FOXP3 Expression in Human Kidney Transplant Biopsies Is Associated with Rejection and Time Post Transplant but Not with Favorable Outcomes

Authors


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

Abstract

Expression of the transcription factor forkhead box P3 (FOXP3) in transplant biopsies is of interest due to its role in a population of regulatory T cells. We analyzed FOXP3 mRNA expression using RT-PCR in 83 renal transplant biopsies for cause in relationship to histopathology, clinical findings and expression of pathogenesis-based transcript sets assessed by microarrays. FOXP3 mRNA was higher in rejection (T-cell and antibody-mediated) than nonrejection. Surprisingly, some native kidney controls also expressed FOXP3 mRNA. Immunostaining for FOXP3 was consistent with RT-PCR, showing interstitial FOXP3+ lymphocytes, even in some native kidney controls. FOXP3 expression correlated with interstitial inflammation, tubulitis, interstitial fibrosis, tubular atrophy, C4d positivity, longer time posttransplant, younger donors, class II panel reactive antibody >20% and transcript sets reflecting inflammation and injury, but unlike these features was time dependent. In multivariate analysis, higher FOXP3 mRNA was independently associated with rejection, T-cell-associated transcripts, younger donor age and longer time posttransplant. FOXP3 expression did not correlate with favorable graft outcomes, even when the analysis was restricted to biopsies with rejection. Thus FOXP3 mRNA expression is a time-dependent feature of inflammatory infiltrates in renal tissue. We hypothesize that time-dependent entry of FOXP3-positive cells represents a mechanism for stabilizing inflammatory sites.

Abbreviations: 
Foxp3

forkhead box P3 [Mus musculus]

FOXP3

forkhead box P3 [Homo sapiens]

PRA

panel reactive antibody

Treg

regulatory T cell

IPEX syndrome

immune dysregulation, polyendocrinopathy, and enteropathy X-linked syndrome

TCMR

T-cell-mediated rejection

ABMR

antibody-mediated rejection

CNIT

calcineurin inhibitor toxicity

GZMB

granzyme B

PBTs

pathogenesis-based transcript set

ATN

acute tubular necrosis

HPRT1

hypoxanthine phosphoribosyltransferase1

g

glomerulitis

cg

glomerulopathy

i

interstitial inflammation

ci

interstitial fibrosis

v

intimal arteritis

cv

fibrous intimal thickening

QCATs

quantitative cytotoxic T-cell-associated transcripts

GRITs

IFN-γ- dependent rejection-induced transcripts

IRITs

injury and repair induced transcripts

KT1s

renal transcripts

Introduction

When regulatory T cells (Tregs) emerged as a mechanism in control of autoimmunity (1–3), considerable interest focused on their role in organ transplantation and their potential for cell-based therapy (4,5). Such studies often incorporate forkhead box P3 (FOXP3), a forkhead-winged helix transcription factor important in the development and function of Tregs (6–8). Foxp3 knockout mice exhibit severe systemic autoimmune-like syndrome (9,10). Humans with mutations of FOXP3 manifest X-linked IPEX syndrome: immune dysregulation, polyendocrinopathy and enteropathy (11). Thus FOXP3 is important in cells that regulate self-tolerance. In humans, both CD4+ and CD8+ T cells can express FOXP3, although FOXP3 is preferentially expressed in CD4+ CD25+ cells (7,12). Retrovirus-mediated expression of FOXP3 in human does not consistently convert human CD4+ T cell into Tregs (13,14). Ex vivo activation of human CD4+ CD25− human T cells generates CD4+ CD25+ T cells expressing FOXP3 and suppressive function (7). However, FOXP3 is not an exclusive marker of regulatory capability: most human T cells transiently express FOXP3 during activation (15).

