Recently, using a global method of T cell repertoire analysis, we showed that purified naive T cells confronted in vitro with allogeneic APCs in a direct pathway-restricted MLR up-regulate their Vβ mRNAs without exhibiting skewing of complementarity-determining region 3 (CDR3) length distribution. In this report, using this approach, we show in vivo that Vβ transcript regulation and CDR3 length distribution follow the same pattern during acute rejection of MHC-incompatible heart allografts. In contrast, in tolerance induction by priming of recipients with donor cells, the vigorous Vβ mRNA accumulation with Gaussian CDR3 length distribution is abolished, providing a possible explanation for the down-regulation of activated T cells in tolerant animals. In addition, tolerated grafts harbor T cells with a highly altered repertoire, suggestive of self-restricted presentation with some patterns corresponding to previously identified regulatory cells.
Although T cells require multiple cosignals to induce clonal expansion and maturation into effector cells, the initial selection of committed cells that follows TCR interaction with processed peptides presented on MHC molecules is the major event of a T cell-mediated response. This TCR selection can be followed by the analysis of altered TCR clonotypic patterns or complementarity-determining region 3 (CDR3)3 sequences (1). Such qualitative analyses (Immunoscope/Spectratype) have provided us with important information in various experimental and clinical situations, including autoimmune (2), infectious (3, 4), and lymphoproliferative (5) diseases. However, they are limited by several constraints. First, due to the variety of MHC restriction determinants that in the thymus differently shape the TCR among individuals, “public” TCR alterations (i.e., those reproduced in different individuals) are rare events, usually restricted to inbred populations (1, 6). Even in such instances, epitope spreading, which develops during T cell responses in vivo (7), further complicates the TCR repertoire patterns. Moreover, the cross-reactive nature of the TCR makes difficult to define a clear structure-function relationship at the TCR level (8). Thus, our understanding of the relevance of TCR alterations should be enhanced by knowledge of the amounts of mRNA involved in these repertoire biases (9, 10). Moreover, in allorecognition the situation is further complicated by superimposing recognition of allogeneic MHC expressed by foreign APC, referred to as direct recognition (11), upon the self-restricted recognition of allogeneic peptides presented by recipient APCs, referred to as indirect recognition (12). Indeed, as for self-restricted patterns, direct recognition can trigger rapid acute rejection of allografts (13, 14).
In this paper, for the first time, we use a new global approach, referred to as TcLandscape (9, 10), to study in vivo T cell mobilization by an allograft through analysis of Vβ mRNA transcription. Qualitative alterations of Vβ use are correlated with the magnitude of accumulation of each Vβ mRNA species involved in a T cell response and can be represented as a Vβ transcriptome “landscape” for each informative time point of a T cell immune response (9, 10). Using this approach, we revisited the complex in vivo T cell response during acute rejection of heart allografts placed in unmodified recipients and during induction of tolerance by donor-specific blood transfusion (DST) in the same combination (15). We show that graft-infiltrating T cells (GITC) in acutely rejected hearts are characterized by vigorously up-regulated, but unaltered CDR3 length distribution patterns of Vβ mRNA, also typical of the pattern observed in vitro during the strict direct recognition of foreign APC by naive T cells (9). Moreover, we show that this strong Gaussian Vβ mRNA accumulation is inhibited in heart allografts during DST tolerance induction where highly altered Vβ mRNA patterns are observed.
Materials and Methods
Adult male rats, 8–12 wk old, from the LEW.1W (RT.1u), LEW.1A (RT.1a), and LEW (RT.1l) congenic strains, were purchased from Janvier (Savigny/Orge, France).
Model of acute rejection and DST-induced tolerance of heart allografts
In most of the studies reported in this work, LEW.1W rats were used as heart and blood donors, and LEW.1A rats were used as recipients. In one study performed on day 5 of the rejection process, LEW.1W rats were used as heart recipients, and LEW.1A rats were used as donors.
