CD103 is an integrin with specificity for the epithelial cell-specific ligand, E-cadherin. Recent studies indicate that CD103 expression endows peripheral CD8 cells with a unique capacity to access the epithelial compartments of organ allografts. In the present study we used a nonvascularized mouse renal allograft model to 1) define the mechanisms regulating CD103 expression by graft-infiltrating CD8 effector populations, and 2) identify the cellular compartments in which this occurs. We report that CD8 cells responding to donor alloantigens in host lymphoid compartments do not initially express CD103, but dramatically up-regulate CD103 expression to high levels subsequent to migration to the graft site. CD103+CD8+ cells that infiltrated renal allografts exhibited a classic effector phenotype and were selectively localized to the graft site. CD8 cells expressing low levels of CD103 were also present in lymphoid compartments, but three-color analyses revealed that these are almost exclusively of naive phenotype. Adoptive transfer studies using TCR-transgenic CD8 cells demonstrated that donor-specific CD8 cells rapidly and uniformly up-regulate CD103 expression following entry into the graft site. Donor-specific CD8 cells expressing a dominant negative TGF-β receptor were highly deficient in CD103 expression following migration to the graft, thereby implicating TGF-β activity as a dominant controlling factor. The relevance of these data to conventional (vascularized) renal transplantation is confirmed. These data support a model in which TGF-β activity present locally at the graft site plays a critical role in regulating CD103 expression, and hence the epitheliotropism, of CD8 effector populations that infiltrate renal allografts.
The mechanisms by which CD8αβ+TCRαβ+ effector populations (CD8 effectors) access and destroy epithelial compartments represents an important gap in our understanding of the immune response to transplanted tissues and organs. The functional elements of most organ allografts are epithelial layers (e.g., renal tubules, pancreatic islets), and the destruction of such structures by alloreactive CD8 effector populations is a defining feature of the rejection process (1). This is particularly apparent in the case of clinical renal allograft rejection, where infiltration of the graft renal tubules with CD8 effectors (tubulitis) is a prime diagnostic indicator of rejection (2).
CD8 effector responses are initiated within draining lymphoid compartments when dendritic cells carrying foreign Ag encounter naive or memory CD8 cells of appropriate MHC/peptide specificity (3). Subsequent to activation, CD8 cells rapidly exit draining lymphoid compartments and extravasate into peripheral inflammatory sites (4) where they eliminate cells expressing their target MHC/peptide complex. Homing of nascent CD8 effector populations to inflammatory sites is mediated by activation-induced up-regulation of integrin family proteins that promote adhesive interactions with the vascular endothelium and extracellular matrix (5). Recent studies indicate that chemokine/chemokine receptor interactions orchestrate initial movement of CD8 effectors into inflammatory sites by regulating integrin avidity and providing directional gradients for chemotaxis (5). However, relatively little is known concerning the downstream events by which CD8 effectors gain access to epithelial compartments.
Previous studies from this laboratory have documented that the integrin heterodimer, αE(CD103)/β7 (herein referred to as CD103), defines a novel CD8 effector subset (6). This observation is pertinent to CD8 effector/epithelial cell interactions, because CD103 confers specificity for the epithelial cell-specific ligand, E-cadherin (7, 8). We and others have shown that CD103 defines a subset of CD8 effectors that infiltrate rejecting clinical renal allografts (6, 9). These graft-infiltrating CD8+CD103+ effectors are preferentially associated with the renal tubular epithelium (9, 10), consistent with a key role for CD103 in promoting the destruction of such structures by CD8 effectors. We recently used mice with targeted disruption of CD103 to demonstrate that CD103 expression is required for destruction of epithelial allografts (pancreatic islets) by CD8+ T cells, strongly implicating CD103+CD8+ effectors in the destruction of graft epithelial elements (11).