In human renal transplantation, the significance of FOXP3 expression in biopsies and body fluids remains unclear. In patients undergoing transplant biopsies for clinical indication, higher FOXP3 mRNA in urine cells was observed in patients with a diagnosis of acute rejection compared to nonrejection and was associated with better graft survival (16). Among the patients with acute rejection episodes, higher FOXP3 mRNA in urine was associated with lower serum creatinine and higher rates of reversal of rejection. The association of FOXP3 with rejection was confirmed by studies of biopsies for clinical indication. By immunostaining, kidneys with T-cell-mediated rejection (TCMR) had more FOXP3+ cells than kidneys with antibody-mediated rejection (ABMR) or calcineurin toxicity (CNIT). Similarly, kidney biopsies with TCMR have higher FOXP3 mRNA expression than control kidneys lacking rejection (17). However, density of FOXP3+ infiltrating cells was not correlated with favorable outcome (12).

In this study we explored the significance of FOXP3 mRNA by quantitative real time RT-PCR in a prospective study of unselected, consecutive kidney biopsies taken for clinical indications. We compared FOXP3 expression to diagnosis, histopathology, demographics and allograft function. In addition, we explored the relationships of FOXP3 expression with recently defined pathogenesis-based transcript sets (PBTs) reflecting major biologic events during allograft rejection (19–23).

Materials and Methods

Patient population and specimens

The study was approved by the institutional review board of the University of Alberta (issue 5299). Written informed consent was obtained from all study patients. We examined histologically normal renal cortical tissues from six native nephrectomies (performed for renal cell carcinoma), 12 implant biopsies and 83 biopsies for cause including: 28 TCMR; 14 ABMR (three acute and 11 chronic active ABMR); three mixed TCMR and ABMR; 14 borderline; three suspicious ABMR, 14 acute tubular necrosis (ATN); seven CNIT from 77 patients between January 2004 and March 2007. Ten patients had more than one biopsy for cause during this time. The biopsies were taken at different time points (often months apart) and represent independent clinical indications for the biopsy, histology scores and gene expression. The diagnosis can be entirely different in two biopsies from the same patient. Thus, biopsies from the same patient were regarded as independent events. Sixty-eight of the biopsies in this study were included in a previous publication from our group (19).

Most patients were maintained on calcineurin inhibitors (92%, mostly tacrolimus), MMF or azathioprine and prednisolone. Implant biopsies were obtained at 1 h after reperfusion during transplantation. With all biopsies, we obtained an extra biopsy core (18-gauge) for expression analysis.

Diagnostic classification

All biopsies were assessed using Banff criteria (24,25), stained for C4d by immunofluorescence, and evaluated by a pathologist (BS) blinded to the molecular results. Data concerning anti-HLA-antibodies in the recipient's serum at time of biopsy were available for all biopsies. All biopsies had adequate cortical tissue for analysis by Banff criteria with the exception of six biopsies (four biopsies had one artery, two biopsies had no arteries). We defined two levels of diagnosis: (1) the histopathologic diagnosis based on Banff criteria; and (2) the clinical rejection episode based on histopathologic diagnosis of borderline change or rejection and retrospective assessment of clinical course as previously described (19).

RNA extraction and processing for real-time RT-PCR and microarrays

RNA extraction and processing was previously described (19). Affymetrix microarray results were available on all biopsies included in this study. Because the microarray probe set for FOXP3 is nonfunctional on affymetrix microarrays, we used real-time RT-PCR to assess FOXP3 and for comparison GZMB expression in the biopsies. We normalized FOXP3 and GZMB mRNA expression against a housekeeping gene (hypoxanthine phosphoribosyltransferase1 (HPRT1) (HPRT1: 4326321E-0601007, FOXP3: HS00203958_m1, GZMB: Hs00188051_m1).

In previous studies we defined pathogenesis-based transcript sets (PBTs), which reflect major biological processes during renal allograft rejection: quantitative cytotoxic T-cell-associated transcripts (QCATs, n = 25) that robustly estimate the effector T-cell burden in the graft (20); IFN-γ−dependent rejection-induced transcripts (GRITs, n = 68) reflecting IFN-γ effects (21); injury and repair induced transcripts (IRITs, n = 1798) (22); renal transcripts (KT1s, n = 1481) reflecting loss of parenchymal integrity (23). Probe sets for each PBT are available at (http://transplants.med.ualberta.ca/).

