Pathological lesions in rejecting kidneys
As previously described (5,6), isografts at days 5 (Figure 1, panel A), 7 and 21 post-transplant appeared normal with little inflammation or tubular necrosis. Allografts showed focal periarterial mononuclear infiltrates at days 3 and 4 (Panel B) and interstitial mononuclear infiltration by day 5, which increased at day 7 and persisted through day 42 (panels C, D and E respectively). Tubulitis was absent at days 3, 4 and 5, mild at day 7, and severe at days 14, 21 and 42. The late grafts at days 14, 21 and 42 showed severe tubular damage with patchy cortical necrosis of 30% of the cortex by day 42 (Panel F). Allografts from days 14 through 42 showed severe loss of tubule diameter, and of E-cadherin and Ksp-cadherin, reflecting parenchymal injury (Fairhead T, Einecke G, Turner P, Zhu L-F, Hadley GA, Famulski KS, Halloran PF: Tubu lesions of kidney transplant rejection are associated with loss of cadherins and do not require CD103. Manuscript in preparation.). The day 42 kidneys resembled some end-stage rejecting human kidneys, which may remain largely viable despite severe tubulitis, arteritis and patchy necrosis. By immunostaining, the infiltrate in kidney allografts at days 5, 7 and 21 contained 40–60% CD3+ T cells. At day 21, T cells were present in the interstitium and tubules, with CD3/CD8+ cells exceeding CD3/CD4+ cells by 8 to 1 (34 ± 4 vs. 4 ± 2 cells per 10 high power field, n = 9). The infiltrate was 35–50% CD68+ (macrophages), with late appearance of 5% CD19+ B cells at day 21. Hosts deficient in mature B cells (Igh6KO or IghJKO) showed similar infiltrate and tubulitis but less necrosis and hemorrhage at day 21 (5), and 19% lower kidney weight (260 ± 58 mg, n = 8 vs. 319 ± 70 mg, n = 6 in wild-type hosts). Detailed histology is shown in supplementary table S1 at http://transplants.med.ualberta.ca/.
Figure 1. Histopathology of rejecting mouse allografts. All panels show PAS staining (magnification 40×). Panel A: isograft (CBA into CBA) at day 5 with normal histology. Panel B: rejecting kidney allograft (CBA into B6) at day 3 showing focal periarterial mononuclear infiltrate. Panel C: rejecting kidney allograft (CBA into B6) at day 5 showing periarterial and beginning interstitial mononuclear infiltration. Panel D: rejecting kidney allograft at D7 (CBA into B6) with mononuclear interstitial infiltration and mild tubulitis. Panel E: kidney transplant (CBA into B6) at day 21 showing heavy tubulitis. Panel F: Allografts show severe tubular damage, with patchy cortical necrosis at day 42.
Download figure to PowerPoint
These kidneys manifested functional impairment. Removing the host kidney in selected mice resulted in elevated serum creatinine at day 7 (data not shown). Moreover, tubulitis per se correlates with loss of function (26), and severe deterioration of the renal epithelium in the tubulitis lesions at days 14, 21 and 42 in our mouse model indicates severe loss of function.
Hierarchical clustering of the global transcript expression in rejecting kidneys, isografts, CTL and d4 MLR
We used unsupervised hierarchical cluster analysis to compare transcript expression between control kidneys, isografts, allografts rejecting in WT and IghKO hosts, d4MLR and CTL. The transcriptomes clustered into three groups (Figure 4): normal kidneys and isografts, with Iso D21 being more similar to NCBA than Iso D5 or Iso D7; allografts, with WT D5, WT D7, IghKO D7 and IghKO D21 in one subcluster and WT D14, WT D21 and WT D42 in a second sub-cluster; and d4MLR and CTL formed a third cluster.
Figure 4. Unsupervised hierarchical clustering of experimental groups. Unsupervised clustering of all transcripts, based on distance, was performed on isografts (ISO), allografts rejecting in wild type hosts (WT) and B-cell deficient hosts (IghKO) and lymphocyte cultures (MLR = mixed lymphocyte culture; CTL = purified CTL from a long-term culture).
Download figure to PowerPoint
CD antigen transcript expression
We analyzed expression of CD transcripts as a reflection of cellular infiltration (Table 1). The 18 CD transcripts in normal kidney may reflect immature dendritic cells in the interstitium (27). T-cell-specific transcripts were absent, making it unlikely that the transcripts detected reflect peripheral blood. Expression of CD transcripts was similar between CTL and d4MLR, with both reflecting effector CD8 cells, with high expression of CD44 (43 and 25-fold vs. NCBA in CTL and MLR, respectively) and no L-selectin. D4MLR contained B-cell-specific transcripts CD79a and CD79b.
Table 1. CD antigen transcripts in isografts and in allografts transplanted into wildtype hosts and B cell deficient hosts. Transcripts had to be at least two fold increased compared to NCBA and flagged as present in at least one of the samples to be considered. The table shows the signal strength for controls and fold changes for the transplants. (−) indicates that a given transcript was not increased; bolded signal values indicate that a transcript was classified as present. In case of multiple probe sets querying the same gene, data obtained from probe sets with suffixes _s_at and _x_at were not considered, and a probe set displaying the most robust signal was selected.
|Symbol||NCBA Signal NCBA||Isografts Fold Change||WT Allografts Fold Change||IghKO Allografts Fold Change||Lymphocytes Fold Change|
Thirty-three CD transcripts were increased at least 2-fold in allografts compared to NCBA kidney. Twenty-one had high expression in d4MLR and CTL (> 5-fold compared to NCBA). CD2f10, CD14 and CD68 macrophage-associated transcripts (28) were increased in rejecting allografts but absent or low in d4MLR or CTL, suggesting that they represent infiltrating activated macrophages, which are not represented in d4MLR or CTL. The B-cell-specific transcripts CD79a and CD79b appeared at days 14, 21, and 42 in WT but not in IghKO hosts, consistent with antibody-producing cells. The analysis of CD transcripts is consistent with an early and sustained CTL/macrophage infiltrate in WT and IghKO hosts, and with late B-cell infiltration in WT hosts. CD4 transcripts were flagged absent in the allografts, indicating a minor presence of CD4 cells in rejecting kidneys, as opposed to d4MLR, which contains both naïve and effector CD4 cells (60-fold increase vs. NCBA).
