Modification of xenoantigens on porcine erythrocytes for xenotransfusion

Modification of xenoantigens on porcine erythrocytes for xenotransfusion Jay Doucet, MD, Zu-hua Gao, MD, Leslie A. MacLaren, PhD, and Vivian C. McAlister, MB, Halifax, Truro, and London, Nova Scotia and Ontario, Canada

Background. Problems of supply and disease transmission with blood transfusion may be controlled by the use of an isolated animal donor pool. However, porcine erythrocytes (PRBCs) usually are destroyed rapidly by preformed antibodies in human serum. We examined the impact on PRBC antigenicity by the removal of cell membrane a-gal(1-3)b-galGlcNac epitopes (called a-gal) and chemical masking of other xenoantigens. Methods. From porcine ‘‘low expressors’’ of a-gal, PRBCs were subjected to (1) enzymatic removal of membrane a-gal with a-galactosidase, (2) covalent attachment of cyanuric acid–linked methoxypolyethylene glycol, or (3) both processes. PRBC integrity was assessed by light microscopy, scanning electron microscopy, osmotic fragility, and determination of oximetric p50. The effects of treatment were measured by hemagglutination, complement fixation, flow cytometric assay of immunoglobulin G/M binding, and clinical cross-match testing to human sera. Results. Cyanuric acid–linked methoxypolyethylene glycol reduced hemagglutination titers moderately, although a-galactosidase treatment reduced hemagglutination titers to levels similar to negative controls. The combination of the treatments was most effective, by the reduction of binding of human immunoglobulin M by 61% compared with controls. RBC morphologic condition, stability, and p50 values were maintained. Clinically used cross-match tests between PRBCs and human sera demonstrated increased compatibility. Conclusions. These data suggest that strategies to remove or mask xenoantigens on PRBCs reduce antigenicity sufficiently to allow in vitro cross-match compatibility to human sera. (Surgery 2004;135:178-86.) From the Department of Surgery, Dalhousie University, Halifax, Nova Scotia; the Department of Surgery, University of Western Ontario, London, Ontario; and the Department of Plant and Animal Sciences, Nova Scotia Agricultural College, Truro, Nova Scotia, Canada

THE THREAT OF DISEASE TRANSMISSION in the human blood supply remains a concern and an impetus in the search for alternatives. Use of cross-linked hemoglobin solutions as a blood substitute has been limited by increased vascular resistance1 and auto-oxidation.2 Even if these problems are overcome, an animal source of hemoglobin may be required.3 Xenogeneic red blood cell (RBC) transfusion may allow for a dependable blood supply from a controlled source. In xenotransplantation research, the pig is the favored source of organs and tissues. A recent review of human recipients of pig tissue found the persistent survival of transplanted tissue without transmission of porcine endogenous retrovirus.4 Continuing research will be required to allay fears regarding disease transAccepted for publication August 1, 2003. Reprint requests: Vivian McAlister, MB, 4-TU40, University Hospital, LHSC-UC, London, Ontario, Canada, N6A 5A5. 0039-6060/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.surg.2003.08.013

178 SURGERY

mission.5 Porcine erythrocytes (PRBC) are rejected hyperacutely by antibodies that are present in human serum. Humans have preformed antibodies to a nonprimate antigen known as gal-a(1-3)bgalGlcNac (a-gal).6 This glycoprotein is synthesized in the pig cell by the a(1,3)galactosyltransferase enzyme, which catalyses the addition of a terminal a-linked galactose to N-acetyl lactosamine.7 In human cells a(1,2)-fucosyltransferase (H transferase) converts the same substrate into fucosylated N-acetyl lactosamine (H substance) that is found on human blood group O cells.8 The expression of a-gal on endothelium has been shown to vary between pigs.9 We have demonstrated variable expression of a-gal on PRBC that correlates with hemagglutination titers of human serum.10,11 Two strategies have been investigated in humans to permit allotransfusion across the ABO blood group barrier. Enzymatic conversion of the B substance back to H (type O) is achieved by a-galactosidase.12 The xenoantigen, a-gal, has a similar structure to B substance, and we hypothesized treatment with a-galactosidase should convert it to

