We utilized mouse models to elucidate the immunologic mechanisms of functional graft loss during mixed antibody-mediated rejection of renal allografts (mixed AMR), in which humoral and cellular responses to the graft occur concomitantly. Although the majority of T cells in the graft at the time of rejection were CD8 T cells with only a minor population of CD4 T cells, depletion of CD4 but not CD8 cells prevented acute graft loss during mixed AMR. CD4 depletion eliminated antidonor alloantibodies and conferred protection from destruction of renal allografts. ELISPOT revealed that CD4 T effectors responded to donor alloantigens by both the direct and indirect pathways of allorecognition. In transfer studies, CD4 T effectors primed to donor alloantigens were highly effective at promoting acute graft dysfunction, and exhibited the attributes of effector T cells. Laser capture microdissection and confirmatory immunostaining studies revealed that CD4 T cells infiltrating the graft produced effector molecules with graft destructive potential. Bioluminescent imaging confirmed that CD4 T effectors traffic to the graft site in immune replete hosts. These data document that host CD4 T cells can promote acute dysfunction of renal allografts by directly mediating graft injury in addition to facilitating antidonor alloantibody responses.
analysis of variance
cluster of differentiation
hematoxylin and eosin
institutional animal care and use committee
laser capture microdissection
mean fluorescence intensity
Despite the now routine nature of clinical renal transplantation, the adaptive immune response to transplanted tissues remains poorly defined. Clearly, both the cellular and humoral arms of the immune response have the potential to contribute to the immunologic destruction of renal allografts, but the relative contributions of the individual pathways remain unclear. There is compelling evidence that antibodies to donor alloantigens are causally related to destruction of clinical renal transplants . For example, deposition of complement split products such as C4d on the graft peritubular capillaries (PTC) correlates closely with the presence of circulating donor-reactive antibodies and eventual development of graft dysfunction [2-5]. Moreover, antibodies reactive with the graft endothelium promote subclinical alterations in graft endothelial cells [6, 7]. However, the vast majority of antibody-mediated rejection (AMR) is accompanied by concomitant T cell infiltration (mixed AMR) , raising the possibility that T cells contribute to development of graft dysfunction. Consistent with this possibility, treatment with anti–T cell reagents reverse mixed AMR episodes . However, the salient mechanisms of graft injury in this common transplant scenario remain largely a matter of speculation.
We have previously defined the mechanisms of AMR of human renal allografts . We herein used mouse models to elucidate the role of host T cells in promoting acute loss of renal allografts during mixed AMR episodes. We provide evidence that CD4 T cells not only play a dominant role in promoting acute graft dysfunction in this rejection scenario by facilitating antidonor antibody responses but also serve as T effectors that directly mediate graft injury. Surprisingly, these data indicate that CD8 T cells play little if any role in promoting graft dysfunction during mixed AMR. These data provide mechanistic insight into an important clinical problem, and have implications for effective management of clinical renal allograft recipients.
Materials and Methods
C57BL/6 (B6, H-2b), BALB/c and DBA/2 (H-2d), FVB/N (H-2q), CD8 KO (B6.129S2-Cd8atm1Mak/J), and RAG−KO (B6;129S7-Rag1tm1Mom/J) mice were purchased from Jackson Laboratories (Bar Harbor, ME). Mice transgenic for firefly luciferase on the B6 background (L2G85.B6) were a kind gift from Dr. Robert Negrin (Stanford, CA). All mice were housed and treated in accordance with Animal Care Guidelines established by the National Institute of Health and The Ohio State University. All experiments described in this manuscript were approved by the OSU IACUC.
Splenic lymphocytes spleen cell [SC] were isolated from skin-primed renal allograft rejectors or controls and CD4 T cells were purified using reagents and columns from Miltenyi Biotec (San Diego, CA). The resulting cells were >95% CD4+CD3+ cells. Unseparated or purified CD4+ T cells were cultured with irradiated DBA/2 splenocytes (SC) for 24 h and assayed for interferon gamma (IFN-γ) or IL-17 production using kits from R&D Systems (Minneapolis, MN). The resulting spots were counted with an ImmunoSpot Series I analyzer (Cellular Technology, Cleveland, OH).
