Altered Distribution of H60 Minor H Antigen-Specific CD8 T Cells and Attenuated Chronic Vasculopathy in Minor Histocompatibility Antigen Mismatched Heart Transplantation in Cxcr3^−/− Mouse Recipients

Alloreactive T cell recruitment into the transplanted organ is a critical step in graft rejection. Adhesion molecules, chemokines and their receptors dictate this recruitment. During antigenic stimulation, alloreactive effector cells up-regulate the receptors for inflammation-induced endothelial adhesion molecules and inflammatory chemokines (1, 2, 3, 4). Thus, they traffic into the target organ via a chemokine gradient that is produced from local inflammation due to ischemia/reperfusion (I/R)² injury followed by stimulation of innate immune cells (4, 5). Distinctive chemokine receptors on Th1 effector cells include Cxcr3 and Ccr5, both of which bind inflammatory chemokines. Rheumatoid arthritis and multiple sclerosis (Th1-related disease) patients show high enrichment of T cells in inflamed areas in that virtually all infiltrating T cells express Ccr5 and Cxcr3 (6, 7). The importance of chemokines and their receptors in transplantation was first noted in studies showing an up-regulation of the chemokine genes in rejecting allografts such as CCl5 (RANTES), CXCL10 (IFN-inducible protein 10, (IP-10)), and CXCL9 (monokine induced by IFN-γ (Mig)) in murine models (8, 9). In accordance with these gene expression patterns, targeting (blocking/deleting) Cxcr3 or Ccr5 has been shown to prolong heart allograft survival (10, 11, 12). Heart grafts showed indefinite survival when they were treated with low-dose cyclosporin A in Cxcr3^−/− and Ccr5^−/− recipients in a murine model. Studies using blocking Abs and small molecular antagonists in the same model demonstrated a survival benefit as well (10, 13).

However, in contrast to these reports, recent studies have shown no cardiac allograft prolongation in Ccr5^−/− mice (14, 15) and unaltered lymphocyte infiltration into the area of inflammation in Cxcr3^−/− mice. Liu et al. (16) showed that Cxcr3^−/− mice exhibited increased severity of experimental autoimmune encephalomyelitis compared with wild-type mice and no difference was observed in CNS-infiltrating leukocytes. Wareing et al. (17) found identical leukocyte infiltration in Cxcr3^−/− mice during influenza infection. Similarly, previous studies from our group and another group showed neither a beneficial effect on graft survival nor any alteration of T cell trafficking in Cxcr3^−/− major mismatched cardiac allograft recipients (18, 19). In these studies, there was no reduction of T cell infiltration at the time of rejection (or on a designated date before rejection) compared with date-matched wild-type recipients after fully MHC mismatched heart transplantation (BALB/c to C57BL/6 cardiac allograft). These results raised questions regarding the impact of Cxcr3 on alloreactive T cell trafficking into the graft. Increased Ccr5⁺ CD8⁺ graft infiltrating cells in Cxcr3^−/− recipients suggested that compensatory T cell accumulation into the graft occurs when a relatively high amount of heterogeneous alloreactive T cells are involved in trafficking, which is a typical characteristic of stringent mouse combinations in transplantation. Therefore, we hypothesized that an attenuated rejection model using strain combinations with smaller repertoire of potentially donor-reactive T cells would allow us to demonstrate an influence of Cxcr3 blockade.

We addressed the roles of the chemokine receptor Cxcr3 using MHC matched/multiple minor Ag mismatched mouse combinations. Several minor H Ags (miHAs) are known as dominant miHA between C57BL/6 and BALB/b mice. Immunization of BALB/b miHAs to B6 mice results in CTL responses mainly to the immunodominant peptides H4^b (SGIVYIHL, H-2K^b-restricted, hemopoietic and neural distribution), H7^b (KAPDNRDTL, H-2D^b-restricted, ubiquitous), H13^b (SSVIGVWYL, H-2D^b-restricted spleen and thymus), H28 (FILENFRPL, H-2K^b-restricted, macrophage and dendritic cells), and H60 (LTFNYRNL, H-2K^b-restricted, MICA homolog and hemopoietic) miHA peptides (20). Among these minor Ags, the H60 miHA predominates in C57BL/6 immune responses to splenocytes from BALB/b throughout primary and secondary challenges (21, 22). In this study, we used a less stringent, multiple minor Ag mismatched mouse cardiac allograft model to investigate the role of Cxcr3 on both acute and chronic rejection. Our studies demonstrate that when compared with control wild-type mice, Cxcr3^−/− mice showed an unaltered recruitment of alloreactive CD8 T cells into the graft with only a marginally prolonged graft survival. However, when we adoptively transferred an equal number of wild-type and Cxcr3^−/− CD8 T cells to allograft-bearing B6 scid recipients, there was less infiltration into the graft, but more cells remained in the peripheral blood in the absence of Cxcr3. Furthermore, Cxcr3^−/− recipients showed significantly less histologic evidence of chronic rejection with less CD8 T cell infiltration. Retention of alloreactive CD8 T cells in the peripheral blood as a result of knocking out Cxcr3 impairs alloreactive T cell trafficking into the transplanted tissue.

