Tubulitis and Epithelial Cell Alterations in Mouse Kidney Transplant Rejection Are Independent of CD103, Perforin or Granzymes A/B

Authors


* Corresponding author: P. F. Halloran, phil.halloran@ualberta.ca

Abstract

One of the defining lesions of kidney allograft rejection is epithelial deterioration and invasion by inflammatory cells (tubulitis). We examined epithelial changes and their relationship to effector T cells and to CD103/E-cadherin interactions in mouse kidney allografts. Rejecting allografts showed interstitial mononuclear infiltration from day 5. Loss of epithelial mass, estimated by tubular surface area, and tubulitis were minimal through day 7 and severe by day 21. Tubules in day 21 allografts manifested severe reduction of E-cadherin and Ksp-cadherin by immunostaining with redistribution to the apical membrane, indicating loss of polarity. By flow cytometry T cells isolated from allografts were 25% CD103+. Laser capture microdissection and RT-PCR showed increased CD103 mRNA in the interstitium and tubules. However, allografts in hosts lacking CD103 developed tubulitis, cadherin loss, and epithelial deterioration similar to wild-type hosts. The loss of cadherins and epithelial mass was also independent of perforin and granzymes A and B. Thus rejection is characterized by severe tubular deterioration associated with CD103+ T cells but not mediated by CD103/cadherin interactions or granzyme-perforin cytotoxic mechanisms. We suggest that alloimmune effector T cells mediate epithelial injury by contact-independent mechanisms related to delayed type hypersensitivity, followed by invasion of the altered epithelium to produce tubulitis.

Introduction

The key features of T-cell-mediated kidney allograft rejection are deterioration in function, interstitial infiltration by mononuclear cells and entry of mononuclear inflammatory cells into the tubule epithelium, designated tubulitis (1). Tubulitis is one of the main lesions for diagnosing rejection in the Banff schema (1) http://tpis.upmc.edu/tpis/schema/KNCode97.html, correlating with functional deterioration (2,3), and may be relevant to the tubular atrophy that often follows rejection. Tubulitis is T-cell mediated: It can develop in mouse hosts lacking B cells and alloantibody (4) and is typically absent in human alloantibody mediated rejection (5). Many infiltrating cells in tubulitis lesions display cytotoxic T lymphocyte (CTL) features, with CD8+ cells expressing perforin 1 (Prf1) being found in the tubular epithelium (6,7). However, tubulitis is not dependent on granule-associated CTL mechanisms, because it can develop in allografts rejecting in hosts lacking Prf1 or granzyme A (GzmA) and granzyme B (GzmB) (8).

Integrin CD103 ('E/β7, Itgae/Itgb7), which engages E-cadherin on the epithelium (9–11), may have a role in tubulitis. Thus CD103 on effector T cells may mediate interaction with epithelial cells (12,13), promoting cytotoxicity (14) and potentially affecting other processes through the role of E-cadherin in adherens junction signaling (15,16) and induction of cell polarity (17). Renal tubules also express other cadherins including kidney specific (Ksp)-cadherin (18), which is not known to engage CD103 (19). CD103+ CD8+ CTL are present in rejecting human renal allografts (14,20,21) and CD103 deficiency impairs islet allograft rejection (22). A recent study of the role for CD103+CD8+ effector cells in rat kidney allograft rejection found that treatment with a monoclonal antibody to CD103 reduced accumulation of CD8+ cells in graft renal tubules and attenuated tubular injury in cyclosporine-treated rats (23). In contrast, in rats with a fully functional immune system, i.e. without immunosuppression, rejection was not altered by anti-CD103. Thus the role of CD103 in kidney epithelial injury remains to be determined.

Alloimmune T cells may mediate epithelial injury in kidney transplants either through direct contact or through contact independent mechanisms. CTL could engage and damage individual epithelial cells, possibly involving CD103-E-cadherin interactions. Alternatively, the interstitial effector T cells and macrophages could injure epithelial cells indirectly due to interstitial inflammation analogous to delayed type hypersensitivity, via release of soluble mediators, alterations in the extracellular matrix, or changes in the blood vessels (endothelial inflammation and injury) and blood supply (vasoconstriction). The present studies were designed to explore the epithelial changes that occur in T-cell-mediated rejection, the development of tubulitis, and their relationship to T-cell effector mechanisms.

Materials and Methods

Mice

Male CBA/J (CBA) and C57Bl/6 (B6) mice were obtained from Jackson Laboratory (Bar Harbor, ME). CD103 deficient mice (13) (CD103−/−), received from Dr C. M. Parker, were bred at the University of Maryland. We confirmed that the CD103−/− mice were homozygous by PCR on genomic DNA using primer sequences flanking the inserted neomycin resistance gene as described in Ref. (13). Prf1 deficient (Prf−/−) and GzmA and GzmB deficient (GzmAB−/−) mice were obtained from Dr. Chris Bleackley (Dept. of Biochemistry, University of Alberta) (8). Mice were maintained in the Health Sciences Laboratory Animal Services at the University of Alberta. All maintenance and experiments conformed to animal care protocols approved by the University of Alberta.

