Mauro S. Sandrin

Galα(1,3)Gal, the Major Xenoantigen(s) Recognised in Pigs by Human Natural Antibodies

The transplantation of pig organs to humans (xenotransplantation) is now receiving serious consideration because of the shortage of human donors for organ transplants of kidney, liver and heart, and of islet cell transplantation for diabetes. The problem with such xenografts would be hyperacute rejection--mediated by natural antibodies in humans to pig antigens, complement fixation to endothelial cells, and the rapid onset of intravascular coagulation. It is now clear that the major target of the natural IgM and IgG antibodies is the terminal carbohydrate epitope Gal alpha(1,3)Gal, formed by the alpha 1,3galactosyl transferase, which places a terminal galactose residue in an alpha-linkage to another galactose. The alpha 1,3galactosyl transferase in the pig gives rise to very high endothelial cell expression of Gal alpha(1,3)Gal, a ready explanation for the hyperacute rejection of vascularized organs. In addition the parenchuma of liver and kidneys have high levels of Gal alpha-(1,3)Gal. These tissues will all fail in a pig-to-human transplant in what can now be precisely defined in terms of antigen and antibody. We have already made some suggestions for removal of anti-Gal alpha(1,3)Gal antibodies and if the procedure were technically feasible xenotransplantation could be attempted now, especially in patients doomed to a certain death because of the absence of a donor (especially for liver where ex vivo perfusion could be performed). However, the immune system is far from simple, as is shown by the healthy status of mice lacking MHC Class I, Class II or both Class I & II molecules. Perhaps the curtain is about to go up to reveal a new scene! Islets differ from the other tissues and may well not undergo acute antibody-mediated hyperacute rejection--it will be of interest to see how these fare in xenotransplantation models or even in patients. Again, normal individuals do not have anti-islet antibodies; but a proportion of diabetic patients do have such antibodies--whether these will cause hyperacute or acute rejection or are markers of immunity of T-cell type, remains to be seen. Whatever, the area is exciting, is progressing rapidly and, as indicated elsewhere, within a few years we should know whether modified pig tissue can be grafted to some patients. The isolation of the cDNA clone encoding the pig alpha 1,3 galactosyl transferase is an essential first step in the production of a transgenic pig lacking the alpha 1,3Galactosyltransferase and therefore the Gal alpha(1,3)Gal epitope, and such animals could serve as donor for human transplantation.

Distribution of the major xenoantigen (gal(α1–3)gal) for pig to human xenografts

We have previously demonstrated that the major epitope in pig tissues detected by naturally occurring human IgM antibodies is galactose (alpha 1-3)galactose. Subsequent biochemical studies demonstrated this epitope to be present on molecules (Mr40-220kDa) on both endothelial cells and lymphocytes. The objective of the present study was to define the distribution of gal(alpha 1-3)gal in different pig tissues, concentrating on those of relevance for the potential transplantation of pig organs or tissues to humans. Adult pig tissues were obtained fresh, fixed, and stained by the immunoperoxidase technique using biotinylated Griffonia simplicifolia lectin (IB4) which binds only to gal(alpha 1-3)gal, and examined histologically. Endothelial cells in all small vessels (capillaries, arterioles and venules) had a unifrom and dense expression of gal(alpha 1-3)gal; in larger vessels, like the aorta, they were less reactive. The highest concentrations were found in the liver parenchyma which stained uniformly, and in the kidney, where the highest amounts were found in the brush border of the proximal convoluted tubules. There was no staining of collecting ducts or glomeruli (except for endothelium) and moderate staining of the distal convoluted tubules. Heart muscle was nonreactive, although the high density of capillaries indicated a reasonable content of gal(alpha 1-3)gal. In contrast to these tissues was the distribution in the pancreas, which, apart from vessels and the lining of ducts, was nonreactive, i.e. islet cells were essentially lacking in gal(alpha 1-3)gal. Other tissues such as the lung contained moderate amounts of material lining the alveoli and bronchioles.(ABSTRACT TRUNCATED AT 250 WORDS)

Switching amino-terminal cytoplasmic domains of alpha(1,2)fucosyltransferase and alpha(1,3)galactosyltransferase alters the expression of H substance and Galalpha(1,3)Gal.