Heterotopic cardiac transplantations were performed as described by Ono and Lindsey (16). DST-treated rats received i.v. injections of 1 ml donor blood, collected by cardiac puncture in a heparinized syringe (20 IU/ml), on days 14 and 7 before cardiac transplantation. In this model, untreated rats reject their grafts in 6.4 ± 0.3 days, whereas DST-treated rats become specifically tolerant, rejecting third-party, but not donor-derived, skin on day 100 (17). Graft function was evaluated daily by abdominal palpation. The first group was composed of rejected grafts from untreated animals, whereas the second group was composed of grafts from DST-induced tolerant animals. Hearts were harvested on days 1, 3, 5, and 7 (i.e., until rejection).
Comparison of the values of Vβ or Cβ-hypoxanthine phosphoribosyl- transferase (HPRT) ratios in spleen vs heart must be avoided, because they reflect dilution of Vβ chain mRNAs by HPRT in the heart. Therefore, proper comparisons must use syngeneic grafts as controls. These syngeneic grafts were performed using LEW.1A rats as heart donors and recipients.
RNA extraction and cDNA synthesis
Total RNA is isolated by the guanidinium isothiocyanate procedure and purified on a cesium chloride gradient (18). RNA (10 μg) was reverse transcribed using a cDNA synthesis kit (Roche, Indianapolis, IN) and was diluted to a final volume of 100 μl.
Qualitative analysis of TCR repertoire alterations
cDNA was amplified by PCR using a Cβ primer and one of the 20 Vβ-specific primers (6). The amplifications were performed in a 9600 PerkinElmer Automate (PE Applied Biosystems, Foster City, CA). PCR amplification conditions were as previously described (19). Each amplification product was used for an elongation reaction using a dye-labeled Cβ primer (6), then heat-denatured, loaded onto a 6% acrylamide-8 M urea gel, and electrophoresed for 5 h using an Applied Biosystems 373A DNA sequencer (PerkinElmer).
Immunoscope software (Institut Pasteur, Paris, France) provides distribution profiles of CDR3 lengths, in amino acids, of the amplified and elongated products (1). Each profile is composed of between seven and 11 peaks, spaced by three nucleotides, corresponding to seven to 11 possible lengths of the CDR3 region. A unique length of the CDR3 is not necessarily associated with the same sequence, and the number of transcripts with a given length of CDR3 is proportional to the area under the peak (1). Some CDR3 length increasing in some Vβ families in the polyclonal T cell background, which shows a Gaussian profile, are defined as alterations. When an expansion is present in all or several individuals, the response is “public,” whereas when it is only present in one individual, the response is defined as “private.” In contrast, the absence of mRNA of a given (or several) Vβ family is referred to as TCR restriction.
Relative quantification of Vβ transcripts
The ABI PRISM 7700 sequence detection application program (PE Applied Biosystems) was used to detect and measure fluorescence emitted during PCR amplification of a given target sequence in a 96-well reaction plate. Data were collected during each PCR cycle, which were conducted in real-time. Direct detection of PCR products was monitored by measuring the increase in fluorescence caused by the binding of SYBR Green (PE Applied Biosystems) to dsDNA. The level of fluorescence was then directly proportional to the level of PCR product. A sample of known concentration, used to draw a standard curve, permitted measurement of the amount of target sequence in each sample. To normalize the levels of the target sequences, the quantity of each Vβ transcript was divided by the quantity of HPRT transcripts obtained from each sample.
Each Vβ and Cβ standard was established by amplifying samples known to contain T cells (data not shown) using specific Vβ and Cβ primers (6). The products of amplification were separated electrophoretically and purified using a gel extraction kit (QIAquick Gel Extraction kit; Qiagen, Hilden, Germany). The OD260 of each standard allowed us to measure the number of copies per milliliter, using the m.w. of the cDNA. A constant amount of cDNA mixed in defined serial dilutions (107, 106, 105, 104, 103, and 102 copies/well) of each standard was amplified to draw the standard curve as described previously (20). The number of copies of the cDNA target sequence was deduced from a comparison of the measured fluorescence with this standard curve. Each sample was analyzed in duplicate.