Surgery Volume 135, Number 2

human ‘‘h’’ (Bombay) antigen, which types as O. Human RBC antigens have also been masked to prevent antibody binding with the use of the covalent attachment of large polymers of cyanuric chloride–activated methoxypolyethylene glycol (PEG).13-15 PRBCs are thought to express non– a-gal xenoantigens and a-gal.16 Recently it has been shown that antibodies to these non–a-gal xenoantigens are induced by vascularized transplantation and not directed at major histocompatibility complex.17 We speculated that treatment with PEG would mask these nonspecific antigens. Therefore, we investigated whether the selection of low expressor PRBC, combined with treatment with the use of a-galactosidase and PEG, would reduce PRBC antigenicity sufficiently to pass clinically used cross-match tests with human sera. METHODS Subjects. Blood was obtained from live pigs in compliance with a protocol approved by the Nova Scotia Agricultural College Animal Care and Use Committee in compliance with the Canadian Council on Animal Care Guidelines (1994). PRBC expression of a-gal has been quantified previously by identification of the number of cells that are required to remove 50% of anti–a-gal antibody from normal human serum (NHS).8,9 Hemagglutination by NHS of PRBC from those pigs that were identified as high expressors of RBC a-gal was significantly stronger compared with RBC from low expressor pigs (titers of 1/160 vs 1/98; P < .01, where 160 was the maximum dilution tested). Hemagglutination titers with NHS of 1 of 128 or less were used therefore to identify porcine donors as low expressors for further RBC treatments. Untreated PRBCs were used as positive controls, and isotype human RBCs were used as negative controls throughout the experiment. Blood from 3 pigs was tested twice in each experiment. Statistical analysis was carried out by the paired Student t test for comparisons of 2 groups and by analysis of variance for multiple groups. Comparison of crossmatch scores was by Kruskal-Wallis procedure with a Tukey multiple comparisons test. A probability value of less than .05 was considered significant. Erythrocyte modification. Low expressor PRBCs at 50% hematocrit were exposed to 0, 0.625, 1.25, 2.5, and 5 l/mL of green coffee bean a-galactosidase (Sigma Chemical Co, St. Louis, Mo) for 30 minutes at 48C in a pH 5.5 phosphate saline solution buffer to determine the enzyme concentration that shows maximum reduction in hemagglutination with NHS without cell lysis during

Doucet et al 179

treatment. Titration of PEG with 0, 2, 4, 6, or 8 mg/mL cyanuric chloride that was activated 5000 d methoxypolyethylene glycol (Sigma Chemical Co) were carried out on low expressor PRBCs at 12.5% hematocrit in a pH 9.2 phosphate citrate buffer for 60 minutes at 48C to determine the concentration that showed the maximum reduction in hemagglutination with NHS without cell lysis during treatment. Hemagglutination. Hemagglutination studies were performed in multiwell plates with repeated dilutions of NHS in saline solution. Whole blood in anticoagulant citrate dextrose was diluted 1:10 in phosphate-buffered saline solution (PBS) that contained 0.09% (weight/volume) gelatin (ICN, Aurora, Ohio) and 20 mmol/L ethylenediaminetetraacetic acid, incubated for 15 minutes at 378C, then centrifuged for 10 minutes at 1250g at 48C. The supernatant and buff coats were discarded. RBCs were then washed 3 times in PBS that contained 0.09% gelatin, 60 lmol/L CaCl2 and 40 lmol/L MgCl2 (PBS-G-Ca-Mg). Twenty-five microliters of a 2% suspension in PBS-G-Ca-Mg were mixed with 25 lL of doubling dilutions of NHS in saline solution in a 96-well plate. After 60 minutes at 258C, wells were scored for agglutination on a 5-point scale: no agglutination, 1; weak agglutination (plus/minus), 2; clear agglutination, 3; moderately strong agglutination, 4; and very strong agglutination, 5. Hemagglutination titer was recorded as the highest dilution that caused an agglutination score of 3 or higher. Complement fixation. Complement fixation was determined by the inhibition of lysis of a standardized preparation of anti–sheep rabbit antibody sensitized sheep RBCs (BioWhittaker, Walkersville, Md) after the incubation of the control and treated PRBCs with guinea pig complement (Cedarlane Laboratories, Ltd, Hornby, Ontario, Canada). Guinea pig complement was first incubated with treated or control PRBCs and NHS in multiwell plates for 30 minutes at 378C. Plates were then centrifuged at 400g, and supernatant was pipetted into plates that contained 2% sensitized sheep cells in veronal buffered saline solution and then incubated at 378C for 30 minutes. Plates were centrifuged again, and supernatants were read in a spectrophotometer at 540 nm to determine the percentage of sheep RBC lysis. The difference in percentage of sheep cell lysis from nonlysed controls was reported as the percentage of guinea pig complement fixation by PRBCs. Flow cytometry. RBCs were incubated at 48C for 30 minutes in 1:50 NHS, washed with PBS, then incubated for 30 minutes in anti-human