The goal of the present study was to elucidate the mechanisms that regulate CD103 expression by CD8 effector populations responding to renal allografts. In vitro studies indicate a key role for TGF-β in regulating CD103 expression by CD8 effector populations (6), but the factors controlling CD103 expression in vivo and the identity of the cellular compartments in which this occurs are unknown. To address these issues we developed a simplified renal allograft model in which allogeneic renal cortical fragments are transplanted into the renal subcapsular space of recipient mice. We report that CD8 effectors elicited in response to renal allografts within draining lymphoid compartments do not initially express significant levels of CD103, but rapidly up-regulate CD103 expression subsequent to entry into the graft site, a process that we document to be dependent on TGF-β activity. The relevance of these findings to rejection of conventional (vascularized) renal allografts is confirmed. These data expand our understanding of CD8 homing events and provide insight into the mechanisms by which CD8 effectors access and destroy the functional elements of organ allografts.
Materials and Methods
1B2 mAb (anti-clonotypic 2C TCR) (12) was purified from hybridoma culture supernatant and conjugated to FITC by Bio Express (West Lebanon, NH). RL172.4 (anti-CD4), J11D.B1 (anti-heat-stable Ag), IM7.8 (anti-CD44), and 25-5-16S (anti-H-2-IAb) were used as hybridoma culture supernatants. Conjugated mAbs to CD8a, CD8b, CD44, CD62L, CD25, CD11a, CD103 (M290), IFN-γ, TNF-α, streptavidin-CyChrome, and isotype controls were purchased from BD PharMingen (San Diego, CA). Biotinylated rat anti-mouse IgG1 was purchased from Caltag Laboratories (Burlingame, CA).
A/J, C57BL/6, DBA/2, BALB/c, and BALB/c-H2dm2/KhEg mice were purchased from The Jackson Laboratory (Bar Harbor, ME). 2C transgenic mice (13) originally obtained from Dr. T. Hansen (Washington University School of Medicine, St. Louis, MO), were maintained by backcrossing heterozygous 2C males to C57BL/6 females and screening the offspring for expression of the clonotypic 2C TCR by FACS of peripheral blood using 1B2 mAb. Mice expressing a dominant negative type II TGF-β receptor (DNRII) 3 (14) and 2C mice bred onto the DNRII background were provided by Dr. R. Gress (National Institutes of Health, Bethesda, MD). All mice were maintained under specific pathogen-free conditions in the animal facility at University of Maryland (Baltimore, MD).
Adoptive transfer experiments
Spleen and lymph nodes from 2C donors were minced and filtered through Nitex filter (Sefar America, Depew, NY) and enriched for naive CD8+ T cells by treatment with a mixture of mAbs to CD4 (RL172.4), heat-stable Ag (J11D.B1), CD44 (IM7.8), and 25-5-16S (anti-H-2-IAb), followed by incubation in 1/10 Low-Tox M rabbit complement (Accurate Chemical and Scientific, Westbury, NY). The resulting cell suspension comprised >60% CD8+TCRαβ+ cells of naive phenotype (CD44−CD62Lhigh), with the remainder comprising residual B cells. 2C CD8 cells (3 × 106) were transferred into C57BL/6 recipients by tail vein injection 1–2 days before tissue transplantation experiments.
Nonvascularized renal allograft model
Renal cortical fragments for transplantation were prepared from kidneys of weanling mice (aged 4–5 wk). After removal of the kidney capsule, vessels, and medulla, the remaining cortex was minced into 1-mm3 pieces, pressed through 60-mesh screen, and washed in HBSS. The resultant cortical fragments were collected in PE-50 tubing and gathered at one end of the tubing by centrifugation. An incision was made to expose the recipient’s left kidney, and ∼0.02 ml of pelleted tissue fragments was transplanted into the renal subcapsular space of recipient mice. After recovery from anesthesia, recipients were maintained under pathogen-free conditions
Vascularized renal transplantation
Orthotopic transplantation of vascularized renal allografts from BALB/c donors to C57BL/6 recipients was performed as described by Zhang et al. (15) using halothane (Halocarbon, River Edge, NJ) inhalation anesthesia. After perfusion with 0.5 ml of cold heparinized saline, the left donor kidney attached to a segment of the aorta, renal vein, ureter, and bladder patch was isolated and removed en bloc. Following removal of the recipient left kidney, the donor aortic segment and renal vein were anastomosed end-to-side with the recipient aorta and inferior vena cava.