FOXP3 immunostaining

The number of FOXP3+ lymphocytes was assessed by immunoperoxidase staining in 7 normal nephrectomies and 21 biopsies for cause (5/7 of normal nephrectomies and 17/21 biopsies for cause were included in mRNA study). Paraffin embedded tissue sections were incubated with mouse monoclonal anti-human FOXP3 primary antibody (Clone 22510, Abcam, Inc., Cambridge, MA). Analysis was done by counting total number of positive nuclei in leukocytic infiltrated areas of interstitium in 1 mm2.

Assessing renal graft outcomes

Graft outcomes were analyzed for the first biopsy for cause of each patient (n = 66). We divided the biopsies into three groups according to FOXP3 or FOXP3/GZMB expression: low (1st–33rd percentiles); intermediate (34th–66th percentiles) and high (67th–100th percentiles).

Renal function was defined by estimated GFR (Cockcroft-Gault): (1) the lowest GFR within 1 week prior to or after the time of biopsy (GFR at biopsy), (2) GFR at 6 months after biopsy (GFR at 6 months) and (3) change of GFR from biopsy to 6 months after (GFR at 6 months—GFR at biopsy). The event in survival analysis was defined as reaching the stage of persistently (>3 months) low GFR (<30 mL/min) or return to chronic dialysis.

Data analysis

For statistical analysis, expression of FOXP3 and GZMB mRNA were defined as the negative ΔCt values (ΔCt = Ct (FOXP3)—Ct (HPRT1)), which were normally distributed. Gene expression within each PBT was assessed by microarray and summarized as the PBT score: the geometric mean of fold changes across all probe sets comparing to normal nephrectomy controls (19).

We compared FOXP3 expression to diagnosis, histopathologic finding and demographics and analyzed FOXP3 expression and FOXP3/GZMB ratio in relationship to PBT scores and graft outcomes. The analysis was performed (a) across the entire data set and (b) in late biopsies only (≥6 months after transplant).

We used log transformation to correct data-skewness in those variables that were not normally distributed including time posttransplant, FOXP3/GZMB and PBT scores. Two independent samples t-test or analysis of variance (ANOVA) with Bonferroni corrections for multiple comparisons were used to compare the mean of FOXP3 expression between the groups of each variable. Kruskal-Wallis test was used to compare the median of GFR between the three groups of biopsies with different FOXP3 or FOXP3/GZMB expression. Correlation was assessed by Pearson's correlation coefficient. The influence of multiple variables on FOXP3 expression was analyzed by multivariate analysis. Univariate analysis of the association between FOXP3 or FOXP3/GZMB and graft outcome in survival analysis (time between biopsy and event or censored for the end of study (Nov 30, 2007), death with functioning graft, or lost to follow-up) was evaluated by Kaplan-Meier survival analysis with log-rank test. The influence of multiple variables on graft outcomes was analyzed by multivariate Cox regression model. A p-value < 0.05 was considered statistically significant.

Results

Demographics and clinical data are shown in Supplementary Table S1. The biopsies for cause were taken from 0.2 to 199.5 months posttransplant (median 12.4 months). The principal indication for biopsy was deteriorating function, and 55% of biopsies were diagnosed as rejection (TCMR, ABMR or mixed). In the first biopsies for cause of each patient, median (min, max) GFR was 45 (8, 97) mL/min at biopsy, 59 (9–97) mL/min at 6 months; change in GFR from biopsy to 6 months was 14 (−22, 74) mL/min. Early biopsies had lower GFR at biopsy and better recovery than late biopsies (Supplementary Table S2).