The 6- to 8-wk-old male BALB/c (H-2^d), BALB/b (H-2^b), C57BL/6 (H-2^b), and B6.CB17-Prkdc^scid/SzJ (H-2^b) were purchased from The Jackson Laboratory. Cxcr3^−/− mice (H-2^b), backcrossed 10 generations onto C57BL/6 background, were obtained from Taconic Farms. Mice were housed in enclosed filter-top plastic cages with controlled light/dark cycles and provided ad libitum with food and water. Experiments were performed in accordance with the National Institutes of Health and U.S. Department of Agriculture guidelines, after the approval of The University of Wisconsin Institutional Animal Care and Use Committee.

BALB/c (H-2^d) or BALB/b (H-2^b) donor hearts were transplanted into C57BL/6 (H-2^b) recipients. Heart transplantation was performed using a modification of the methods described by Corry et al. (23). Briefly, the recipient aorta and vena cava were prepared before the donor operation. Donor hearts were transplanted into the abdominal cavity of the recipients after a short period of cold ischemia in Euro-Collins solution. Donor aorta and pulmonary artery were anastomosed in an end-side fashion to the infrarenal aorta and vena cava as the inflow and outflow vessels for circulation, using running 10/0 nonabsorbable monofilament sutures. The grafts were monitored by daily palpation and graded from 4+ (strong beat) to 0 (no beat), which was confirmed by laparotomy at the time of sacrifice. Animals were sacrificed on the day of rejection (cessation of beating) or a designated day.

Transplanted mouse heart grafts were procured and used for paraffin and frozen sectioning. Paraffin sections were stained with routine H&E. Frozen samples were embedded in OCT compound (Sakura Finetek), cut at 5 microns, and fixed with acetone or 4% paraformaldehyde for 5 min. Sections were blocked for nonspecific staining using 10% normal donkey serum in TBS with 1% BSA for 1 h followed by 5% nonfat milk in PBS, 30 min. Rat anti-mouse primary Abs against T cell markers CD4, CD8 (BD Pharmingen) were incubated overnight at 4°C. Endogenous peroxidase quenching was performed postprimary Ab incubation with 1% hydrogen peroxide in TBS for 10 min. A donkey anti-rat IgG-HRP conjugated secondary Ab from The Jackson Laboratory was used for detection and incubated for 45 min at room temperature. The signal was visualized using diaminobenzidine chromogen (DakoCytomation) followed by counterstaining with hematoxylin. Slides were viewed using light microscopy. The International Society for Heart and Lung Transplantation classification was used for scoring the presence and the degree of rejection (24). The number of cells staining positive was counted from five random fields per graft sections. Seven different grafts were used in each group.

Spleen was harvested and placed into a single cell suspension in RPMI 1640 (Invitrogen Life Technologies) by passing through a cell strainer (BD Labware). Explanted heart grafts were minced and digested for 30 min in liberase CI purified enzyme blend (0.45 mg/ml; Roche Diagnostics) in RPMI 1640 medium at 37°C. Mononuclear cells were isolated via density gradient centrifugation using lymphocyte separation medium (Mediatech) according to the manufacturer’s instructions. ACK lysing buffer (Cambrex) was applied when it was necessary. Lymphocytes were counted using a hemocytometer under light microscope. Cells were resuspended in FACS buffer (PBS containing 2% FBS and 0.09% NaN₃) and stained with conjugated biotin, PE, FITC, PerCp, or allophycocyanin Abs directed at mouse CD4 (H129.19), CD8 (53-6.7), Cxcr3 (220803), NK1.1, CD44 (IM7), CD11a (2D7), CD62L (MEL-14), and isotype controls (BD Pharmingen). PE-conjugated MHC class I tetramer (Beckman Coulter) or Pentamer (Proimmune) was costained. Cytometric analysis was performed using a FACSCaliber cytometer (BD Biosciences) and analyzed using CellQuest (BD Biosciences) and FlowJo (Tree Star) software. Absolute cell number of stained cells was calculated based on total lymphocyte count.