T cell depletion
CD8 T cells were depleted by treatment of mice with 100 mg mixture of mAbs to CD8 (TIB 105 and YTS 169) on days −3, −2, −1, +5 and +10 relative to renal allograft transplantation. CD4 T cells were depleted by treatment with 100 mg of mAb GK1.5 on days −3, −2 and −1 relative to renal allograft transplantation. Treatment strategies were confirmed to be effective in eliminating the corresponding cells from the graft site.
Three-micrometer sections of paraffin embedded grafts were stained with hematoxylin and eosin (H&E), trichrome or immunostained for CD4d deposition using an antibody provided by Dr. Wink Baldwin of the Cleveland Clinic. For immunohistochemistry (IHC) studies, 4 μm frozen sections of rejecting renal allografts were stained singly with mAb to mouse CD8 (BD Biosciences, San Diego, CA) or doubly stained with anti-CD4 mAb (BD Biosciences), in combination with biotinylated rabbit antirat IgG, mouse adsorbed (Vector Laboratories, Burlingame, CA) visualized with the chromagen vina green (Biocare, Concord, CA) followed byprimary rabbit polyclonal antibodies to FasL (Abcam ab15285; Cambridge, MA), Granzyne B (Abcam ab4059), perforin (Santa Cruz Biotechnology sc-9105; Dallas, TX) or IFN-γ (Life Technologies AMC4034; Grand Island, NY) followed by antirabbit IgG-alkaline phosphatase (Biocare) visualized with warp red chromagen (Biocare), then counterstained with Richard Allan Hematoxylin (Fisher Scientific, Pittsburgh, PA). All studies were performed in the Pathology Core Laboratory at the Ohio State University Wexner Medical Center.
Isolation of graft infiltrating lymphocytes
After removal of the renal capsule, the graft cortex was cut into approximately 1 mm3 pieces, rinsed to remove contaminating peripheral blood, then incubated for 30 min in DMEM:F12 (50:50) containing 0.1% collagenase (type IV; Worthington Biochemical, Lakewood, NJ), 0.1% soybean trypsin inhibitor (Sigma–Aldrich, St. Louis, MO) and 0.01% DNase I (Boehringer, Mannheim, Indianapolis, IN). Following vigorous agitation, the resulting cell suspension was centrifuged on Lympholyte-M (Cedarlane Laboratories, Burlington, NC) to isolate lymphocytes.
Flow cytometry was performed using a FACSCalibur (BD Biosciences). Data were analyzed using FlowJo (Tree Star, Inc., Ashland, OR) and WinMDI software (Scripps Institute, Lo Jolla, CA). The percentage of positive cells for a given marker was based on cutoff points chosen to exclude >99% of the negative control population. Fluorochrome-conjugated mAbs to mouse CD3e (145-2C11), CD4 (GK1.5), CD8a (53–6.7), CD8b (H35-17.2), CD44 (IM7) and CD103 (M290) and the respective species- and isotype-matched negative control mAbs were purchased from BD Biosciences.
The presence of donor-reactive antibodies was determined by the ability of sera to bind to DBA/2 splenocytes detected with goat antimouse IgG (γ chain specific) antibodies (Southern Biotechnology, Birmingham, AL) measured as mean fluorescence intensity (MFI) of DBA/2 SC targets. Background fluorescence was determined for each subtype by taking the MFI plus three standard deviations of sera from five naïve C57BL/6 mice. Negative controls included SC plus each detection antibody; MFI values ranged from 25 to 35.
Purified CD4 T cells were isolated from DBA/2-primed L2G85.B6 mice and then adoptively transferred into syngeneic recipients previously primed to donor (DBA/2) alloantigens and transplanted with either renal isografts or DBA/2 renal allografts. Imaging was performed 3–5 days after transplantation to localize CD4 T cells by injecting mice intravenously with 4 mg D-luciferin. An In Vivo Imaging System 100 (PerkinElmer, Waltham, MA) was used for data acquisition. Living Image software (PerkinElmer) was used for data analysis. All bioluminescent imaging studies were performed at the Small Animal Research Imaging Core at the OSU College of Veterinary Medicine.