Isolation of cell populations
Graft-infiltrating lymphocytes (GIL) were isolated by mincing the graft and incubating the resulting fragments for 30 min in medium containing collagenase (Worthington Biochemical, Freehold, NJ), soybean trypsin inhibitor (Sigma-Aldrich, St. Louis, MO), and DNase (Roche, Indianapolis, IN). Lymphocytes were isolated by centrifugation on Lympholyte-M (Cedarlane Laboratories, Hornby, Ontario, Canada). Cells were isolated from the draining lymph node (lateral aortic) and spleen of allograft recipients by mincing with forceps and passage of the resulting cell suspension through nylon mesh of 100-μm pore size.
For two-color FACS analyses, cells were stained with PE-conjugated mAb to CD8b in combination with FITC-conjugated mAbs to markers of interest. For three-color analyses, cells were stained with anti-CD8a-CyChrome in combination with FITC- and PE-conjugated mAbs. Species- and isotype-matched mAbs of irrelevant specificity were used as controls for nonspecific fluorescence. After staining, cells were fixed with 0.5% paraformaldehyde, and 3–6 × 104 cells were analyzed using a FACScan (BD Biosciences, San Jose, CA). Intracellular cytokine staining (ICS) for FACS was performed using an ICS kit (no. 2040KK) purchased from BD PharMingen. Briefly, GIL isolated as described above were cultured in medium containing PMA, ionomycin, and monensin for 4 h, and stained with anti-CD8a-FITC, then permeabilized, followed by PE-coupled mAbs to IFN-γ or TNF-α. Lymphocyte populations were gated by forward/side scatter analysis to exclude monocytes and debris. WinMDI 2.8 software (downloaded from http://facs.scripps.edu/software.html) developed by Dr. J. Trotter (Scripps Institute, San Diego, CA) was used for analysis and graphic display of flow cytometry data. The percentage of positive cells for a given marker was based on cutoff points chosen to exclude >99% of the negative control population.
Graft-bearing kidneys were harvested at the designated times, fixed in 10% buffered formalin, and embedded in paraffin. Sections (6 μm) were stained with H&E.
Characterization of nonvascularized mouse renal allograft model
The goals of this study were to elucidate the mechanisms operating to control CD103 expression by CD8 effector populations that infiltrate renal allografts and to identify the in vivo compartments in which this occurs. To expedite this investigation, we initially developed a simplified (nonvascularized) mouse renal allograft model in which renal cortical tubule fragments from DBA/2 donors (H-2d) were transplanted under the left kidney capsule of fully allogeneic C57BL/6 (H-2b) recipients. Such grafts comprise renal tubular fragments almost exclusively, with no detectable contamination with glomeruli (not shown). Normal kidneys do not contain significant numbers of CD8 cells (<0.1 × 104/kidney; Fig. 1,A). Similarly, only trace numbers of CD8 cells (<0.1 × 104/kidney) infiltrate renal isografts or the contralateral (nontransplanted) kidneys of allograft recipients (not shown). However, as shown in Fig. 1,A, there is massive recruitment of CD8 cells into host kidneys transplanted with renal allografts, which reaches 2–4 × 104 CD8 cells/kidney by day 7 and persists at this level through day 21 post-transplantation. As shown in Fig. 1 B, infiltrating lymphocytes in the kidney were strictly confined to the site of transplantation, the renal subcapsule.
