC) on EC after perfusion. Immunohistochemistry staining for MAC was 3+ intensity on non-transgenic EC following perfusion while it was focal and trace on transgenic EC on average. Therefore, the human CD59 which had transferred from RBCs to EC was functional.

4.7 Construction of DAF Transgenic Expression System

A human DAF cDNA was cloned from human placenta mRNA. The CDNA was obtained by using PCR of a first strand cDNA synthesis reaction. Primers DAF 5 and DAF 3, shown in Table 2, were designed specifically to include only the coding region of the cDNA with some restriction sites engineered into the ends for ease of cloning. The DAF cDNA was cloned by PCR and digested with Xba I and

Pst I prior to ligation into pGem-4Z (Promega Biotechnology

Madison WI) to create pGem-4Z DAF. The DAF clone was removed from pGem-4Z DAF with Xba I, Pst I and blunted with Mung bean nuclease in preparation for cloning into pSELECT alpha. pSELECT alpha was digested with Nco I and blunted with Mung bean nuclease prior to ligation with the blunted DAF clone. Once again, the alpha promoter could then be utilized to direct expression and the alpha gene could provide mRNA stability sequences. The LCR alpha DAF plasmid was digested with Sea I and Kpn I to release the LCR alpha DAF linear fragment used in microinjection.

4.8 Construction and Use of CD59/DAF Transgenic Co-Expression System The LCR alpha CD59 Sac II fragment previously described was used in conjunction with the alpha DAF fragment to make a LCR alpha CD59, alpha DAF fragment. To accomplish rhis, pBluescript

(Stratagene, LaJolla, CA) was cut with endonuclease Eco RV. LCR alpha CD59 was obtained as a Sac II fragment as previously described and blunted prior to ligation into the blunted Evo RV site in pBluescript. The pBluescript LCR alpha CD59 was then cut with Eco RI and blunted. A Cla I, Kpn I alpha DAF fragment was removed from pSelect alpha DAF and blunted prior to ligation into the blunted Eco RI site in pBluescript LCR alpha CD59. The LCR alpha CD59, alpha DAF fragment was removed with a Sal I complete, Sac II partial digest and purified for microinjection.

Transgenic mice were identified which contained the LCR alpha CD59, alpha DAF construct. Erythroid specific expression of both CD59 and DAF was identified in these mice by FACS analysis. CD59 and DAF were also detected on endothelial cells in the transgenic mice by immunoflouresence.

4.9 Construction of MCP:DAF Artificial GPI-link-.d Fusion Protein Transgenic Expression System

A fusion of MCP and DAF cDNAS was performed to generate a GPI-linked MCP which would possess the same mobile properties of CD59 and DAF. MCP is a C-inhibitor which, in nature, is not GPI- linked. To accomplish this, PCR primers MCP 5 and MCP-Sea I were used to clone the first 825 bases of MCP cDNA. PCR primers DAF- Sca I and DAF 3 were used to clone the last 279 bases of DAF. The DAF PCR fragment was digested with Seal and Pst I while the MCP fragment was digested with Sea I and Xba I prior to ligation into pGem cut with Xba I and Pst I. A 1104 MCPrDAF fragment was isolated in pGem and sequenced to confirm proper alignment of codons. The MCP:DAF Xba I, Pst I fragment was cloned into pSelect alpha as previously described for DAF to make pSelect alpha MCP:DAF.

4.10 Construction of CD59/DAF/MCP:DAF Co-Expression System A clone containing alpha CD59, alpha DAF and alpha MCP:DAF was constructed in pBluescript. To accomplish this, pbluescript was first cut with Eco RI and blunted. Alpha MCP:DAF was removed from pSelect alpha MCP:DAF by digestion with Cla I and Kpn I and blunted prior to ligation into the blunted pBluescript to create pBluescript alpha MCP:DAF. PBluescript alpha MCP:DAF was then cut with Sac II and blunted with Eco RI linkers added at the old Sac II site. It was then cut with Cla I and blunted before being cut with Sal I in preparation for adding in the LCR alpha CD59, alpha DAF sequences. The pBluescript alpha CD59, alpha DAF previously described was partial cut with Sac II and blunted and then subsequently cut with Sal I to release a LCR alpha CD59, alpha DAF fragment that was blunt on one end and contained a Sal I overhang on the other. This fragment was ligated into the prepared pBluescript alpha MCP:DAF to create pBluescript LCR alpha CD59, alpha DAF, alpha MCP:DAF. The 17.25 kb LCR alpha CD59, alpha DAF, alpha MCP:DAF fragment was removed with a partial Eco RI, Sal I digest and prepared for microinjection. Transgenic mice and a transgenic pig were produced which contained this construct.

4.11 Construction and Use of CD59 Transgenic Expression System with Epsilon Globin Regulatory DNA

A construct which more closely resembles the endogenous globin gene locus arrangement was designed. Human epsilon globin gene, obtained from Grosveld was ligated into pGem as a Kpn I fragment. The pGem epsilon was digested with Bgl II and Bam HI to release a major portion of the coding region. After gel purification, the digested pGem epsilon was relegated producing a mutant epsilon gene (EΔ) which contained the promoter region but little coding region, therefore, the mutant epsilon could not produce viable epsilon protein. EΔ was removed as a Kpn I fragment and blunted before ligation into the Eco RV site of pBluescript. The human beta globin gene, obtained from Grosveld, was ligated into pSelect as a Kpn I, Mlu I fragment. The human DAF cDNA was ligated into the Nco I site of beta globin as described for alpha globin. Beta globin DAF was removed as a Dpn I, Mlu I fragment and blunted before ligation into the blunted Eco RI site of pBluescript EΔ. A LCR alpha CD59 fragment was blunt ligated into the Eco RV site of pBl escript. This pBluescript LCR alpha CD59 plasmid was then cut with Eco RI and blunted before the blunt ligation with a EΔ beta DAF fragment. The pBluescript LCR alpha CD59 EΔ beta DAF fragment was removed by a partial Sac II, Sal I digest before preparation for microinjection into mouse eggs. Transgenic mice were produced which expressed CD 59 and DAF on RBCs and transferred to endothelial cells. EXAMPLE V: TRANSPLANTATION OF A TRANSGENIC ORGAN INTO A PATIENT

