HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Borenstein, S. H. Articles by Chamberlain, J. W. Search for Related Content PubMed PubMed Citation Articles by Borenstein, S. H. Articles by Chamberlain, J. W.

CD8⁺ T Cells Are Necessary for Recognition of Allelic, But Not Locus-Mismatched or Xeno-, HLA Class I Transplantation Antigens

Steven H. Borenstein^*, Jeremy Graham, Xiao-Li Zhang and John W. Chamberlain^2,*,,

^* Institute of Medical Science and Department of Immunology, University of Toronto; and Research Institute, Program in Infection, Immunity, Injury, and Repair, The Hospital for Sick Children, Toronto, Canada

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
Although HLA transgenic mice (HLA TgM) could provide a powerful approach to investigate human MHC-specific T cell responsiveness, the extent to which these molecules are recognized by the mouse immune system remains unclear. We established TgM expressing HLA class I alleles A2, B7, or B27 in their fully native form (HLA^nat) or as hybrid molecules (HLA^hyb) of the HLA 1/2 domains linked to the H-2K^b 3, transmembrane, and cytoplasmic domains (i.e., to maintain possible species-specific interactions). Comparison of each as xeno- (i.e., by non-TgM) vs allo- (i.e., by TgM carrying an alternate HLA allele) transplantation Ags revealed the following: 1) Although HLA^hyb molecules induced stronger xeno-CD8⁺ T cell responses in vitro, additional effector mechanisms must be active in vivo because HLA^nat skin grafts were rejected faster by non-TgM; 2) gene knockout recipients showed that xenorejection of HLA^nat and, unexpectedly, HLA^hyb grafts doesn’t depend on CD8⁺ or CD4⁺ T cells or B cells; 3) each HLA^hyb strain developed tolerance to "self" but rejected allele- (-B27 vs -B7) and locus- (-B vs -A) mismatched grafts, the former requiring CD8⁺ T cells, the latter by CD8⁺ T cell-independent mechanisms. The finding that recognition of xeno-HLA^hyb does not require CD8⁺ T cells while recognition of the identical molecule in a strictly allo context does, demonstrates an 1/2 domain-dependent difference in effector mechanism(s). Furthermore, the CD8⁺ T cell-independence of locus-mismatched rejection suggests the degree of similarity between self and non-self 1/2 determines the effector mechanism(s) activated. The HLA Tg model provides a unique approach to characterize these mechanisms and develop tolerance protocols in the context of human transplantation Ags.

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
The MHC class I and II molecules are highly polymorphic cell surface glycoproteins centrally involved in development and responsiveness of the immune system through their capacity to present self and foreign Ag-derived peptides to T lymphocytes (1, 2). MHC class I molecules (HLA-A, -B, and -C in humans; H-2K, -D, and -L in mice) present peptides to CD8⁺ (cytotoxic) T cells, whereas MHC class II molecules (HLA-DR, -DP, and -DQ in humans; I-A and I-E in mice) present peptides to CD4⁺ (helper) T cells (1, 3, 4). Class I molecules are expressed in association with a smaller nonpolymorphic chain called ß₂-microglobulin (ß₂m)³ on most somatic cells at levels that vary significantly between tissues and cell types (5, 6). Class II molecules are found mainly on B cells, macrophages, and other APCs as well as on thymic stromal cells.

Although conventional T cell recognition of foreign (i.e., viral) Ag is self-MHC- restricted (7), CD8⁺ and CD4⁺ T cells also respond vigorously when confronted with allogeneic cells expressing non-self class I and II molecules, respectively. This potent cell-mediated immune response is believed to be the primary event involved in allograft rejection, and results from the high frequency of T cells that are alloreactive (i.e., 1–10% of the T cell repertoire) (8, 9). Although it was long believed that such cells directly recognized polymorphic differences between MHC alleles, now known to reside largely within the Ag-binding cleft (10, 11, 12), more recent evidence also suggests an important role for indirect recognition of donor MHC (or non-MHC) Ag-derived peptides in association with recipient MHC (13, 14, 15). In addition, the finding that different MHC alleles have different peptide-binding specificities (16) suggests that some alloreactive T cells may be influenced by the distinct array of self-peptides presented by foreign MHC molecules at the surface of transplanted tissues (9). Although both the direct and indirect pathways are now generally believed to be important, the relative contribution of each to allograft rejection in vivo is not clear.

In contrast to allografts, the role of the T cell response in rejection of grafts between species (i.e., xenografts) is much less characterized, in part because it occurs only after two earlier stages of rejection, namely, hyperacute rejection and delayed acute vascular rejection (17, 18, 19, 20, 21, 22). Studies conducted in vitro circumvent these earlier events and have generally detected a very low T cell response to xenogeneic cells (21, 23). This could be interpreted as resulting from either a low frequency of T cells bearing TCRs able to interact with the polymorphic domains of xenogeneic MHC and/or inefficient cross-species interactions involving coreceptor, costimulatory or accessory molecules, or other incompatibilities between species that influence T cell responsiveness. Given the multiple molecular interactions necessary for T cell activation to foreign Ag, it has been difficult to distinguish which of these is correct. This is important to resolve because, although progress has been made toward eliminating the hyperacute response (22), successful in vivo xenotransplantation protocols will require overcoming all three barriers.

Most investigations of the mechanisms of allograft rejection have been conducted with mouse strains differing at multiple major (class I and II) and/or minor MHC loci. With the exception of the nonallelic (i.e., mutant) class I (e.g., K^bm1) or class II molecules (24) and a limited number of H-2 transgenes (25, 26), few studies have addressed the influence of individual natural class I or II alleles on their own to mediate rejection. It is also unclear whether individual mouse or human alleles are functionally equivalent as transplantation Ags or whether a hierarchy exists. There are also a number of questions that need to be clarified about the cellular response to xenografts, including how complex the xenoreactive T cell population is and whether anti-xeno-MHC T cells respond to the same types of allelic polymorphisms seen by alloreactive T cells or whether there are other species-specific nonallelic differences that dominate. It will also be important to determine whether xeno-MHC molecules induce additional non-T cell effector mechanisms that contribute to rejection.