Lasar capture microdissection
RAG KO mice received DBA/2 renal allografts and after 1 week were injected intravenously with 15 million purified CD4 T cells from CD8 KO donors on the C57BL/6 background that had previously rejected DBA/2 skin allografts. For laser capture microdissection (LCM), CD4 T cells in the graft were identified using anti-CD4 mAb (Abcam) followed by captures. RNA was then isolated from CD4+ cells in the graft and analyzed by qRT-PCR to determine mRNA levels. Data were normalized to values obtained for kidneys from naïve DBA/2 mice.
Analyses were performed using Prism 5 program (GraphPad Software, LaJolla, CA). Graft survival curves were compared using the log-rank test. Other analyses performed included Student's t-test and one-way analysis of variance. Differences between means of experimental groups were considered significant at p < 0.05.
Key role for host CD4 T cells in promoting acute loss of renal allografts during mixed AMR
We have previously described a mouse model of mixed AMR in which B6 hosts are primed to DBA/2 alloantigens with a DBA/2 skin allograft 14–21 days prior to receiving a DBA/2 renal allograft . As shown in Figure S1, identical results were obtained with recipients primed to donor alloantigens at times ranging from 11 to 91 days prior to kidney transplantation, indicating that the time of renal transplantation relative to the priming stage was not critical. As shown in Figure 1A, FACS analyses of DBA/2 renal allografts undergoing acute rejection revealed a predominance of CD8 T cells in the kidney allografts accompanied by a minor population of CD4 T cells at the time of rejection. To assess the functional contributions of CD4 and CD8 T cells in this model, we depleted CD4 or CD8 T cells in sensitized mice at the time of kidney transplantation. Surprisingly, CD4 depletion was highly effective at preventing graft loss whereas CD8 depletion was totally ineffectual (Figure 1B). Moreover, sensitized CD8 KO mice rejected grafts acutely (at days 9–10), documenting that CD8 T cells are not required for acute graft dysfunction in this model. It should be noted that mortality of hosts depleted of CD4 T cells was nearly 40% (Figure 1B). The vast majority of these deaths occurred immediately after the final injection suggesting the involvement of a toxic agent or an immune reaction rather than rejection per se.
A disadvantage of the above model is that DBA/2 renal allografts are spontaneously accepted in B6 hosts, a scenario quite unlike that seen clinically. However, as shown in Figure 2A, depletion of CD4, but not CD8, T cells also prevented graft loss when BALB/c renal allografts were transplanted into naïve B6 hosts, a model in which grafts undergo slow spontaneous rejection in naïve hosts, similar to the clinical situation. Note that rejection in this model also exhibits classic features of mixed AMR with deposition of complement split products on the PTC in the context of massive graft inflammation (Figure 2B–D). These data are consistent with the possibility that CD8-independent rejection extends to mixed AMR occurring in nonsensitized recipients.
Figure 3A shows that sensitization to donor alloantigens followed by renal transplantation was accompanied by an approximately fourfold increase in the frequency of CD4 T cells capable of responding to donor alloantigens via production of IFN-γ. The response was specific in that the same cells failed to mount responses in lieu of stimulation (responders alone). As shown in Figure 3B, mice primed to skin allografts alone exhibited an increase in frequencies of CD4 T cells producing IFN-γ in response to DBA/2 alloantigens approximately one-half of that of hosts receiving both skin and renal allografts, suggesting that the two allograft types make comparable contributions to the increased frequency. The response was highly specific to DBA/2 alloantigens as the frequency of splenic CD4 cells responding to syngeneic or third party stimulators did not significantly increase. Note that a third party MHC allogeneic stimulator is not shown in Figure 3A as the BALB/c and DBA/2 strains share the H-2d haplotype.
Note that splenic CD4 T cells mounted comparable responses to MHC-matched BALB/c (H-2d) (MHC matched with the donor strain but disparate for multiple minor histocompatibility antigen) stimulator cells (Figure 3) indicating specificity for MHC-encoded alloantigens via the direct pathway of allorecognition. As shown in Figure S2, the frequency of T cells producing IL-17 in response to alloantigen mirrored that of CD4 T cells producing IFN-γ, suggesting that all potential mediators of graft injury are increased following combined skin/renal allotransplantation.