A patient in need of a heart transplant is matched to a suitably sized transgenic pig prepared in accordance with Example IV. Other patients in need of different organs may simultaneously be matched to the same transgenic animal. Tissue samples from the pig are assayed by the method of Example III to determine if the organs are sufficiently accommodated to humans. The pig is then sacrificed and its heart removed. The heart is surgically transplanted into the patient.

EXAMPLE VI: MODIFICATION OF ORGANS AND TISSUES IN VIVO

The blood is withdrawn before and after the transgenic pig from Example IV is sacrificed. Alternatively blood is obtained from transgenic pigs not being sacrificed. Sodium citrate or EDTA is added to the recovered blood to prevent clotting.

Normal pigs are matched to patients in need of a kidney (or other organ) . The normal pigs are anesthetized and their jugular veins catheterized. A liter of the pig's blood is withdrawn from each animal and replaced with one liter of transgenic pig blood. The treated pigs are separated from other animals and are given food and water for one week. After seven days, a sample of the pig tissue is removed and assayed according to Example I to determine if the organs have become sufficiently accommodated for transplantation. The pigs are sacrificed and the kidneys (or other organs) are removed and surgically transplanted into human patients.

EXAMPLE VII; XENOGRAFT ACCOMMODATION IN VIVO A patient currently undergoing dialysis therapy in need of a kidney transplant is matched to a suitably sized pig. The pig is first anesthetized and then immunosuppressed by a combination of plasmapheresis with removal of natural preformed antibodies to human cells and complement by passing serum through a protein A-bound column and a complement receptor-bound column. The resulting serum is checked for complete removal by removing a sample and incubating it with washed human red blood cells and observing for the presence of lysis.

One liter of blood is removed from the pig and replaced with one liter of washed human red blood cells. The human red blood cells are obtained from a blood bank and are chosen on the basis of being cross-matched to the recipient patient. The process is repeated daily for eight days.

At the end of eight days, the pig is sacrificed and the kidneys- removed. Prior to transplantation, a. portion of the organ to be transplanted or an unrelated tissue in the animal may be tested in accordance with the assay in Example III to determine whether the kidney has sufficiently accommodated to the recipient host to prevent hyperacute rejection. The organ or tissue is then surgically transplanted into a recipient human. EXAMPLE VIII: ANALYSIS OF INTERCELLULAR TRANSFER OF GPI- LINKED PROTEINS

The mechanism of GPI transfer is not clear. Some have suggested that transfer occurs through the spontaneous release of free GPI-linked protein. If this were the case then with a high frequency of spontaneous release you would expect to find free GPI-linked protein such as DAF, CD59 Alkaline phosphatase and 5' -nucleotidase (all expressed on RBCs) present in the serum. This is not the case, suggesting that these proteins are efficiently absorbed into other membranes, or that the frequency of GPI release is very low, or that the free proteins are rapidly hydrolyzed by serum PIPLD. This latter possibility may explain the presence of some soluble GPI proteins.

The mechanism of GPI transfer is important with respect to its impact on Xenotransplantation. If transfer occurs via free protein then the efficiency of GPI transfer, both in transgenic animals, and in human patients which had received xenotransplants might be improved through the administration of PIPLD inhibitors. GPI transfer via vesicles, or "flipping" would not benefit from this treatment.

There are three possible mechanisms:

(i) GPI-linked proteins may be spontaneously released from the membrane surface of a cell and then integrate into membrane of another cell by virtue of the GPI tail. In this mechanism the transfer vehicle is free protein.

(ii) Cells may produce lipid vesicles which may contain, amongst other proteins, GPI-linked proteins embedded in the membrane. In this mechanism, the vesicle, which would then fuse to another membrane, is the vehicle of GPI-linked protein transfer. This mechanism might be expected to transfer transmembrane proteins as efficiently as GPI-linked proteins.

(iii) GPI-linked protein transfer may occur only with physical contact between cells. As a result of this contact, the GPI-linked protein of one cell may "flip" so that it is now embedded in the membrane of the other cell. This latter mechanism is distinct from the first since it would require cell-cell contact to initiate the GPI transfer and therefore does not necessarily require any finite rate of spontaneous release of GPI-linked proteins from the cell surface.

Our in vitro culture system provides an opportunity through which GPI transfer between cells can be investigated. For example, it is well known in the literature that phosphatidylinositol specific phospholipase C is capable of removing GPI-linked protein from the surface of cells. In contrast phosphatidylinositol specific phospholipase D which appears to be a serum protein (see Davitz et al (1989) cannot remove GPI-linked proteins from the cell membrane, but does hydrolyze the GPI linkage of free protein in solution. These upases therefore provide the necessary tools to distinguish between two alternative mechanisms for GPI transfer.