Relationship between FOXP3 mRNA expression and diagnoses (univariate analysis)

Kidneys with rejection (TCMR, ABMR and mixed) had higher FOXP3 expression than kidneys without rejection (ATN, CNIT and implant) (mean negative ΔCt of FOXP3 ± 1SD: −6.00 ± 1.81 vs. −7.85 ± 1.21, p < 0.001). Kidneys with a clinical rejection “episode” had higher FOXP3 expression than kidneys without a clinical episode (−5.94 ± 1.88 vs. −7.33 ± 1.45, p < 0.001). To define whether FOXP3 was differentially expressed in TCMR versus ABMR, we excluded three mixed cases. Due to small numbers, three biopsies suspicious for ABMR were excluded. There was no significant difference between ABMR and TCMR (Figure 1). When the analysis was restricted to late biopsies, the result was similar except biopsies with ATN were not different from rejecting kidneys, probably due to small number of two in the ATN group (data not shown).

Figure 1.

Relationship between FOXP3 mRNA expression and histopathologic diagnoses. Each panel represents the scatter-plots and the mean of log base 10 of FOXP3 expression defined by % HPRT of FOXP3 of the biopsies in each histopathologic diagnosis category: TCMR = T-cell-mediated rejection; ABMR = antibody-mediated rejection; BL = borderline rejection; ATN = acute tubular necrosis; CNIT = calcineurin inhibitor toxicity; Imp = Implant biopsies; Control = normal nephrectomies. Numbers of biopsies in each diagnosis were shown under the scatter-plots. P-values were obtained by one-way ANOVA with Bonferroni corrections. *p ≤ 0.05 =**p ≤ 0.01.

Biopsies with borderline rejection (borderline TCMR and suspicious ABMR) showed intermediate values between kidneys with rejection and no rejection, with no significant differences from both of them.

Interestingly, some native kidney controls (specimens from nephrectomies for kidney cancer ruled to be within normal limits by histopathology) had FOXP3 expression intermediate between kidneys with rejection and implant biopsies or nonrejecting kidneys (Figure 1).

Relationship between FOXP3 mRNA expression and Banff lesions (univariate analysis)

Higher FOXP3 expression was associated with more interstitial inflammation (i-score) and tubulitis (t-score) (Figure 2A, B). (FOXP3 was not different between t1 and t2 scores, in keeping with our previous report that t1 and t2 have similar molecular findings (19). FOXP3 was positively associated with C4d positivity, interstitial fibrosis (ci-score) and tubular atrophy (ct-score) (Figure 2C–E). There was no association of FOXP3 expression with glomerulitis (g-score), intimal arteritis (v-score), transplant glomerulopathy (cg-score) and fibrous intimal thickening (cv-score) (p > 0.05).

Figure 2.

Relationship between FOXP3 mRNA expression and Banff lesions. The relationship between log base 10 of FOXP3 expression defined by % HPRT of FOXP3 and the degree of interstitial inflammation (i-score) is shown in panel A, tubulitis (t-score) in panel B, C4d positivity in panel C, interstitial fibrosis (ci-score) in panel D and tubular atrophy (ct-score) in panel E. Due to small numbers, samples with the ci and ct-scores of 2 and 3 were merged. P-values were obtained by two independent samples t-test or one-way ANOVA with Bonferroni corrections. *p ≤ 0.05, **p ≤ 0.01. Note that we grouped some Banff lesion scores with less than 10 cases to make the comparisons more robust.

When the analysis was restricted to late biopsies, higher FOXP3 expression was associated with i-score, t-score and v-score (Supplementary Figure S1).

Relationship between FOXP3 mRNA expression and demographics (univariate analysis)

FOXP3 expression increased with time posttransplant. This correlation was observed when the analysis included all biopsies for cause (Figure 3A) or was restricted to biopsies with rejection (r = 0.34, p = 0.02), but was not seen in biopsies without rejection (r =−0.12, p = 0.60). Late biopsies had greater FOXP3 expression than early biopsies, whether the analysis was restricted to all biopsies for cause (−6.15 ± 1.80 vs. −7.58 ± 1.52, p < 0.001) or to rejecting kidneys (−5.60 ± 1.64 vs. −7.01 ± 1.86, p = 0.03). If the analysis was confined to late biopsies, there was no further effect of time. Thus the effect of time on FOXP3 expression was observed between biopsies before 6 months and after 6 months, but was not progressive in biopsies beyond 6 months. In contrast to FOXP3 mRNA, the principal PBTs associated with inflammation (QCATs or GRITs) did not correlate with time posttransplant (Figure 3B, C).