With these issues in mind, and because current mouse models are limited in what they can reveal about T cell recognition in the context of the human MHC molecules, we and others have explored the possibility that HLA class I and II molecules expressed in transgenic (Tg) mice might provide a useful model for studying HLA-dependent immune function in vivo (5, 27, 28, 29, 30, 31, 32, 33). For instance, characterization of the non-Tg mouse T cell response to human MHC expressed in tissues of otherwise genetically identical HLA Tg mice (TgM) should make it possible to identify the specific immune mechanisms involved in xeno-MHC recognition and rejection as well as the importance of MHC-dependent vs -independent interactions in the apparent reduced xenogenic cellular response detected in vitro. Furthermore, comparison of the non-Tg (i.e., xeno-) and HLA Tg (i.e., allo-) T cell responses to alternate (non-self) HLA Tg alleles could provide a unique approach to compare the structures recognized by T cells that respond to xeno- vs allo-MHC. However, despite some efforts in these directions, the extent to which Tg HLA molecules are functionally recognized by non-TgM or TgM T cells remains relatively unclear. Some reports suggest that fully human (native) HLA class I molecules are recognized only inefficiently at best by the mouse immune system as either xeno- or alloantigens, or as restriction elements (30, 31, 34, 35, 36, 37, 38, 39). In contrast, other reports, often using apparently similar alleles and strains of mice, suggest that Tg HLA class I molecules are recognized by the mouse immune system and T cells much the same as alternate mouse class I alleles (28, 29, 32, 33, 40, 41, 42, 43, 44, 45, 46, 47, 48).

It is unclear whether these inconsistencies are actually due to differential function of distinct HLA alleles in the mouse background, as opposed to other quantitative or qualitative aspects of expression, or possibly differences in specific functional assays. To distinguish between these possibilities, with the longer-term objective being development of this model as an accurate reflection of HLA function in humans, we have established a panel of TgM that express the class I alleles HLA-A2, -B7, or -B27. One set of mice expresses the fully human native heavy chain in conjunction with human ß₂m (hß₂m), whereas the other set expresses a hybrid form of each allele, consisting of the exons encoding the human 1 and 2 polymorphic domains (i.e., the peptide binding cleft) linked to the 3, transmembrane, and cytoplasmic domains of the mouse H-2K^b protein. Development of the HLA/H-2K^b hybrid TgM was based on results of our own and others, which suggested the possibility that species-specific molecular interactions may influence how efficiently a human class I molecule expressed in Tg cells is able to undergo intracellular interactions (i.e., with ß₂m, chaperones) and transport, have access to a suitable array of self and foreign peptides, or interact with the T cell coreceptor CD8 at the cell surface (6, 27, 30, 36, 37, 49, 50). As at least some of these interactions are known to depend partly or completely on class I domains outside the Ag-binding 1/2 domains (i.e., ß₂m-3; CD8-3), the rationale for the HLA/H-2 hybrid construct was that the encoded molecule should retain the peptide-binding specificity of the human allele and be able to efficiently undergo these other interactions when expressed in a mouse background.

The studies in this paper investigate the immune mechanisms activated in vitro and in vivo by non-TgM and TgM in response to these three well-characterized human MHC class I alleles expressed as fully native HLA (HLA^nat) vs HLA/H-2 hybrid (HLA^hyb) Tg xeno- or allotransplantation Ags. The results demonstrate that distinct effector mechanisms are involved in recognition of, first, the identical HLA allele as a xeno- vs an allo-MHC molecule, and second, different locus-matched alleles (HLA-B7 vs -B27) vs locus-mismatched products (HLA-B vs HLA-A).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
Inbred, knockout (KO), and HLA Tg Mice

(B6/SJL)F₁ (H-2^b/s), C57BL/6J (BL/6) (H-2^b), DBA/2J (H-2^d), B6.CH-2bm1 (H-2K^bm1) (24), and IgM^-/- (H-2^b) (51) mice were obtained from The Jackson Laboratory (Bar Harbor, ME). CD8^-/- (52), CD4^-/- (53), and CD4^-/-CD8^-/- (54) mice were obtained from Dr. Tak Mak (Amgen Institute and Ontario Cancer Institute, Toronto, Ontario, Canada).

The DNA constructs used to generate the HLA TgM are shown in Fig. 1. For each construct, multiple founder mice were generated by microinjection of (B6/SJL)F₂ embryos (5, 55, 56), bred with non-Tg C57BL/6J mates to establish lines, and characterized with respect to Tg expression by RNA blot hybridization of tissue RNAs and by flow cytometry (5, 56). The HLA Tg lines used for these studies were selected on the basis of a normal breeding and transgene transmission rate, an appropriate tissue distribution of Tg HLA RNA (5, 56), and cell surface expression at a level similar to each other and to endogenous H-2 class I (5, 39). All constructs were genomic clones containing MHC class I exons, introns, and 5' and 3' flanking DNA in the genomic configuration. Each construct contained several hundred base pairs of HLA gene 5' flanking DNA, which we and others have shown previously to include all MHC class I cis-active transcriptional regulatory information sufficient to direct appropriate HLA class I Tg expression in TgM (55, 56). The HLA-B7^native (B7^nat) mice carried a 6.0-kb EcoRI-BamHI fragment encoding the fully human HLA-B7 molecule (5). The HLA-B7^nat/hß2m (B7^nat/hß2m) double TgM were derived by breeding the B7^nat mice above with TgM carrying a 14-kb PvuI-SaII fragment encoding hß₂m as described (5). The HLA-B27^nat/hß2m (B27^nat/hß2m) and HLA-A2^nat/hß2m (A2^nat/hß2m) mice were derived by coinjecting the human ß₂m gene together with either a 6.5-kb EcoRI fragment encoding HLA-B27 (57) or a 6-kb EcoRI fragment encoding HLA-A2 (58). For both the B27^nat/hß2m and A2^nat/hß2m Tg lines used here, the hß₂m and HLA genes were cotransmitted to offspring, indicating that both genes were cointegrated at a single chromosomal site.

View larger version (15K): [in this window] [in a new window]   FIGURE 1. Schematic representation of HLA class I native (HLAnat) and hybrid (HLAhyb) transgene constructs. The HLAnat class I -A2, -B7, and -B27 TgM were generated by pronuclear microinjection of (B6/SJL)F2 embryos with DNA constructs as represented in A encoding HLA-A2, -B7, or -B27, respectively. Each construct was fully human containing all exons, introns, 5' and 3' flanking DNA, and 5' transcriptional control sequences in their native genomic configuration. Aside from their natural allelic differences, the three constructs differed from each other only slightly in the amount of 5' and/or 3' flanking DNA. The HLA-A2nat construct was a 6.0-kb EcoRI fragment, whereas the -B7nat construct was a 6.0-kb EcoRI-BamHI fragment and the -B27nat construct was a 6.5-kb EcoRI fragment. Each of these HLAnat fragments was introduced into TgM along with a cloned genomic fragment encoding hß2m as described (see Materials and Methods for details) (5 ). The HLAhyb class I -A2/H-2Kb (or -A2hyb), -B7/H-2Kb (or -B7hyb), and -B27/H-2Kb (or -B27hyb) TgM were generated by pronuclear microinjection of (B6/SJL)F2 embryos with DNA constructs, as represented in B, encoding the hybrid proteins HLA-A2/H-2Kb (A2hyb), -B7/H-2Kb (B7hyb), or -B27/H-2Kb (B27hyb), respectively. The A2hyb construct contained HLA-A2 5' flanking DNA and -A2 exons 1–3 on a 2.68-kb fragment linked in intron 3 to a 2.35-kb fragment containing exons 4–8 and 3' flanking DNA of the self H-2Kb gene (the approximate position of the HLA/H-2Kb junction is indicated by a small vertical arrowhead). The B7hyb construct contained HLA-B7 5' flanking DNA and B7 exons 1–3 on a 1.9-kb fragment linked in intron 3 to the 2.35-kb exon 4–8 H-2Kb fragment. The B27hyb construct contained HLA-B27 5' flanking DNA and B27 exons 1–3 on a 1.7-kb fragment linked in intron 3 to the 2.35-kb exon 4–8 H-2Kb fragment. For both the native (A) and hybrid (B) constructs, the human exons are depicted as , the mouse H-2Kb exons as , and 5' and 3' flanking DNA and introns as a thin line. The exons 1–8 are numbered below the map together with the corresponding protein domains (1, 2, 3, transmembrane (Tm), and cytoplasmic (Cyt)). The scale is approximate.