Transfer between cells mediated by small lipid vesicles containing GPI-linked proteins would be resistant to PIPLD which would not affect the GPI-linked protein in the vesicle, but inhibited by PIPLC which would remove the GPI-linked protein from vesicles. Conversely, transfer via free GPI-linked protein would be sensitive to both PIPLC, removing the GPI-linked protein from the cell surface, and PIPLD, hydrolizing the free GPI-linked protein in solution. By incorporating these enzymes in our in vi tro co-culture assay these alternative mechanisms (numbers (i) and (ii) above) might be resolved. To investigate the necessity of cell-cell contact, donor cells RBCs) could be cultured alone so as to condition the media. This conditioned media could then be centrifuged at low speed to remove all cells, and at high speed to remove macromolecular components such as vesicle. Conditioned media, treated in this manner could then be applied to target cells (e.g., endothelial cells) and the transfer of GPI-linked protei ns measured as described in the patent. If transfer occurs with conditioned media after a low speed centrifugation, then cell-cell contact is not necessary and transfer could occur via free protein or vesicles. If transfer of GPI-linked proteins also occurs after high speed centrifugation, then the mechanism;- of transfer is likely to be due to free protein. This experiment could also be combined with PIPLC and PIPLD treatment of the conditioned media. As above, PIPLC is expected to inhibit transfer via vesicles whereas PIPLD should have no effect on vesicle mediated transfer. Similarly transfer of free protein should be inhibited by both PIPLC and PIPLD.

GPI transfer might also be examined in a modified culture system in which the target cells (endothelial cells) are separated from the donor cells (RBCs) by a selectively permeable membrane. The pore size of the membrane coulri be selected to allow macromolecules, such as GPI-linked proteins, and small vesicles, or, with an alternative membrane, only macromolecules, to exchange between the two cell types. The presence of the membrane should place severe constraints on intimate cell-cell contact. This modified culture system could be used along with PIPLC and PIPLD enzymes treatments to determine the need for intimate cell-cell contact, and the maximum size (that permitted by the separating membrane) and chemical nature of the transfer vehicle.

Miscellaneous

The foregoing description of the specific embodiments reveal the general nature of the Invention so that others can, by applying current knowledge, readily modify and/or adapt for various applications such specific embodiments without departing from the generic concept, and, therefore, such adaptations and modifications should and are intended to be comprehended within the meaning and range of equivalents of the disclosed embodiments. It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation. All references mentioned in this application, including patents, patent applications, and articles, are hereby incorporated by reference.

CITATIONS

Adams et al. (1985) "The c-myc oncogene driven by immunoglobin enhancers induces lymphoid malignancy in transgenic mice". Nature 381:533-538

Agishi, T. , Takahashi, K. , Ota, K. and T. Yagisawa, (1992), "Comparative evaluation of immuno-absorption and double filtration plasmapheresis for removal of anti-A or -B antibody in ABO-incompatible kidney transplantation". Transplantation Proceedings. 24:557-558.

Auchincloss, H. , Jr., (1988), "Xenogeneic transplantation". Transplantation 16:1-20.

Bach, et al., "Accommodation: A Working Paradigm for Progressing Toward Clinical Discordant Xenografting, " Transplant. Proc, 23:205-7 (1991) .

Behringer, et al., (1989) "Synthesis of Functional Human Hemoglobin in Transgenic Mice". Science, 245:971-973.

Bonifer, et al. (1991) "Dynamic chromatin:The regulatory domain organization of eukaryotic gene loci". J. Cellular Biochem. 47:99-108.

Bonifer, et al. (1990) "Tissue specific and position independent expression of the complete gene domain for chicken lysozyme in transgenic mice". EMBO J. 9:2843-2848.

Bouma, S.R., Drislane, F.W. and W.N. Huestis, (1977), "Selective extraction of membrane-bound proteins by phospholipid vesicles". J.Biol. Chem. 252:6759-6763.

Chomczynski, Piotr and Sacchi, Nicoletta, (1987) , "Single-Step Method of RNA Isolation by Acid Guanidinium Thiocyanate-Phenol- Chloroform Extraction". Anal. Biochem. 162:156-159.

Cook, S.L., Bouma, S.R. and W.H. Huestis, (1980), "Cell to vesicle transfer of intrinsic membrane proteins: Effect of membrane fluidity". Biochem. 19:4601-4607.

Cross, G.A.M. (1990) , "Glycolipid anchoring of plasma membrane proteins". Annu. Rev. Cell Biology. 6:1-39.

Dalmasso, A.P., Vercellotti, G.M. , Platt, J.L. and F.H. Bach (1990) , "Inhibition of complement-mediated endothelial cell cytotoxicity by decay-accelerating factor". Transplantation. 52:530-533.

Ferguson and Williams (1988) , "Cell Surface Anchoring of Proteins via Glycosyl-Phosphatidylinositol Structures, " Ann. Rev. Biochem., 57:285-320.

Ferguson, M.A.J., (1991), "Lipid anchors on membrane proteins". Current Opinion in Structural Biology. 1:522-529.

Fishcel, R.J., Matas, A.J., Perry, E., Dalmasso, A., Noreen, N. and R.M. Bolman, III (1992) , "Plasma exchange, organ perfusion, and immunosuppression reduce "natural" antibody levels as measured by binding to xenogeneic endothelial cells and prolonged discordant xenograft survival". Transplantation Proceedings. 24:574-575.

Gianello, P., Latinne, D., Squifflet, J.P., Carlier, M. Lambotte, L., Cosyns, J.P., Dewaele, A. and G.P.J. Alexandre, "Long term survival in discordant renal xenotransplantation (pig to baboon)". First International Congress on Xenotransplantation. Abstracts: p.18.