Rapid acquisition of CD103 expression by graft-infiltrating CD8 cells
As shown in Fig. 1,C, a major subset of CD8 cells that infiltrated the graft site expressed CD103 by day 20 post-transplantation. Note that CD103 expression by allograft-infiltrating lymphocytes was restricted almost exclusively to the CD8 subset (Fig. 1,C). This distribution is similar to that previously described for rejecting clinical renal allografts (10) and mouse pancreatic islet allografts transplanted into the renal subcapsule (11). Three-color analyses showed that CD103+CD8+ cells infiltrating renal allografts exhibit a classic CD8 effector phenotype: CD44high, TCR-αβ+, CD11ahigh, CD62Llow (Fig. 1 C), perforin+ (not shown). Thus, this model replicates the salient characteristic of clinical renal allograft rejection, i.e., accumulation of CD8 effectors expressing high levels of CD103 at the graft site.
To determine when and where CD8 cells responding to renal allografts first acquire CD103 expression, we performed a kinetic analysis of CD103 expression by CD8 cells in the graft and draining (lateral aortic) lymph node of renal allograft recipients. As shown in Fig. 2, CD8 cells that initially infiltrated allografts (day 7) were predominantly CD103−, with only a minor subset (<20%) expressing significant levels of CD103. However, the proportion of CD8 cells in the graft expressing CD103 progressively increased with time, such that the majority of CD8 cells expressed CD103 by day 21 post-transplant (Fig. 2). Note that the level of CD103 expressed by CD103+CD8+ cells was initially low, but progressively increased to levels 5- to 10-fold higher than those expressed by CD8 cells in the draining lymph node. As shown in Fig. 1 B, accumulation of CD103+CD8+ cells in the graft was delayed compared with the rate of total CD8 infiltration, suggesting that up-regulation of CD103 expression by CD8 effectors occurs subsequent to recruitment into the graft.
Selective localization of CD8+CD103+ effectors at the graft site
As previously reported (6, 16), 40–60% of CD8 cells in the draining node and spleen of normal mice constitutively expressed low levels of CD103 (Fig. 2, day 0). However, three-color analyses revealed that the vast majority (>90%) of CD8+CD103+ cells in the draining node of allograft recipients retained a naive (CD44low) phenotype (Fig. 3,B). Conversely, CD8 effector/memory (CD44high) cells in the draining node were predominantly CD103−, with a small subset (<10%) expressing CD103 at 5- to 10-fold lower levels than those in the graft (Fig. 3, B vs A). In contrast to the progressive accumulation of CD103+CD8+ cells observed in the graft, total numbers of CD103+CD8+ cells in the draining node increased only transiently after transplantation, then rapidly returned to pretransplant levels (Fig. 1 B, right).
To directly compare CD103 expression by functional CD8 effector populations in the graft vs draining node, we used the technique of ICS for IFN-γ-producing cells. In this assay, freshly isolated lymphocytes are stimulated with PMA/ionomycin, which induces rapid IFN-γ production by memory/effector T cells, but not by naive T cells. As shown in Fig. 3,C, the majority (>60%) of CD103+CD8+ cells in the graft produced copious amounts of IFN-γ after stimulation. In contrast, only a small fraction (∼6%) of CD103+CD8+ cells in the draining node produced significant levels of IFN-γ, and the level of IFN-γ produced per cell was substantially lower than those in the graft (Fig. 3 D). Nearly identical results were obtained for TNF-α production (data not shown). These data confirmed that CD8+CD103+ effectors are selectively localized to the graft site. These data suggest that naive CD8 cells activated in the draining lymph initially lose CD103 expression, then up-regulate CD103 to high levels after migration into the graft site.
Expression of CD103 by donor-specific CD8 cells
Due to the nonspecific nature of T cell recruitment, inflammatory sites such as the graft and draining node contain large numbers of effector/memory cells of irrelevant specificity (5, 17). To monitor CD103 expression by a donor-specific CD8 population, trace numbers of purified naive (CD44low) CD8+ T cells from 2C TCR transgenic mice were adoptively transferred into syngeneic C57BL/6 hosts that subsequently received renal allografts from BALB/c (H-2d) donors. 2C mice express a TCR with exquisite specificity for H-2Ld in association with a ubiquitously expressed 8-mer peptide (p2Ca) on >99% of peripheral CD8 cells (13). An mAb (1B2) directed to the clonotypic 2C TCR (12) was then used to monitor acquisition of CD103 expression by Ld-specific CD8 cells responding to renal allografts.
