Effects of blocking helper T cell induction in vivo with anti-Ia antibodies. Possible role of I-A/E hybrid molecules as restriction elements.

To examine the role of Ia antigens in controlling T cell activation in vivo, unprimed (CBA X B6)F1 (H-2k X H-2b) T cells were positively selected to sheep erythrocytes (SRC) for 5 d in irradiated F1 mice in the presence of large doses of anti-Iak antibody. With selection in the presence of broad-spectrum anti-Iak antibody (A.TH anti-A.TL antiserum), the activated T cells were markedly reduced in their capacity to collaborate with either B10.BR (I-Ak I-Bk I-Jk I-Ek I-Ck) (kkkkk) or B10.A(4R) (kbbbb) B cells but gave good helper responses with B10 (bbbbb) and (B10 X B10.BR)F1 B cells. Because there was no evidence for suppression, these findings were taken to imply that the anti-Iak antibody bound to Ia determinants on radioresistant macrophagelike cells of F1 host origin and blocked the activation of the IGk-restricted subgroup of F1 T cells but did not affect activation of the Iab-restricted T cell subgroup. Analogous experiments in which F1 T cells were selected to SRC in F1 mice in the presence of monoclonal anti-I-Ak antibody gave different results. In this situation, the reduction in T cell help for Iak-bearing B cells applied to B10.A(4R) B cells but not to B10.BR B cells. With selection of F1 T cells in B10.A(4R) mice, by contrast, anti-I-Ak antibody blocked T cell help for both B10.A(4R) and B10.BR B cells. These data suggested that genes telomeric to the I-A subregion were involved in controlling T cell activation and T-B collaboration. Because no evidence could be found that I-B through I-C determinants per se could act as restrictions elements, the working hypothesis for the data is that Iak-restricted T cells consist of two subgroups of cells: one subgroup is restricted by I-A-encoded molecules, whereas the other is restricted by I-A/E hybrid molecules encoded by two separated genes situated in the I-A and I-E subregions, respectively. The notion that A/E hybrid molecules serve as restriction elements is in line with the findings of other workers that these molecules can act as alloantigens and control responses to certain antigens under double Ir gene control.

Restricted helper function of F1 leads to parent bone marrow chimeras controlled by K-end of H-2 complex.

F1 leads to parent bone marrow chimeras were prepared by transferring F1 hybrid marrow cells into heavily irradiated parental strain mice. When unprimed, donor-derived F1 T cells from the chimeras were activated to sheep erythrocytes (SRC) for 5 days in irradiated normal F1 mice, high IgM and IgG anti-SRC responses were observed with F1 B cells, and with B cells H-2-compatible with the strain in which the T cells were raised from stem cells. Significantly, however, responses with B cells of the opposite parental strain were either absent or very low. The restriction in T-helper function mapped to the K-end of the H-2 complex and could not be attributed to active suppression.

Role of the H-2 complex in induction of T helper cells in vivo. I. Antigen-specific selection of donor T cells to sheep erythrocytes in irradiated mice dependent upon sharing of H-2 determinants between donor and host.

When purified CBA lymph node T cells were mixed with sheep erythrocytes (SRC) and filtered from blood to lymph through irradiated syngeneic mice for 1-2 days, the donor cells lost their capacity to stimulate anti-SRC responses by CBA B cells; the response to a third-party antigen (horse erythrocytes) was unaffected and active suppression was not involved. This process of specific negative selection to SRC also occurred when semiallogeneic mice were used as filtration hosts. By contrast, when allogeneic hosts were used the helper function of the donor cells was not reduced; this applied to both primed and unprimed T cells. Studied with congeneic resistant strains indicated that negative selection to SRC occurred only when the donor and host shared H-2 determinants. Studies with T cells depleted of alloreactive lymphocytes showed that negative selection to SRC in irradiated F1 hybrid mice was followed by a stage of positive selection where the donor cells gave greatly increased responses to the injected antigen. Positive selection did not occur in H-2-different mice, however, and the helper function of the donor cells remained unchanged. By these parameters it was concluded that homozygous T helper cells have no detectable capacity to recognize antigen in an H-2-different environment.

Restricted helper function of F1 hybrid T cells positively selected to heterologous erythrocytes in irradiated parental strain mice. I. Failure to collaborate with B cells of the opposite parental strain not associated with active suppression.