Figure 3.

Relationship between FOXP3 mRNA expression and demographics. A–C Each panel represents the scatter-plots between log base 10 of FOXP3 expression defined by % HPRT of FOXP3 or log base 10 of PBT scores on Y axis and time posttransplants in log scale on X axis. The straight line in each panel represents linear regression line. FOXP3 expression correlated with the time posttransplant in the analysis of biopsies for cause (n = 83) (panel A). In contrast to FOXP3 expression, cytotoxic T-cell-associated transcripts (QCATs) (panel B) and IFN-γ effect-associated transcripts (GRITs) (panel C) showed no correlation with the time posttransplant. FOXP3 expression correlated with younger donor age (n = 77) (panel D) and associated with panel reactive antibody (PRA) class II at the time of biopsy greater than 20% (panel E). Correlations and p-values were obtained by Pearson's correlation and other p-values were obtained by two independent samples t-test. *p ≤ 0.05, **p ≤ 0.01.

Among all biopsies for cause, higher FOXP3 expression correlated with younger donor age (Figure 3D) and class II panel reactive antibody (PRA) >20% at biopsy (Figure 3E). FOXP3 expression did not correlate with previous transplants, calcineurin inhibitor or sirolimus-based regimens, recipient gender, recipient age, donor type, donor gender, anti-rejection treatment before biopsy or recipient ethnicity (Caucasian vs. non-Caucasian).

Relationship of FOXP3 mRNA expression and PBTs (univariate analysis)

FOXP3 expression correlated with QCAT, GRIT, IRIT and KT1 scores (Figure 5A–D). When the analysis was restricted to late biopsies, the correlation between FOXP3 expression and PBTs was stronger: QCATs, r = 0.68, p < 0.001; GRITs, r = 0.62, p = < 0.001; IRITs, r = 0.43, p = 0.001; KT1s, r =−0.43, p = 0.001. Thus FOXP3 expression is associated with inflammation, injury and repair and loss of parenchymal integrity.

Figure 5.

Relationship between FOXP3 mRNA expression and PBTs. Each panel represents the scatter-plots between log base 10 of FOXP3 expression defined by % HPRT of FOXP3 on the Y axis and log base 10 of PBT scores on X axis. Cytotoxic T-cell-associated (QCATs) (panel A) and IFN-γ effect-associated (GRITs) (panel B), injury and repair (IRITs) (panel C) and renal parenchyma integrity transcripts (KT1s) (panel D). The straight line in each panel represents linear regression line. Correlations and p-values were obtained by Pearson's correlation. *p ≤ 0.05, **p ≤ 0.01.

The positive association of FOXP3 with measures of inflammation was lost when the FOXP3 value was divided by the GZMB mRNA value, probably because GZMB mRNA is a measure of infiltration. Thus FOXP3/GZMB ratio did not correlate with IRITs or KT1s (p > 0.05), and correlated inversely with QCATs (r =−0.51, p = <0.001) and GRITs (r =−0.58, p = <0.001).

Multivariate analysis of the association of FOXP3 mRNA expression with histopathology, PBT expression and demographics

Multivariate analysis of the relationship between FOXP3 expression and other molecular and clinical features was performed in two models: (1) the association of FOXP3 expression with Banff scores, PBT scores and demographics in a regression analysis; (2) the association of FOXP3 expression with histopathologic diagnosis, PBT scores and demographics in an ANOVA. Due to the fact that histopathologic diagnoses are based on histologic lesion scores and thus do not represent independent measurements, they could not be included in the same multivariate model. In the first model, higher FOXP3 expression independently correlated with higher QCAT-score, younger donor age and longer time posttransplant. In the second model, higher FOXP3 expression correlated with rejection and longer time posttransplant (Table 1). Thus in both multivariate models, FOXP3 correlated with time posttransplant and inflammation.