Goodnow, et al. (1988)

"Altered immunoglobin expression and functional silencing of self-reactive B lymphocytes in transgenic mice". Nature 334:676-

682

Greaves, et al. (1989)

"Human CD2 3 '-flanking sequences confer high-level, T cell- specific, position-independent gene expression in transgenic mice". Cell 56:979-986

Grosveld, F.G. , (1989), "Vector for integration site independent gene expression in mammalian host cells". International Patent Publication Number WO89/01517.

Hanson, et al. (1991)

"The 5'-flanking region of the human CGL-1/Granzyme B gene targets expression of a reporter gene to activated T-lymphnocytes in transgenic mice". J. Biol. Chem. 266:24433-24438

Hebell, T. , Ahearn, J.M. and D.T. Fearon, (19S1) , "Suppression of the immune response by a soluble complement receptor of β lymphocytes". Science 254:102-105.

Higgs, et al (1990)

"A major positive regulatory region located far upstream of the human alpha-globin gene locus". Genes and Dev. 4:1588-1601

Huestis, W.H. and Newton, A.C., (1986), "Intermembrane protein transfer". J. Biol. Chem. 261:16274-16278.

Hundt, et al. (1992)

"The glycosylphosphatidylinositol-linked FCy receptorlll represents the dominant receptor structure for immune complex activation of neutrophils". Eur. J. Immunol. 22:811-816

Johnstone, R.M.. Adam, M. , Hammond, J.R., Orr, L. and C. Turbide,_. (1987) , "Vesicle formation during reticulocyte manturation" . J. Biol. Chem. 262:9412-9420.

Kinoshita, T. , (1991), "Biology of complement The overture." Immunology Today. 12:291-295.

Kisielow, et al. (1988)

"Tolerance in T-cell-receptor transgenic mice involves deletion of non ature CD4+8+ thymocytes". Nature 333:742-746

Lang, et al. (1991)

"Deletion analysis of the human CD2 gene locus control region in transgenic mice". Nucleic Acids Res. 19:5851-5856

Leong, et al. (1992)

"Classical transplantation tolerance in the adult: the interaction between myeloablation and immunosυppression". Eur. J. Immunol. 22:2825-2830.

Lexer, G., Cooper, D.K.C., Wicomb, W.N., Rose, A.G. , Rees, J. , Keraan, M. , Reiehart, B. and E. Du Toit (1987), "Cardiac transplantation using discordant xenografts in a non-human primate model". Transplantation Proceedings. 19:1153-1154.

Lisanti, et al, (1990) , "Emerging Functional Roles for the Glycosyl-Phosphatidylinositol Membrane Protein Anchor," J. Membrane Biol., 117:1-10.

Lutz, H.U., Liu, S.C., Palek, J. , (1977), "Release of spectrin- free vesicles from human erythrocytes during ATP depletion". J. Cell Biol. 73:548-560.

Medof, M.E., Kinoshita, T. and V. Nussenzweig (1984), "Inhibition of complement activation on the surface of cells after incorporation of decay-accelerating factor (DAF) into their membranes". J. Exp. Med. 160:1558-1578.

Meri, et al., (1990) "Human protectin (CD59) , an 18,000-20,000 MW complement lysis restricting factor, inhibits C5b-8 catalysed insertion of C9 into lipid bilayers". Immunology, 71:1-9.

Moberg, A.W. , Shons, A.R., Gewurz, N. , Mozes, M. and Najarian, J.S., (1971), "Prolongation of renal xenografts by the simultaneous sequestration of preformed antibody, inhibition of complement, coagulation and antibody synthesis" . Transplantation Proceedings. 3:538-541. Monostori, et al. (1991)

"Human CD2 is functional in CD2 transgenic mice". Immunol.

74:369-372

Platt, J.F., Vercelotti, G.M. , Dalmasso, A.P. Matas, A.J., Bolman, R.M. , Najarian, J.S. and F.H. Bach, (1990), "Transplantation of discordant xenografts: A review of progress". Immunology Today. 11:450-456.

Platt, J.L. and F.H. Bach, (1991) , "Discordant xenografting: Challenges and controversies". Current Opinion in Immunology. 3:735-739

Pruitt, S.K., Baldwin, W.M. , III, Marsh, H.C., Lin, S.S., Yeh, C.G. and R.R. Bollinger, (1992), "Effect of soluble complement receptor type I on natural antibody levels during xenograft rejection". Transplantation Proceedings 24:477-478.

Rifkin, M.R. and F.R. Landsberger, (1990) , "Trypanosome variant surface glycoprotein transfer to target membranes: A model for the pathogenesis of trypanosomiasis" . P.N.A.S. (U.S.A.) 87:801- 805.

Rose, E., Sanchez, J. , Watkins, J. , Xu, H. , and C. Berger (1991), "Lymphocyte xenoantigens recognized by preformed antibodies". First International Congress on Xenotransplantation. Abstracts: p. 47

Rosse, W.F. (1990) "Paroxysmal nocturnal hemoglobinuria and decay-accelerating factor." Annu. Rev. Med. 41:431-436.

Skalnik, et al. (1991)

"Targeting of transgene expression to monocyte/macrophages by the gp91-phox promoter and consequent histiocytic malignancies". Proc. Natl. Acad. Sci. USA 88:8505-8509

Stief, et al. (1989) "A nuclear DNA attachment element mediates eleva'red and position- independent gene activity". Nature 341:343-345

Swanson, et al., BIO/TECHNOLOGY, 10:557-59 (Ma.*, 1992).