All Tg strains were maintained by backcrossing to C57BL/6J (H-2^b) mates. The strains used in the experiments described here had been backcrossed at least 8–10 generations. The HLA-B7^hyb/CD8^-/- and HLA-B27^hyb/CD8^-/- TgM were generated by breeding the corresponding HLA^hyb+/-/CD8^+/+ line (-B7^hyb or -B27^hyb) with CD8^-/- (H-2^b) (52) mice and subsequently breeding the HLA^hyb+/-/CD8^+/- heterozygous offspring with CD8^-/- mice. All inbred, KO, and Tg mice were housed in a pathogen-free animal facility at The Hospital For Sick Children in accordance with the current regulations and standards of the Canadian Council of Animal Care.

Transfection of P815 cells

The DBA/2 (H-2^d)-derived mastocytoma cell line P815 (59) was cotransfected by electroporation of the bacterial neomycin (G418) resistance gene pHußpr-Neo (60) with individual HLA^nat or HLA^hyb gene constructs using a Bio-Rad gene pulser (Bio-Rad, Richmond, CA). Electroporated cells were cultured and selected in MEM supplemented with 5% newborn calf serum (Sigma, St. Louis, MO) and 200 µg/ml G418 (Life Technologies, Rockville, MD) at 37°C and 5% CO₂. Clones expressing high comparable levels of cell surface HLA for each transfected population were isolated by limiting dilution and characterized by flow cytometry.

Flow cytometry

The mAbs used, their specificities, and their sources were as follows: ME-1 (specific for HLA-B7, -B27, and -Bw22) was obtained from American Type Culture Collection (ATCC; Manassas, VA) (5, 61); MA2.1 (specific for HLA-A2 and -B17) was purchased from ATCC (62); B9.12.1 (pan-HLA class I-specific) was a generous gift of F. Lemmonier (Pasteur Institute, Paris, France; Ref. 63); and 28-14-8S (specific for D^b, L^d, and D^q) (64) was obtained from ATCC. For flow cytometry, single cell suspensions from lymph nodes were prepared, and 1 x 10⁶ cells were stained with the primary mAb for 45 min at 4°C in 75 µl of staining buffer (1x PBS containing 5% BSA and 0.1% sodium azide). Then, cells were washed and incubated for 30 min with FITC-conjugated F(ab')₂ goat anti-mouse IgG (Fc-specific) (Accurate Chemical and Scientific, Westbury, NY). Cells were fixed in 1% paraformaldehyde in PBS containing 0.1% sodium azide and then analyzed on a Becton Dickinson FACScan flow cytometer (Mountain View, CA).

Cell-mediated lympholysis (CML) assay

CML assays were performed essentially as described (65). Mouse responder lymph node cells (LNCs) (1 x 10⁴ to 3 x 10⁵ cells/well) were cultured in 96-well round-bottom microtiter plates (Nunc, Naperville, IL) with 3 x 10⁵ irradiated (2000 rad) stimulator cells for 5 days at 37°C and 5% CO₂ in MEM supplemented with 10% FCS (Sigma), 10 mM HEPES, 5 x 10^-6 M 2-ME, and penicillin/streptomycin (Life Technologies). Stimulator cells were obtained from the spleens of (B6/SJL)F₁ (H-2^bs), C57BL/6 (H-2^b), H-2K^bm1, and DBA/2J (H-2^d) mice or HLA^nat or HLA^hyb class I TgM. Targets were either spleen cells that had been stimulated for 3 days with Con A (see below) or HLA-transfected P815 cells. After 5 days of in vitro stimulation, 100 µl of the culture supernatant was removed from each well and ⁵¹Cr-labeled targets (3 x 10³) were added to the effector cells at the ratios indicated in a 200-µl final volume for 4 h at 37°C. Subsequently, 100 µl of supernatant was removed from each well and counted on a Wallac gamma counter (Gaithersburg, MD). Specific lysis was calculated as [(experimental - spontaneous release)/(maximal - spontaneous release)] x 100%. Spontaneous release was estimated by adding 3 x 10^{3 51}Cr-labeled target cells to 100 µl of media containing stimulator but not responder cells. Maximal release was estimated by adding 3 x 10^{3 51}Cr-labeled target cells to 100 µl of media containing 1% acetic acid.

⁵¹Cr labeling of target cells

Con A-stimulated lymphoblasts were generated by incubating 1 x 10⁷ spleen-derived cells at 37°C, 5% CO₂ for 2–3 days in a vertical T-25 flask in 10 ml of MEM containing 10% FCS (Sigma), 10 mM HEPES, 5 x 10^-6 M 2-ME, penicillin/streptomycin, and 2 mg/ml Con A (generously provided by Dr. R. Miller, Ontario Cancer Institute). HLA-transfected P815 cells were cultured in T-25 Falcon flasks (Becton Dickinson, Franklin Lakes, NJ) at 37°C and 5% CO₂ in MEM containing 5% newborn calf serum (Life Technologies) and penicillin/streptomycin. Cells were passed over Lympholyte-M (Cedarlane, Hornby, Ontario, Canada), washed three times, resuspended in 10 ml of medium, and counted. Con A lymphoblasts (1 x 10⁶) or HLA-transfected P815 cells were centrifuged, resuspended in 200 µl FCS, and labeled with 150 µCi of Na⁵¹CrO₄ (NEN, Boston, MA) at 37°C, 5% CO₂ for 1.5 h.

Skin grafts

Skin grafting was conducted essentially as described (66). Briefly, recipients were anesthetized and shaved. Two graft beds were prepared with a skin bridge between them by removing two sections of skin from the posterior chest wall of each recipient. Full-thickness skin grafts (1–1.5 cm) harvested from the tails of donor and syngeneic control mice were engrafted onto the graft beds. Collodian was applied at the junction of the skin graft and donor skin to secure the grafts in place. Graft sites were covered with petroleum jelly gauze and a circumferential bandage, following which the mice were allowed to recover for 7 days. On day 7, the bandages were removed and the skin grafts were evaluated daily for redness, hair growth, hemorrhagic spots, presence of scales, and status of graft borders. Grafts were considered rejected when <10% of the graft bed contained viable grafted skin (66).