When α(1,2)fucosyltransferase cDNA is expressed in cells that normally express large amounts of the terminal carbohydrate Galα(1,3)Gal, and therefore the α(1,3)galactosyltransferase (GT), the Galα(1,3)Gal almost disappears, indicating that the presence of the α(1,2)fucosyltransferase (HT) gene/enzyme alters the synthesis of Galα(1,3)Gal. A possible mechanism to account for these findings is enzyme location within the Golgi apparatus. We examined the effect of Golgi localization by exchanging the cytoplasmic tails of HT and GT; if Golgi targeting signals are contained within the cytoplasmic tail sequences of these enzymes then a “tail switch” would permit GT first access to the substrate and thereby reverse the observed dominance of HT. Two chimeric glycosyltransferase proteins were constructed and compared with the normal glycosyltransferases after transfection into COS cells. The chimeric enzymes showed Km values and cell surface carbohydrate expression comparable with normal glycosyltransferases. Co-expression of the two chimeric glycosyltransferases resulted in cell surface expression of Galα(1,3)Gal, and virtually no HT product was expressed. Thus the cytoplasmic tail of HT determines the temporal order of action, and therefore dominance, of these two enzymes.

Isolation and characterization of cDNA clones for Humly9: the human homologue of mouse Ly9

Ly9 is a mouse cell membrane antigen found on all lymphocytes and coded for by a gene that maps to chromosome 1. We previously described the isolation and characterization of a full-lenght cDNA clone for mouse Ly9. Using cross-species hybridization we isolated cDNA clones encoding the human homologue Humly9. Analysis of the predicted protein sequence suggests that the extra-cellular portion of the Humly9 molecules is composed of four Ig-like domains: a V domain (V) without disulphide bonds and a truncated C2 domain (tC2) with two disulphide bonds, a second V domain without disulphide bonds and a second tC2 with two disulphide bonds, i.e., as V-tC2-V-tC2. The gene encoding Humly9 was mapped to chromosome 1 by analysis of human/hamster hybrids, and more specifically to the 1q22 region by in situ hybridization. The protein sequence data support the view that Humly9 belongs to the immunoglobulin-superfamily subgroup which includes CD48, CD2, and LFA-3.

Reduction of the major porcine xenoantigen Galoc(1,3)Gal by expression of α(1,2)fucosyltransferase

Abstract: Although removal of the Galα(1,3)Gal antigen from pigs would prevent hyperacute graft rejection, the technique of homologous recombination to knock out the α 1,3 galactosyltransferase gene is not available for pigs, and an alternative strategy is presented. As both α 1,3 galactosyltransferase and α 1,2 fucosyltransferase use the same substrate (N‐acetyl lacto‐samine), competition between the transferases in vitro and in vivo was examined. The data show that there is indeed a hierarchy of these gly‐cosyltransferases competing for the same substrate, and that α 1,2 fuco‐syltransferase takes precedence over α 1,3 galactosyltransferase: a) COS cells simultaneously transfected with cDNA clones encoding α, 2 fuco‐syltransferase and α 1,3 galactosyltransferase show preferential expression of the H substance (synthesised by α 1,2fucosyltransferase) rather than Galα(1,3)Gal (synthesised by α 1,3galactosyltransferase), even though α 1,3galactosyltransferase mRNA and functional enzyme was present, b) In a pig kidney cell line that expressed both the Galα(1,3)Gal and H, the increased expression of H induced by the transfection and stable expression of α 1,2fucosyltransferase resulted in decreased expression of Galα(1,3)Gal. c) Coexpression of α 1,2fucosyltransferase and α 1,3galactosyltransferase in either COS cells or the pig cell line resulted in decreased human antibody binding and complement‐mediated cell lysis, d) Transgenic mice, ubiquitously expressing α 1,2fucosyltransferase show a major decrease in Galα‐(1,3)Gal expression and a decrease in natural human antibody binding. These findings have important implications for xenotransplantation in that α,2fucosyltransferase transgenic pigs could be a source of donors for xenotransplantation to humans.

Mutation in the ia subregion of the murine h-2 complex.