Table 1.  Multivariate analysis of the association of increased FOXP3 expression with histopathologic diagnosis, Banff score, PBT score and demographics
 p-ValueR2
Model 1. Association of increased FOXP3 with Banff score, PBT score and demographics
 Analysis in the whole data set 0.602
   QCAT score<0.001 
   Longer time posttransplant (≥6 months) 0.001 
   Younger donor age 0.031 
 Analysis in late biopsies (≥6 months posttransplant) 0.679
   QCAT score<0.001 
   Intimal arteritis (v) score<0.001 
Model 2. Association of increased FOXP3 with histopathologic diagnosis, PBT score and demographics
 Analysis in the whole data set 0.337
   Histopathologic diagnosis of rejection<0.001 
   Longer time posttransplant (≥6 months) 0.005 
 Analysis in late biopsies (≥6 months posttransplant) 0.271
   Histopathologic diagnosis of rejection 0.007 

When the analysis was restricted to late biopsies, the effect of time was weakened. In the first model, QCAT-score and v-score independently correlated with FOXP3. In the second model, histopathologic diagnosis was the only variable associated with FOXP3 (Table 1).

Univariate analysis of FOXP3 mRNA expression and graft outcomes

We observed no relationship between the level of FOXP3 or FOXP3/GZMB expression with future function (defined as change in GFR from the time of biopsy to 6 months post-biopsy), regardless of the presence or absence of rejection (Table 2). In early biopsies, the group with high FOXP3/GZMB had greater improvement in GFR after biopsy compared to the intermediate group. However, low FOXP3/GZMB group showed intermediate improvement in GFR between intermediate and high groups with no significant differences between them (Table 1). In late biopsies, no significant differences of change in GFR after biopsy between the three groups of FOXP3 or FOXP3/GZMB expression was found (Table 1).

Table 2.  Lack of association between FOXP3 expression or FOXP3/GZMB and renal function
 Change in GFR(mL/min) from biopsy to 6 months later [median (min, max)]
FOXP3FOXP3/GZMB
Low (1st–33rd)Intermediate (34th–66th)High (67th–100th)Low (1st–33rd)Intermediate (34th–66th)High (67th–100th)
  1. aDue to small numbers in early biopsies with rejection (n = 9) and with no rejection (n = 12), sub-analysis was not done.

  2. bDue to small numbers in late biopsies with no rejection (n = 7), sub-analysis was not done.

  3. All the comparisons between low, intermediate and high FOXP3 or FOXP3/GZMB show no significant differences except †: p = 0.03.

1st biopsies for cause
 All (N = 59)19 (−9, 74)13 (−22, 52)9 (−15, 43)12 (−3, 52)7 (−15, 43)18 (−22, 74)
 With no rejection (N = 17)23 (−4, 41)7 (−9, 74)37 (19, 52)25 (5, 52)11 (−9, 41)27 (−5, 74)
 With rejection (N = 31) 6 (−22, 35)12 (−12, 43)7 (−8, 32)4 (−3, 27)7 (−12, 43)10 (−22, 43)
1st early biopsies for causea
 All (N = 21)25 (−5, 74)20 (3, 52)8 (−2, 34)21 (−2, 52)3 (−5, 31)27 (17, 74)
1st late biopsies for causeb
 All (N = 38) 9 (−22, 41)4 (−12, 43)12 (−15, 43)10 (−3, 43)12 (−15, 32)9 (−22, 43)
 With rejection (N = 23)3 (−22, 9)13 (−12, 43)7 (−8, 32)4 (−3, 27)7 (−22, 43)9 (−8, 21)

In Kaplan-Meier analysis, biopsies with high FOXP3 expression had lower graft survival compared to biopsies with low or intermediate FOXP3 expression (Figure 6A). However, sub-analysis in biopsies with rejection (Figure 6B) or in late biopsy (Figure 6C) showed no association between FOXP3 expression and graft survival. FOXP3/GZMB expression was not associated with graft survival (p > 0.05) even when the analysis was restricted to biopsies with rejection or late biopsies. There was no event in the biopsies without rejection and only one event in early biopsies, so the survival analysis could not be done in these groups.

Figure 6.