Unprimed (CBA X C57BL/6)F1 lymph node T cells were transferred with sheep erythrocytes (SRC) into heavily irradiated F1 or parental strain mice and recovered from thoracic duct lymph or spleens of the recipients 5 days later. To study their helper function, the harvested F1 T cells were transferred with antigen into irradiated F1 mice plus B cells from either the two parental strains or from F1 mice. F1 T cells activated in F1 mice gave high IgM and IgG anti-SRC responses with all three populations of B cells. By contrast, F1 T cells activated in mice of one parental strain collaborated well with B cells of this strain, but poorly with B cells of the opposite strain. Active suppression was considered an unlikely explanation for this result since (a) good responses were found with F1 B cells, and (b) addition experiments showed that the poor response with B cells of the opposite parental strain (which was equivalent to that produced by unprimed F1 T cells) could be converted to a high response by a supplemental injection of F1 T cells activated in F1 mice. The phenomenon (a) was specific for the antigen used for activation (criss-cross experiments were performed with horse erythrocytes), (b) was reflected in levels of serum hemagglutinins as well as in numbers of splenic plaque-forming cells, (c) applied also to comparable activation of (DBA/2 X C57BL/6)F1 T cells, and (d) could be prevented by activating F1 T cells in mice of one parental strain in the presence of peritoneal exudate cells of the opposite parental strain. The hypothesis was advanced that F1 T cells contain two discrete subpopulations of antigen-reactive cells, each subject to restrictions acting at two different levels: (a) during T-macrophage interactions and (b) during T-B collaboration. It was suggested that when F1 T cells are activated to antigen in a parental strain environment, radioresistant macrophages activate only one of the two subgroups of T cells, and this subgroup is able to collaborate with B cells of the strain used for activation (and with F1 B cells) but not with B cells of the opposite parental strain. The other subgroup of T cells remains in an unprimed (nonactivated) state.

Restricted helper function of F1 hybrid T cells positively selected to heterologous erythrocytes in irradiated parental strain mice. II. Evidence for restrictions affecting helper cell induction and T-B collaboration, both mapping to the K-end of the H-2 complex.

Studies with H-2-congenic and recombinant strains showed that when F1 hybrid T cells were activated to sheep erythrocytes in irradiated mice of parental strain or related strain, a population of helper cells was generated which collaborated only with B cells sharing the K-end of the H-2 complex with the strain used for activation. No evidence was found that the restriction in helper function (a) reflected a deficiency of appropriate macrophages during T-B collaboration, or (b) was influenced by the Ig allotype of the B cells. It was concluded that the results signified restrictions acting at both the level of helper cell induction (presumed to be a reflection of T-macrophage interactions in the irradiated intermediate hosts) and during T-B collaboration. With (CBA X C57BL/6)F1 cells, the restrictions at each level mapped to the same region i.e. to the left of the I-B subregion. Consequently, one gene (or set of genes) might control restriction at both levels. If so, T-cell recognition of major histocompatibility complex-associated antigen on macrophages and on specific B cells would be either identical or very similar. The fact that genes mapping to the K-end of the H-2 complex also control the restrictive interactions of homozygous T cells implies that F1 T cells behave functionally as a mixture of T cells derived from the two parental strains. Positive selection to antigen in parental strain mice appears simply to alter the ratio of these two populations.

Cell-to-cell interaction in the immune response. VI. Contribution of thymus-derived cells and antibody-forming cell precursors to immunological memory.

Collaboration between thymus-derived lymphocytes and nonthymus-derived antibody-forming cell precursors occurs in the primary antibody response of mice to heterologous erythrocytes and serum proteins. The purpose of the experiments reported here was to determine whether collaboration took place in an adoptive secondary antibody response. A chimeric population of lymphocytes was produced by reconstituting neonatally thymectomized CBA mice soon after birth with (CBA x C57BL)F(1) thymus lymphocytes. These mice could be effectively primed to fowl immunoglobulin G (FgammaG) and their thoracic duct lymphocytes adoptively transferred memory responses to irradiated mice. The activity of these cells was impaired markedly by preincubation with CBA anti-C57BL serum and to a lesser extent by anti-theta-serum. Reversal of this deficiency was obtained by adding T cells in the form of thoracic duct cells from normal CBA mice. Cells from FgammaG-primed mice were at least 10 times as effective as cells from normal mice or from CBA mice primed to horse erythrocytes. These results were considered to support the concept that memory resides in the T cell population and that collaboration between T and B cells is necessary for an optimal secondary antibody response. Poor antibody responses were obtained in irradiated mice given mixtures of thoracic duct cells from primed mice and of B cells from unprimed mice (in the form of spleen or thoracic duct cells from thymectomized donors). In contrast to the situation with T cells, the deficiency in the B cell population could not be reversed by adding B cells from unprimed mice. It was considered that memory resides in B cells as well as in T cells and that priming probably entails a change in the B cell population which is fundamentally different from that produced in the T cell population.