180 Doucet et al

Surgery February 2004

Fig 1. Effect of treatment with different concentrations of a-galactosidase on hemagglutination of porcine erythrocytes by NHS.

Fig 2. Effect of treatment with different concentrations of polyethylene glycol on hemagglutination of porcine erythrocytes by NHS.

immunoglobulin M–fluorescein isothiocyanate (FITC) or immunoglobulin G–FITC (Cedarlane Laboratories, Ltd) for 50 minutes. Cells were washed twice with buffer, resuspended in 200 lL,

and passed through a FACScan flow cytometer (Becton Dickinson, Franklin Lakes, NJ) to determine the relative fluorescence of the cells (expressed as mean channel shift).

Surgery Volume 135, Number 2

Doucet et al 181

Fig 3. Effect of treatment of porcine erythrocytes with optimal concentrations of polyethylene glycol and a-galactosidase alone or in combination on hemagglutination by NHS.

Clinical cross-match study. Low expressor PRBCs that were exposed to a combination of the optimal PEG and a-galactosidase treatments were cross-match tested with a clinical standard antihuman globulin/polyacrylamide gel system (IDMTS Gel Test; Ortho-Clinical Diagnostics, Inc, Pompano Beach, Fla) with random group A, B, and O human sera. Test results were scored by an experienced blood bank technician who was blinded to the RBC source and treatments with a 5-point scale; a 0 score indicated no agglutination reaction, and a 4+ score indicated the strongest possible agglutination reaction. Structure, osmotic fragility, oximetry. RBC structure was determined by light and scanning electron microscopy. RBC (20 lL) were mixed in cuvettes with serial dilutions of saline solution (0.85%-0%) for 20 minutes at room temperature. After centrifuge at 250g for 5 minutes, the optical densities of supernatants were read on a spectrophotometer at a wavelength of 540 nm. Distilled water was used as a blank. The fraction of RBC that was lysed was estimated with the optical density of RBC in 0.85% saline solution as 0% lysis and in distilled water as 100% lysis. Mean corpuscular fragility was taken to be the percentage of saline

solution that caused 50% hemolysis. Oxygen affinity was determined by the measurement of p50 in a blood gas analyzer and co-oximeter (ABL 505 and OSM 3; Radiometer, Copenhagen, Denmark). RESULTS A concentration of 2.5 units of a-galactosidase enzyme per milliliter of 50% PRBCs gave an optimal reduction in hemagglutination with NHS without observable hemolysis (Fig 1). Titration of PEG demonstrated an optimal concentration of 6 mg PEG per milliliter of 12.5% PRBC in reducing hemagglutination with NHS without visible cell lysis (Fig 2). Both PEG and a-galactosidase treatments significantly reduced hemagglutination with untreated control PRBC that had a mean inverse titer of 143 ± 2.9 versus 37 ± 27 (P = .029) for PEG-treated PRBCs and 3.5 ± 1 (P = .015) for a-galactosidase–treated PRBCs (Fig 3). Used alone, PEG was less effective than a-galactosidase in reducing hemagglutination. Combined treatment gave a mean inverse titer of 2.75 ± 1.5, but this was not significantly better than a-galactosidase treatment alone. PEG-treated PRBCs reduced immunoglobulin M binding to 79.5% of the untreated control PRBC

182 Doucet et al

Surgery February 2004

Fig 4. Mean channel fluorescence of FITC-labeled rabbit anti-human immunoglobulin M–bound treated and control porcine erythrocytes after incubation for 30 minutes in 1:25 diluted pooled human serum at 25oC.