Wagner, et al., (1981) , "Microinjection of a rabbit β-globin gene into zygotes and its subsequent expression in adult mice and their offspring", Proc. Natl. Acad. Sci. USA, 78:10:6376-6380.

Weisman, H.F., Bartour, T., Leppo, M.K. Marsh, H.C., Jr., Carson, G.R., Concino, M.F., Boyle, M.P., Roux, K.H., Weisfeldt, M.L. and D.T. Fearon, (1990) , "Soluble human complement receptor type 1: in vivo inhibitor of complement suppressing post-ischemic myocardial inflammation and necrosis". Science 249:146-151.

White, D., Oglesby, T.J., Tedja, I., Liszewski, K. , Wallwork, J., Van Den Bogaerde, J. and J.P. Atkinson, (1992), "Expression of human decay-accelerating factor or membrane cofactor protein genes on mouse cells inhibits lysis by huirian complement". Transplantation Proceedings. 24:474-476.

Wilcox, L.A. , Ezzell, J.L. Bernshaur, N.J. and C.J. Parker (1991) , "Molecular basis of the enhanced susceptibility of the erythrocytes of paroxysmal nocturnal hemoglobinuria to hemolysis in acidified serum. " Blood 78:820-829

Yannoutsos, N. , Shelton, A., White, D. and F. Grosveld, (1991), "Transgenic mice with human DAF and MCP DNA constructs". First International Congress on Xenotransplantation. Abstracts: p.7.

Ye, Y., Cooper, D.K.C., Niekrasz, M. , Rolf, L.L., Koren, E., Baker, J. , Martin, M. , Smith, J. and N. Zhuhdi, (1992), "Removal of dog antipig antibody by absorption with pig red blood cell stroma columns". Transplantation Proceedings. 24:563-565.

Xia, W. , Fearon, D.T., Moore, F.D., Schoen, F.J., Oritz, F. and R.L. Kirkman, (1992), "Discordant complement systems as a factor in hyper-actue xenograft rejection". Transplantation Proceedings. 24:479-480. TABLE 1 * COMPLEMENT INHIBITORS

Component Expression Function

DAF Erythrocytes, T- and B-cells, Inhibits C3 fibroblasts, vascular and C5 endothelium, some epithelium, convertase. glandular cells (ocular, lacrimal, and salivary) . Also present in saliva, tears and urine in soluble form.

CD59 Erythrocytes, lymphoid cells, Blocks the glomeruli and tubular epithe¬ formation lium, endothelium of renal of the cortex, capillary endothelium. membrane Also present in urine in attack soluble form. complex.

HRF Erythrocytes, vascular Blocks the endothelium, Schwann cells, formation Ependymal cells, Acinar of the cells, Bronchial epithelium, membrane Renal tubules, and attack squamous epithelium. complex.

MCP B and T cells, monocytes, granulocytes, platelets, endo¬ thelial cells, epithelial cells and fibroblasts.

Factor H serum protein

C4bp serum protein

CR1 (C3b/C4b Found only in primates and human. receptor) Found on erythrocytes, monocytes, granulocytes, B-cells and some T cells CR2 Mature B cells, some T cells and epithelial cells.

Vaccinia virus gp 35 major viral secretory product

Herpes simple virus, gC-1 and gC-2 viral secretory products

*for review see, Kotwal et al., 1990; Kotwal and Moss, 1988; McNearney et al., 1987; and Kinoshita, 1991.

Table IA (References for Table 1)

DAF

Caras, I.W., Davitz, M.A. , Rhee, L., Weddell G. , Martin, D.W., Jr. and Nussenzweig, V. (1987); Nature; 325:545-548 Ewulonu, Ku, Ravi, L. and Medof, M.E. (1991); PNAS; 88:4675-4679

CD59

Davies, A., Simmons, D.L., Hale, G., Harmon, R.A. , Tighe,

H. , Lackmann, P.J., and Waldman, H. (1989); J. Exp. Med.;

170:637-654

Sawada, R. , Ohashi, K. , Anaguich, H., Okazak, H., Hattori,

M. , Minato, N. , and Naruto, M. (1989); Nucleic Acids

Research; 17:6728

CR2

Kurtz, C.B., O'Toole, E., Christensen, S.M., and Weis, J.H.

(1990); J. Immunol.; 144:3581-3591

Fujisaku, A., Harley, J.B., Frank, M.B., Guner, B.A.,

Frazier, B., and Holers, V.M. , J. Biol. Chem. 264:2118-2125

(1989)

C46-binding protein

Aso, T. , Okamura, S., Matsuguchi, T., Sakamoto, N. , Sata, T. , and Niho, Y. (1991); Biochem. Biophys. Res. Commun.; 174:222-227

MCP

Purcell, D.F., Russell, S.M. , Deacon, N.J., Brown, M.A. , Hooker, D.J., and McKenzie, I.F. (1991); Immunogenetics; 33:335-344

Lublin, D.M., Liszewski, M.K., Post, T.W.. Arce, M . A . , LeBeau, M.M. , Rebentisch, M.B., Lemons, R.S., Seya, T. , and Atkinson, J.P. (1988); J. Exp. Med.; 168:181-194

FactorH

Estaller, C, Koistinen, V., Schwaeble, W. , Dierich, M.P., and Weiss, E.H. (1991); J. Immunol.; 146:3190-3196 GP-35

Kotwal, G.J. and Moss, B. (1988); Nature; 335:176-178

CR-1

Klickstein, L.B., Wong, W.W. , Smith, J.A. , Weis, J.H. , Wilson, J.G., and Fearon, D.T. (1987); J. Exp. Med.; 165:1095-1112

C-1 (HSV; Possible C-inhibitor

Frink, R.J., Eisenberg, R., Cohen, G. and Wagner, E.K. (1983); J. Virol.; 45:634-647

Draper, K.G., Costa, R.H., GT, Y. , Spear, P.G., and Wagner, E.K. (1984); J. Virol.; 51:578-585

Draper, K.G., Frink, R.J., Devi, G.B., Swain, M. , Galloway, D. and Wagner, E.K. (1984); J. Virol.; 52:615-623.