Features of T cells causing H-2-restricted lethal graft-vs.-host disease across minor histocompatibility barriers.

Evidence is presented that T cells that produce lethal graft-vs.-host disease (GVHD) to minor histocompatibility antigens (minor HA) comprise discrete subgroups of H-2K- and H-2D-restricted T cells; double negative selection of T cells in irradiated H-2 recombinant mice was used to separate these two subgroups. No evidence could be found that I-restricted T cells contributed to GVHD, either as effector cells or helper cells. The (unprimed) precursor cells for GVHD expressed the Thy-1+, Lyt-1+/-2, Ia- phenotype. Studies in which H-2-semiallogeneic bone marrow chimeras were used as hosts for negative selection suggested that presentation of minor HA to T cells during the induction phase is controlled by marrow-derived cells; indirect evidence was obtained that these latter cells can "process" minor HA presented on H-2 different cells and thereby render the antigens immunogenic. Studies in which minor HA-different, H-2-compatible chimeras were re-irradiated and then injected with donor-vs.-host T cells suggested that the effector phase of lethal GVHD involves contact of antigen on non-marrow-derived cells.

Physiology of B cells in mice with X-linked immunodeficiency. II. Influence of the thymus and mature T cells on B cell differentiation.

Evidence is presented that the in vivo differentiation of B cells expressing X-linked immunodeficiency (xid) is controlled by mature T cells. Normal (C57BL/6 X CBA/J)F1 mice were thymectomized (ATx), heavily irradiated, and reconstituted with CBA/N (xid) or CBA/Ca (nondefective) marrow. In contrast to sham-operated mice, ATx recipients of xid marrow showed an almost total absence of Ig+ B cells in lymph nodes (LN) and thoracic duct lymph at 2 mo post-reconstitution ; B cells were markedly reduced in the spleen in some mice but only moderately in others. Addition of mature T cells soon after marrow reconstitution substantially abrogated the B cell depletion. In control experiments with nondefective B cells, the number of B cells developing in ATx irradiated recipients of normal (xid-) marrow cells was not detectably lower than in sham-operated recipients. These data imply that a subset of T-dependent B cells is either missing in normal mice or present in only very small numbers.

Properties of purified T cell subsets. I. In vitro responses to class I vs. class II H-2 alloantigens.

In light of the widely accepted view that Ia-restricted L3T4+ T helper cells play a decisive role in controlling the differentiation of Lyt-2+ cells, experiments were designed to examine whether Lyt-2+ cells can respond to antigen in the absence of L3T4+ cells. The results showed that highly purified Lyt-2+ cells gave high primary mixed lymphocyte reactions (MLR) to various class I differences, including both mutant and allelic differences; responses to class II (Ia) differences were generally undetectable with Lyt-2+ cells. The intensity of MLR to class I differences was not affected by addition of anti-L3T4 monoclonal antibodies (mAb) to the cultures or by removing T cells from the stimulator populations. Negative selection experiments showed that Lyt-2+ cells could respond to class I differences across Ia barriers. MLR of purified Lyt-2+ cells peaked on days 3-4 and then fell sharply; background responses with syngeneic stimulators (auto-MLR) were virtually absent. Parallel experiments with purified L3T4+ cells showed that this subset responded in MLR only to class II (Ia) and not class I differences, reached peak responses only on day 6 rather than days 3-4, and often gave high auto-MLR. Within the first 3-4 d of culture, MLR were generally higher with Lyt-2+ cells than L3T4+ cells. Although no evidence could be found that Ia-restricted L3T4+ cells were required for the response of Lyt-2+ cells, presentation of antigen by Ia+ cells appeared to be essential. Thus, responses were ablated by pretreating stimulator cells with anti-Ia mAb plus C'. Significantly the failure of Lyt-2+ cells to respond to anti-Ia plus C'-treated stimulators could not be restored by adding syngeneic spleen cells; addition of IL-2 led to only a minor (15%) restoration of the response. It is suggested that Ia+ cells provide an obligatory second signal required by Lyt-2+ cells.