Transplants

Non-life-supporting renal transplants were performed as previously described (24) using wild-type CBA (H-2k) mice as donors and wild-type B6 (H-2b) and BALB/c (H-2d) or Prf−/−, GzmAB−/− (B6 background) or CD103−/− (BALB/c background) as recipients. Hosts did not receive immunosuppression. Contralateral host kidney and naïve CBA kidney served as controls. Kidneys were harvested on days 5, 7 and 21 post-transplant and stored as previously described (25).

Antibodies

For immunohistochemistry and Western blotting, we used the following monoclonal antibodies (mAb): rat mAb to E-cadherin (Calbiochem-Novabiochem Corporation, San Diego CA); mouse mAb to Ksp-cadherin (Zymed Laboratories Inc., San Francisco, CA); rat mAb to CD3 (Serotec, Oxford, UK); rat mAb to CD4 (BD Pharmingen, Mississauga, ON, Canada); rat mAb to CD8 (Serotec, Oxford, UK); rat mAb to CD45 (BD Pharmingen, Mississauga, ON, Canada); rat mAb to CD68 (Serotec, Oxford, UK); HRP-conjugated goat affinity purified F(ab')2 to rat IgG (ICN Pharmaceuticals, Inc., Aurora, OH); HRP-conjugated rabbit anti-rat and HRP-conjugated goat anti-mouse antibody (Jackson Immunoresearch Laboratories Inc., West Grove, PA). For flow cytometry, we used anti-mouse FcγRIII/II (BD Pharmingen, Mississauga, ON, Canada); anti-CD3ε, anti-CD103, anti-CD44 and anti-CD62L (eBioscience, San Diego, CA); anti-CD4 and anti-CD8 (BD Pharmingen).

Histology and electron microscopy

For each sample, paraffin sections (2 μm) were stained with periodic acid-Schiff (PAS) and subjected to histologic analysis as described previously (4). Electron microscopy was performed on glutaraldehyde-fixed tissue by the Department of Laboratory Medicine and Pathology, University of Alberta. Infiltrating mononuclear cells were stained for CD markers (CD3, CD4, CD8, CD45, CD68) and were counted in 10 ocular grid areas at ×400 (n = 6 for D5 allografts, n = 4 for D7 allografts).

Image analysis of tubular epithelial surface area

We estimated tubular epithelial surface area by using the ScanScope® T3 (Aperio Technologies, Inc., Vista, CA). Microscope slides were scanned to produce a high quality resolution digital image (0.25 μm/pixel at 40×) (ScanScope® Console software) and analyzed using ImageScope™ viewing software (Aperio Technologies, Inc., Vista, CA). We measured tubular surface area using the “positive pixel count algorithm”. This algorithm quantifies the amount of a user-defined stain (PAS stain) present in a selected area by counting the number of pixels which satisfy the color specification and calculates the corresponding surface area. On PAS stained virtual slides, we randomly selected 100 tubular sections per sample and calculated the mean surface area per sample, based on the stained tubular cells. In order to standardize the analysis, the selection of tubules was restricted to the cortex area and followed a route from on kidney pole to the other, selecting non-overlapping areas along the cortex. We analyzed all circular cross-sections in the selected areas but excluded longitudinal sections from the analysis. Because both distal and proximal nephrons are represented in the cortex, we chose to analyze 100 tubular sections per sample; this relatively large number ensured that the ratio of distal and proximal tubular sections was similar across all samples. The algorithm measured only tubular epithelial surface area, excluding the tubular lumen. Necrotic areas or atrophic tubules were excluded from this analysis.

Immunohistochemistry

Cryostat sections (4 μm) were incubated with primary antibodies to E-cadherin or Ksp-cadherin or isotype IgG as control, followed by secondary HRP-conjugated antibodies. E-cadherin staining was scored in the cortex area as the percentage of tubules with positive staining (<25%: 1; 25–75%: 2; >75%: 3) as well as average staining intensity (weak: 1; intermediate: 2; strong: 3). Separate scores were given for staining of the basolateral and the apical tubular membrane.

Flow cytometry

Kidneys (n = 3) were homogenized, placed in 10 mL of PBS containing 2% BSA and 2 mg/mL collagenase (Sigma-Aldrich), and incubated (37°C for 1 h) with occasional pressing through a syringe plunger. Cells were strained, washed and resuspended in PBS containing 0.5% FCS. Prior to flow cytometry, Fc receptors were blocked with anti-mouse FcγRIII/II antibody, and 1 × 106 cells were stained using anti-CD3ε, anti-CD103, anti-CD4, anti-CD8, anti-CD44, and anti-CD62L antibodies (diluted in 0.5% FCS/PBS).