Graphic representation of integrated qualitative and quantitative TCR alterations
For each animal, the CDR3 length profiles obtained from the Immunoscope analysis were normalized, so that the total area was equal to 1. The profiles obtained from three naive rats used as a control were measured and the mean profile for the 20 Vβ families was used as reference. Then, for each CDR3 length profile experiment, the normalized profile was compared with the sample one, and the difference was plotted on a landscape, according to the method used by Gorochov et al. (4). Percentages of alterations are represented as a color code on the landscapes. This first analysis step only gives qualitative information, because there is no indication of the amount of Vβ altered. To combine the qualitative alterations with the level of Vβ mRNAs involved, the values obtained from the quantitative TaqMan analysis are included in these landscapes, where the tops of the peaks represent the amount of a given CDR3 size among each Vβ family. MatLab software was used to compute and display the data. In the integrated landscapes, referred to as a TcLandscape (for T cell landscape) developed in our laboratory (9, 10), the x-axis displays the 20 Vβ families analyzed. The z-axis shows the ratio of the number of Vβ transcripts to the number of HPRT transcripts (Vβ transcripts/HPRT transcripts). The y-axis gives the 10 possible CDR3 lengths, and colors represent the percentage of alterations. The color range is deep blue (value, −50%) to dark red (50%) and is indicated on the right of the figure. As ratio values are always higher than zero, down-regulated mRNA species are compacted to the baseline level and visually underscored by comparison with strongly accumulated Vβ mRNA.
TcLandscapes quantify CDR3-altered Vβ transcripts and also detect global Vβ mRNA regulation of nonaltered transcript populations
Most of the TCR Vβ chain analyses were performed using qualitative methods (Immunoscope/Spectratype, the latter developed in our laboratory). However, only a combined quantitative analysis of Vβ transcripts allows understanding the meaning of altered or unaltered Vβ profiles. Two contrasted examples are provided in Fig. 1 only to illustrate the type of information made available by TcLandscape analysis. These examples correspond, at the left of the figure, to highly altered repertoire found in tolerant animals (Fig. 1, a–c, and see During induction of DST-mediated tolerance, TCR landscapes are dominated by highly altered patterns, with down-regulation of the accumulation of unaltered Vβ transcripts) and, at the right of the figure, to Gaussian Vβ up-regulation following Con A stimulation (9) (Fig. 1, d–f).
Immunoscope analysis (1) allows identifying, in each Vβ family separately, modifications (Fig. 1,a) in the Gaussian CDR3 length distribution pattern typically observed in normal animals (Fig. 1,d). These CDR3 length alterations observed in some Vβ families can be visualized in graphics (Reperturb) (4) that allow a global assessment of total Vβ alterations (Fig. 1, b and e) in each sample (4). TcLandscape, by combining this qualitative study (color code) with the quantitative analysis (height of the peaks) of each Vβ family, gives a global visual assessment of the amount of altered and unaltered Vβ transcriptome (Fig. 1, c and f). For instance, in the landscape shown in Fig. 1,c, the Vβ18 family, although strongly altered (see Fig. 1,a given by Immunoscope and Fig. 1,b given by Reperturb) and thus colored red (see Materials and Methods), is slightly accumulated (see z-axis; Vβ18:HPRT ratio for each of the 10 CDR3 length possibilities, y-axis). In contrast, Fig. 1,f shows an example in which T cells are polyclonally activated (such patterns can be observed in Con A stimulation, or under direct MLR conditions (9)). Neither Immunoscope (Fig. 1,d) nor Reperturb (Fig. 1,e) can differentiate a resting pattern from strongly and polyclonally stimulated T cells (19). In this situation our method (Fig. 1, c and f) allows a global visual appreciation of the magnitude of Vβ mRNA accumulation (z-axis) of altered and unaltered profiles.