To obtain alloreactive CD8 T cells, wild-type and Cxcr3^−/− C57BL/6 mice were i.v. immunized with 2 × 10⁷ BALB/b splenocytes 10 days before harvesting the spleen. T cells were enriched from immunized mouse spleen. Briefly, spleens were mechanically dissociated, and cell suspension was subjected to hypotonic lysis of RBCs. CD3⁺CD8⁺ T cells (purity >98%) were sorted by FACSVantage (BD Biosciences). A total of 2 × 10⁶ enriched cells were i.v. injected into posttransplantation on day (POD)10 B6.CB17-Prkdc^scid/SzJ (B6 background SCID mouse) heart graft recipient, which were subject to sacrifice 48 h later.

Total RNA was isolated from cardiac allograft tissue sample by using TRIzol (Invitrogen Life Technologies). Primers (TIB MoBiol) used in PCR were previously described (25) as follows: Mig (CXCL9) 5′-gaa cgg aga tca aac ctg cct-3′ and 5′-tgt agt ctt cct tga acg acga-3′; IP-10 (CXCL10) 5′-cga tga cgg gcc agt gag aatg-3′ and 5′-tca aca cgt ggg cag gat agg ct-3′; IFN-inducible T cell α chemoattractant (I-TAC, CXCL11) 5′-aat tta ccc gag taa cgg ctg-3′; and GAPDH 5′-caa tgt gtc cgt cgt gga-3′ and 5′-gat gcc tgc ttc acc acc-3′. RNA quantification was conducted by real-time RT-PCR with a LightCycler RNA amplification kit SYBR Green I adapted for one-step RT-PCR in glass capillaries using a LightCycler instrument (Roche Diagnostics). For LightCycler reaction a mastermix of the following reaction components was prepared to the indicated end-concentration: 8.0 μl of water, 1.6 μl of MgCl₂ (5 mM), 2 μl of forward primer (0.5 μM), 2 μl of reverse primer (0.5 μM), 4 μl of PCR mix, 0.4 μl of Enzyme mix, and 2 μl of RNA (100 ng). The following LightCycler experimental run protocol was used: reverse transcription (55°C for 20 min, 95°C for 30 s), 40 cycles of denaturation (95°C for 0 s, 20°C/s slope), amplification (55°C for 8 s, 20°C/s slope), extension (72°C for 6 s, 2°C/s slope), melting curve program (65–95°C with a heating rate of 0.1°C/s and continuous fluorescence measurement) and cooling program to 40°C. For the mathematical model it is necessary to determine the crossing points for each transcript. Crossing point is defined as the point at which the fluorescence rises appreciably above the background fluorescence. The “Fit Point Method” was performed in the LightCycler software 3.3 (Roche Diagnostics), at which the crossing point would be measured at constant fluorescence level. Each reaction runs with controls (RNA alone or Templates alone). The relative expression ratio was calculated by the mathematical model of the ΔΔ method using the equation (2^{−[ΔCp(sample − control)]} = 2^−ΔΔCp), where Cp is the crossing point. The amount of mRNA for each gene was normalized to the amount of mRNA for GAPDH in each sample. The ratio (×10³ or ×10²) between each gene of interest and GAPDH is used for comparison.

The grafts from long-term recipients were harvested on day 100 after transplantation. The explanted grafts were bisected sagittally and were fixed in 10% formalin. They were evaluated using H&E and elastic trichrome staining for analyses of arterial intimal lesions (26). Briefly, six grafts from wild-type and Cxcr3^−/− recipients were examined for morphometric analysis. Microscopic images of all vessels (over 50 μm in diameter) in a plane were analyzed and measured with computer-based software (Metamorph 6.1). The lumen area was determined by tracing the perimeter of the lumen, and the intima area was determined by tracing the internal elastic lamina and subtracting the lumen area. Intimal thickening was calculated from captured vessels for each graft and determined as the mean of the neo-intimal index using the equation (intima area/lumen + intima area) × 100). Data are expressed as mean ± SD of six grafts in each group. Vessels with a neo-intimal index over 20% were considered diseased. Their presence was expressed as a percentage for each graft.

Experimental results were analyzed by a GraphPad Prism (GraphPad Software), using the log-rank test for differences in graft survival and unpaired Student t test for other data. All data are presented as mean ± SD. Values of p < 0.05 were considered to be statistically significant.