Table 2 PRIMER NUCLEOTIDE SEQUENCES

DAF 5 GGA GCT CTA GAT CAT GAC CGT CGC GCG GC

DAF 3 GGA TCC TGC AGC TAA GAA ACT AGG AAC AGT CTG

CD 59 5 GGA TCC CCA TGG GAA TCC AAG GAG GGT C

CD 59 3 CGA CCC ATG GTT AGG GAT GAA GGC TCC

MCP 5 CTC GAG TCT AGA CCA TGG AGC CTC CCG GCC

MCP 3 CTG CAG CTA GCT CAG CCT CTC TGC TCT GCT G

MCP Sea I CTC AGT ACT GTT ACT GTC ACA GAC AAT TG

DAF Sea I CTC AGT ACT TCC AAG GTC CCA CCA ACA G

TABLE 3: PROMOTERS WITH SPECIFICITY FOR MOBILE CELLS

Gene/Promoter Specificity Reference

β-globin LCR RBC Grosveld (1989)

alpha-globin RBC Higgs et al. (1990)

B-cells* Adams et al. (1985) Goodnow et al. (1988) Grosschedl et al. (1984) ∞ o

CD-2, TCR T-cells Greaves et al. (1989) , Lang et al (1991), Monostori et al . (1991) , Kisielow et al. (1988)

Human CGLl/Granzyme B activated T-cells Hanson et al. (1991]

chick Lysozyme macrophage Bonifer et al. (1991) , Stief et al, (1989) , Bonifer et al . (1990)

FcRIII** PMN Hundt (1992)

gp91-phox monocytes/macrophages Skalnik (1991)

* Expression is predominant in B-cells, but there is some expression in T-cells.

**Sιte of endogenous gene expression. Specificity not yet demonstrated in transgenic animals.

TABLE 4

NATURALLY OCCURRING GPI ATTACHMENT SIGNALS

Protein Amino acid sequence Hydrophilicity

T. brucei VSG MIT117a ACKD SSILVTKKF^ALTWSAAFVALLF-2. 3

T. brucei VSG MIT221 TGSS NSFVISKTPLWLAirLLF-2. 4

T. congolense VSG YNatl . l HLPS GSSHGTKA" IRSILHVALLM- 1 . 7

T. brucei VSG MITllβa KCRN GSFLTSKQFA^FSWSAAFVALLF-2. 4

T. brucei PARP PEPG AATLKSV~ALPFAIAAAALVAAF-2 . 2

Fola te binding protein EEVA RFYAAAMSGAG^" PWAAWPFLLSLALMLLWLLS -

1 . 6

THY1 LVKC GGISLLVQNTS^* WLLLLLLSLSFLQA TDFISL -

1 . 7

AP GTTD AAHPGRS" WPALLPLLAGTLLLLETATAP -1 .

5 ' -Nucleotidase IKFL AASHYQGS^FPLIILSFWAVILVLYQ- 1 . 7

MRC OX -45 LARS SGVHWIA"ANLWTLSSIIPSILLA -2 . 2

Hamster prion protein GGRS SAVLFSSPPVILLISFLIFLMVG-2 . 5

AChE (Torpedo) ATAC DGELSSSGTSSS~kGIIFYVLFSILYLIFY- l -

CEA TVSA SGTSPGL^* SAGATVGIMIGVLVGVALI -2. 3

Dictylostelium PsA TTTG SASTWA"SLSLIIFSMILSLC-2 . 4

RT6. 2 (ra t) CLYS SAGARES"CVSLFLVQLLCLAEP-1 . 4

DAF GTTS GTTRLLSGHTC^* FTLTGLLGTLVTMGLLT- 1 , 4

This table, adapted from Cross, lists the carboxy amino acid sequence predicted from cDNAs. The Vertical lines indicate the position of post translational GPI cleavage and attachment. Hydrophilicity was calculated by the method of Kyte and Dooli ttle, J. Mol . Biol . , 157: 105-132

(1982) using a 7 amino acid window. The ^Λ indicates the starting posi tion of the sequence used to calculate hydrophilici ty, i . e. , the beginning of the presumed hydrophobic domain. Only in the case of DAF has this hydrophobic domain been experimentally mapped. The other domains were selected on the basis of their spacing from the GPI attachment site (minimum distance was 8 amino acids) and on the hydrophilici ty plot of the entire sequence, selecting that amino acid which exhibited the greatest negative change in slope. Hydrophilicity values represent the mean of the carboxy terminal residues from the ^A to the end.