Statistical analysis

Statistical significance was determined with unpaired Student’s t test for the comparison of means with unequal variances (Microsoft Excel software; Redmond, WA). Differences between groups were considered to be significant if p < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
HLA^nat and HLA^hyb-A2, -B7, and -B27 TgM

To further characterize and develop the HLA Tg model for studies of human MHC-dependent T cell recognition and responsiveness, we established a panel of TgM that expresses the HLA class I alleles HLA-A2, -B7, or -B27 in either the fully native form (i.e., HLA^nat) together with hß₂m or as a hybrid molecule of the human peptide-binding 1/2 domains linked to the mouse H-2K^b 3, transmembrane, and cytoplasmic domains (HLA^hyb). Fig. 1 gives a schematic representation of the three HLA^nat (Fig. 1A) and three HLA^hyb (Fig. 1B) gene constructs. Of multiple founder mice generated and characterized, one representative line was selected for each construct for further detailed analysis and the studies reported here. Lines were selected on the basis of displaying an appropriate tissue distribution of expression at levels similar to each other and to endogenous H-2 class I.

Expression of HLA^nat and HLA^hyb transgenes in TgM

Cell surface expression of Tg HLA^nat and HLA^hyb class I was analyzed for LNCs by flow cytometry with several anti-HLA class I mAbs, including B9.12.1 (reacts with all HLA class I alleles) (63), ME-1 (reacts with HLA-B7, -B27, and -Bw22) (61), and MA2.1 (reacts with HLA-A2 and -B17) (62). Endogenous H-2 class I expression was examined with the H-2D^b-reactive mAb 28-14-8S (64). Following analysis of all Tg lines, single representative lines for each construct were selected for the studies described in this article on the basis of the most similar levels of Tg HLA expression to each other and to endogenous H-2 class I. As already described (5), surface expression of HLA-B7^nat increased 10-fold in LNCs of mice coexpressing hß₂m (B7^{nat/hß2 m}), indicating suboptimal interaction of the human heavy chain with mouse ß₂m in mice Tg for the HLA^nat class I gene alone. As a result, and to ensure the appropriate conformation of the Tg HLA^nat heavy chains, the HLA-A2^nat and HLA-B27^nat TgM were derived by coinjection of fertilized eggs of the heavy chain gene along with the hß₂m gene. Fluorescent in situ hybridization analyses (67), as well as cotransmission of both the coinjected HLA and hß₂m genes to offspring, demonstrated that both genes were cointegrated at a single chromosomal site (results not shown). The HLA-B7^nat/hß₂m mice were derived from the breeding of singly Tg parental HLA-B7^nat and hß₂m lines (5). The HLA^hyb gene constructs (Fig. 1B) were designed such that the encoded molecules would associate much more efficiently than the HLA^nat class I proteins with endogenous mouse ß₂m, thereby obviating the need for coexpression of hß₂m.

Fig. 2A shows flow cytometry staining results for LNCs from HLA-A2^nat/hß2m and HLA-A2^hyb TgM, whereas Fig. 2B shows similar analyses for HLA-B7^nat, -B7^nat/hß2m, -B7^hyb, -B27^nat/hß2m, and -B27^hyb TgM. Relative to the background levels of fluorescence observed for non-Tg LNCs stained with the same anti-HLA-A2.1 (Fig. 2A), anti-HLA-B7/B27 (Fig. 2B), or anti-pan-HLA class I (Fig. 2C) mAbs, significant surface expression of both the native and hybrid molecules for each allele was detected on Tg cells (Fig. 2). A high level of hß₂m was also detected at the surface of cells from HLA-A2^nat/hß2m, -B7^nat/hß2m, and -B27^nat/hß2m TgM (not shown). The level of endogenous H-2D^b (and K^b, not shown) class I expression was similar in all HLA Tg strains compared with non-TgM (Fig. 2A, and not shown).

View larger version (20K): [in this window] [in a new window]   FIGURE 2. Flow cytometric analysis of HLAnat and HLAhyb class I cell surface expression on Tg LNCs. A, LNCs from non-Tg (B6/SJL)F1 and HLA-A2nat/hß2m and HLA-A2hyb class I TgM were stained with the HLA-A2-specific mAb MA2.1 (thick solid line) and the H-2Db-reactive mAb 28–14-8S (dashed dotted line). B, LNCs from non-Tg (B6/SJL)F1 and HLA-B7nat, HLA-B7nat/hß2m, HLA-B7hyb, HLA-B27nat/hß2m, and HLA-B27hyb class I TgM were stained with the HLA-B7/B27-specific mAb ME1 (thin solid line). C, LNCs from non-Tg (B6/SJL)F1 and HLAhyb (-A2, -B7, and -B27) TgM were stained with the pan-HLA class I-specific mAb B9.1.2.1 (dotted line). In all cases, the secondary Ab was FITC-conjugated F(ab')2 goat anti-mouse IgG.

Enhanced recognition of Tg HLA^hyb class I molecules in vitro

To evaluate the ability of non-Tg H-2-matched T cells to recognize and respond to Tg HLA^nat and HLA^hyb class I molecules as transplantation Ags in vitro, CML assays were performed. As the immune system of the responder strain used in these initial assays had not previously been exposed to the HLA molecule during immune development (i.e., in vivo), we consider this situation to represent a mouse anti-human MHC class I xenoresponse. Following a 5-day primary in vitro stimulation of non-Tg LNCs from (B6/SJL)F₁ mice with spleen cells from each of the three HLA^hyb or three HLA^nat/hß2m TgM, ⁵¹Cr release assays were performed using Con A-stimulated spleen cells from various non-TgM and TgM strains as targets. Fig. 3, A and D, shows the results obtained for the anti-HLA^nat and anti-HLA^hyb responses, respectively. For all alleles, the hybrid molecule (Fig. 3D) induced a higher level of killing than its native counterpart (Fig. 3A). This increase was most apparent for B7 and B27 because the native forms of these induced only very weak responses (Fig. 3A). The elevated anti-HLA^hyb response was not due simply to higher cell surface expression because flow cytometry (Fig. 2), quantitative measurements (39), and RNA analyses (not shown) showed that each of the HLA^nat molecules in HLA^nat/hß2m TgM was expressed at similar or greater levels than the corresponding HLA^hyb molecule (Fig. 2, A and B). In contrast to the very low level of killing detected for HLA-B7^nat or -B27^nat, the A2^nat molecule induced a stronger response, which was further increased when A2 was expressed as a hybrid molecule (Fig. 3, A and D). The higher response to HLA-A2^nat compared with -B7^nat or -B27^nat was not due to differences in expression level because B7^nat/hß2m was expressed at a higher level on the cell surface than A2^nat/hß2m, and B27^nat/hß2m expression was only slightly less (Fig. 2, A and B, and not shown). The magnitude of the anti-HLA^hyb response for all alleles was close to that generated against the strong mouse class I allotransplantation Ag H-2K^bm1 (68, 69), whereas the response to HLA-A2^nat was only slightly less (Fig. 3, A and D).