We previously observed no beneficial effect of targeting Cxcr3 on heart graft survival in a BALB/c to C57BL/6, full MHC mismatch combination (18). To reduce alloimmunity, we performed multiple minor mismatched (BALB/b to C57BL/6) heart transplantation. We found that the grafts in fully MHC mismatched heart transplantation (BALB/c into C57BL/6) were acutely rejected within 9 days, whereas the multiple minor histocompatibility Ag mismatched combination (BALB/b into C57BL/6) produced spontaneous long-term allograft survival without any manipulation in ∼50% of animals with pathological features of chronic rejection. Surprisingly, allograft survival in Cxcr3^−/− mice was not significantly prolonged compared with wild-type mice (56.9 ± 44.0 vs 60.7 ± 41.1; p > 0.05) (Fig. 1,A). Also, the acceptance rate of the heart grafts remained unchanged. However, graft survival between animals experiencing rejection in each group was modestly but significantly prolonged (13.9 ± 3.5 vs 21.34 ± 11.0; p < 0.05). These data indicate that acute rejection is marginally delayed in the absence of Cxcr3. Tissue sections (H&E) prepared from date-matched (POD 10) rejecting allografts showed similar grades of acute rejection, and frozen sections showed similar amounts of CD4 and CD8 T cell infiltration (Fig. 1, B and C).

Allograft rejection is mediated by T cells that accumulate in the transplanted organ. Because Cxcr3 has an important role in chemotaxis of T cells, we hypothesized that T cells are differently distributed in anatomical compartments of allografted mice due to Cxcr3 deficiency. To test this hypothesis, we evaluated three different immune compartments, which are spleen, peripheral blood, and the graft on POD 10. Differential compartmentalization of T cell subpopulations was identified by flow cytometric analysis. Most evidently, CD8 T cells in peripheral blood were significantly increased in Cxcr3^−/− recipients (p < 0.001), which could represent retention of CD8 T cells in blood due to lack of Cxcr3. However, we also found increased numbers of CD8 T cells in spleen of Cxcr3^−/− recipients compared with wild-type recipients (p < 0.01). Interestingly, neither relative ratio nor actual number of intragraft CD8 T cell infiltration was significantly different (Fig. 2). These results show that unlike the fully mismatched model, altered distribution of CD8 T cells in anatomical compartments is seen in the minor mismatched combination in the absence of Cxcr3.

Following transplantation, it has been shown that Cxcr3-expressing CD8 T cells are proportionately increased (27, 28, 29, 30). However, Cxcr3 expression on alloantigen-specific CD8 T cells has not been evaluated. Alloreactive CD8 T cell might have different Cxcr3 expression compared with naive or bystander-activated T cells. BALB/b donor hearts were transplanted into wild-type C57BL/6 recipients. Recipients were sacrificed 10 days after transplantation and splenocytes were stained with Cxcr3 mAbs along with H60 tetramer and CD8 mAbs. Multiple minor mismatched heart transplantation showed increased Cxcr3 expression on CD8 T cells in the spleen after transplantation (Fig. 3,A). Expression of Cxcr3 was dramatically increased on CD8⁺ T cells. However, expression of Cxcr3 on CD4⁺ T cells was not affected significantly by transplantation. These data suggest that minor Ag mismatch does not promote Cxcr3 expression any less than major MHC mismatch. To address the mechanisms responsible for the alloreactive T cell accumulation into the graft, we also measured intragraft Cxcr3 ligands in this multiple minor mismatch combination. Total RNA was extracted from day 10 grafts, and mRNA expression of Mig, IP-10, and I-TAC was analyzed by real-time quantitative PCR. As shown in Fig. 3,B, expression of CXCL9 (Mig) and CXCL10 (IP-10) were significantly increased compared with isograft controls. These data indicate that Cxcr3-driven transmigration exists in the multiple minor mismatched heart transplantation. The BALB/b to C57BL/6 strain combination has an advantage in that their mismatched Ags recognizing CD8 T cells can be traced by multimer technology. It is known that H60 miHA is immunodominant over other miHAs in the BALB/b to C57BL/6 strain combination (21, 22). However, H60 immunodominance has not yet been evaluated after heart transplantation. Therefore, we next determined the number of alloreactive CD8 T cells present in the spleen after transplantation in Cxcr3^−/− and wild-type recipients. We measured H4-, H7-, H13-, and H28-specific CD8 T cells along with H60-specific CD8 T cells using Ag-specific PE-conjugated multimers (tetramers/pentamers). We found that the H60-speciifc CD8 T cells are the most prevalent alloreactive CD8 T cells after BALB/b to C57BL/6 heart transplantation. Interestingly, the hierarchy of immune response to these minor Ags was not altered in the absence of Cxcr3, in that both Cxcr3^−/− and wild-type C57BL/6 recipients showed the order of H60>H4>H7>H13, H28 (Fig. 3,C). In other words, the potency of CD8 responses as measured by minor Ag specific CD8 T cells detected by multimers varied according to this minor Ag hierarchy. Next, the expression of chemokine receptors on immunodominant alloreactive CD8 T cells during transplant immunization was evaluated by using H60 tetramer with a CD8 mAb. We used H60 (LTFNYRNL)/MHC class I tetramer to track alloreactive CD8 T cells. H60 tetramer-positive CD8 T cells were not found in naive C57BL/6 mice but were dramatically expanded after BALB/b heart transplantation. H60-specific CD8 T cells from the recipient splenocytes on POD 10 showed that it was a traditional effector T cell phenotype (CD62L^lowCD11a^highCD44^high) (Fig. 3,D). Uniform and high expression of Cxcr3 on H60 tetramer-positive CD8⁺ T cells was observed in the lymph node, spleen, and blood whereas overall CD8⁺ T cells showed varying degrees of Cxcr3 expression in these locations. This observation showed H60 miHA-specific CD8 T cell population was more homogeneous than the overall CD8 T cell subset. Interestingly, we found that Cxcr3 expression was decreased both on CD8⁺ and H60 tetramer-positive CD8⁺ T cells in the graft compared with other immune compartments (Fig. 3 E). It is possible that Cxcr3 is internalized or down-regulated in the inflamed site (graft). Our results demonstrate that after transplantation, alloreactive CD8 T cells underwent activation, proliferation and modification of surface markers such that they preferentially migrate into the graft.