Real-time RT-PCR

Expression of CD103, E-cadherin, and Ksp-cadherin was assessed by TaqMan real-time RT-PCR. Total kidney RNA was extracted using CsCl density gradient. Two micrograms of RNA were reverse transcribed using M-MLV reverse transcriptase and random primers. For laser capture microdissection (LCM), we stained frozen sections (8 μm) with HistoGene LCM Frozen Section Staining kit (Arcturus, Mountain View, CA), captured tubules and interstitial material from day 21 transplants with the LCM instrument (Arcturus, Mountain View, CA) and extracted total RNA from 150 tubules and interstitial areas using the PicoPure RNA isolation kit (Arcturus). Purified RNA was reverse transcribed and amplified using the TaqMan One-Step RT-PCR kit (Applied Biosystems, Foster City, CA) in a multiplex reaction for 48 cycles. TaqMan probe/primer combinations were obtained as assay on demand (Applied Biosystems) (Ksp-Cadherin) or designed using Primer Express software version 1.5 (PE Applied Biosystems) (CD103: forward: 5′-CAGGAGACGCCGGACAGT; reverse: 5′-CAGGGCAAAGTTGCACTCAA; probe: 5′-AGGAAGATGGCACTGAGATCGCTATTGTCC; E-Cadherin: forward: 5′-CTGCCATCCTCGGAATCCTT; reverse: 5′-TGGCTCAAATCAAAGTCCTGGT; probe: 5′-AGGGATCCTCGCCCTGCTGATTCTGATC). We quantified gene expression as previously described (26). HPRT and ribosomal RNA were used as controls; because ct values for HPRT were in the same range as those for E-cadherin and varied within 1-2 ct values, we normalized the data to the more highly expressed and thus more stable ribosomal 18S-RNA (100× diluted). Data are expressed relative to the expression in normal kidneys.

Microarrays

We performed microarray analysis on normal kidneys (NCBA), CBA into BALB/c wild-type and CBA into CD103−/− allografts at day 21. RNA extraction, dsDNA and cRNA synthesis, hybridization to MOE430 2.0 oligonucleotide arrays (GeneChip, Affymetrix®), washing and staining were carried out according to the Affymetrix Technical Manual (http://www.affymetrix.com), as previously described (7). Equal amounts of RNA from three mice for NCBA and allografts in wild-type recipients and RNA from two mice for allografts in CD103−/− recipients were pooled for each array. Two or three biological replicates were tested for allografts and NCBA, respectively. We normalized and analyzed data using GeneChip Operating Software 1.2 (Affymetrix) and GeneSpring™ software (Version 7.2, Silicon Genetics, CA) as described previously (7).

Western blots

Approximately 40 mg of kidney was homogenized in buffer (0.1% Nodinet P-40, 0.05% sodium deoxycholate, 0.01% SDS, 150 mM NaCl, 40 mM Tris-HCl pH 7.6, 10 mM 2-mercaptoethanol), treated with PMSF then centrifuged. 150 μg of protein were run on 7.5% SDS-PAGE mini-gels (Bio-Rad, Mississauga, ON, Canada) and wet-transferred to Hybond C+ membranes (Amersham Biosciences, Baie d'Urfe, QB, Canada). Quality of transfer and evenness of loading was confirmed with Ponceau S (Sigma-Aldrich). Samples were destained in TBST (140 mM NaCl, 40 mM Tris-HCl pH 7.6, 0.1% Tween 20) and blocked with 5% milk-TBST. To preserve E-cadherin epitopes, all solutions contained 10 mM CaCl2. Blots were incubated with primary antibodies overnight (3 μg/mL, 4°C), followed by secondary antibodies (1:5000 in 1% milk/TBST; 1 h at room temperature). Immune complexes were detected with ECL reagent (Amersham Biosciences) using Fuji Super RX films. Developed films were scanned using GS-800 densitometer and quantified using Quantity One software (Bio-Rad).

Results

Development of interstitial infiltrate and tubulitis

Host kidneys and isografts at days 5, 7 and 21 displayed normal histology with minimal inflammation and no tubulitis (Figure 1A,B). As previously described (4,7), allografts showed focal periarterial mononuclear infiltration at days 3 and 4, and interstitial mononuclear infiltration by day 5 (Figure 1C). Tubulitis was absent at day 3, 4 and 5 (Figure 1D). By day 7, interstitial mononuclear infiltration increased (Figure 1E), accompanied by veinitis and edema with increased kidney weight (308 ± 58 mg vs. 195 ± 20 mg in normal kidney (↑ 58%)), while tubulitis was minimal, tubule structure was preserved and arteritis was absent (Figure 1F). We characterized the cellular infiltrate at days 5 and 7 by immunostaining; the results are shown in Table 1. In the late allografts at day 21 interstitial infiltration persisted (Figure 1G), with edema (increase in kidney weight by 100% compared to normal kidney) and severe tubulitis with distortion and shrinkage of tubule cross-sections (Figure 1H), accompanied by veinitis and endothelial arteritis, with patchy cortical necrosis.

Figure 1.

Histologic lesions in allografts (CBA into B6 allograft; PAS staining). (A) Isograft controls (CBA into CBA) do not develop any histologic lesions at any timepoint, as shown here in day 21 isografts as an example (magnification 20×). (B) Day 21 isografts at magnification 100×. (C) Day 5 allograft with periarterial infiltration (magnification 20x). (D) Day 5 allografts show no tubulitis (magnification 100×). (E) In day 7 allografts infiltration increases and extends to the interstitium (magnification 20×). (F) Day 7 transplants have preserved tubular structure (magnification 100×). (G) Day 21 allografts show severe interstitial infiltration and edema (magnification 20×). (H) Day 21 allograft with marked tubulitis (arrows) and distorted tubules (magnification 100×).

Table 1.  Leukocyte markers by immunohistochemistry in CBA into B6 allografts at days 5 and 71
 CBA/B6 allograft D5CBA/B6 allograft D7
  1. 1CD markers were counted in 10 ocular grid areas at ×400 (n = 6 for D5 allografts, n = 4 for D7 allografts). Numbers are shown as mean ± SD.