Highly purified splenic CD4 T cells were used as responders in the above assays thereby precluding the detection of indirect alloantigen presentation due to the absence of host antigen presenting cells. However, as shown in Figure 3C, unseparated splenocytes (SC) mounted vigorous responses to subcellular alloantigen. Thus, these data indicate that sensitization to donor alloantigens followed by renal transplantation increases the frequency of T cells capable of responding to renal allografts by both the direct and indirect pathways of allorecognition.
Antibody responses in CD4-depleted recipients
To assess the impact of CD4 depletion on the development of alloantibodies in this model, serum from CD4-depleted recipients was collected over time and analyzed for donor-reactive antibodies by FACS analyses. Consistent with the findings of Steele et al , CD4 depletion completely abrogated donor-reactive IgG antibody levels elicited in response to skin allografts (Figure 4A). Staining for complement split products (C4d) in the PTC (Figure 4B) failed to reveal C4d deposition in the vast majority (14/16) of grafts (Figure 4B, left). In contrast, grafts in CD4 replete (untreated) mice showed strong positivity (Figure 4B, right). These data indicate that the loss of antibodies detected in the circulation in this model does not reflect absorption by the graft.
CD4 T effectors but not alloantibodies promote acute injury to renal allografts
To further assess the contribution of antidonor alloantibodies to acute destruction of renal allografts, we adoptively transferred alloantibodies into RAG KO mice (B6;129S7-Rag1tm1Mom/J) bearing DBA/2 renal allografts. RAG KO mice lack T and B cells  and therefore do not reject transplanted tissues without transferred antibodies or T cells. Antibodies examined in these studies included sera from sensitized WT mice that had rejected DBA/2 renal allografts, and an mAb (HB159) directed to donor MHC class I (H-2Kd) alloantigens. As shown in Figure 5A, sera from mice that had rejected renal allografts did not cause graft dysfunction nor did antidonor MHC I mAb even when administered at high doses. Long-term follow-up confirmed the absence of detectable injury out to day 30, and identical (negative) results were obtained when antibodies were transferred prior to renal transplantation (data not shown). Graft survival was not due to an absence of antibodies because these were detectable in recipient sera throughout the course of the experiment (Figure 5B) and led to antibody deposition (Figure S3A) and fixation of complement (Figure S3D) in the graft. Notably, comparable doses of HB159 mAb elicit vascular injury to cardiac allografts , arguing that the failure to promote injury to renal allografts does not reflect an inadequate dose or specificity of the mAb. Rejector sera did not cause detectable immunoglobulin or complement deposition in RAG KO mice (Figure S3B and E), probably because of the low concentration of anti-MHC antibodies that could be achieved through the transfer of sera.
In contrast, transferred primed CD4 T cells were highly effective at destroying renal allografts in RAG KO mice as evidenced by rapid graft loss (Figure 5A and C) and histologic injury to the graft (not shown). Note that the combination of HB159 mAb and CD4 T effectors did not change renal allograft survival relative to that of CD4 T effectors alone at either high or low doses (Figure 5A). In experiments not shown, purified CD4 T cells from sensitized CD8 KO mice yielded identical results, documenting that the destructive capacity observed was not due to contamination with CD8 T cells. As shown in Figure 5D, CD4 T cells mediating rejection of renal allografts transplanted into RAG KO recipients elicited production of the effector cytokines tumor necrosis factor (TNF)- and IFN-γ in the absence Fox-P3 expression consistent with the possibility that CD4 T cells entering the graft were T effectors rather than regulatory T cells. These data suggest that CD4 T effectors are fully capable of promoting renal allograft dysfunction without the involvement of host B cells.
Graft infiltrating CD4 T cells express effector molecules
To gain insight into the mechanisms by which CD4 T cells infiltrating renal allografts cause graft dysfunction, we used LCM. In these experiments, purified CD4 T cells from sensitized CD8 KO mice were adoptively transferred into RAG KO mice. As shown in Figure 6A, graft infiltrating CD4 T cells produced high levels of FasL (CD95L), granzyme B, perforin and IFN-γ following entry into the graft. Note that the donor cells in these experiments are homozygous for targeted disruption of the CD8a locus, so do not contain CD8a-expressing cells including CD8 T cells and NK cells, but that the RAG KO recipients potentially do as do DBA/2 renal allografts. However, most if not all leukocytes (CD45-postive cells) in the graft co-expressed CD4 (data not shown) suggesting that non-CD4 T cells do not make a significant contribution to gene expression in these experiments. As shown in Figure 7, IHC studies confirmed that subsets of CD4 cells expressing the above effector molecules infiltrate the graft site concomitant with acute graft dysfunction, strongly suggesting that these cells are key mediators of the rejection process.