To assess the trafficking of alloreactive CD8 T cells, but not bystander or nonspecific CD8 T cells, during multiple minor mismatch cardiac allograft rejection, we evaluated H60-specific CD8 T cells in the three immune compartments. The level of H60-specific CD8 T cells remained <3% of the total lymphocytes in spleen and PBMC, but H60-specific CD8 T cells dramatically accumulated at the graft site (up to 12%) in wild-type recipients (Fig. 4, A and B). These data were consistent with previous studies of Ag-specific CD8 T cell responses, which showed rapid movement of recently activated CD8 cells from the draining lymph node to the graft (1) or to the peripheral site of infection (31) without detectable accumulation in lymphoid compartments. However, H60-specific CD8 T cell staining of peripheral blood showed a significant increase of H60-specific CD8 T cells (p < 0.01) in Cxcr3^−/− recipients (Fig. 4,B). Because the retention of alloreactive T cells in the peripheral blood might indicate interference with trafficking of these cells into the graft, we evaluated the H60-specific CD8 T cells in the graft. Despite increased numbers of H60-specific CD8 T cell in the peripheral blood, intragraft infiltration of H60-specific CD8 T cells was not significantly altered in Cxcr3^−/− recipients (Fig. 4, B and C). These data suggest that without Cxcr3, donor-specific CD8 T cells migrate to the peripheral blood but do not efficiently traffic into the inflamed site (graft). To confirm the role of Cxcr3 on CD8 T cell infiltration into the graft, we used an adoptive transfer model for T cell graft infiltration. Wild-type or Cxcr3^−/− CD8 T cells were adoptively transferred into the B6.CB17-Prkdc^scid/SzJ recipients bearing minor histocompatiblity complex mismatched BALB/b heart transplants. The cells were transferred 10 days after transplantation to minimize early surgical and I/R injuries following transplantation. A total of 2 × 10⁶/200 μl of purified CD8 T cells were adoptively transferred to the B6.CB17-Prkdc^scid/SzJ recipients via tail vein (Fig. 5,A). After 48 h, grafts were recovered from the SCID recipients and CD8 T cell infiltration was evaluated. Allografts with adoptive transfer of Cxcr3^−/− CD8 T cells showed significantly less (p < 0.05) CD8 T cell graft infiltration compared with wild-type CD8 T cells (Fig. 5,B). Interestingly, more CD8 T cells were found in the peripheral blood from Cxcr3^−/− CD8 T cells transferred recipients (Fig. 5 C). These data suggest that alloreactive CD8 T cells are dependent on Cxcr3-driven trafficking in a dampened inflammatory milieu.