View larger version (24K): [in this window] [in a new window]   FIGURE 3. Comparison of primary in vitro-stimulated CML response of non-Tg LNCs against Tg cells expressing xenogeneic HLAnat vs HLAhyb class I molecules. Responder LNCs from non-Tg (B6/SJL)F1 (H-2b/s) mice were stimulated in vitro with irradiated spleen cells from HLAnat class I (HLAnat/hß2m) (A, B, and C) or HLAhyb class I TgM (HLAhyb) (D and E). The response against cells from mice carrying the H-2Kbm1 class I alloantigen was included for comparison in all experiments and is shown in (A). After 5 days of stimulation, the responding cultures were tested for specific cytotoxicity on target cells in 4-h 51Cr-release assays (65 ). A, (B6/SJL)F1 (H-2b/s) responders stimulated by HLA-A2nat, HLA-B7nat, or HLA-B27nat class I-expressing Tg cells (on a H-2b-matched background) were tested on Con A-stimulated mouse lymphoblast targets from HLA-A2nat, HLA-B7nat, or HLA-B27nat TgM (H-2b-matched), respectively. All anti-HLAnat responders were also tested on non-Tg targets and are shown for anti-HLA-A2nat responders. B, all three anti-HLAnat responding cultures were tested on HLA-A2nat-, HLA-B7nat-, or HLA-B27nat-expressing H-2-mismatched P815 (H-2d) cells respectively. Analogous experiments are shown in (D) and (E) for the anti-HLAhyb class I responders tested on HLAhyb-matched Con A-stimulated (D; H-2b-matched) or P815 (E; H-2d-mismatched) targets. C, responders stimulated by HLA-B7nat were tested on HLA-B7nat- and HLA-B7hyb-expressing P815 cells. The E:T ratio and the percentage of specific lysis are indicated on the x- and y-axes, respectively. The data represent the means of three individual experiments.

To investigate the basis of the apparent weak anti-HLA^nat response further (Fig. 3, A and B), cultures derived from stimulating non-Tg (B6/SJL)F₁ LNCs with HLA-B7^nat Tg spleen cells were tested on HLA-B7^hyb- vs HLA-B7^nat-expressing P815 target cells (Fig. 3C). As in Fig. 3B, the killing of B7^nat-P815 was very low and only slightly above the background level of lysis observed for parental P815 (not shown). In contrast, these same HLA-B7^nat-stimulated cultures gave a much higher level of killing of HLA-B7^hyb-P815 cells (Fig. 3C). Similar experiments conducted with cultures derived from stimulating (B6/SJL)F₁ LNCs with HLA-A2^nat or -B27^nat Tg spleen cells also showed a significantly increased level of killing of allele-matched HLA^hyb-P815 compared with HLA^nat-P815 targets (i.e., specific lysis increased from 34.6% for HLA-A2^nat-P815 to 77.3% for A2^hyb-P815 at an E:T ratio of 100:1 (p = 0.03), and from <10% for HLA-B27^nat-P815 to 71.3% for B27^hyb-P815 at an E:T ratio of 100:1 (p = 0.001)). This increased killing of allele-matched HLA^hyb-expressing targets by HLA^nat-stimulated (B6/SJL)F₁ LNCs indicates that the low level of killing of HLA^nat-expressing targets is not due to the absence of xenoreactive T cells in the normal non-Tg mouse repertoire able to recognize and be stimulated by the HLA^nat molecule. Rather, the results suggest that some aspect of the structure of each HLA^nat molecule on target cells important for mediating lysis is suboptimal compared with the HLA^hyb molecule. This observation implies differential HLA domain-dependent interactions occurring at the induction vs effector phases of these in vitro CML assays.

Rapid xenorejection of both HLA^hyb and HLA^nat class I Tg skin grafts

To extend assessment of the relative immunogenicity of each of the HLA^hyb and HLA^nat class I alleles in vivo, tail skin from each Tg strain was grafted onto non-Tg H-2-matched ((B6/SJL)F₁; H-2^b/s) recipients, and the mean survival time (MST) was determined (Fig. 4). Although grafts from HLA^hyb- or HLA^nat-Tg-negative offspring (HLA Tg^neg) or from mice Tg solely for hß₂m (hß₂m⁺) were indefinitely accepted (Fig. 4A and Table I), grafts from HLA-A2^hyb, -B7^hyb, or -B27^hyb TgM were rejected after 11.5 ± 2.4, 16.4 ± 1.8, and 17.2 ± 1.3 days, respectively (Fig. 4B and Table I). These rates of rejection were similar to that for skin from mice carrying the known strong mouse class I alloantigen H-2K^bm1 (13.8 ± 1.5 days, Fig. 4A), whereas a full MHC class I/class II strain mismatch (DBA/2; H-2^d) was rejected slightly faster at about 9.4 ± 0.5 days (Fig. 4A).

View larger version (15K): [in this window] [in a new window]   FIGURE 4. Rejection of Tg HLAhyb and HLAnat class I tail-skin xenografts by H-2b-matched non-Tg recipients. Groups of non-Tg (B6/SJL)F1 recipient mice (see Table I for number of mice per group) were grafted with full-thickness skin from the following donor mice: HLAhyb or HLAnat transgene-negative [Tgneg] offspring from backcrosses of heterozygous HLA Tg parents with C57BL/6J, from hß2m TgM (hß2m+), from DBA/2 (H-2d), or from bm1 (H-2Kbm1) (A); HLA-A2hyb, -B7hyb, or -B27hyb TgM (B); HLA-A2nat/hß2m, -B7nat, -B7nat/hß2m, or -B27nat/hß2m TgM (C). The plots show the percentage of mice within each group with surviving grafts as a function of days posttransplantation. The skin-graft protocol and criteria for rejection were as described in Materials and Methods.

In the case of the HLA^hyb alleles, these results demonstrate that the immune system of non-TgM is able to recognize and respond to the xenogeneic human 1/2 polymorphic domains of all three alleles such that graft rejection occurs at a similar rapid rate as for the strong allotransplantation Ag H-2K^bm1. In the case of the HLA^nat alleles, despite only being recognized weakly in cytotoxicity assays in vitro, these molecules are equal or more potent than the corresponding HLA^hyb alleles as xenotransplantation Ags in vivo.