CD3248 ± 235432 ± 25
CD4102 ± 10445 ± 37
CD8135 ± 137221 ± 153
CD45555 ± 374664 ± 402
CD68167 ± 156260 ± 139

Cadherin staining is decreased in rejecting kidneys

We analyzed expression of E-cadherin, the ligand for CD103, in rejecting allografts. By immunostaining, E-cadherin was expressed on the basolateral membrane of tubular epithelial cells in normal kidneys and host kidneys (data not shown). Isografts at day 7 and day 21 showed the same staining intensity and distribution as normal kidneys. All tubules were positive for E-cadherin, although the intensity was highly variable among tubules. (Figure 2A,C). In allografts at day 7, staining intensity was unchanged compared to normal kidneys and isografts (Figure 2B), but by day 21 E-cadherin staining was both severely decreased and redistributed, with loss of basolateral staining and staining of the luminal membrane in some tubules (Figure 2D). Distribution of Ksp-cadherin in control kidneys (normal kidneys and isografts) was similar to that for E-cadherin, with strong staining of the basolateral membrane (Figure 2E,G). In allografts, Ksp-cadherin staining intensity was decreased at day 7 (Figure 2F) and greatly diminished at day 21 (Figure 2H), similar to E-cadherin.

Figure 2.

Immunohistochemical staining of E-cadherin and Ksp-cadherin (magnification 100×). Arrows show localization of cadherins. (A) In isografts (CBA into CBA) at day 7 post-transplant, E-cadherin was localized to the basolateral membrane. (B) In allografts rejecting in wild-type hosts (CBA into B6) at day 7 post-transplant, E-cadherin remained unchanged. (C) Isografts at day 21 showed no change in E-cadherin staining compare to normal kidneys or isografts at day 7, while (D) in allografts (CBA into B6) at day 21 post-transplant, E-cadherin staining was decreased and redistributed to the apical membrane. (E) Ksp-cadherin was localized to the basolateral membrane in isografts (CBA into CBA) at day 7 post-transplant. (F) In allografts rejecting in wild-type hosts (CBA into B6) at day 7 Ksp-cadherin staining was decreased. (G) In isografts at day 21 Kap-cadherin staining was unchaged compare to normal kidneys or isografts at day 7. (H) Allografts at day 21 post-transplant showed severe loss of Ksp-cadherin staining. (I) Allografts rejecting in CD103−/− hosts (CBA into CD103−/−) showed loss and redistribution of E-cadherin staining and (J) loss of Ksp-cadherin staining, indistinguishable from day 21 allografts rejecting in wild-type hosts. (K) E-cadherin staining was decreased in allografts rejecting in Prf1−/− hosts (CBA into Prf1−/−) and (L) in allografts rejecting in GzmAB−/− hosts (CBA into GzmAB−/−), indistinguishable from day 21 allografts in wild-type hosts.

In Western blot analysis, E-cadherin protein expression was unchanged in day 7 allografts and decreased to 58% at day 21 compared to control kidneys. Ksp-cadherin protein decreased early to 76% at day 7 and 50% at day 21 (Figure 3A). E-cadherin mRNA levels fell transiently in allografts at day 5 and recovered by day 21. Ksp-cadherin mRNA decreased by 50% in day 5 allografts and remained depressed through day 21 (Figure 3B). Thus E-cadherin and Ksp-cadherin expression were severely altered by immunostaining, with decreases in mRNA and protein expression. However, the changes in immunostaining were more extensive than the mRNA and protein changes.

Figure 3.

Figure 3.

E-cadherin and Ksp-cadherin in rejecting allografts. (A) Western blot analysis of E-cadherin and Ksp-cadherin protein expression. Fold changes were calculated from the band intensity ratio of allografts (CBA into B6) versus controls (C) (contralateral kidney: B6). Numbers represent mean ± SD, n = 3. Basal levels of cadherins did not differ significantly between normal (CBA) and contralateral kidneys (B6). (B) Real time RT-PCR analysis of cadherin mRNA expression in allografts (CBA into B6) and isografts (CBA into CBA) at days 5, 7 and 21. Values are fold changes relative to normal CBA kidney (NCBA), expressed as mean ± SD (n = 2, three kidneys in each pool). Cadherin mRNA levels were normalized to ribosomal RNA. Assays were done in duplicate. (C) Real-time RT-PCR analysis of E-cadherin and Ksp-cadherin mRNA expression in allografts in wild-type Balb/c (WT) or CD103−/− hosts at day 21.

Figure 3.

Figure 3.

E-cadherin and Ksp-cadherin in rejecting allografts. (A) Western blot analysis of E-cadherin and Ksp-cadherin protein expression. Fold changes were calculated from the band intensity ratio of allografts (CBA into B6) versus controls (C) (contralateral kidney: B6). Numbers represent mean ± SD, n = 3. Basal levels of cadherins did not differ significantly between normal (CBA) and contralateral kidneys (B6). (B) Real time RT-PCR analysis of cadherin mRNA expression in allografts (CBA into B6) and isografts (CBA into CBA) at days 5, 7 and 21. Values are fold changes relative to normal CBA kidney (NCBA), expressed as mean ± SD (n = 2, three kidneys in each pool). Cadherin mRNA levels were normalized to ribosomal RNA. Assays were done in duplicate. (C) Real-time RT-PCR analysis of E-cadherin and Ksp-cadherin mRNA expression in allografts in wild-type Balb/c (WT) or CD103−/− hosts at day 21.