Primed CD4 T cells migrate to the site of renal allografts
An important limitation of the transfer models described above is that naïve T cells can artifactually acquire effector function following transfer into immunodeficient hosts [15, 16]. This was confirmed in our model by transfer of purified CD4 T cells from naïve mice (Figure 5C). To circumvent this limitation, bioluminescent imaging was used to track migration of donor-primed CD4 T cells transferred into immune replete WT recipients. As shown in Figure 8A, by day 5 posttransplantation, the majority of primed CD4 T cells following renal transplantation resided in the allograft, with the remainder residing in the host spleen. Such accumulation was specific in that transferred CD4 cells failed to accumulate within renal isografts (Figure 8B). These data indicate that host CD4 T cells responding to donor alloantigens in immune replete hosts specifically migrate to the site of the renal allograft at the time of graft injury, consistent with a key role for these cells in direct attack of the graft.
The salient finding of the present study is that CD4 T effector cells promote acute dysfunction of renal allografts during mixed AMR, serving both as T helpers for antibody responses and as CD4 T effectors that directly attack the graft. However, we find no evidence that antibodies contribute to acute graft dysfunction in this model. That CD4 T effectors directly promote injury to renal allografts is supported by the results of adoptive transfer studies indicating that host CD4 T effectors are highly efficient at eliciting functional destruction of renal allografts in both immune deficient (Figure 5A and C) and immune sufficient hosts (Figures 1B and 2A). We further show that host CD4 T cells migrate to the site of renal allografts but not isografts in immune replete hosts (Figure 8), and produce cytotoxic effector molecules therein (Figures 6 and 7), consistent with a role in direct attack of the graft.
Our data provide novel insight into mechanisms of renal allograft rejection. Surprisingly, our data indicate CD8 T cells are not necessary in promoting acute dysfunction of renal allografts in either naïve (Figure 2) or sensitized (Figure 1) recipients. This finding is in accord with results in mouse cardiac allograft models in which CD8 depletion has a barely discernible impact on rejection events, simply altering the type of CD4 T cell response (Th1 vs. Th2) leading to graft destruction . These data also are in line with prior studies showing that CD4 T cells alone are adequate to reject mouse skin and cardiac allografts , and that CD4 T effectors can be more efficient in killing activity than CD8 T cells even across MHC I disparities . These data argue that classical CD8 T effector populations play little if any role in promoting graft injury during mixed AMR. Our LCM and IHC data showing that a subset of graft infiltrating CD4 T cells express CD95L, perforin, granzyme B and IFN-γ following entry into the graft (Figures 6 and 7) provide insight into the mechanisms by which these cells promote acute graft dysfunction. Notably, IFN-γ production identifies these cells as Th1 cells (as opposed to Th17 or Th2 cells), known to play critical roles in regulating immunopathology [20, 21]. Our ELISPOT data (Figure 3) indicate that host CD4 T cells comprise a mixed population responding to donor alloantigens by both the direct and indirect pathways of allorecognition. The presence of an IgG response, which is known to depend on CD4 T cells recognizing donor alloantigen by the indirect pathway of allorecognition , further supports the involvement of the latter cells in our model. Brennan et al  documented that indirect pathway cells are initially a minor population but undergo rapid expansion following transplantation, and there is evidence that that such cells contribute to chronic injury to renal allografts . We postulate that direct pathway CD4 T cells directly attack the graft, whereas indirect pathway CD4 T cells provide help for B cell responses, but our data do not discriminate between these possibilities.