The I region of the major histocompatibility complex (MHC) of the mouse contains genes coding for a number of functions: Ir genes, Ia specificities, histocompatibility genes, and other products (Shreffler and David 1975). The interrelationships of these genes, particularly of Ir genes and cell-membrane Ia antigens, has been studied mainly by the use of H-2 recombinant strains and, by this means, a large number of complexities in the expression of lr-gene functions has been revealed (Shreffler and David 1975, Klein 1975). A second approach to the fine dissection of these genetic relationships would involve studies of / region mutations such as those already used in studies of the H-2K, H-2D, and H-2L alloantigens. Of the 22 H-2 mutations previously described, all have involved H-2K, H-2D and H-2L loci IKohn et al. 1978) except one. We report here the first H-2 mutation involving the I -A subregion. The B6.C-H-2 bmaz congenic mutant strain carries a spontaneous mutation (H-2 bma2, formerly H-2 bin) which originally occurred in the H-2 b haplotype of a (C57BL/6Kh x BALB/cKh)F1 female mouse. The mutation is of the gain and loss type, as parental and mutant strains reciprocally reject skin grafts within 14 davs a n d gene complementation studies mapped the bm12 mutation to the K I A region of the H-2 complex, although the ability of the H 2 K b and bml2 mutants to complement each other for C57BL/6 grafts indicated that the mutation had not occurred in the H 2 K b gene (McKenzie et al. 1979) Serological studies indicated that the H-2K/D specificities of bml 2 were the same as those of C57BL/6 including H-2.33, the private H-2K b specificity, although the initial studies suggested that H-2.33 was defective in bml2 as the antisera used contained both anti-H-2.33 and anti-Ia antibodies. However, subsequent studies indicated that the bml2 mutant lacked Ia specificities (McKenzie et al. 1979). The C57BL/6 (H-2 b] strain carries the Ia specificities 3, 8, 9, 15, 20, all coded for by the I -A subregion Ino I-E/C products

Fetal pig islet xenografts in NOD/Lt mice: The effect of peritransplant anti‐CD4 monoclonal antibody and graft immunomodification on graft survival, and lack of expression of Gal(α1–3)Gal on endocrine cells

Abstract: Transplantation of tissues that are vascularized by ingrowth of host vessels, as opposed to primarily vascularized grafts, may be one way of avoiding hyperacute rejection (HAR). Pretransplant treatment of the graft to eliminate from it highly immunogenic cells may also improve graft survival. We examined organ cultured fetal pig pancreas in prediabetic NOD/Lt female mice treated with peritransplant anti‐CD4 MAb (GK 1.5) and in separate experiments also studied the expression of Gal(α 1–3)Gal, an epitope that may be a major target of a human anti‐pig response. Immunocytochemistry to detect insulin, glucagon, and somatostatin and Gal(α 1–3)Gal showed that differentiated (i.e., hormone containing) endocrine cells were Gal(α 1–3)Gal negative but hormone‐negative ductal cells showed strong Gal(α1–3)Gal staining on their luminal border and interstitial cells were also positive particularly early after transplantation. Rejection occurred in all animals, but its rate was markedly altered by transient immunosuppression with the depleting anti‐CD4 MAb. Cell infiltration was first detected in controls 2 days posttransplant, was pronounced by 4 days, and grafts showed advanced rejection by 7 days whether or not they had been “immunomodified” by 2 days exposure to 90% O2. In contrast, peritransplant immunosuppression with GK 1.5 improved graft survival but rejection was advanced by 28 days. The control graft sites had many eosinophils, macrophages and mast cells by 4 days, but little infiltration was seen until after 14 days in the immunosuppressed mice and was predominantly by mononuclear cells. Use of “immunomodified” grafts was of no benefit. Thus, graft immunomodification per se, that often allows islet allografts to survive in rodents, had no effect on xenograft survival in this model. A major potential target antigen for natural Ab in humans, Gal(α 1–3)Gal, is not present on differentiated graft endocrine cells.

The human Ia system: Definition and characterization by xenogeneic antisera

The production of xenogeneic anti-Ia serum against Ia antigens in serum has been previously described in the mouse and we now describe the production of xenogeneic anti-human Ia antisera using similar methods. With an indirect resetting technique, Ia-like antibodies were shown to react with the majority of B cells (95%), a subpopulation of T cells, with carbonyl iron adherent cells, and with some E−Ig− null cells, but there was no reaction with red cells and platelets. These reactions were the same as those obtained with DRW antisera using cytotoxicity testing. In addition, antigens detected with xenogeneic antisera were also found in serum, where they were found to exist in a low molecular weight, dialyzable form. By the selective removal of different cell surface markers by cocapping, no association could be found with the specifities detected with the xenogeneic anti-Ia antisera and with surface Ig,β2-microglobulin, or HLA-A and B specificities. Alloantibodies to DRW specificities (but not HLA-A, B specificities) were able to specifically block the binding of the rabbit anti-Ia antibodies to B cells, and reciprocal blocking of rabbit antisera by DRW antibodies was also observed. Several xenogenic antisera were produced by immunizing rabbits with the serum of different individuals. Each antiserum was shown to contain a number of different specificities, as they gave different reaction patterns with different individuals when testing was done both directly and by absorption. These xenogeneic anti-la sera also segregated in a family with HLA-A and B specificities. The detection of a polymorphic antigenic system segregating with the HLA complex, distinct from HLA-A and B specificities, and whose antigens occur predominantly on B cells is therefore described. Because of the similarity of the reactions of the xenogeneic antisera in man to those found in the mouse, and because of the close relationship to the DRW specificities, the system has been provisionally called the ‘H.Ia’ system.