TCR landscape patterns of resting T cells are comparable in inbred animals
Fig. 2 shows examples of landscapes of resting splenocytes in normal adult rats. Several congenic strains sharing the complete Lewis background, but differing at their MHC loci (RT1a, RT1u, and RT1l for the LEW.1A, LEW.1W, and LEW rat examples, respectively), were studied. As shown by the Vβ/HPRT transcript ratio reliefs (z-axis) and from their distribution on the y-axis, all patterns are Gaussian in term of CDR3 length usage, with a uniform six-nucleotide dominant length. This characteristic is also illustrated by the uniform green color of the reliefs in the graphic representation, which corresponds to <10% perturbation (very low level). This is the well-established pattern for resting T lymphocytes obtained by qualitative analyses (1). When the magnitude of mRNA amplification is considered (z-axis), an almost stereotyped topology is observed in LEW.1A rats, involving dominant Vβ2, -4, -6, -8, -10, -14, and -19 families, with two major species (Vβ10 and -14; Fig. 2,a). The LEW.1W rats show typical dominant Vβ2, -4, -5, -10, -14, and -19 families with two major species (Vβ4 and -14; Fig. 2,b), whereas Vβ3, -4, -6, -12, -14, -16, and -19 families, with three other major species (Vβ3, -6, and -14), are dominant in LEW rats (Fig. 2 c). Thus, despite almost stereotyped topologies within an inbred strain, some differences are observed when congenic strains with different MHC are compared.
Acute rejection of MHC-mismatched allografts by unmodified recipients mobilizes a majority of Vβ families of the T cell repertoire with a Gaussian CDR3 length distribution of Vβ mRNA
First, we studied in vivo the global regulation of TCR Cβ mRNA in rejected hearts in the LEW.1W→LEW.1A combination (through Cβ/HPRT transcript ratios, Fig. 3 a). During the first 3 days following transplantation, the level of Cβ transcript accumulation in T cells of rejecting hearts was roughly of the same magnitude as that observed in T cells infiltrating syngeneic grafts (LEW.1A→LEW.1A combination) that were used as controls. Then Cβ accumulation rose to peak on day 5 and fell at rejection, probably reflecting the extensive tissue necrosis.
Fig. 3,b shows, on two representative examples from four experiments, the kinetics of TCR Vβ landscape patterns during acute heart rejection. As assessed by Vβ/HPRT transcript ratios and fitting with the Cβ global profile, Vβ mRNA accumulation was observed after day 3 and was maximum on day 5 (when most of the T cell activation also takes place in this model (21, 22)), regressing on day 7. The maximum accumulation of Vβ mRNA immediately preceded (day 5) the definitive rejection (6.4 ± 0.3 days). However, not all Vβ family transcripts contributed similarly to the global Cβ accumulation. Indeed, some families (Vβ2, -4, -5, -9, -14, -15, and -16) represented most of the Vβ accumulation, whereas others (Vβ1, -7, -8, -18, and -20, for example) were less mobilized. Fig. 4 shows that in the patterns observed in the reverse combination (LEW.1A→LEW.1W), the strong Vβ mRNA accumulation observed on day 5 was also Gaussian. However, interestingly, the dominant families involved were different from those found in the LEW.1W→LEW.1A combination, indicating that Vβ families, despite their Gaussian CDR3 profile, are not mobilized at random. Interestingly, the strong unaltered Vβ mRNA accumulation observed in the graft on day 5 mimicked the profiles we recently observed in vitro in MLR performed in conditions only allowing the direct pathway of allorecognition to occur (9), and this suggests that direct recognition pathways are instrumental in acute allograft rejection. However, in the in vivo situation, up-regulation of Vβ mRNA families engaged in recognition of allopeptide presented by self APC (indirect pathway of recognition) is also likely to occur, but is probably masked by the superimposed Gaussian accumulation of Vβ transcripts. Nevertheless, a few examples of altered (CDR3 length biases) Vβ families were detected, indicated by the appearance of red spots on the TcLandscape representation (Fig. 3 b). However, they involved only extremely low levels of Vβ mRNA accumulation. Two kinds of altered patterns were observed, among which very early alterations (day 1) hardly conform to a classical indirect recognition pathway for timing reasons. Those alterations probably result from the small number of T cells in the infiltrate on day 1 in allogeneic grafts, also observed in syngeneic grafts (data not shown), which may be related to a phenomenon of dilution of TCR mRNA in poorly infected hearts. Alternatively, some pre-existing memory T cells could also lead to such patterns. The second type involves minor and inconstant modifications that occur later (day 7), when the rejection process has destroyed the graft, and probably reflect the expanding self-restricted pathway of allorecognition.