Rejection of HLA^hyb or HLA^nat class I Tg xenoskin grafts does not depend on CD8⁺ T cells, CD4⁺ T cells, or B cells

Given the apparent discrepancy between the in vitro and in vivo results on recognition of Tg HLA^nat vs HLA^hyb class I molecules as transplantation Ags, we wished to examine the role of conventional cellular and humoral immune effector mechanisms in HLA Tg skin graft rejection. Therefore, HLA^hyb or HLA^nat class I Tg skin for each allele was grafted onto H-2^b-matched gene-KO recipient mice that were deficient for either CD8⁺ T cells (CD8^-/-KO) (52) (Fig. 5A), CD4⁺ T cells (CD4^-/-KO) (53) (Fig. 5B), CD4⁺ and CD8⁺ T cells (CD8^-/-CD4^-/- double KO) (54) (Fig. 5C), or B cells (IgµM^-/-KO) (51) (Fig. 5D). Although five of seven CD8^-/-KO recipients retained mouse class I disparate H-2K^bm1 allografts for longer than 200 days (Fig. 5A), grafts from TgM for all three HLA^nat alleles as well as two of the three HLA^hyb alleles (A2^hyb and B7^hyb) were rejected at rates (Fig. 5A) very similar to those observed for the non-KO recipients (i.e., as in Fig. 4; see Table I). The only exception was that rejection of B27^hyb skin was somewhat delayed to 29.9 ± 8.1 days compared with wild-type recipients (p = 0.01) (Fig. 5A). Similarly, the allele-specific rejection rates of all HLA^nat and HLA^hyb grafts on CD4^-/- and IgµM^-/-KO recipients (Fig. 5, B and D; Table I) were virtually identical with those observed in the non-KO recipients. Grafts from TgM for all three HLA^hyb alleles as well as HLA-A2^nat/hß₂m were rejected by CD4^-/-CD8^-/- double KO recipients at rates similar to non-KO mice, whereas rejection of HLA-B7^nat/hß₂m and -B27^nat/hß₂m grafts were slightly delayed to 27.2 ± 12.5 days (p = 0.02) and 22.0 ± 5.2 days (p < 0.001) (Fig. 5C and Table I). Taken together, these data suggest that although the relative contribution of CD8- and CD4-dependent T cell mechanisms to HLA^hyb or HLA^nat class I Tg skin graft rejection demonstrate some variability depending on the allele, neither CD8⁺ or CD4⁺ T cells or B cells are absolutely required for graft rejection. This lack of dependence on CD8⁺ T cells for rejection of HLA^hyb grafts was particularly surprising in view of the in vitro CML results above as well as previous results of others (30, 36, 37).

View larger version (21K): [in this window] [in a new window]   FIGURE 5. Rejection of Tg HLAhyb and HLAnat class I tail-skin xenografts by gene KO recipient mice. Groups of CD8-/- KO (A), CD4-/- KO (B), CD4-/- CD8-/- double KO (C), and IgµM-/- KO (D) mice (see Table I for number of mice per group) were grafted with tail skin from HLAnat (HLA-A2nat/hß2m, -B7nat/hß2m, -B27nat/hß2m) or HLAhyb (HLA-A2hyb, -B7hyb, or -B27hyb) TgM, non-Tg siblings of the above HLA Tg donors (Non-Tg), or bm1 mice (H-2Kbm1). The plots show the percentage of mice within each group with surviving grafts as a function of days posttransplantation. The skin-graft protocol and criteria for rejection were as described in Materials and Methods.

The above results show that rejection of HLA^hyb-expressing grafts in vivo does not depend on CD8⁺ T cells. This contrasts to the clear CD8 dependence of rejection of grafts expressing a single mouse class I alloantigen (i.e., H-2K^bm1, Fig. 5A). These observations may indicate that even when the "xeno" component of the foreign MHC molecule is limited to only the 1/2 domains as in the HLA^hyb molecules, there is still an inherent difference between the immune effector mechanisms that respond to this molecule compared with those that respond to an allo-MHC class I molecule. The HLA Tg model provides a unique opportunity to investigate this issue directly because it is possible to compare recognition of a given HLA^hyb class I allele as a xenotransplantation Ag (i.e., by the immune system of non-TgM) with recognition of the identical molecule as an allotransplantation Ag (i.e., by the immune system of mice that are Tg for an alternate HLA^hyb class I allele and thus developed in this human MHC-expressing environment). To this end, in vitro CML and in vivo graft rejection studies similar to the above were conducted, but rather than the responder LNCs and recipient mice being non-Tg, responder LNCs and graft recipients that expressed an alternate HLA^hyb class I Tg allele relative to the stimulator and donor strain were used.

Following primary in vitro stimulation with HLA-B7^hyb Tg spleen (Spl) cells, LNCs from HLA-matched B7^hyb TgM did not lyse HLA-B7^hyb-P815 targets above the background levels observed for P815 parental cells (Fig. 6A). Compared with the strong xenogeneic anti-HLA-B7^hyb response generated by non-Tg LNCs (Fig. 3, D and E), this lack of killing by B7^hyb Tg LNCs indicates that CTL in these mice are tolerant to the HLA-B7^hyb molecule. When LNCs from these same mice were stimulated with spleen cells from H-2K^bm1 mice and assayed on H-2K^bm1 targets, a high level of lysis was observed (Fig. 6A) similar to that seen with non-Tg H-2^b responders (Fig. 3A). Following stimulation with HLA-B27^hyb Tg spleen cells, LNCs from HLA-B7^hyb TgM gave a significant level of lysis of both B27^hyb Tg spleen Con A blast targets (B27^hyb Tg Spl (H-2^b)) and B27^hyb-P815 targets (Fig. 6B). This lysis appeared to be allele-specific as these same B27^hyb-stimulated responders did not lyse HLA-A2^hyb-expressing P815 cells (or A2^hyb Tg spleen Con A-blasts; not shown) above background levels observed for parental P815 or C57BL/6J Con A blast targets (Non Tg Spl (H-2^b); Fig. 6B). These results demonstrate that the peripheral T cell repertoire of HLA-B7^hyb TgM has become tolerant to the self-B7^hyb allele but is alloreactive to alternate related human alleles such as HLA-B27^hyb. Additional experiments of the peripheral repertoire of HLA-B27^hyb TgM yielded similar results, demonstrating tolerance to the self-HLA-B27^hyb allele and allele-specific alloreactivity to related non-self HLA alleles (i.e., B7^hyb; not shown).

View larger version (28K): [in this window] [in a new window]   FIGURE 6. Comparison of primary in vitro-stimulated CML response of HLAhyb/CD8+/+ and HLAhyb/CD8-/- KO LNCs against H-2-matched Tg cells expressing alternate HLAhyb class I alloantigens. Responder LNCs from HLA-B7hyb CD8+/+ mice (A and B) were stimulated in vitro with irradiated HLA-syngeneic (HLA-B7hyb) or non-Tg H-2Kbm1 (A) or HLA-allogeneic (HLA-B27hyb) (B) spleen cells and assayed on Con A-stimulated blasts or HLA-B7hyb-, HLA-B27hyb-, or HLA-A2hyb-expressing P815 target cells. Con A-stimulated blasts from C57BL/6J or bm1 mice or P815 cells were included as control targets. C, Responder LNCs from HLA-B7hyb/CD8-/- KO mice were stimulated with HLA-B27hyb spleen cells and assayed on HLAhyb-mismatched targets. D, Responder LNCs from non-Tg CD8-/- KO or HLA-B7hyb/CD8-/- KO mice were stimulated with irradiated H-2Kbm1 or HLA-B7hyb Tg spleen cells and assayed on bm1 Con-A blast targets or HLA-B7hyb-expressing P815 cells. The E:T ratios and the percentage of specific lysis are indicated on the x- and y-axes, respectively.