Figure 3.

Figure 3.

E-cadherin and Ksp-cadherin in rejecting allografts. (A) Western blot analysis of E-cadherin and Ksp-cadherin protein expression. Fold changes were calculated from the band intensity ratio of allografts (CBA into B6) versus controls (C) (contralateral kidney: B6). Numbers represent mean ± SD, n = 3. Basal levels of cadherins did not differ significantly between normal (CBA) and contralateral kidneys (B6). (B) Real time RT-PCR analysis of cadherin mRNA expression in allografts (CBA into B6) and isografts (CBA into CBA) at days 5, 7 and 21. Values are fold changes relative to normal CBA kidney (NCBA), expressed as mean ± SD (n = 2, three kidneys in each pool). Cadherin mRNA levels were normalized to ribosomal RNA. Assays were done in duplicate. (C) Real-time RT-PCR analysis of E-cadherin and Ksp-cadherin mRNA expression in allografts in wild-type Balb/c (WT) or CD103−/− hosts at day 21.

CD103 is associated with but not required for tubulitis

T cells expressing integrin αEβ7 (CD103) are associated with tubulitis (27). We confirmed the presence of CD103+ T cells in allografts at day 21 by flow cytometry: 25 ± 4% of T cells were CD103+, with 54 ± 4.8% of CD103+ T cells being CD4+ and 44 ± 4.8% being CD8+. By RT-PCR analysis of RNA isolated from whole kidney, CD103 mRNA increased 4-fold at day 5 and 14-fold at day 7 compared to normal kidneys, and remained 12-fold elevated at day 21. Using LCM, we confirmed the presence of CD103 mRNA in the interstitium as well as in the epithelium of allografts with established tubulitis at day 21. Compared to normal kidney, CD103 mRNA increased 42-fold in the interstitial infiltrate and 91-fold in tubules in day 21 allografts (Figure 4).

Figure 4.

RT-PCR analysis of CD103 expression in host kidneys and in the interstitial infiltrate and tubules (laser capture microdissection) of CBA into B6 allografts at day 21. Numbers represent fold change vs. normal CBA kidney.

To assess the role of CD103 in tubulitis lesions and epithelial deterioration, we studied histologic lesions in the absence of CD103. Allografts transplanted into CD103−/− hosts developed histologic lesions similar to allografts in wild-type hosts (Figure 5A and B), with edema, distortion of tubules and florid tubulitis. Electron micrographs of tubulitis lesions in allografts in CD103−/− hosts were similar to those in wild-type hosts, with intra-epithelial inflammatory cells tightly applied to the epithelial cell membranes (Figure 5C and D). Semi-quantitative assessment of day 21 allografts revealed no differences in histologic lesions, including tubulitis, edema (assessed by kidney weight), and loss of epithelial mass (as indicated by decreased tubular epithelial surface area) (Table 2). Epithelial surface area was reduced by 51% in allografts transplanted into CD103−/− hosts at day 21 (846 ± 37 μm2) compared to normal CBA kidneys (1736 ± 165 μm2) (p = 0.003). The decrease in tubular cross-sectional area in allografts rejecting in CD103−/− hosts was similar to wild-type hosts (840 ± 49 μm2) (p = NS) (Figure 6).

Figure 5.

Histology of allografts rejecting in wild-type (CBA into Balb/c) or CD103−/−(CBA into CD103−/−) hosts at day 21. (A) Allografts in wild-type hosts show interstitial edema, marked tubulitis (arrows) and distorted tubules (PAS staining, magnification 60×). (B) Allografts in CD103−/− hosts show interstitial edema, marked tubulitis (arrows) and distorted tubules, indistinguishable from wild-type (PAS staining, magnification 60×). (C) Electron microscopy of tubulitis lesions in allografts rejecting in wild-type hosts (D) Electron microscopy of tubulitis lesions in allografts rejecting in CD103−/− hosts. (inline image Lymphocytes within the tubular epithelial cells;inline image lymphocytes in the interstitium;inline image tubular basement membrane.)

Table 2.  Histopathologic changes in kidneys in wild-type Balb/c (WT.Balb/c) and in CD103−/− hosts at day 21 post-transplant1
HistologyWT.Balb/c D21 (n = 5)CD103−/− D21 (n = 4)
  1. 1Interstitial infiltrate, tubulitis, graft necrosis and peritubular capillary congestion (PTC) were recorded as a percentage positive of the whole cortex area. Glomerulitis lesions were scored from 0 to 3 (0 = no change, 1 = 0-25%, 2 = 25–75%, and 3 = 75–100% of total parenchyma involved). Arteritis and veneitis lesions were counted and given as the mean number of involved vessels per kidney section. Numbers shown are mean ± SD. Numbers in parentheses indicate the % decrease compared to control kidneys. There were no significant differences between wild-type and CD103−/− hosts.