Fig 5. Mean channel fluorescence of FITC-labeled rabbit anti-human immunoglobulin G–bound treated and control porcine erythrocytes after incubation for 30 minutes in 1:25 diluted pooled human serum at 25oC.

level versus 43.2% for a-galactosidase treatment, which was significantly better (P = .033; Fig 4). Combined treatment showed a further improvement to 39.0% of untreated control PRBC; however, the difference from a-galactosidase treatment

alone was not statistically significant (Fig 4). In contrast, immunoglobulin G binding did not seem to be affected by either treatment, but the mean channel fluorescence was low in all groups compared with immunoglobulin M binding (Fig 5).

Doucet et al 183

Surgery Volume 135, Number 2

Fig 6. Complement fixation as measured by the reduction (%) in lysis of standardized hemolysin-treated sheep erythrocytes that were mediated by guinea pig complement by serum after incubation with porcine erythrocytes.

Untreated PRBCs that were incubated with NHS and guinea pig complement demonstrated removal of more than 99% of complement-mediated sheep RBC hemolysis, which suggested near total binding of complement by RBC surface antibody (Fig 6). Treatment with PEG reduced complement fixation to 57% ± 6% of control (P = .02), but treatment with a-galactosidase was significantly better at 31% ± 9% of control (P = .008). Combined treatment showed a further reduction in complement fixation to 19% ± 5.5%, which was significantly better than a-galactosidase alone (P = .04). Anti-human globulin polyacrylamide gel crossmatch testing of porcine red cells with human sera from random patients demonstrated a marked reduction in reaction when cells were treated with a combination of PEG and a-galactosidase (Table I). Untreated PRBCs invariably showed a maximal 4+ reaction to human type A, B, or O sera, but treated cells all had some reduction in agglutination. Group O serum had significantly less reactivity to treated PRBCs than either group A or B. Compatible cross-matches were found in 5 of the 9 tests that were undertaken with blood group O sera. Serum from 1 patient did not react to treated PRBCs from 3 different animals. RBC structure as measured by light or scanning electron microscopy was unchanged by either treatment method alone or in combination. Osmotic fragility was unaffected by treatment

Table I. Mean anti-human globulin/ polyacrylamide gel cross-match scores between porcine erythrocytes (PRBC) that were treated with a combination of polyethylene glycol and a-galactosidase and random human sera* Serum blood type

Untreated PRBCs (n = 3)

Treated PRBCs (n = 3)

A ( n = 3) B (n = 3) O (n = 3)

4+ 4+ 4+

2.1 + (1-3)yz§ 1.7 + (1-2)yz 0.7 + (0-2)y§

*Scored on a scale 0 to 4+, expressed as mean (range). yTreated versus untreated: P < .01. zTreated type A versus type B: P < .05. §Treated type A versus O: P < .05.

(Fig 7). Oxygen dissociation shift was to the left by a small yet statistically significant margin with PEG treatment, either alone or in combination with agalactosidase (Table II). Treatment with a-galactosidase alone did not affect the p50. Co-oximetry did not detect any significant amount of methemoglobin in any group. DISCUSSION Treatment of PRBC with either a-galactosidase or PEG reduced the in vitro reaction of human serum to the cells. The dominance of a-gal among xenoantigens is the likely reason that the specific treatment with a-galactosidase alone was more effective than the treatment with PEG. Treatment

184 Doucet et al

Surgery February 2004

Fig 7. Osmotic fragility of treated porcine erythrocytes (PRBC). Untreated PRBCs are more susceptible to osmotic lysis than human RBCs but are unaffected by treatment.