Although CD8⁺ T cells from non-Tg (xeno) and HLA^hyb Tg (allo) mice are able to respond in vitro to Tg cells expressing a foreign HLA^hyb allele, CD8⁺ cells are not required for the rapid xenorejection by non-TgM of grafts from HLA^hyb TgM in vivo. To examine whether expression of the HLA class I molecule as a self-MHC allele influences the mechanisms by which the immune system responds in vivo to grafts from mice expressing alternate HLA class I alleles (i.e., human allo-MHC Ags), a series of skin-grafting experiments were conducted (Fig. 7). Compared with rejection of (xeno) HLA-B27^hyb skin grafts by non-TgM at about 17.2 ± 1.3 days (Fig. 4B, Table I), HLA-B27^hyb TgM accepted HLA-B27^hyb (i.e., syngeneic) Tg skin indefinitely (Fig. 7A, Table II). However, these same animals rejected grafts from both HLA-B7^hyb and -A2^hyb TgM at rates similar to those for non-Tg recipients (i.e., 17.2 ± 4.0 and 11.7 ± 2.9 days for HLA-B27^hyb recipients (Fig. 7A, Table II) vs 16.4 ± 1.8 and 11.5 ± 2.4 days for non-Tg recipients (Table I)). A similar pattern was seen for HLA-B7^hyb TgM in that, compared with non-TgM, which rejected B7^hyb grafts at 16.4 ± 1.8 days (Table I), B7^hyb recipients indefinitely accepted these grafts (Table II) but rejected HLA-B27^hyb and -A2^hyb Tg skin after 17.3 ± 3.4 and 13.3 ± 2.9 days (Fig. 7B, Table II) (compared with 17.2 ± 1.3 and 11.5 ± 2.4 days, respectively, for non-Tg recipients).

View larger version (25K): [in this window] [in a new window]   FIGURE 7. Rejection of Tg HLAhyb allele- or locus-mismatched class I tail-skin allografts by HLAhyb/CD8+/+ and HLAhyb/CD8-/- recipients. A and B, Results of grafting HLA-B27hyb/CD8+/+ (A) or HLA-B7hyb/CD8++ (B) mice with HLAhyb allele-matched or HLAhyb allele- or locus-mismatched skin grafts. C and D, Results of grafting HLA-B27hyb/CD8-/- (C) or HLA-B7hyb/CD8-/- (D) mice with HLAhyb-allele-mismatched (B7hyb, C; B27hyb, D) or HLAhyb-locus-mismatched (A2hyb, C and D) Tg skin grafts. The plots show the percentage of mice within each group with surviving grafts as a function of time (days) posttransplantation. The skin-graft protocol and criteria for rejection were as described in Materials and Methods. See Table II for number of recipient mice per group and graft MSTs.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

 
Although HLA TgM have the potential to provide a useful model for studies of human MHC-dependent immune function in vivo, the extent to which mouse T cells are able to functionally respond to human class I or II alleles in this system was not clear (27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48). We established a panel of TgM carrying different HLA alleles in both the fully human (HLA^nat) and hybrid human/mouse (HLA^hyb) forms to address several aspects of immune recognition to clarify this issue. Here we have examined whether the same in vitro and in vivo immune mechanisms are responsible for recognition of a given MHC class I allele as a xeno- as opposed to an allotransplantation Ag. To minimize any influence of quantitative differences in expression level on differential function of the three HLA alleles in either the native or hybrid form, we selected single lines for each that expressed the most similar cell surface levels of each Tg product to each other and to endogenous mouse H-2 class I (Fig. 2). The close to physiological expression levels of all Tg HLA^hyb and HLA^nat molecules, together with reactivity with all anti-HLA class I mAbs specific for polymorphic or monomorphic determinants tested, implies that each Tg HLA^hyb and HLA^nat (in the presence of hß₂m) molecule is able to efficiently undergo intracellular trafficking and association with endogenous peptides and to adopt an appropriate conformation at the cell surface.

Despite physiological surface levels of Tg HLA^nat class I, the in vitro cytotoxicity studies in Fig. 3, A and B, together with previous results of our own and others, implied a limitation at some level in the efficiency of recognition of fully human MHC class I molecules by non-Tg cytotoxic T cells. To investigate this issue, a number of groups, including ours, have studied TgM expressing HLA^hyb Tg constructs to distinguish whether inefficient xenorecognition of HLA^nat class I was due to the low frequency of mouse TCRs able to interact with the HLA class I 1/2 domains (i.e., "holes" in the repertoire) as opposed to other species-specific molecular incompatibilities involving interactions outside the peptide-binding 1/2 domains. The results in Fig. 3, D and E, demonstrate that the in vitro primary xenoresponse of non-TgM T cells to HLA class I molecules is enhanced for all three alleles when cells from HLA^hyb rather than HLA^nat TgM are used as stimulators and targets. In the HLA^hyb form, all alleles induced levels of killing that were close to that against the strong mouse class I transplantation Ag H-2K^bm1 (68, 69). In the HLA^nat form, only HLA-A2 induced a reproducibly significant level of killing. It is unclear why this HLA^nat allele induces a stronger response than the others but, given the similar cell surface expression levels, it must be due to structural polymorphic differences. This type of HLA^nat allelic difference may explain some of the differing results reported by groups working with various HLA^nat class I Tg strains.

With regard to the non-Tg response against HLA^hyb class I, it is important to recognize that the strong lysis of H-2-mismatched HLA^hyb-transfected P815 targets indicates that killing is due largely to direct recognition of intact MHC molecules. The induced responses were also allele-specific, indicating that they were directed primarily at the polymorphic regions as opposed to shared human-specific determinants. Together, these results imply that the weak anti-HLA^nat response detected in vitro is not due to a low frequency of xeno-MHC reactive mouse T cells in the non-Tg repertoire, but rather mainly to species-specific interactions outside of the Ag-binding cleft. A similar conclusion was reached previously by others studying HLA-A2 (30, 37) and -B27 (36), but not for -B7 (41).