Weight (mg)340 ± 123338 ± 48
Necrosis (%)4.0 ± 8.923 ± 21
PTC (%)20 ± 2453 ± 25
Glomerulitis2.8 ± 0.43.0 ± 0.0
Tubulitis (%)64 ± 5.563 ± 10
Interstitial infiltrate (%)56 ± 5.543 ± 10
Arteritis1.2 ± 1.11.8 ± 1.0
Arterial thrombosis0.00.0
Veneitis1.8 ± 1.31.8 ± 1.7
Mean tubular860 ± 49846 ± 37
 surface area (μm2) (↓ 50%) (↓ 51%)
Figure 6.

Figure 6.

Tubular epithelial surface area in allografts rejecting in wild-type and CD103−/−, Prf−/−and GzmAB−/−hosts. (A) Tubular epithelial surface area was measured on PAS stained virtual slides. We randomly selected 100 tubular cross sections per sample (green circles). The pixel algorithm calculates the area of selected sections, measuring only areas that meet the specified color criteria (regardless of color intensity; blue and yellow coloring). The algorithm measured only tubular epithelial surface area, excluding the tubular luminal surface area. Necrotic areas or atrophic tubules were excluded from this analysis. (B) Mean tubular surface area was assessed by image analysis in normal kidneys (NCBA), day 21 isografts (CBA into CBA), and day 21 allografts. Allografts rejecting in hosts deficient in CD103 (CD103−/−), granzymes A and B (GzmAB−/−) or perforin (Prf−/−) are compared to allografts rejecting in wild-type (WT) hosts of the same strain. For each sample, we calculated the mean tubular surface area of 100 cortical tubular cross sections. Numbers represent mean ± SD for each experimental group (allografts and NCBA: n = 4, isografts: n = 3).

Figure 6.

Figure 6.

Tubular epithelial surface area in allografts rejecting in wild-type and CD103−/−, Prf−/−and GzmAB−/−hosts. (A) Tubular epithelial surface area was measured on PAS stained virtual slides. We randomly selected 100 tubular cross sections per sample (green circles). The pixel algorithm calculates the area of selected sections, measuring only areas that meet the specified color criteria (regardless of color intensity; blue and yellow coloring). The algorithm measured only tubular epithelial surface area, excluding the tubular luminal surface area. Necrotic areas or atrophic tubules were excluded from this analysis. (B) Mean tubular surface area was assessed by image analysis in normal kidneys (NCBA), day 21 isografts (CBA into CBA), and day 21 allografts. Allografts rejecting in hosts deficient in CD103 (CD103−/−), granzymes A and B (GzmAB−/−) or perforin (Prf−/−) are compared to allografts rejecting in wild-type (WT) hosts of the same strain. For each sample, we calculated the mean tubular surface area of 100 cortical tubular cross sections. Numbers represent mean ± SD for each experimental group (allografts and NCBA: n = 4, isografts: n = 3).

The decrease and redistribution of E-cadherin and Ksp-cadherin staining in allografts rejecting in CD103−/− hosts was similar to that in wild-type hosts (Figure 2I,J). Cadherin mRNA levels were also similar to wild-type hosts, with a decrease in Ksp-cadherin mRNA and persistence of E-cadherin mRNA (Figure 3C).

By microarray analysis, gene expression in allografts in CD103−/− hosts correlated strongly with that in allografts rejecting in wild-type hosts (r= 0.91). Figure 7 shows the regression plot of gene expression in CD103−/− vs. wild-type hosts. This is similar to the regression between biological replicate pools of either wild-type or CD103−/− hosts results). These results illustrate the transcriptome changes in rejecting allografts compared to controls, but no difference between allografts rejecting in wild-type hosts or CD103−/− hosts.

Figure 7.

Microarray analysis of allografts rejecting in wild-type hosts or in CD103−/−hosts at day 21. Correlation of gene expression between normal kidneys (NCBA) and CBA allografts in wild-type Balb.c hosts (WT), between WT and CBA allografts in CD103−/− hosts (CD103−/−), and between replicate pools of either WT or CD103−/−. Numbers represent fold change in expression compared to normal CBA kidney. Lines represent a 2-fold increase or decrease in CD103−/− hosts compared to wild-type hosts.

We previously reported a set of cytotoxic T-cell associated transcripts (CATs) in rejecting kidneys reflecting T cells recruited to the graft (7). Expression of CATs correlated strongly between allografts in wild-type hosts and CD103−/− hosts (r= 0.95). Compared to normal kidneys, geometric mean of CAT expression in wild-type was 4.3-fold increased, similar to allografts in CD103−/− hosts (5.0-fold increased), indicating a similar T-cell burden (Figure 8). Statistical analysis of all transcripts on the array (n = 45 102) by Bayesian t-test (false discovery rate 0.05) did not identify any transcripts as differentially expressed between allografts in wild-type or CD103−/− hosts (p < 0.05). Thus the transcriptomes confirm the histopathologic observations, indicating the CD103 deficiency has little effect in these allografts.

Figure 8.

Microarray analysis of CTL-associated transcripts (CATs, n = 287) in allografts rejecting in wild-type hosts or in CD103−/−hosts at day 21. Average CAT-expression in allografts in wild-type hosts and CD103−/− hosts. The boxes represent the median and 25th and 75th percentile; numbers represent fold change in gene expression compared to normal CBA kidney.