As shown in Fig. 1,C, 50% of grafts were spontaneously accepted in this multiple minor mismatched heart transplant mouse model. Interestingly, grafts that survived long-term developed chronic rejection at POD 100 (see Fig. 7,A). We hypothesized that the abolishment of Cxcr3-driven alloreactive T cell trafficking would alleviate the development of chronic rejection due to a lower number and more homogeneous T cell population participating and a less complex inflammatory milieu. Longitudinal evaluation of H60-specific CD8 T cells in PBMC showed reduction of Ag-specific CD8 T cells over time after the expansion phase. Even though not significantly different, H60-specific CD8 T cell levels in peripheral blood were always higher in Cxcr3^−/− recipients compared with wild-type recipients (data not shown). When we examined frozen sections to assess infiltration of CD4 and CD8, there was a striking decrease in infiltration by CD8 T cells but not CD4 T cells into the grafts retrieved from Cxcr3^−/− recipients (Fig. 6,B). The hypothesis is based on the assumption that Cxcr3 ligands are still expressed from the long-term surviving graft. We measured intragraft expression of Cxcr3 ligands Mig (CXCL9), IP-10 (CXCL10), and I-TAC (CXCL11). We found Mig and IP-10 were expressed significantly higher than in date-matched isograft controls (Fig. 6,A). To investigate the effect of Cxcr3 on chronic rejection, we evaluated cardiac allografts from long-term (POD 100) recipients. In the MHC-matched but multiple minor-mismatched combinations, luminal occlusion was eminent without any treatment (Fig. 7,A). To quantify chronic allograft vasculopathy, we performed morphometric analysis for neo-intimal hyperplasia. The mean NI percentage of the individual hearts within each group was compared. Elastic trichrome staining revealed significant neo-intimal hyperplasia at POD 100, whereas isografts showed no signs of neo-intimal hyperplasia (<15%). Neo-intimal hyperplasia was significantly reduced in Cxcr3^−/− mice. The degree of luminal occlusion was 60.0 ± 14.0% for wild-type recipients and 39.9 ± 12.6% for Cxcr3^−/− recipients (p = 0.026) (Fig. 7,B). However, diseased vessel percentage of individual grafts was not significantly different (p > 0.05) between groups (Fig. 7 C).

The chemokine receptor Cxcr3 is expressed on different cell types, including T cells, B cells, NK cells, monocytes, dendritic cells, tubular cells, and vascular smooth muscle cells (6, 27, 32, 33, 34, 35, 36, 37, 38, 39, 40). It is known that markedly increased numbers of CD8 T cells express Cxcr3 upon activation (6, 27, 28, 29, 30). In experimental transplant models of skin, cardiac, lung, and small bowel allografts, Cxcr3 played a critical role in allograft rejection, and blockade or absence of Cxcr3 in the host reduced effector T cell infiltration and prolonged survival of the transplanted organs (10, 41, 42, 43). However, we observed unaltered leukocyte infiltration in the graft after fully mismatched heart transplantation from Cxcr3^−/− recipients (18). This observation was consistent with studies from several groups indicating unaltered leukocyte infiltration into the inflammatory site in the absence of Cxcr3 (16, 17). Interestingly, we found that the fraction of Ccr5⁺ cells in the CD8 T cell infiltrate population was increased in the graft. This suggests that there is compensatory trafficking via Ccr5 in the absence of Cxcr3. We conjectured that not only Ccr5 but also other inflammatory chemokine receptors would participate in this compensatory trafficking due to their redundant expression. Abdi et al. (44) also speculated on the compensatory role of other chemokine systems in the absence of a dominant chemokine receptor. Perhaps the characteristics of a fully mismatched allograft, such as high numbers of alloreactive T cells, involvement of direct allorecognition, addition of ischemic injury, heterogeneity of the T cell population with complex chemokine/cytokine milieu overwhelms the effect of abolishment of Cxcr3.

In minor mismatched heart transplantation, genetic removal of Cxcr3 did not affect graft acceptance, or severity of rejection, and only modestly delayed acute rejection (13.9 ± 3.5 vs 21.34 ± 11.0) (Fig. 1,A). Both immunohistochemistry and flow cytometric analysis revealed a similar amount of CD4 and CD8 T cell infiltrations in the graft on POD 10 (Figs. 1 and 2). Interestingly, we found that CD8 T cells were increased in the peripheral blood and spleen of Cxcr3^−/− recipients, whereas CD4 T cells were similar, compared with wild-type recipients (Fig. 2). In this study, we did not measure the recruitment of other cell phenotypes that express Cxcr3 and that could potentially contribute to prolonged graft survival in Cxcr3^−/− recipients. It is possible that less infiltration of other cell types into the graft, despite comparable T cell infiltration on day 10, resulted in a marginal prolongation of graft survival in the absence of Cxcr3.