Prolonged survival in vivo of unprimed B cells responsive to a T-independent antigen.

Despite earlier evidence to the contrary, it has recently been claimed that most B lymphocytes, including lymph node (LN) and thoracic duct B cells, are short-lived cells of recent marrow origin. To seek direct information on this question, we transferred unprimed LN or thoracic duct B cells from normal mice to xid mice, i.e., mice unresponsive to the T-independent antigen, trinitrophenyl (TNP)-Ficoll. At varying periods after B cell transfer the recipients were challenged with TNP-Ficoll; anti-TNP plaque-forming cells were assayed in the spleen 6 d later. The results showed that the B cell recipients retained responsiveness to TNP-Ficoll for at least 3 mo after transfer. Responsiveness increased within the first 3 wk but then remained relatively constant. These findings imply that, at least for TNP-Ficoll-reactive cells, B cells residing in LN and thoracic duct lymph are not short-lived cells of recent marrow. Indeed, the data suggest that once the pool of recirculating B cells is fully formed in adult mice, further input of newly formed cells from the marrow into the recirculating pool is very limited.

Downregulation of T cell responses by antibodies to the T cell receptor.

We have derived a T cell clone that recognizes and responds to three different types of antigen: self + X (fowl gamma globulin + H-2d), allo-H-2p,b, and minor lymphocyte-stimulating (Mlsa,d) determinants. Anti-TcR mAb and their F(ab')2 and Fab fragments were tested for their capacity to block the response of this clone. When responses were assayed on day 4 or later, addition of KJ16 or F23.1 mAb caused a marked inhibition of the response to each of the three antigens recognized by the clone. Responses measured at earlier time points however were unaffected or enhanced. This finding suggested that the inhibitory effects of anti-TcR mAb that followed the phase of enhancement might have reflected downregulation of the cells rather than simple blockade of TcR. In support of this possibility it was found that addition of anti-TcR mAb caused marked inhibition of the response of the clone to IL-2, i.e., a response that is not known to involve the TcR.

Physiology of B cells in mice with X-linked immunodeficiency (xid). III. Disappearance of xid B cells in double bone marrow chimeras.

Evidence is presented that B cells from mice with X-linked immunodeficiency (xid) differentiate at a slower rate than normal B cells. This conclusion stems from studies in which (B6 X CBA/J)F1 mice were heavily irradiated (1,000 rads) and reconstituted with a mixture of T-depleted marrow cells taken from (a) nondefective B6 mice (H-2b) and (b) xid CBA/N or nondefective CBA/Ca mice (both H-2k). With transfer of CBA/Ca plus B6 marrow cells, the irradiated recipients become repopulated with B cells derived from both parental marrow sources; except for an early imbalance (probably reflecting Hh resistance), the degree of chimerism remained relatively stable over a period of more than 6 months. Very different results occurred with transfer of a mixture of xid CBA/N and normal B6 marrow. Within the first 2 months after marrow reconstitution, a low but significant proportion of the B cells in both spleen and lymph nodes were of CBA/N origin. Thereafter the proportion of these cells fell progressively, and by 6-9 months virtually all of the B cells were of B6 origin. This gradual decline in CBA/N-derived cells did not apply to other cell types, i.e., T cells or pluripotential stem cells. Analogous results were obtained with transfer of CBA/N vs. CBA/Ca marrow cells into sublethally irradiated (750 rads) (CBA/N X DBA/2)F1 male vs. female mice. For example, CBA/N-marrow derived B cells differentiated effectively and survived for long periods in F1 male mice (xid----xid) but not in F1 female mice (xid----normal). The finding that xid B cells eventually disappear in the presence of normal B cells strengthens the view that xid B cells are an abnormal population not represented in normal mice.

Variable capacity of L3T4+ T cells to cause lethal graft-versus-host disease across minor histocompatibility barriers in mice.

Highly purified populations of L3T4+ and Lyt-2+ T cell subsets were compared for their capacity to cause lethal GVHD in six different H-2-compatible, multiple minor histocompatibility antigen-different murine strain combinations. In four of these combinations (C3H.SW----B6, DBA/2----B10.D2, B10.BR----CBA, and B10.S----SJL), lethal GVHD appeared to be caused almost entirely by Lyt-2+ cells; the injection of L3T4+ cells resulted in low mortality even when these cells were presensitized to the recipient antigens. In the remaining two combinations (B10.D2----DBA/2 and B10.D2----BALB/c), L3T4+ T cells were able to cause a high incidence of GVHD and were more potent than the Lyt-2+ cells. The implications of these findings are discussed.