Granzyme-perforin cytotoxic mechanisms are not required for epithelial deterioration

We previously showed that tubulitis is T-cell dependent and occurs in hosts lacking mature B cells (4), but is unchanged in hosts lacking Prf1 or GzmA and GzmB (8). We studied whether kidney allografts lose cadherin staining and epithelial mass (decreased tubular surface area) in hosts lacking Prf1 or GzmA and GzmB. Loss of E-cadherin (Figure 2K,L) and Ksp-cadherin (data not shown) was unchanged in Prf−/− or GzmAB−/− hosts compared to wild-type hosts at day 21. The percentage of tubules with positive staining as well as average staining intensity was decreased, with re-distribution to the apical membrane similar to wild-type hosts (Table 3).

Table 3.  Histology, E-cadherin staining and tubular surface area in normal kidneys (NCBA), isografts and allografts rejecting in wild-type hosts (WT) or in hosts lacking granzyme A and B (GzmAB−/−) or perforin (Prf−/−) at day 211
 E-Cadherin stainingTubular surface area (μm2)
Staining scoreStaining intensity
BasolateralApicalBasolateralApical
  1. 1E-cadherin staining was scored as the percentage of tubules in the cortex area with positive staining (<25%: 1; 25–75%: 2; >75%: 3); staining intensity was scored as weak (1), intermediate (2) or strong (3). Separate scores were obtained for staining of the basolateral or apical membrane. Staining was evaluated in the cortex area of n = 4 (NCBA), n = 6 (WT), n = 5 (GzmAB−/−) and n = 4 (Prf−/−) allografts. Numbers represent mean values. Tubular surface area was measured in four animals for each experimental condition. Numbers shown are mean ± SD.

NCBA 3.0 ± 0.00.0 ± 0.03.0 ± 0.00.0 ± 0.01736 ± 165
Isograft D7Tx3.0 ± 0.00.0 ± 0.03.0 ± 0.00.0 ± 0.01577 ± 88
Isograft D21Tx3.0 ± 0.00.0 ± 0.03.0 ± 0.00.0 ± 0.01548 ± 167
WT D7Host3.0 ± 0.00.0 ± 0.03.0 ± 0.00.0 ± 0.0N.D.
Tx3.0 ± 0.00.0 ± 0.02.7 ± 0.60.0 ± 0.01327 ± 130
WT D21Host3.0 ± 0.00.0 ± 0.03.0 ± 0.00.0 ± 0.0N.D.
Tx1.8 ± 0.81.7 ± 0.51.2 ± 0.41.8 ± 0.41018 ± 128
GzmAB−/− D21Host3.0 ± 0.00.0 ± 0.03.0 ± 0.00.0 ± 0.0N.D.
Tx1.4 ± 0.61.4 ± 0.61.0 ± 0.01.8 ± 0.5843 ± 71
Prf −/− D21Host3.0 ± 0.00.0 ± 0.02.8 ± 0.50.0 ± 0.0N.D.
Tx1.8 ± 0.51.8 ± 0.51.0 ± 0.01.4 ± 0.61116 ± 227

In normal kidneys tubular epithelial mass was 1736 ± 165 μm2. At day 21, epithelial mass was decreased in allografts rejecting in Prf−/− (832 ± 71 μm2, decreased by 51%) and in GzmAB−/− hosts (1116 ± 227 μm2, decreased 35%),(p < 0.05), similar to wild-type hosts (1018 ± 128 μm2 decreased by 41%) (Table 3; Figure 6).

Discussion

We studied the evolution of epithelial changes in kidney allograft rejection and their relationship to immunologic effector mechanisms, particularly cytotoxic mechanisms and CD103. At days 5 and 7 post-transplant, the interstitial mononuclear infiltrate was excluded from the epithelium, whereas day 21 allografts manifested severe tubulitis and decreased tubular surface area. E-cadherin and Ksp-cadherin were reduced and redistributed when tubulitis developed, indicating loss of polarity. CD103 mRNA was increased in kidneys before tubulitis developed, peaking at day 7 and persisting through day 21, associated with tubule cross-sections by LCM at day 21. However, tubulitis and shrinkage of tubules developed in allografts in hosts lacking CD103, and the reduction and redistribution of E-cadherin were not dependent on CD103, Prf1, or GzmA and GzmB. The independence of the epithelial deterioration from CD103 and Prf1-Gzm suggests that the interstitial effector T cells and macrophages produce the epithelial changes that are the central lesion of graft rejection via a contact-independent mechanism e.g. delayed type hypersensitivity.

The loss of cadherins is probably a late stage in the response of the epithelium to interstitial inflammation and could reflect an early stage of epithelial mesenchymal transition (28). Loss of cadherins may be triggered by TGF-β1, which is produced in alloimmune response and tissue injury, and induces loss of epithelial characteristics such as cadherin and cytokeratin expression and acquisition of mesenchymal features (29). The loss of cadherin staining probably involves post-translational mechanisms, indicated by the modest decrease in mRNA levels compared to severe alterations in immunostaining. Enzymatic cleavage could play a role: E-cadherin is cleaved by MMP3 and MMP7 (30,31) or dysadherin (Fxyd5) (31), and is degraded during ischemic or toxic stress in kidney cells (32,33). By microarray analysis, MMP3, MMP7 and Fxyd5 mRNA as well as other MMPs, are increased in the rejecting kidneys in the present studies, supporting this hypothesis. Cadherins participate in adhesion (16) and signaling (34,35) and sustain the terminally differentiated tubular epithelial phenotype (19). Thus the loss of cadherins could alter epithelial function, induce transformation and mobilize epithelial cells to enter the interstitium, while permitting inflammatory cells to enter the epithelium.