During induction of DST-mediated tolerance, TCR landscapes are dominated by highly altered patterns, with down-regulation of the accumulation of unaltered Vβ transcripts
The kinetics of Cβ/HPRT transcript ratios observed in DST-induced tolerance were very different from the rejecting ones, with an early and sustained mRNA accumulation until day 5, which decreased thereafter (Fig. 5,a). The difference in mRNA accumulation between tolerant and rejecting hearts was highly significant on day 5 at the Cβ level (p < 0.05). This difference was also observed at the level of most of the Vβ families, as shown in Fig. 6, where each mRNA species (from rejecting and tolerated hearts) is compared with Vβ transcript accumulation in syngeneic grafts. Significantly regulated Vβ mRNA values (p < 0.05) are plotted as a filled square on a grid encompassing the entire TCRβ transcriptome (9). Twenty percent of the possible CDR3 lengths of the Vβ transcripts was significantly regulated in GITC from rejecting hearts (Fig. 6,a), whereas only 2% of the possible mRNA species underwent significant changes in DST-induced tolerance (Fig. 6 b).
Again, the global Cβ profile did not reflect the complex Vβ transcriptome patterns. Fig. 5,b shows two representative examples from four experiments of TcLandscape of GITC of tolerated hearts following the DST induction protocol. As a first general statement, the patterns were in sharp contrast with those observed during rejection, with strong and early CDR3 length alterations in different Vβ families. This accumulation of altered Vβ families was highly suggestive of a self-restricted presentation pathway following DST priming (exemplified on the graph by the red peaks, which are almost absent in patterns of acute rejection in unmodified hosts, as shown in Fig. 3,b). These altered patterns of different Vβ families, serially observed in time, are reminiscent of a spreading-type phenomenon. Interestingly, we have previously used RNase protection assays to quantify a unique clonotypic alteration (Vβ18 Dβ1 Jβ2.7), with the same CDR3 nucleotide sequence, identified in all tolerated hearts of DST-treated animals tested under the same conditions (6). TcLandscape shaped by DST also showed Vβ18 alterations triggered by LEW.1W DST priming, with the appearance of red spots as early as day 1 following LEW.1W heart graft and sustained until day 7 (Fig. 5,b). The possibility that such a self-restricted pattern is related to the presence of regulatory T cells that play a role in the induction of tolerance is also suggested by our recent observation that vaccination using Vβ18-encoding cDNA abrogates the tolerance state (23). In addition, another modification observed in TCR landscapes of graft-infiltrating cells from tolerated organs is the concomitant absence of the vigorous Gaussian-type Vβ mRNA accumulation, normally associated with the rejection process (see Fig. 3 b, day 5). Therefore, the profile of Vβ mRNA accumulation during tolerance induction is congruent with a down-regulation of direct stimulation. It is possible that this down-regulation precedes (and explains) the broad inhibition (involving both Th1 and Th2 cells) of T cell activation previously reported in this model of tolerance induction (21, 22).