As shown in Fig. 4,A, donor-specific (1B2+) CD8 cells dramatically accumulated at the graft site (GIL) in recipients of Ld-expressing allografts. Few 1B2+ cells (<4% of total CD8 cells) had entered the graft by day 4 but this increased rapidly to nearly 16% by day 7 and declined slightly thereafter. In contrast, the level of 1B2+ cells remained at <2% of the total CD8 cells in the lymph nodes (both draining and nondraining), spleen, and blood (Fig. 4,A). Accumulation of CD8+1B2+ cells in the graft was alloantigen specific, because 1B2+ cells did not accumulate within H-2dm2 allografts, which express all H-2d-encoded alloantigens except H-2Ld (Fig. 4,B). As shown in Fig. 5 B, 1B2+CD8+ cells that entered H-2Ld-expressing grafts transiently up-regulated CD25 and converted to a CD44high phenotype, consistent with these representing recently activated (donor-specific) CD8 cells. These data are consistent with previous studies of Ag-specific CD8 responses, which show rapid movement of recently activated CD8 cells from the draining lymph node to the graft (4) or peripheral site of infection (18) without detectable accumulation in lymphoid compartments.
Before transplantation, adoptively transferred 1B2+CD8+ cells in peripheral lymph nodes possessed the CD103low phenotype of naive peripheral CD8 cells (not shown). However, as shown in Fig. 5,B, 1B2+CD8+ cells recruited into H-2Ld-expressing grafts rapidly and uniformly up-regulated CD103 expression. In contrast, 1B2+CD8+ cells in peripheral lymphoid compartments (blood, spleen, and nodes) remained CD103− to CD103low at all time points examined despite undergoing marked activation, as evidenced by a dramatic increase in the expression of CD25 (draining node) and CD44 (all compartments; Fig. 5 C). These data document that CD8 cells responding to donor alloantigens do not initially express significant levels of CD103, but rapidly up-regulate CD103 expression to high levels subsequent to entry into the graft site. The absence of donor-specific CD103+CD8+ effectors in the peripheral lymphoid compartments before or after their accumulation in the graft further suggests that these cells originate at the graft site and remain sequestered therein.
CD103 phenotype of donor-reactive CD8 effectors is controlled by TGF-β
We have previously shown that bioactive TGF-β promotes CD103 expression by in vitro-generated CD8 effector populations (6). Together with the selective localization of CD103+CD8+ effectors at the graft site, these data strongly suggested that TGF-β in the local environment of the graft site regulates CD103 expression by donor-specific CD8 effectors. To examine this issue, we used a mutant line of 2C mice that express DNRII on a T cell-specific promoter (2C-DNRII) (14). CD8 cells in such mice are Ld-specific, but are unable to respond to TGF-β1 or any of its isoforms (not shown). As shown in Fig. 6, allospecific CD8 cells expressing the mutant TGF-β receptor were highly deficient in CD103 expression after migration into Ld-expressing renal allografts (Fig. 6, A vs B). On the average, 19.8 + 4.6% (n = 4) of 2C-DNRII (3) CD8 cells within the graft on day 14 post-transplantation expressed significant levels of CD103 compared with 72.9 + 3.1% (n = 3) of wild-type 2C cells, a >70% reduction (Fig. 6 C). The total number of CD8+1B2+ cells at the graft site was consistently higher for recipients of 2C-DNRII cells (range, 1.0–2.6 × 105 cells/graft; n = 3) compared with recipients of wild-type 2C cells (range, 0.2–0.9 × 105 cells/graft; n = 3), consistent with the documented role of TGF-β-dependent signaling in limiting expansion of peripheral CD8+ T cells (14). These results provide compelling evidence that TGF-β is a dominant factor regulating CD103 expression by CD8 effectors elicited in response to renal allografts.