Response of mature unprimed CD8+ T cells to Mlsa determinants.

Contrary to existing dogma, evidence is presented that proliferative responses of mature unprimed T cells to Mlsa antigens involve CD8+ cells as well as CD4+ cells. The response of CD8+ cells to Mlsa antigens proved to be heavily dependent on help from CD4+ cells, and responses were stronger in three I-E+ strain combinations than in an I-E- combination. In I-E+ combinations, CD8+ blast cells accounted for 20-25% of the blasts generated from unseparated T cells responding to Mlsa-bearing stimulator cells in vitro; similar findings applied to blast cells generated in vivo. The observation that the majority (greater than or equal to 50%) of Mlsa-stimulated CD8+ cells (and CD4+ cells) were V beta 6+ indicated that CD8+ cells respond to Mlsa antigens, per se, rather than to nonspecific stimuli. Whether CD4+ and CD8+ cells use the same or different H-2-restricting elements to respond to Mlsa antigens has yet to be resolved.

Selection of cytotoxic T-cell precursors specific for minor histocompatibility determinants. I. Negative selection across H-2 barriers induced with disrupted cells but not with glutaraldehyde-treated cells: evidence for antigen processing.

Intravenous injection of CBA mice with H-2-compatible irradiated B10.BR spleen cells led to a sequence of negative and positive selection of the host T-cell response against the multiple foreign minor histocompatibility antigens (HA) on the injected cells. By 1 d posttransfer, thoracic duct lymphocytes (TDL) of the host had lost the capacity to differentiate in vitro into cytotoxic cells specific for the injected minor HA; spleen and lymph node cells, by contrast, gave normal or enriched responses at this time. By 5 d posttransfer, TDL were hyperresponsive to the injected antigens. Selection with disrupted (sonicated) cells gave similar findings. With injection of either irradiated of disrupted spleen cells, the H-2 haplotype of the minor HA-bearing cells had no apparent effect on the magnitude of selection. By contrast, treatment of spleen cells with glutaraldehyde before injection led to H-2 restriction of selection, i.e., negative selection of the CBA response to B10.BR was marked with injection of glutaraldehyde-treated H-2-compatible B10.BR cells but was minimal with H-2-different B10 or B10.D2 cells. These data are taken to imply that, at least in H-2-incompatible situations, the minor HA-bearing cells must be processed by host cells, i.e., to allow the antigens to become associated with self H-2 determinants. Circumstantial evidence from studies on the specificity of selection induced with glutaraldehyde-treated cells from mice of the B10 recombinant strains suggested that I region-restricted T cells may control the induction of H2K, D-restricted cytotoxic precursor cells.

Negative selection of T cells causing lethal graft-versus-host disease across minor histocompatibility barriers. Role of the H-2 complex.

With a model in which CBA T cells cause lethal graft-versus-host disease (GVHD) in irradiated B10.BR mice (H-2-compatible mice that express multiple minor histocompatibility antigen [HA] differences), information was sought on whether the induction phase of GVHD to minor HA is H-2 restricted. When unprimed CBA (H-2k) T cells were recirculated from blood to lymph for 1 d through irradiated H-2-compatible B10.BR or B10.K mice, the T cells underwent specific negative selection to the minor HA of the host, i.e, the filtered T cells failed to cause GVHD after transfer to B10.BR mice. With filtration through totally H-2-different B10 (H-2b), B10.D2 (H-2d), or B10.S (H-2s) mice, by contrast, no selection occurred, i.e., the filtered cells were unimpaired in their capacity to kill B10.BR mice. Selection was marked after filtration through H-2-semiallogeneic (B10 X CBA)F1 mice. These data, together with the results of filtering T cells through various H-2 recombinant strains, indicated that selection depended upon the donor and filtration host sharing determinants encoded by both the K- and D-ends of the H-2 complex. Compatibility only in the I region failed to cause demonstrable selection.

Absence of H-2 restriction in primary and secondary mixed-lymphocyte reactions to strong M1s determinants.