The failure of antibodies to elicit acute allograft dysfunction (Figure 5A) was an unexpected finding. This was not due to inadequate doses of antibody because the present study used mAb levels comparable or higher than those used in cardiac allograft studies which document the capacity of this mAb to induce graft injury . Moreover, we documented in our studies that excess antibody binding activity (Figure 5B) was present in circulation for >30 days, with clear evidence of antibody and complement deposition at the graft site (Figure S3). Thus, while there is evidence that the historic failure of passively transferred antidonor antibodies to elicit acute graft loss reflects inadequate antibody titers [24, 25] or an inability of antidonor mAbs to fix complement , this is not a viable explanation in our experiments. We speculate that this difference reflects the fact that unlike the renal allografts examined in this study, cardiac allografts are not life supporting—i.e. these are heterotopic grafts that do not pump blood—and thus are susceptible to immune mechanisms distinct from those that mediate rejection of renal allografts.
Wasowska et al  showed that transfer of donor-specific mAb into Ig KO mice results in acute rejection of mouse cardiac allografts. Given that the Ig KO hosts used in this study were undergoing cellular rejection at the time of transfer, these data suggest that an ongoing cellular immune response is critical for development of acute renal allograft dysfunction. Our data indicate that transfer of antibody plus T cells into RAG−/− hosts does not increase the pace of graft loss (Figure 5A). Based on these data, we postulate that interaction between antibody and non–T cells—i.e. macrophages and/or neutrophils as described by Bickerstaff et al —is necessary for reconstitution of acute AMR of renal allografts. Our studies indicate that non–T cells are not significantly present during CD4-dependent rejection of renal allografts in RAG KO recipients.
Our data indicate that the peak of the donor-reactive antibody response occurs in this model at the time of the skin allograft rejection then gradually disappears in about 2 weeks following CD4 depletion (Figure 4A). Given that we observe no evidence of C4d deposition in the graft PTC in the vast majority of grafts (14/16) the simplest explanation for these data is that the gradual disappearance of circulating antibodies reflects a lack of CD4 help for antidonor B cell responses rather than absorption by the graft. It should be noted, however, that the interpretation of these data is complicated by the fact that the time of CD4 depletion (day 10 postpriming) in our model coincides with the peak of germinal center formation potentially abrogating the process of affinity maturation and the generation of donor-reactive alloantibodies.
The clinical relevance of our data is uncertain given that the time intervals for priming used in our mouse model are potentially much shorter than those that occur in clinical renal allograft recipients. Thus, there may be inadequate time for development of true memory B cell responses, potentially resulting in incomplete elimination of alloantibodies following CD4 depletion in the clinical situation. While our data (Figure S1) indicate that the time of priming does not have a significant impact on graft survival time, it is not clear that the mechanisms of injury are the same given the likely differences in populations of memory cells. This limitation to data interpretation is compounded by the fact that the mechanistic studies reported in this manuscript were performed using samples from mice primed shortly before (14 days or less) renal transplantation.
Our data provide an alternative explanation for the observation that anti–T cell reagents successfully reverse antibody-associated graft injury in clinical renal allograft recipients [9, 27]. In this scenario, the capacity of anti–B cell therapies such as rituximab and IVIg to reverse rejection episodes may be explained by an impact on antidonor CD4 T cell responses. In addition to their well-documented role in antibody production, B cells play important roles as antigen presenting cells for alloreactive CD4 T cell responses [28, 29], and IVIg preparations contain specificities that impact T cell activation events . Based on these data, we postulate that immunosuppressive strategies in current clinical usage ameliorate-mixed AMR episodes because they inhibit CD4 T cell responses. Our data support ours and others' experience that in acute AMR, unless vascular thrombosis develops, subsequent/concomitant cellular rejection may be the most important direct pathogenic factor causing graft injury. However, the role of antidonor antibodies should not be neglected; there is mounting evidence implicating antibodies in the development of chronic graft injury in human renal allografts, a condition now called chronic AMR . It will be important to develop good animal models for chronic AMR to define mechanisms of late loss of renal allografts.
We gratefully acknowledge the outstanding contributions of Cherri Bott, Viy McGaughy and Kristin Kovach in providing invaluable expertise and guidance in analyses of graft histology and IHC. We would also like to thank Dr. Wink Baldwin for his kind gift of polyclonal antibodies with specificity for CD4d. These studies were supported by a grant from the National Institutes of Health to GAH (AI036532).
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.