Relevance to vascularized renal transplantation
Nonvascularized renal allografts lose typical renal morphology and probably undergo massive necrosis (Fig. 1,B); thus, they may be subject to an usual degree of local inflammation compared with conventional (vascularized) renal allografts. To validate the relevance of our findings in this model to conventional renal transplantation, we transferred CD8 cells from wild-type 2C or 2C-DNRII donors into C57BL/6 hosts that subsequently received primarily vascularized (orthotopic) renal allografts from fully allogeneic BALB/c (H-2d) donors. As shown in Fig. 7,A (left panel), a major subset (31.3%) of wild-type 2C cells that infiltrated vascularized renal allografts on day 14 post-transplantation expressed high levels of CD103. In contrast, as shown in Fig. 7,A (right panel), few (<4%) wild-type 2C cells in the spleen, lymph node, and blood expressed CD103 despite the fact that CD8+1B2+ cells in all compartments were comparably activated, as evidenced by a uniform CD44high phenotype. As shown in Fig. 7 B, 2C-DNRII cells were strikingly deficient in the capacity to up-regulate CD103 expression after entry into the graft, with <8% of CD8+1B2+CD44high cells in the graft expressing significant levels of CD103 compared with >31% of wild-type cells. These data confirm that the salient findings in the nonvascularized kidney fragment allograft model are relevant to CD8 responses elicited by primarily vascularized renal allografts.
This study provides insight into the important clinical problem of renal allograft rejection. We (6) and others (9) have reported that CD103 is expressed at high levels by CD8 effectors that infiltrate rejecting clinical renal allografts, Moreover, we recently demonstrated that CD103 promotes CD8-mediated destruction of epithelial allografts (pancreatic islets) transplanted into the renal subcapsular site (11). The salient finding of the present study is that CD8 cells responding to donor alloantigens within draining lymphoid compartments are initially CD103−, but rapidly up-regulate CD103 expression subsequent to entry into the graft site. Taken together, these data support a model in which the local environment of renal allografts regulates CD103 expression by graft-infiltrating CD8 effectors and thereby promotes their capacity to access and destroy graft epithelial compartments.
Our data implicate local TGF-β activity in regulating CD103 expression by graft-infiltrating CD8 effectors. We demonstrate that an intact TGF-β signaling pathway is required for efficient induction of CD103 expression by graft-infiltrating CD8 effectors (Figs. 6 and 7). That the requisite TGF-β activity operates at the local as opposed to systemic levels is supported by the lack of CD103 expression by donor-specific CD8 cells in host lymphoid compartments (draining node, spleen, and blood) despite dramatic induction of CD103 expression on CD8 effectors that enter the graft site (Figs. 5 and 7). In experiments not shown, 1B2+CD8+ effectors generated in vitro in the absence of TGF-β activity do not express significant levels of CD103, but rapidly and uniformly up-regulate CD103 expression after exposure to bioactive TGF-β (G. A. Hadley, unpublished observations). The striking similarity of such induction to that exhibited by graft-infiltrating 1B2+CD8+ cells (Fig. 5, B and C) provides compelling evidence that TGF-β activity within the local environment of the allograft induces CD103 expression by graft-infiltrating CD8 effectors. This hypothesis is consistent with the well-documented association of TGF-β with the renal environment after injury and disease (19), including renal allograft rejection (20, 21, 22).