Epithelial deterioration in kidney allograft rejection is associated with CD103+ T cells and perforin-granzyme positive T cells but tubulitis and shrinkage of tubules (decreased tubular surface area) and cadherins are not dependent on these mechanisms. While CD103+ T cells are present in rejecting kidneys and associated with tubulitis lesions, CD103 deficient hosts had a similar T-cell burden and were not protected against the loss of cadherins and epithelial mass, tubulitis, and other severe morphologic changes in the epithelium. These findings are consistent with previous observations that although CD103+ CD8+ CTL are present in rejecting human renal allografts (14,20,21), β7-integrins are not essential for entry of antigen-specific CD8+ lymphocytes into epithelium (13,36).

While indicating that CD103 is not essential for tubulitis, the present studies do not contradict previous studies showing roles for CD103 in allograft rejection. In mouse islet transplantation, CD103 deficiency impaired rejection (22), and adoptive transfer of CD103+ effector T-cells-elicited rejection. There are some limitations to all these models, leaving open the question of whether the effector mechanisms are model specific, differing between islet and solid organ transplantation. A recent study (23) investigating the role of CD103+CD8+ effector cells in rat kidney allografts, using a monoclonal antibody to CD103 to reduce accumulation of CD8+ cells in the graft, also concluded that CD103 was not essential for tubulitis or tubular injury, similar to the present studies. However, rats treated with cyclosporine showed attenuated tubular injury, indicating that effector systems redundant under conditions of full immune responsiveness may not be redundant when the immune response is impaired.

The present model probably reflects the mechanisms that operate in human T-cell-mediated rejection. The histology mimics the Banff lesions of kidney rejection, with progressive and devastating damage to the epithelium, simulating the lesions of clinical transplantation (4,8). The remaining host kidney prevents early death due to uremia, an essential condition for studying the evolution of histologic lesions. But the model must be interpreted with reservations, as with all animal models. The percentage of CD8+ T cells expressing CD103 is higher in human than in mouse kidney transplants, which has a relatively low number of CD3+ cells (25%) in the allograft expressing CD103, including both CD4+ and CD8+. Moreover, immunosuppression in humans may reduce redundancy by limiting effector generation and maximize the quantitative contribution of CD103 and cytotoxic molecules. Thus while CD103-epithelial interactions and Prf1-GzmA-GzmB mechanisms are not essential for tubulitis and epithelial deterioration, they may still contribute to injury when other effector mechanisms are limiting.

From these observations, we propose that epithelial deterioration, the central aspect of renal transplant rejection, is mediated by T-cell-macrophage effector mechanisms in the interstitium related to delayed type hypersensitivity, acting on the tubule epithelium without direct contact and without using CD103, Prf1 or GzmA or GzmB. Studies in progress in our laboratory are examining the role of other cytotoxic mechanisms such as Fas—Fas ligand. However, studies on Fas—Fas ligand in allograft rejection are ambiguous: overexpression of FasL leads to a heightened immune response, but Fas—Fas ligand interactions may have a beneficial role through apoptosis of infiltrating lymphoid cells (37). Contact independent mechanisms akin to delayed type hypersensitivity would explain independence of tubulitis of Prf1, GzmA, GzmB and CD103. We also have evidence that loss of epithelial transporters is demonstrable long before any invasion of the epithelium by lymphocytes (‘Epithelial deterioration in kidney transplant rejection: relationship to tubulitis and cytotoxic T-cell molecules’; manuscript in preparation). Tubulitis may thus be a relatively late change in the epithelium, reflecting loss of epithelial integrity that permits entry of lymphocytes. This would explain the lack of requirement for cytotoxic mechanisms and also why tubulitis frequently develops in atrophic tubules independent of rejection. Loss and redistribution of cadherins as a consequence of such tubule changes may contribute to the entry of inflammatory cells into the epithelium. Mechanisms directly altering the epithelium could include soluble effector T-cell or macrophage products (cytokines, Tnf superfamily members e.g. TNFSF, FasL, LTB acting on epithelial receptors, possibly in synergy, reactive oxygen species, nitric oxide, eicosanoids and enzymes). The epithelium may change in response to changes in the extracellular matrix (e.g. synthesis of hyaluronic acid) or the microcirculation. Nevertheless, even if tubulitis reflects compromised epithelial integrity, this does not reduce the diagnostic value of tubulitis as an indicator of this T-cell-mediated rejection.

Acknowledgments

Authors thank Kara Allanach, Jennifer Coppens and Joan Urmson for technical assistance, and Dr. Deborah James for valuable assistance with the manuscript. This research is supported by Genome Canada; Kidney Foundation of Canada; Roche Organ Transplant Research Foundation; Roche Canada; Astellas Canada; University of Alberta Hospital Foundation; Muttart Chair in Clinical Immunology. Dr. Halloran holds a Canada Research Chair in Transplant Immunology.

Ancillary