∞

TABLE 5: NATURALLY GPI-LINKED PROTEINS

Hydrolytic enzymes alkaline phosphatase acetylcholinesterase

5 ' -nucleotidase trehalase alkaline phosphodiesterase renal dipeptidase aminopeptidase P lipoprotein lipase Parasite proteins p63 protease Leishmania

Merozoite protease Plasmodium variant surface glycoprotein Trypanosoma surface proteins Paramecium

195 - kDa antigen Plasmodium

Scrapie prion protein

Tegument protein Schistosoma

Ssp-r Typanosoma Immunol ogi cal

Thy-1^*

RT- 6^*

Qa -2^*

Ly- 6^*

Blast-1

CD 14'

Fell I receptor^* Cancer related

Carcinoembryonic antigen nonspecific crossreacting antigen Complement

Decay accelerating factor^*

CD59 Cell adhesion

LFA-3^* neural cell adhesion molecule

Heparin sulfate proteoglycan PH-20 guinea pig sperm Miscellaneous

130 -kDa hepatoma glycoprotein

34 -kDa placental growth factor

GP-2

Folate receptor

Oligodendrocyte-myelin protein

Antigen 117 Dictyostelium

125 -kDa glycoprotein Saccharomyces

Elongation factor EF-1 alpha

* Denotes GPI-linked proteins expressed on migratory cells such as erythrocytes, macrophages , lymphocytes and fibroblasts .

Table 6

The amino acids in order of hydropathy (hydrophobicity₎ ,

Kyte-Doolittle Scale

Table 7

Amino Acids in Order of Size

------ Volume

(angstroms) Gly(G) 60.1

Ala (A) 88.6

Ser (S) 89.0

Cys(C) 108.5

Asp(D) 111.1

Thr(T) 116.1

Asn(N) 117.7

Pro (p) 122.7

Glu(E) 138.4

Val(V) 140.0

Gln(Q) 143.9

His(H) 153.2

Met(M) 162.9

He (I) 166.7

Leu(L) 166.7

Lys(K) 168.6

Arg(R) 173.4

Phe(F) 189.9

Tyr (Y) 193.6

Trp(W) 227.8

SEQUENCE LISTING

(1) GENERAL INFORMATION:

(i) APPLICANT: BYRNE, GUERARD W. KOOYMAN, DAVID L. LOGAN, JOHN S.

(ii) TITLE OF INVENTION: PREACCOMMODATED XENOGENEIC ORGAN TRANSPLANTS

(iii) NUMBER OF SEQUENCES: 12

(iv) CORRESPONDENCE ADDRESS:

(A) ADDRESSEE: Browdy & Neimark

(B) STREET: 419 Seventh Street, N.W.

(C) CITY: Washington

(D) STATE: DC

(E) COUNTRY: USA

(F) ZIP: 20004

(v) COMPUTER READABLE FORM:

(A) MEDIUM TYPE: Floppy disk

(B) COMPUTER: IBM PC compatible

(C) OPERATING SYSTEM: PC-DOS/MS-DOS

(D) SOFTWARE: Patentln Release #1.0, Version #1.25

(vi) CURRENT APPLICATION DATA:

(A) APPLICATION NUMBER: US 07/948,521

(B) FILING DATE: 22-SEP-1992

(C) CLASSIFICATION:

(viii) ATTORNEY/AGENT INFORMATION:

(A) NAME: Cooper, Iver P.

(B) REGISTRATION NUMBER: 28,005

(C) REFERENCE/DOCKET NUMBER: BYRNE 1

(ix) TELECOMMUNICATION INFORMATION:

(A) TELEPHONE: 202-628-5197

(B) TELEFAX: 248633

(C) TELEX: 202-737-3528

[2) INFORMATION FOR SEQ ID NO:l:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 29 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA (xi) SEQUENCE DESCRIPTION: SEQ ID NO:l:

GGAGCTCTAG ATCATGACCG TCGCGCGGC 29

(2) INFORMATION FOR SEQ ID NO:2:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 33 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:2

GGATCCTGCA GCTAAGAAAC TAGGAACAGT CTG 33

(2) INFORMATION FOR SEQ ID NO:3:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 28 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:3

GGATCCCCAT GGGAATCCAA GGAGGGTC 28

(2) INFORMATION FOR SEQ ID NO:4:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 27 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:4

CGACCCATGG TTAGGGATGA AGGCTCC 27

(2) INFORMATION FOR SEQ ID NO:5:

(i) SEQUENCE CHARACTERISTICS: (A) LENGTH: 24 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

( i) SEQUENCE DESCRIPTION: SEQ ID NO:5:

GAGAACCCAC ATGACCGTCG CGCG 24

(2) INFORMATION FOR SEQ ID NO:6:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 24 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:6:

CGCGCGACGG TCATGTGGGT TCTC 24

(2) INFORMATION FOR SEQ ID NO:7:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 23 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:7:

CTAGTTTCTT AGCATGGTGC TGT 23

(2) INFORMATION FOR SEQ ID NO:8:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 23 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA (xi) SEQUENCE DESCRIPTION: SEQ ID NO:8

ACAGCACCAT GCTAAGAAAC TAG 23

(2) INFORMATION FOR SEQ ID NO:9:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: .23 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:9

GAGAACCCAC CATGGGAATC CAA 23

(2) INFORMATION FOR SEQ ID NO:10:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 23 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:10

TTGGATTCCC ATGGTGGGTT CTC 23

(2) INFORMATION FOR SEQ ID NO:11:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 23 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS : single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:11:

TTCATCCCTA ACCATGGTGC TGT 23 (2) INFORMATION FOR SEQ ID NO:12:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 23 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:12

ACAGCACCAT GGTTAGGGAT GAA 23

(2) INFORMATION FOR SEQ ID NO:13:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 29 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: DNA

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:13: SGTTRLLSGH TCFTLTGLLG TLVTMGLLT 29

Claims

What is claimed is:

1. A method of delivering a protein of interest to a tissue of a target animal which comprises

(a) providing in said animal mobile cells which bear the protein of interest on their membranes, said protein being attached to the membrane by a linked glycophosphatidyl inositol (GPI) anchor,

(b) causing said cells to circulate in said animal in such a manner as to come into contact with said tissue, and to transfer the protein of interest to said tissue by intermembrane transfer, where said protein is transferred to said tissue by said mobile cells, by intermembrane transfer, to a substantially greater degree than occurs naturally in said target animal .