The concept that TGF-β can promote differentiation of CD8 effectors is apparently at odds with the well-documented immunosuppressive properties of this molecule (23). Indeed, TGF-β efficiently blocks CD8 activation and early maturation events at multiple levels (24, 25); introduction of even minute amounts of bioactive TGF-β at the initiation of MLC cultures completely blocks CTL generation (6). It is important to note, however, that exposure of CD8 cells to bioactive TGF-β subsequent to their progression to effector status induces high level CD103 expression without compromising CD8 effector function (6). The present in vivo findings strongly argue that CD8 cells encounter significant TGF-β activity only subsequent to entry into renal allografts, a time at which they have already progressed to an advanced stage of differentiation and thus are not susceptible to TGF-β-mediated suppression. These data are consistent with the pleiotropic and sometimes paradoxical effects of TGF-β in inflammation and wound repair, where its mode of action (inhibition vs stimulation) is dependent upon the nature of the target cell, the state of cellular differentiation, and the context of action (26).
The present data challenge prevailing dogma in the field of immune monitoring that donor-reactive T cells present in peripheral lymphoid compartments (e.g., blood) of allograft recipients are representative of those that infiltrate the graft. Thus, the overwhelming majority of CD103+CD8+ cells in peripheral lymphoid compartments of renal allograft recipients are naive CD8 cells (Fig. 3). We found no evidence that CD103+CD8+ effectors generated at the graft site recirculate to other compartments; i.e., allospecific CD8+CD103+ cells were not detectable in draining or peripheral lymph nodes, spleen, or blood at any time point examined (Fig. 5, B and C). It is unlikely that this reflects rapid down-modulation of CD103 expression subsequent to emigration from the graft site, because in vitro studies indicate that CD103 expression by TGF-β-induced effectors is highly stable even in the absence of exogenous TGF-β (G. A. Hadley, unpublished observations). We postulate that sequestration of CD103+CD8+ effectors at the site of renal allografts reflects the epitheliotropic properties of CD103; i.e., CD103 may promote adhesion of donor-specific CD103+CD8+ effectors to graft renal tubules, which are known to express high levels of its ligand, E-cadherin (27).
CD8-mediated destruction of epithelial layers is a hallmark of diverse pathologic processes, including not only organ allograft rejection (1), but also organ-specific autoimmune disorders (28) and graft-vs-host disease pathology (29). Similarly, physiologic host defense against intracellular parasites (30) and tumors (31) often requires destruction of infected/neoplastic epithelial cells by CD8 effector populations. Based on the present findings, we postulate that local TGF-β activity plays a general role in promoting such interactions. Consistent with this possibility, TGF-β is an important mediator of tissue remodeling and repair at sites of injury and is ubiquitous at peripheral sites of inflammation (26). Furthermore, there is compelling evidence that CD103 expression by mucosal T cells in the intestinal intraepithelial compartment is controlled by local TGF-β activity (4, 32). Combined with the documented capacity of CD103 to promote destruction of epithelial layers by CD8 effector populations (11), this model provides a plausible mechanistic basis for diverse tissue-specific immune phenomena including the extreme immunogenicity of skin, lung, and pancreas allografts compared with nonepithelial grafts, such as heart and hemopoietic tissue (33, 34), preferential targeting of epithelial layers in the skin, gut, and liver in graft-vs-host disease pathology (29), and the local nature of organ-specific autoimmune disorders (35). It will now be important to determine whether the capacity to promote CD103 expression by CD8 effector populations is a general property of peripheral inflammatory sites or, alternatively, is unique to specialized microenvironments such as the gut and kidney.
In summary, the salient finding of the present study is that CD8 cells responding to renal allografts do not initially express significant levels of CD103, but dramatically up-regulate CD103 expression subsequent to entry into the graft site. We demonstrate that such induction is dependent on TGF-β activity and provide evidence that the requisite TGF-β activity operates at the local level of the graft. These data expand our understanding of CD8 homing events and provide general insight into the mechanisms that promote destruction of epithelial layers by CD8 effector populations.
We are indebted to Dr. Jan Cerny for critical reading of this manuscript and helpful comments.
This work was supported by National Institutes of Health Research Grant AI36532 (to G.A.H.).
Abbreviations used in this paper: DNRII, dominant negative type II TGF-β receptor; GIL, graft-infiltrating lymphocyte; ICS, intracellular cytokine staining.