Negative and positive selection procedures were used to establish whether the strong proliferative response of T cells to M1sa determinants is H-2 restricted. After negative selection of H-2 determinants in vivo, it was shown that T cells give high primary mixed lymphocyte reactions in vitro to M1sa determinants presented on H-2-incompatible stimulator cells. Other studies demonstrated that (a) negative selection of T cells to M1sa determinants on H-2-incompatible cells removed T cells with specificity for M1sa-bearing H-2-compatible cells, and (b) T cells primed in vitro or in vivo to M1sa determinants on H-2-compatible cells gave high secondary responses to M1sa determinants presented either on H-2-compatible or H-2-incompatible stimulator cells. From these data we conclude that T cells recognize M1sa determinants per se rather than an association of M1sa plus self or allo-H-2 determinants.

Several different cell surface molecules control negative selection of medullary thymocytes.

Repeated attempts to show that costimulation for negative selection is controlled by a single cell surface molecule have been unsuccessful. Thus, negative selection may involve multiple cell surface molecules acting in consort. In support of this idea, we show here that at least three cell surface molecules, namely CD28, CD5, and CD43, contribute to Fas-independent negative selection of the tolerance-susceptible population of heat-stable antigen (HSA)hiCD4+8- cells found in the medulla. The costimulatory function of these three molecules can be blocked by certain cytokines, IL-4 and IL-7, and coinjecting these cytokines with antigen in vivo abolishes negative selection; Fas-dependent negative selection, however, is maintained. The results suggest that efficient negative selection requires the combined functions of at least four cell surface molecules: CD28, CD5, CD43, and Fas.

Role of the H-2 complex in induction of T helper cells in vivo. II Negative selection of discrete subgroups of T cells restricted by I-A and I-A/E determinants.

Previous studies have shown that negative selection of T cells to sheep erythrocytes (SRC) after adoptive transfer to irradiated mice requires a sharing of H-2 determinants between the donor T cells and the selection hosts. This paper examines which part of the H-2 complex controls selection. The results show that, in the case of T cells of the H-2k haplotype, complete selection occurs with donor host matching limited to the I-A through I-E subregions of the H-2 complex. Selection to SRC was partial in I-A compatible, I-E incompatible hosts, minimal or not detectable in I-A incompatible, I-E compatible hosts, but near-complete in hosts matched at both the I-A and I-E subregions. Consecutive selection in hosts matched solely at (a) the I-A subregion and (b) the I-E subregion led to incomplete selection. From these and other findings it is argued that H-2k T cells comprise a mixture of T cells restricted by I-A and I-A/E hybrid molecules.

Role of the H-2 complex in induction of T helper cells in vivo. III. Contribution of the I-E subregion to restriction sites recognized by I-A/E-restricted T cells.

Previous studies on negative selection of T cells to sheep erythrocytes in irradiated mice showed that CBA (I-Ak,I-Ek) (kk) T cells comprise two subgroups of cells restricted by I-A (A alpha-A beta) and I-A/E (E alpha-E beta) molecules. Selection of the I-A/E-restricted by I-A (A alpha-A beta) and I-A/E (E alpha-E beta) molecules. Selection of the I-A/E-restricted subset requires that the donor T cells and the selection host share both I-A (E beta) and I-E (E alpha) gene products; only the I-A-restricted cells undergo selection in B10.A(4R) (kb) mice. This paper demonstrates that negative selection of the I-A/E-restricted subgroup of CBA T cells can occur in F1 hybrids between B10.A(4R) and various Ia.7+ (E alpha+) I-E-incompatible strains; selection does not occur in hybrids between B10.A(4R) and Ia.7- (E alpha-) strains. These data suggest that, despite the fact that E alpha chains display detectable structural allelic variations, these chains are functionally nonpolymorphic. This conclusion applies to E alpha k,d,p,r,j chains. With F1 hybrids between B10.A(4R) and another Ia.7+ strain, B10.PL (H-2u), in contrast, only intermediate selection is observed. This finding is consistent with recent evidence that cell surface expression of E alpha-u-E beta dimers displays strong cis preference. In contrast to E alpha+ CBA T cells, E alpha- B10.A(4R) (kb) T cells undergo complete negative selection in hosts matched only in the I-A (and H-2K) subregion, i.e., B10.BR (kk) mice; no selection occurs in B10 (bb) mice. These data imply that Ia-restricted T cells in E alpha- strains are probably restricted solely by I-A molecules.