Although HLA^nat-stimulated non-Tg responders were unable to give significant levels of killing of HLA^nat allele-matched targets, these same responders gave much higher levels of killing of HLA^hyb allele-matched targets for HLA-B7 (Fig. 3C), -A2, and -B27 (not shown). These results suggest that the low killing observed in the experiments of Fig. 3, A and B, is not necessarily because the xeno-HLA^nat molecules are unable to induce a cellular response but rather may be due in part to a suboptimal ability of the induced cells to recognize the HLA^nat molecule on target cells. In studies of the anti-influenza T cell response in analogous strains of HLA-A2^nat and A2/K^b TgM, Sherman et al. (37) observed that A2^nat-restricted effectors from HLA-A2^nat TgM could lyse A2K^b (A2^hyb) targets in the presence or absence of flu peptide. The interpretation of this result was that the A2^nat-restricted CTLs had only low affinity for the A2^nat molecule but an increased affinity for the A2K^b (A2^hyb) molecule as a result of incorporating mouse CD8 in the interaction by inclusion of the mouse 3 domain in the A2^hyb molecule. Although our studies have examined the non-Tg T cell response to xeno-HLA-A2 rather than the usage of HLA-A2 as a restriction element by Tg T cells, the killing of HLA-A2^hyb targets by HLA-A2^nat-induced T cells may also be the result of an increased affinity of interaction resulting from involvement of mouse CD8 with the A2^hyb vs A2^nat molecule during the killing of HLA-A2^hyb vs A2^nat targets.

It is generally accepted that in vitro CML assays used above are an in vitro correlate of MHC class I disparate graft rejection in vivo (72, 73, 74, 75). Based on this, it would be expected that if recognition of HLA Tg class I molecules as xenotransplantation Ags by the non-TgM immune system is similar to that of mouse H-2 class I molecules, then the mechanisms mediating HLA Tg and H-2 class I disparate graft rejection should be similar and be reflected by these assays. However, based on our skin graft rejection studies, we believe that some of the previous inconsistencies among studies using this type of model result from the apparent breakdown of this correlation. For example, given the very low level of killing observed following stimulation with each Tg HLA^nat product (Fig. 3), the rapid rejection of skin grafts from these same HLA^nat TgM was not expected (Fig. 4C). However, although Van Twuyver et al. (76) also reported rapid rejection of skin grafts from mice Tg for HLA-B27^nat, others appear to have limited their analyses to in vitro primary CML assays with the assumption that the in vitro results reflected in vivo graft recognition and rejection (36).

The mechanisms underlying the very rapid rejection of HLA^nat Tg grafts are not obvious. Clearly, it is not due to the influence that coexpressing hß₂m has on either the quantitative level of expression or the conformation of the HLA^nat class I heavy chain as skin grafts from singly Tg HLA-B7^nat mice (i.e., hß₂m-negative) are also rapidly rejected (Fig. 4C) and those from singly Tg hß₂m mice (i.e., HLA-negative/hß₂m-positive) are indefinitely accepted (Fig. 4A). Also, the use of gene KO graft recipient mice deficient for either CD8⁺ or CD4⁺ T cells or B cells showed that these populations on their own were not responsible for rejection of Tg HLA^nat grafts. Either other effector mechanisms (i.e., NK cells), or multiple mechanisms as suggested by the somewhat prolonged survival of grafts from two of the three HLA^nat Tg strains in CD4^-/-CD8^-/- double KO recipients, are operative in the rapid rejection of these grafts. Studies to investigate these possibilities are in progress.

The second instance in which there is a discrepancy between results obtained from in vitro CML vs in vivo graft rejection assays was with xenorecognition of the HLA^hyb class I molecules. The in vitro CML assays shown here, together with results from others (36), demonstrate that each HLA^hyb class I allele is recognized as efficiently as the mouse class I alloantigen bm1, presumably due to improved interaction of the HLA^hyb, compared with the HLA^nat class I molecule, with mouse CD8 as a result of including the mouse 3 domain (36, 37). However, our results showed that when CD8^-/- KO mice were engrafted with donor HLA^hyb Tg skin, the allele-specific rejection rates (Fig. 5A) were very similar to those observed for wild-type (CD8^+/+) recipients. Thus, despite the differences detected in vitro for the HLA^nat and HLA^hyb xeno-MHC molecules, these effects are of no significance in vivo in at least two respects: first, despite poor recognition in vitro, HLA^nat grafts are rejected very rapidly in vivo at rates that are equal to or faster than those for the corresponding HLA^hyb grafts or even bm1 grafts; and second, despite improved recognition in vitro of HLA^hyb class I molecules by non-TgM CD8⁺ T cells, this effect is completely irrelevant to rejection of HLA^hyb grafts in vivo. In contrast, the absence of CD8⁺ T cells had a significant effect on the survival of H-2K^bm1 skin grafts. This finding of prolonged survival of mouse class I allogeneic, but not HLA^hyb class I Tg, skin grafts in CD8^-/- KO recipients argues that CD8⁺ T cells play an important role in the rejection of murine allografts disparate at a single MHC class I molecule but not in class I-disparate xenografts. A similar independence of HLA^hyb Tg graft rejection on CD4⁺ T cells was revealed when CD4^-/- KO recipients were grafted with HLA Tg skin (Fig. 5B). Thus, similar to HLA^nat grafts, these data indicate that neither CD8⁺ or CD4⁺ T cells on their own are necessary for rejection of xeno-HLA^hyb Tg grafts and that either other or multiple mechanisms must be involved. As the only difference between the HLA^hyb molecules and the "self" H-2K^b class I molecule is in the 1/2 domains, our results show that although the human (xeno) 1/2 domains are sufficient for recognition as a very strong major histocompatibility transplantation Ag, the immune mechanisms induced by this xeno-MHC class I molecule are distinct from those induced by an allo-MHC (mouse) class I molecule.

The strong non-Tg anti-HLA^hyb xenoresponses detected in vitro and in vivo were not detected when immune cells from HLA^hyb TgM were stimulated with HLA^hyb allele-matched cells (Fig. 6A) or grafts (Fig. 7, A and B). In contrast, stimulation of HLA^hyb Tg cells with HLA^hyb allele-mismatched (i.e., allogeneic) Tg cells led to strong lysis of target cells expressing the mismatched HLA^hyb allele but not the HLA^hyb self allele or an alternate third party allele (Fig. 6, A and B). Similarly, although HLA^hyb TgM were tolerant to grafts expressing their self-HLA^hyb allele, they rapidly rejected grafts from mice expressing a HLA^hyb allele-mismatched (rejection of B7^hyb grafts by B27^hyb TgM, Fig. 7A; rejection of B27^hyb grafts by B7^hyb TgM, Fig. 7B) as well as locus-mismatched grafts (rejection of HLA-A2^hyb grafts by -B27^hyb TgM, Fig. 7A; and by -B7^hyb TgM, Fig. 7B). Thus, given that the non-Tg response to HLA^hyb class I molecules as xeno-MHC Ags involves both a CD8⁺ T cell-dependent component (based on the in vitro CML assays) and a non-CD8⁺/non-CD4⁺