Table II. Mean ( ± SD) p50 (mm Hg oxygen) of human, porcine, and treated porcine erythrocytes Mean ± SD Human erythrocytes Porcine erythrocytes (PRBC) a-Galactosidase PRBC PEG PRBC a-Galactosidase + PEG PRBC

26.5 28.9 27.7 24.0 23.4

± ± ± ± ±

0.23 1.54 0.75 1.15 1.24

with a-galactosidase reduces the cytotoxicity of porcine endothelial cells by human or monkey serum.18 In a pig-to–rhesus monkey vein transplantation model, a-galactosidase delayed the onset of but did not prevent hyperacute rejection.19 Treatment with increasing concentrations of PEG increasingly reduced hemagglutination. We were unable to measure actual PEG binding. Recently it was suggested that alternative methods to activate PEG binding may increase its binding and its immunoprotective effect.20 In addition, the full

impact of PEG treatment may not be appreciated until transfusion in a primate is tested because antibodies to non–a-gal xenoantigens are induced by exposure to antibodies.19 The combination of specific and nonspecific treatments had an additive effect in reducing complement fixation. The combined strategy reduced the degree of cross-match positivity of all PRBC samples. Treated blood from some pigs appeared to be ‘‘compatible’’ with some patients, according to the clinical test used. However, immunoglobulin M binding was not eliminated. This suggests that treated PRBCs would still induce an immune response, which would be most pronounced on repeat exposure. In previous experiments, response of human sera to porcine RBCs varied with human blood type.10 However, in this study anti-human globulin augmented polyacrylamide gel cross-match in this study showed strong agglutination when sera of human blood types A, B, and O was mixed with untreated PRBCs. Clinically, the sensitivity of the gel test to clinically exclude donor-recipient

Doucet et al 185

Surgery Volume 135, Number 2

matches with major blood group incompatibility exceeds 99%.21 All untreated PRBCs were maximum positive cross-matched with human serum, regardless of ABO type. However, when treatment of PRBC reduced the reaction with human serum, the varying degrees of cross-match positivity from each blood group became apparent, with types A, B, and O appearing in descending order. These differences may be due to different impacts of treatment on PRBC antigens of interest to several human antibodies. Treatment of PRBC reduced reactivity to a greater extent for human type B serum than type A serum, which would suggest that the targets of anti-A antibody were affected to a greater extent than those of anti-B antibody. Pigs have been characterized as A-like or O-like blood type according to their responsiveness to human type A reagent.22 In humans, the expression of blood group substances is variable so that the concentration of A antigen is much less on blood group A2 compared with group A1. In a similar fashion, the concentration of blood group antigen on the cell surface is far greater for group B than group A. The expression of a-gal is variable on PRBCs, but we selected low expressor animals for this experiment and used saturation exposure to a-galactosidase. Little is known regarding the variation in non a-gal antigens on PRBCs and their relationship to the human ABO substances or whether this type of variation would explain our findings. Human blood group B substance is identical to the a-gal antigen, except that the penultimate galactose is fucosylated. Humans of blood type B or AB have naturally occurring antibodies to a-gal that do not cross-react with B antigen.23 On the other hand, those antibodies of blood type A or O (non– B expressors) have a single antibody that has a shared specificity for B and a-gal antigens.22 B antigen expressors (blood type B or AB) have lower titers than non–B expressors of anti–a-gal immunoglobulin G.24,25 The greatest reduction in augmented cross-match reactivity occurred for human blood group O sera, compared with groups A and B. Group O serum differs from groups A and B in it has more anti-A and anti-B immunoglobulin G than other groups.26 The predominant anti–agal antibody class is thought to be immunoglobulin M, and it does not vary between the human blood types.27 Treatment with a-galactosidase and PEG was more effective in preventing immunoglobulin M, compared with immunoglobulin G, binding and markedly reduced the degree of cross-match positivity of PRBC with human serum. In the studies of ABO incompatible allotransfusion, the finding that some individuals were