Less stringency (minor mismatch) does not abolish the expression level of Cxcr3 on CD8 T cells in spleen or intragraft Cxcr3 chemokine expression (Fig. 3, A and B). We found that H60-specific CD8 T cells are the predominant alloreactive T cell population after heart transplantation (Fig. 3,C) similar to donor splenocyte transfusion (21, 22). Extending this observation, we now find that most H60-specific CD8 T cells are CD11a^high, CD44^high, and CD62L^low, indicating that they represent effector cells (Fig. 3,D). Changes in surface molecules are inevitable as T cells leave lymphoid tissue and migrate into inflamed tissue (45, 46). Another notable observation in this study is the lowered expression of Cxcr3 on alloreactive CD8 T cells in the graft compared with corresponding cells in lymph node, blood, and spleen (Fig. 3 E). Lowered Cxcr3 expression on infiltrating T cells was observed in models of eye inflammation (47, 48). Previously, rapid Cxcr3 internalization (49, 50) in the presence of Cxcr3 ligands and Cxcr3 down-regulation (51, 52) via TCR signaling in vitro were identified as mechanisms for this lowered expression. Possibly, the down-regulation of Cxcr3 could allow recruited Ag-specific T cells to be detained in the inflamed site.

We evaluated whether Cxcr3 deficiency had an impact on donor-specific CD8 T cell expansion, trafficking and immunodominance. Surprisingly, alloreactive H60-specific CD8 T cells accumulated in the peripheral blood in Cxcr3^−/− recipients (Fig. 2). However, we also found these cells were increased in the spleen. This phenomenon was not seen in the fully mismatched combination (BALB/c to C57BL/6). We suspect that in full MHC mismatch, alloreactive T cell expansion is saturated while there is some margin to expand more in minor mismatch. More proliferation of both CD4 and CD8 T cells from Cxcr3^−/− mice has been reported (16, 53, 54). However, when we evaluated whether the absence of Cxcr3 altered (limiting/enhancing) CD8 T effector function, ⁵¹Cr release CTL assay showed that the cytotoxic effect of wild-type and Cxcr3^−/− CD8 T cells was not significantly different (data not shown). There was no difference in intragraft infiltration of H60 miHA-specific CD8 T cells despite increased numbers in spleen and peripheral blood, indicating retention of H60 miHA-specific CD8 T cells in Cxcr3^−/− recipients. And enrichment of H60-specific CD8 T cells in peripheral blood in the absence of Cxcr3 could be direct evidence that Cxcr3 is important for alloreactive T cells to enter the graft but not important to dislodge themselves from lymphoid organs. It has been shown that immobilized chemokines (including IP-10 and RANTES) presented by endothelial cells are critical for haptotaxis (55, 56). More recently, oligomeric but not monomeric CXCL10 presented by endothelial cells via GAG was found to be essential for effector T cell recruitment (57). In this study, retention of alloreactive CD8 T cells in the peripheral blood might represent less efficient arresting of alloreactive CD8 T cells on site of extravasation without Cxcr3. One implication of this study is that alloreactive T cells in peripheral blood are not an accurate indication of sensitization or risk of rejection, but rather a measure of interference in trafficking.

Nevertheless, we did not observe less infiltration of CD8 T cells or alloreactive CD8 T cells (H60-specific) in the graft. Adoptive transfer of wild-type vs Cxcr3^−/− T cells has been used to examine the requirement of Cxcr3 expression on T cells for allograft rejection (43). We adoptively transferred wild-type or Cxcr3^−/− CD8 T cells into the B6.CB17-Prkdc^scid/SzJ recipients bearing minor histocompatiblity complex mismatched BALB/b heart grafts. To minimize the I/R and early surgical injuries, the cells were transferred 10 days after transplantation (Fig. 5,A). CD8 T cells were analyzed from the graft and peripheral blood, 48 h after adoptive transfer. Finally, there was significantly less infiltration (p < 0.05) of CD8 T cells into the graft, whereas more CD8 T cells remained (p < 0.01) in the blood in the absence of Cxcr3 (Fig. 5), indicating that Cxcr3 is indispensable in the migration of alloreactive CD8 T cells into the graft.