CATs in rejecting mouse kidney allografts
CATs (287 transcripts) were defined by high expression in both CTL and in d4MLR but absent in normal kidney. CAT expression was lower in d4MLR than in CTL (median 87%, expressed as a percentage of CTL signal). In NCBA, expression of CATs was called ‘absent’ by Affymetrix definition. CAT signal was unchanged in isografts (Iso D5, 1.0; Iso D7, 1.1; Iso D21, 1.0-fold vs. NCBA). In host kidneys of transplanted animals at day 5, CAT signals were not detectable (median 1.1 fold vs. NCBA), showing that circulating blood does not contribute to CAT expression in rejecting transplants.
In contrast to NCBA and isografts, CATs were strongly expressed in allografts. CATs were detectable by days 3 and 4, with a median 2.2-fold change in WT D3 and 3.3-fold change in WT D4 compared to NCBA. Expression of CATs was well established by day 5 with a median 3.8-fold increase compared to NCBA (Figure 5a). After day 5, expression of CATs was stable and persisted throughout the rejection process (D7, 4.3-fold; D14, 4.8-fold; D21, 4.6-fold; D42, 3.3-fold compared to the signal in NCBA) (Figure 5b).
Figure 5. Expression of CTL associated transcripts (CATs) in isografts and WT allografts. CATs were defined as absent in normal kidney, but highly expressed in lymphocyte cultures. (A) CATs were readily detected by day 3 although at a lower level than at day 5.
Download figure to PowerPoint
Figure 5. (B) Expression of CATs was low in isografts but high in rejecting kidneys at day 5 and persisted throughout the rejection process.
Download figure to PowerPoint
The level of CAT expression gives us an indication of the T-cell burden in the graft. However, the quantification of T-cell infiltration based on fold changes over a low background signal of CATs in NCBA may not be accurate. We therefore took a different approach: as CAT expression by definition was very high in CTL, the signal in this sample is much more reliable than the low background in NCBA. Given that the RNA from purified CTL is not diluted by RNA from other cell populations, the signal in this sample should exclusively represent transcript expression in CTL. Therefore, we chose to set the signal in CTL as 100% and quantified CAT expression in the rejecting transplants relative to the expression in CTL as an estimate for the relative amount of lymphocyte RNA in rejecting transplants.
The background signal of CATs in NCBA was median 4% of that in CTL and unchanged in isografts (D5, 4%, D7, 5%, D21, 4% of expression in CTL). At day 5 post-transplant, the signal for CATs was median 14% of that in CTL. After day 5, CAT expression remained very stable (D7, 16%; D14, 16%; D21, 16%; D42, 12% expressed as median% of expression in CTL). When we diluted RNA from d4MLR with kidney RNA in a ratio 1:4, the resulting signal was similar to that in all rejecting kidneys (median 15% of expression in CTL) and adequately reflected the dilution (median 20% of the d4MLR). This suggests that the RNA from lymphocytes infiltrating the kidney is diluted about 7-fold compared to CTL.
We analyzed the consistency of expression of individual CATs in all conditions by a nonparametric regression test. The CAT signal correlated well in all transplants, MLR and CTL, indicating robust maintenance of CAT expression in vivo and in vitro. The d4MLR correlated with diluted MLR (r= 0.91), despite 80% decrease in signal, and with the CTL (r= 0.81; p < 0.001). In rejecting transplants in WT hosts, the CAT signals correlated among all days (day 5 throughout day 42; r= 0.90–0.96; p < 0.001), indicating consistency of CAT expression over many days in all allografts. The CATs in the d4MLR correlated with the transplants at all days (r= 0.70–0.78; p < 0.001), as did the expression in CTL (r= 0.66–0.74).
K-means cluster analysis of CATs, based on their expression level in WT allografts relative to purified CTL, resulted in five clusters (Figure 6). Expression values for each cluster are given in supplementary tables S2–6. Cluster 1 included 140 transcripts (e.g. CD2, CD3g, GzmB, Tcrb, mes), and displayed low expression in allografts relative to CTL but was stable throughout the time course, with median 3.6-fold increase versus NCBA at day 5. Cluster 2 included 23 transcripts, including Ccl3, CD44, Il16, Il10ra, Il21r and Runx3. Expression of cluster 2 CATs was higher in d4MLR and in rejecting allografts relative to CTL than in cluster 1, with median 5-fold increase in rejecting kidneys at day 5, a further 2-fold increase from day 5 to day 14, and stable thereafter. Cluster 3 (n = 74) CAT expression was relatively high in d4MLR versus CTL but low in rejecting kidney. Cluster 4 (n = 46) CATs were less expressed in d4MLR than CTL, increased 1.9 fold between day 5 and day 14 and decreased thereafter by 1.3-fold. Cluster 5 consisted of Pdcd1, Socs1 and Stat1, which were as highly expressed in rejecting grafts as in CTL or d4MLR.
Figure 6. K-means clustering of CATs in WT allografts. Expression in WT allografts is shown as % of expression in CTL. CATs (n = 287) cluster in five groups according to their relative expression in the allografts versus CTL. The boxplots represent the median and quartiles of expression of CATs for each timepoint.
Download figure to PowerPoint