cross-match negative to altered ABO-incompatible RBC was a prelude to a clinical trial with increasingly large transfusions.28,29 Although there is a reasonable chance that a maiden xenotransfusion with compatible combination-treated PRBCs would not be hemolysed in humans, a clinical trial is not yet envisaged because the risks of infection and sensitization are not understood conclusively. A recent study of PRBC xenotransfusion in a primate model showed that a-galactosidase treatment prolonged the survival of transfused PRBCs.30 However, the lifespan of the xenotransfused cells was much reduced compared with autologous transfusion. This may be due to residual immunity, splenic sequestration, and PRBC fragility. In our study, the addition of PEG treatment to agalactosidase treatment further reduced complement fixation, which may prolong RBC survival. Additional strategies may well be required to prevent in vivo destruction, despite in vitro compatibility. The enzymatic and chemical treatments that were used here are compatible with other donordirected strategies that are being investigated currently to allow xenotransplantation. These strategies include the use of transgenic technology on swine to knock-out a-gal, to convert a-gal to H (blood group 0) substance or to add human complement regulatory proteins such as decay accelerating factor.31,32 PRBCs are more susceptible to osmotic lysis than human RBCs.33 Treatment with PEG did not increase the robustness of the cells. This may prove to be a limiting factor for the handling and storage of PRBCs. We have begun investigations of alternative animal sources of RBC, with profiles more suitable for phlebotomy and storage.33 RBCs possess advantages as candidate grafts to develop xenotransplantation. Being nonnucleated, they are less likely to harbor certain viruses and will be less able to overcome antigen-directed strategies by the creation of new antigen. In addition, expectations of graft survival would be considerably shorter for xenotransfusion than for organ or cellular xenotransplantation. However, the excellent quality of clinical allotransfusion requires that alternative strategies meet or exceed that standard of quality before implementation.

REFERENCES 1. Winslow RM. Artificial blood: ancient dream, modern enigma. Nat Biotechnol 1998;16:621-2. 2. Kawano T, Hosoya H. Oxidative burst by acellular haemoglobin and neurotransmitters. Med Hypotheses 2002;59: 11-5.

186 Doucet et al

3. Cohn SM. Blood substitutes in surgery. Surgery 2000; 127:599-602. 4. Paradis K, Langford G, Long Z, Heneine W, Sandstrom P, Switzer WM, et al. Search for cross-species transmission of porcine endogenous retrovirus in patients treated with living pig tissue. Science 1999;285:1236-41. 5. Blusch JH, Patience C, Martin U. Pig endogenous retroviruses and xenotransplantation. Xenotransplantation 2002; 9:242-51. 6. Galili U, Rachmilewitz EA, Peleg A, Flechner I. A unique natural human IgG antibody with anti-alpha-galactosyl specificity. J Exp Med 1984;160:1519-31. 7. Blanken WM, Van den Eijnden DH. Biosynthesis of terminal Gal alpha 1–3Gal beta 1–4GlcNAc-R oligosaccharide sequences on glycoconjugates. J Biol Chem 1985;260:12927-34. 8. Watkins WM. Biochemistry and genetics of the ABO, Lewis, and P blood group systems. Adv Hum Genet 1980;10:1-136, 379-85. 9. Alvarado CG, Cotterell AH, McCurry KR, Collins BH, Magee JC, Berthold J, et al. Variation in the level of xenoantigen expression in porcine organs. Transplantation 1995;59: 1589-96. 10. MacLaren L, Lee TD, Anderson D, Nass M, McAlister VC. Variation in porcine red blood cell alpha-galactosyl expression and agglutination by human serum. Transplant Proc 1998;30:2468-9. 11. MacLaren LA, Riggs CM, Johnstone J, Doucet J, McAlister VC. Evaluating porcine red blood cell and platelet a-galactosyl expression. Transfusion 2002;42:1184-8. 12. Goldstein J, Siviglia G, Hurst R, Lenny L, Reich L. Group B erythrocytes enzymatically converted to group O survive normally in A, B, and O individuals. Science 1982;215: 168-170. 13. Abuchowski A, van Es T, Palczuk NC, Davis FF. Alteration of immunological properties of bovine serum albumin by covalent attachment of polyethylene glycol. J Biol Chem 1977;252:3578-81. 14. Scott MD, Murad KL, Koumpouras F, Talbot M, Eaton JW. Chemical camouflage of antigenic determinants: stealth erythrocytes. Proc Natl Acad Sci U S A 1997;94:7566-71. 15. Jeong ST, Byun SM. Decreased agglutinability of methoxypolyethylene glycol attached red blood cells: significance as a blood substitute. Artif Cells Blood Substit Immobil Biotechnol 1996;24:503-11. 16. Zhu A. Binding of human natural antibodies to non-alphagal xenoantigens on porcine erythrocytes. Transplantation 2000;69:2422-8. 17. Buhler L, Xu Y, Li W, Zhu A, Cooper DK. An investigation of the specificity of induced anti-pig antibodies in baboons. Xenotransplantation 2003;10:88-93. 18. LaVecchio JA, Dunne AD, Edge AS Enzymatic removal of alpha-galactosyl epitopes from porcine endothelial cells diminishes the cytotoxic effect of natural antibodies. Transplantation 1995;60:841-7.