Interestingly, the TcLandscape patterns obtained from the spleen of rejecting or tolerant hearts do not differ from those observed in naive animals (data not shown). This suggests that most of altered (potentially regulating) cells need the graft structure for differentiation and expansion.
In this report, we revisited the T cell mobilization in rejected and tolerated allografts using a combination of qualitative and quantitative assessments of all CDR3 length-restricted mRNAs in each Vβ family. The method used has several advantages over classical qualitative approaches (9, 10). Indeed, without concomitant quantitative information, an analysis restricted to qualitative CDR3 length alterations does not discriminate polyclonal regulation (observed in Con A or superantigen stimulation) in which no CDR3 length biases are expected from a resting stage (9). Using this method, which allows quantifying both CDR3-altered and polyclonal responses, we were able to show that rejection is associated with a vigorous Gaussian accumulation of Vβ chain mRNA in TCR GITC, whereas a down-regulation of Cβ transcripts and a strongly altered pattern are observed in tolerated grafts.
The TcLandscape resting patterns of inbred animals are of an almost stereotyped Gaussian polyclonal type, as expected from previous studies of CDR3 length distribution in resting T cell populations (1). However, landscape topologies (Vβ mRNA accumulation) vary somewhat among congenic rat strains with LEW background but different MHCs. These different patterns could have been shaped by thymus selection processes as well as by subtle preimmune constraints in the TCR repertoire (24). Interestingly, whereas T cell activation is characterized by a down-regulation of membranous TCR (25), TCR chain mRNA accumulation is observed (9, 26, 27). However, Vβ transcript accumulation is the result of a composite regulation depending on both activation (26) and kinetics of T cells infiltrating the graft. In the model used in this paper, the infiltrate is maximum on day 5 during rejection and tolerance induction following DST priming (22, 28), and T cells are found at similar levels in rejected and tolerated grafts (22, 28, 29). Thus, as would also be the case for study of the transcripts of T cell-related cytokines, this approach gives an integrated value of the flux of immigrant T cells in the graft (similar in rejected and tolerant grafts (21, 22)) as well as an idea of the magnitude of their Vβ transcription regulation.
A first important observation that emerges from this global appraisal of TCR during acute heart rejection is that strong Vβ chain accumulation occurs in the absence of alterations of CDR3 length distribution among all tested Vβ families. Interestingly, the Gaussian CDR3 length mRNA distribution and the transcript accumulation observed in vivo during acute rejection closely mimic what we recently described in vitro in an MLR performed to select experimental conditions of strict direct allorecognition pathway (9). This is in agreement with the fact that TCR recognition of alloligands may involve more molecular interactions with MHC framework determinants (30, 31) than in the CDR3-restricted recognition of self-MHC molecules (31, 32, 33). Such peptide-dependent (but not peptide-specific) interactions with foreign MHC framework moieties would be the basis of the high frequency of alloreactive T cells. Finally, allo-MHC molecules have been shown to elicit an alloreactive cytotoxic response even in the absence of peptide (34, 35). However, we do not exclude that the diversity and the amounts of the allodeterminants from a heart allograft could result in a polyclonal-type pattern. The unaltered CDR3 patterns observed during the rejection process are in agreement with the hypothesis of an unexpected proportion of responding naive T cells, possibly amplified by TCR cross-reaction (8). However, and in contrast to direct-restricted MLR, both direct and indirect pathways are involved in vivo during acute rejection (14). It is therefore possible that the expected altered pattern of self-restricted allorecognition that develops when an acellular extract of allogeneic tissue is injected into a LEW.1A rat (9) is masked by the vigorous Gaussian (direct-type) response observed in rejection, or that it has not had time to occur due to the rapid rejection process (days 6 and 7). Our observations allow a better understanding of why the response of a naive (i.e., unprimed) recipient against a graft is so vigorous and can result in a definitive acute rejection of a large transplant within a very short time (6–7 and 5 days for heart and kidney allografts, respectively, in this combination). In fact, a large proportion of T cells driven by the direct recognition pathway are probably involved in this process, whereas only a few specifically committed T cells, with a strongly altered TCR pattern that depends on self-restricted presentation by self APC, would be involved in the delayed or chronic rejection process, as observed in both rats (19) and humans (20). However, not all Vβ family mRNAs are overexpressed. Some families have unchanged mRNA levels, contrasting with the strong accumulation of others, and the profiles, despite Gaussian, are not produced randomly.