2. The method of claim 1 wherein the mobile cells are red blood cells.

3. The method of claim 1 wherein the mobile cells are migratory cells.

4. The method of claim 3 wherein the migratory cells are macrophages.

5. The method of claim 3 wherein the migratory cells are fibroblasts.

6. The method of claim 1 wherein the GPI-linked protein is produced by the mobile cell .

7. The method of claim 6 wherein the target animal is a transgenic or chimeric animal genetically engineered so that one or more of its mobile cells produce the GPI-linked protein.

8. The method of claim 1 wherein the mobile cells are introduced into the target animal .

9. The method of claim 8 in which the GPI-linked protein is produced by the mobile cell.

10. The method of claim 8 in which the GPI-linked protein is attached to the mobile cell prior to its introduction into the target animal.

11. A method for obtaining cells of one species which are preaccommodated for xenogeneic transplantation to an animal of a second, discordant species which comprises causing transplantable cells from a first species of animal to come into proximity with resistant cells bearing a transferrable complement inhibition factor capable of inhibiting the complement of a second, discordant species of animal, under conditions effective for transfer of complement resistance from said resistant cells to said transplantable cells.

12. A method of transplanting an organ from a donor animal to a recipient animal of a second and discordant species, which comprises (a) preaccommodating at least the vascular endothelial cells of said organ to such xenogenic transplantation by the method of claim 11, and then (b) transplanting said preaccommodated cells to said recipient animal.

13. The method of claim 12 wherein, between steps (a) and (b) , a sample of said transplantable cells are assayed for the presence of complement inhibition factors active against complement of the recipient animal.

14. The method of claims 12 or 13 wherein the transplantable cells are preaccommodated by co-cultivating them in vitro with said resistant cells.

15. The method any of claims 11-14 wherein the resistant cells are red blood cells.

16. The method of any of claims 12-15 in which the resistant cells are introduced into the body of the do-ior animal under conditions effective for transfer of complement resistance.

17. The method of claim 16 in which the resistant cells are blood cells which are transfused into the donor animal.

18. The method of claim 16 in which the donor animal is immunosuppressed so that it does not reject the resistant cells.

19. The method of claim 12 in which the donor animal is a transgenic animal, and the resistant cells are primarily mobile cells of said donor animal which express a transgene that encodes a complement inhibition factor, said transgene being expressed primarily in said mobile cells.

20. The method of claim 19 in which the resistant cells are red blood cells.

21. The method of claim 20 in which the expression of the transgene is primarily erythroid-specific.

22. The method of claim 19 in which the resistant cells are migratory cells.

23. The method of claim 22 in which the migratory cells are macrophages.

24. The method of claim 22 in which the migratory cells are fibroblasts.

25. The method of claim 14 in which the resistant cells are obtained from a transgenic animal, at least some of whose cells produce complement inhibition factor in transferrable form.

26. The method of claim 14 in which the resistant cells are obtained from the recipient animal.

27. The method of any of claims 12-26 in which the recipient animal is a human.

28. The method of any of claims 12-27 in which the donor animal is a pig.

29. The method of any of claims 11-28 in which the complement inhibition factor is a membrane associated molecule.

30. The method of claim 29 in which the factor is a protein with a GPI anchor.

31. The method claim 30 in which the factor is a chimeric protein wherein the complement inhibiting domain corresponds to that of a protein which in nature is not GPI-linked.

32. The method of claim 31 wherein the factor is an analogue of MCP which bears a GPI anchor.

33. The method of any of claims 11-28 in which the complement inhibition factor is selected from the group consisting of DAF, CD59, MRF, MCP, Factor H, C4bp, CR1, CR2, vaccinia virus gp35, and herpes simplex virus gC-1 and gC-2.

34. The method of claim 11 in which the transfer is an intermembrane transfer between contacting cells.

35. The method of claim 11 wherein said conditions comprise contacting the cells for a period of at least about six days.

36. An organ whose cells are preaccommodated by the method of claim 11 so that it is not rejected by hyperacute rejection when transplanted into a predetermined immunocompetent discordant animal in which said factor inhibits hyperacute rejection.

37. The organ of claim 36 which is selected from the group consisting of kidney, eye, heart, skin, liver, bone arrow, intestine, blood vessels, joints, pancreas, lung, bronchi, brain tissue, glands, hormone producing tissues, and portions of these tissues .

38. A transgenic animal of a first species which comprises mobile cells which express a transgene encoding a complement inhibition factor effective against the complement system of a second, normally discordant animal species, said mobile cells transferring said factors to other, non-mobile cells, whereby organs of said transgenic animal may be transplanted into a recipient animal of the second species without rejection by hyperacute rejection.

39. The animal of claim 38 in which said transgene is expressed essentially only in said mobile cells.

40. The animal of claim 39 in which said mobile cells are red blood cells.

41. An assay to determine whether or not a tissue is accommodated to a discordant animal comprising; obtaining a sample of said tissue or other tissue similarly treated; contacting and incubating said sample with serum from said discordant animal; and, measuring the binding of antibody from said serum to said sample.

42. The assay according to claim 39 wherein antibody binding to said sample is measured by complement fixation, adding a labeled anti-immunoglobulin antibody or observing cells in said sample.