Surgery February 2004

19. Luo Y, Wen J, Luo C, Cummings RD, Cooper DK. Pig xenogeneic antigen modification with green coffee bean alpha-galactosidase. Xenotransplantation 1999;6:238-48. 20. Bradley AJ, Murad KL, Regan KL, Scott MD. Biophysical consequences of linker chemistry and polymer size on stealth erythrocytes: size does matter. Biochim Biophys Acta 2002;1561:147-58. 21. Langston MM, Procter JL, Cipolone KM, Stroncek DF. Evaluation of the gel system for ABO grouping and D typing. Transfusion 1999;39:300-5. 22. Oriol R. Tissular expression of ABH and Lewis antigens in humans and animals: expected value of different animal models in the study of ABO-incompatible organ transplants. Transplant Proc 1987;19:4416-20. 23. Galili U, Buehler J, Shohet SB, Macher BA. The human natural anti-gal IgG: III, the subtlety of immune tolerance in man as demonstrated by crossreactivity between natural anti-gal and anti-B antibodies. J Exp Med 1987;165:693-704. 24. McMorrow IM, Comrack CA, Nazarey PP, Sachs DH, DerSimonian H Relationship between ABO blood group and levels of -gal alpha, 3-galactose-reactive human immunoglobulin G. Transplantation 1997;64:546-9. 25. Neethling FA, Cooper DK. Serum cytotoxicity to pig cells and anti-alpha-gal antibody level and specificity in humans and baboons. Transplantation 1999;67:658-65. 26. Rawson AJ, Abelson NM. Studies of blood group antibodies: IV, physico-chemical differences between iso-anti-A, B and iso-anti-A or iso-anti-B. J Immunol 1960;85:640-7. 27. Thibaudeau K, Anegon I, Lemauff B, Soulillou JP, Blanchard D. Human natural antibodies to porcine platelets. Transplantation 1994;57:1110-5. 28. Lenny LL, Hurst R, Goldstein J, Benjamin LJ, Jones RL. Single-unit transfusions of RBC enzymatically converted from group B to group O to A and O normal volunteers. Blood 1991;77:1383-8. 29. Lenny LL, Hurst R, Zhu A, Goldstein J, Galbraith RA. Multiple-unit and second transfusions of red cells enzymatically converted from group B to group O: report on the end of phase 1 trials. Transfusion 1995;35:899-902. 30. Eckerman JM, Buhler L, Awwad M, Zhu A, Cooper DKC. Alpha-galactosidase-treated pig red blood cell transfusion in primates: initial studies [abstract]. Transplantation 2002; 74:630. 31. Cozzi E, Bhatti F, Schmoeckel M, Chavez G, Smith KG, Zaidi A, et al. Long-term survival of nonhuman primates receiving life-supporting transgenic porcine kidney xenografts. Transplantation 2000;70:15-21. 32. Costa C, Zhao L, Burton WV, Rosas C, Bondioli KR, Williams BL, et al. Transgenic pigs designed to express human CD59 and H-transferase to avoid humoral xenograft rejection. Xenotransplantation 2002;9:45-57. 33. Johnstone JE, Maclaren L, Doucet J, McAlister VC. In-vitro studies regarding the feasibility of bovine erythrocyte xenotransfusion. Xenotransplantation. In press.