In contrast to the acute rejection, the patterns exhibited by tolerated hearts following DST are strongly altered, suggesting the development of a recall indirect pathway type of T cell response following transplantation after DST priming. This interpretation is in agreement with the early kinetics of these TCR alterations, present as early as day 3, with strongly altered patterns on day 5, in contrast to the Gaussian activation profile observed in rejected hearts at the same time. In addition, and possibly instrumental in the mechanism of graft acceptance, the vigorous day 5 Gaussian Vβ mRNA accumulation consistently observed in unmodified rejecting recipients is inhibited in DST-primed rats despite a strong T cell infiltrate (22, 36) and the absence of clonal deletion of specific alloreactive cells in this model (37, 38). This pattern fits with (and possibly controls) the inhibition of Th1- and Th2-related cytokines observed on day 5 following transplantation in this model of tolerance induction (21, 22). The idea that the T cells with altered CDR3 length distribution that accumulate in tolerated hearts are regulatory cells is an attractive possibility. Graft-resident dendritic cells, the only class II+ APC of normal rat hearts (39), are likely to be involved in this process, as they have been shown to be required in the graft for tolerance induction following DST (40). Alternatively, putative regulatory cells with strongly skewed Vβ usage may regulate the capacity of graft dendritic cells to stimulate naive T cells. Indeed, regulatory T cell clones can also impact on stimulating properties of APC (11). A role for these clones is shown by the fact that vaccination using the plasmid encoding Vβ18 cDNA, one of the most strongly altered families (6), with identical CDR3 sequences, found in DST-treated recipients from days 1–7 (Fig. 5), prevents tolerance induction in this model (23). Interestingly, the altered patterns were only found in the heart and not in the spleen. This is in agreement with the concept that regulatory T cells require the presence of the tolerated tissues. Indeed, only retransplantation of the tolerated allogeneic tissue into T cell-deprived mice restores T cells with regulatory capabilities (41). The exact mechanism of such regulation remains elusive. However, our data cannot rule out the possibility that other T cell subsets, such as memory or effector cells, are also present in the graft and contribute to build the altered profiles observed. Indeed, despite the findings that DST-treated tolerant recipients can specifically accept a LEW.1A skin graft and can, as described in the model used by Zhai et al. (42), generate splenocytes able to transfer tolerance to naive hosts, they nevertheless present unambiguous chronic rejection at histological examination over the long term (K. Renaudin and R. Josien, manuscript in preparation), indicating that the two populations may coexist.
We thank H. Smit, V. Proust, and C. Usual for performing the rat heterotopic allotransplantations, and J. M. Heslan for technical help with real-time PCR. We thank G. Benichou (Harvard Medical School, Boston, MA), A. Saoudi (Institut National de la Santé et de la Recherche Médical, Unité 28, Toulouse, France), B. Rocha (Institut National de la Santé et de la Recherche Médical, Unité 345, Paris, France), and R. Josien (Institut National de la Santé et de la Recherche Médical, Unité 437, Nantes, France) for discussion of the data and criticism of the manuscript.
This work was supported in part by Fondation Transvie, Société Française de Transplantation, Etablissement Français des Greffes, and Association Française de Lutte contre la Mucoviscidose.
Abbreviations used in this paper: CDR3, complementarity-determining region 3; DST, donor-specific transfusion; GITC, graft-infiltrating T cell; HPRT, hypoxanthine phosphoribosyltransferase; LDA, limiting dilution assay; LEW, Lewis.