The lesser degree of stringency in this minor mismatch combination allowed us to evaluate the role of Cxcr3 on long-term allograft function. As mentioned, a high expression of Cxcr3 on alloreactive H60 miHA-specific CD8 T cells might be important for chronic rejection because prolonged exposure to a graft produces continuous inflammatory signals, including chemokines. Chronic rejection is the primary limitation to long-term success in organ transplantation; therefore, understanding the pathogenesis of this process is of major clinical importance. Previous studies have identified Cxcr3 and its ligands, CXCL9 (Mig) and CXCL10 (IP-10), as important mediators of chronic rejection in a lung transplantation model, and obliterative bronchiolitis was reduced in Cxcr3^−/− recipients (58, 59). We also observed increased intragraft Mig and IP-10 mRNA expression from the grafts that had long-term survival (Fig. 6,B). In a mouse heart transplantation model, neutralization of Mig (CXCL9) with antiserum attenuated both neo-intimal hyperplasia and fibrosis (60). More recently, Heller et al. (61) showed the role of IP-10 (CXCL10) and effector T cell accumulation on the development of atherosclerosis. We found that Cxcr3^−/− dampens the development of chronic rejection in minor mismatched heart transplantation (Fig. 7). Neo-intimal hyperplasia was significantly reduced in Cxcr3^−/− compared with wild-type recipients (39.9 ± 12.6% vs 60.0 ± 14.0%, p < 0.05). We found the possibility of peripheral enrichment for H60-specific CD8 T cells maintained in long-term Cxcr3^−/− recipients (not statistically significant). It suggests that Cxcr3 is important for continuous trafficking of recently activated alloreactive CD8 T cells. Interestingly, only CD8 T cell infiltration but not CD4 T cell infiltration was significantly reduced in Cxcr3^−/− recipients compared with wild-type recipients (Fig. 6). As a consequence, these two groups showed different pathologic features in the graft. However, Cxcr3 is also expressed on endothelial cells. Thus its vascular related effects may participate in the vessel growth process (or angiogenesis). We did not rule out an effect of Cxcr3^−/− on vascular smooth muscle or nonhemopoietic cells.

Our studies raise the question of why there is no significant prolongation in Cxcr3^−/− recipients after heart transplantation. It is believed that signaling through Cxcr3 not only attracts T cells but also suppresses T cell responses (62). In the absence of Cxcr3, Ag-specific T cell-APC conjugation potentially stabilizes signal integration, resulting in more proliferation, which overwhelms trafficking interference. Decreased IFN-γ production in Cxcr3^−/− mice also suggests multiple roles of Cxcr3 (63). In other words, knocking out Cxcr3 may alter functions other than chemotaxis. The results of the current study show that the absence of Cxcr3 expression on recipient cells skewed donor-specific CD8 T cell proliferation as well as the localization of alloreactive CD8 T cells after multiple minor mismatched heart transplantation. Recent literature has also shown that Cxcr3^−/− T cells were more proliferative yet produced less IFN-γ against Ag-specific stimulation in a mouse experimental autoimmune encephalomyelitis model (16). It is possible that increased expansion of alloreactive CD8 T cells compensates for inefficient trafficking and results in unaltered alloreactive T cell graft infiltration and only marginally delayed acute rejection. However, we have demonstrated that knocking out Cxcr3 alone has a significant effect on the T cell graft infiltration without any factors compensating for it (I/R injury, expansion, etc.), and that it reduces the development of chronic allograft vasculopathy in the long-surviving allograft. Given the heterogeneity of alloreactive T cells in MHC mismatched transplantation and redundancy of chemokines/chemokine receptors, these influences may overwhelm the effect of knocking out Cxcr3 expression with respect to graft survival. The apparent effect of Cxcr3^−/− in a less stringent combination suggests chemokine targeting as a strategy to overcome chronic rejection. Nevertheless, our data indicate that Cxcr3 plays an indisputable role in the trafficking of the alloreactive T cells, and that this role persists when small numbers of alloreactive T cells are involved in minor mismatch responses. Targeting Cxcr3 may have a therapeutic potential in clinical transplantation to prevent chronic rejection, but is likely be of less use in preventing acute rejection.

We thank Dr. Robert Fairchild (Department of Immunology, Cleveland Clinic Foundation, Cleveland, OH) for helpful suggestions. We also thank Glen Leverson for consulting statistical analysis, Jose Torrealba for histological analysis, Kathleen Schell and the University of Wisconsin Comprehensive Cancer Center Flow Cytometry Facility (Madison, WI) for assistance in cell sorting.

The authors have no financial conflict of interest.