SEQUENCE AND STRUCTURE OF THE MOUSE THB GENE

The mouse Thb locus is located on chromosome 15 and encodes low molecular weight, phosphatidyl inositol-anchored cell surface glycoproteins with unique expression on B cells, thymocytes, and a subset of thymic epithelial cells (Ledbetter and Herzenberg 1979; Matossian-Rogers et al. 1982; Steirnberg et at. 1987; Gumley et al. 1992). Thb is tightly linked to the Ly-6 loci, and the Ly-6 gene products (Ly-6A/E, Ly-6B, Ly-6C, Ly-6F, Ly-6G, and Sca 2) are structurally related to ThB by amino acid homology, including 10 common cysteine residues which are located in highly conserved positions (Gumley et al. 1992). This homology extends to similar molecular sizes (12000-18000 Mr and glycosylation patterns, a phosphatidyl inositol linkage to the cell surface and similar tissue distribution profiles (Shevach and Korty 1989). Southern blot analysis of genomic DNA probed with ThB cDNA demonstrates the presence of a single gene locus (Gumley et al. 1992), in contrast to Ly-6 cDNA, which detects at least 30 tightly linked genes, including Ly-6A/E, Ly-6C, Ly6F, and Ly-6G (LeClair et at. 1986; Kamiura et al. 1992; Fleming et at. 1993). We now report on the genomic sequence and structure of the Thb locus. To examine the organization of Thb, a 10.6 kilobase (kb) ThB genomic clone (pgThB-7.2) was isolated from a BALB/c genomic library using the ThB cDNA clone, pThB-A, as a probe (Gumley et al. 1992). A restriction fragment pattern identical to that observed with BALB/c genomic DNA was obtained after probing Eco RI-, Hin dIII-, Pst I-, and Barn HI-digested pgThB-7.2 DNA with pTbB-A cDNA (Gumley et at. 1992). In particular, after digestion of pgThB-7.2 with Eco R1, two fragments of approximately 4.2 kb and 2.2 kb hybridized with the pThBA probe, and there were additional 2_8 kb and 1.4 kb

Production of a cytotoxic Ia.6-specific antibody

The I region of the murine H-2 complex is currently divided into five subregions, I-A, I-B, I-J, I-E, and l-C, each defined by either an immune response (It) gene or by serologically detected, cell-surface Ia antigenic specificities (McKenzie et al. 1981). The I-C subregion has been defined by the Ia.6 specificity, detected as a cytotoxic antibody present in the antiserum B10.A(4R) anti-B10.A(2R), and initially this antiserum was reported to react only with the d and p H-2 haplotypes (David et al. 1974). The Ia.7 specificity was later also mapped to the I-C subregion and the Ia-3 gene was postulated to control both these specificities (David and Shreffler 1974). More recently, the I-E subregion has been defined by the description of the Ia.22 specificity of the k haplotype (David et al. 1978, Sachs 1978), and specificity Ia.7 was later shown by coprecipitation to be present on the same molecule as Ia.22 (Cullen et al. 1976), leaving only the Ia.6 specificity to define the I-C subregion. In the initial description, Ia.6 was found to be present on T cells (Okuda and David 1978), unlike the I-A and I-E specificities which are found on B cells. However, because of several difficulties, the separate definitions of I-C and I-E subregions have become blurred and these problems have led many investigators to refer to these regions as the I-E/C subregion rather than as two separate subregions. The major problem, encountered by several laboratories, has been to reproduce Davids original antiserum containing the Ia.6 antibody (David et al. 1978, Okuda and David 1978, Sachs 1978). However, other studies have pointed to the existence of the I-C subregion. For example, immune response genes (Dorf and Benacerraf 1975), alloantigen-mediated, mixed lymphocyte response suppressor factor(s) (Rich and Rich 1976), determinants on Fc-receptor-positive T cells (Stout et al. 1977), control of immune suppression (Benacerraf and Germain 1978), mixed-lymphocyteresponse-stimulating determinants (Okuda and David 1978), and a histocompatibility gene (McKenzie and Henning 1976) have all been mapped to the I-C