Relationship between FOXP3 mRNA expression and graft survival Kaplan-Meier estimates of graft survival based on the level of FOXP3 mRNA expression at the time of biopsy: low (1st–33rd percentiles); intermediate (34th–66th percentiles); high (67th–100th percentiles) and time postbiopsy to time of reaching the stage of persistently low GFR < 30 mL/min or the end of study (death and lost to follow-up censored) in the first biopsies for cause (panel A), in the first biopsy for cause with rejection (panel B) and in the first biopsy for cause that were taken ≥6 months after transplant (panel C). Vertical dash line indicates time of the first event.

Multivariate analysis of FOXP3 mRNA expression and graft outcome

Univariate analysis of the association of clinical and molecular parameters (histopathology, PBT expression and demographics in all first biopsies for cause) with graft survival showed a significant relationship for g, cg and t-score, diffuse C4d positivity, PRA class II at the time of biopsy >20%, time posttransplantation and GRIT and IRIT scores with graft survival. In a multivariate analysis, only C4d positivity and GRIT score, but not FOXP3 mRNA expression, were independently associated with graft survival (Table 3).

Table 3.  Multivariate analysis (Cox regression model) of the association of FOXP3, histopathology, PBT expression and demographics with graft outcome (failure or persistently low GFR)
ModelsVariablesHR (95% CI)p-Value
  1. Note: The significant of diagnosis of rejection cannot be assessed in this model due to no events in the biopsies with no rejection.

Without FOXP3
 Diff C4d positivity7.10 (1.80–27.98)0.005
 GRIT score3.84 (1.22–12.12)0.022
With FOXP3
 FOXP3 expression 
 Low1.00 (Reference)  
 Intermediate0.69 (0.05–9.17) 0.781
 High1.27 (0.11–14.18)0.848
 Diff C4d positivity6.31 (1.28–31.08)0.024
 GRIT score3.96 (1.18–13.34)0.026

FOXP3 immunostaining

Immunostaining for FOXP3 confirmed the associations of FOXP3 mRNA with interstitial mononuclear cells. Biopsies with TCMR or ABMR had more interstitial FOXP3+ lymphocytes than nonrejecting biopsies (5.5 ± 8.4 vs. 1.8 ± 2.5 cells/mm2) or normal renal tissue (0.2 ± 0.4 cells/mm2) (after exclusion of one outlier) (Figure 4A). Although these differences were not significant in a nonparametric test (p > 0.05), likely due to the small number of cases with immunostaining and high variability in the biopsies with positive cell counts. Most FOXP3+ cells were interstitial around non-atrophic tubules and adventitia, with occasional FOXP3+ cells infiltrating tubular epithelium or within nodular aggregates (Figure 4C, D). Normal nephrectomy tissue showed minimal numbers of FOXP3+ cells (0–1 FOXP3+ cells /mm2) With the exception of one outlier, which had 18 interstitial FOXP3+ cells/mm2 (Figure 4B). This sample was within normal limits by histopathology but showed focal patchy inflammation in serial sections. The correlation between FOXP3+ cell count and FOXP3 mRNA expression across all biopsies with immunostaining was 0.30, p = 0.17. The lack of correlation is probably due to the patchiness of the infiltrate and the limited number of samples with suitable material for immunostaining.

Figure 4.

FOXP3 immunostaining (A) The distribution of interstitial FOXP3+ cell numbers/mm2 according to histopathology diagnoses: TCMR = T-cell-mediated rejection (n = 9); ABMR = antibody-mediated rejection (n = 6); NR = nonrejecting biopsies (n = 5); Control, normal nephrectomies (n = 7). (B) Immunoperoxidase staining for FOXP3 in a normal nephrectomy shows an aggregation of lymphocytes in the cortex with FOXP3+ cells (arrows) (400×, original magnification, immunoperoxidase). C and D Immunoperoxidase staining for FOXP3 in a renal biopsy with ABMR. (C) The nuclear staining for FOXP3 (arrows) in interstitial infiltration. (D) Another area showing occasional intratubular FOXP3+ lymphocytes (arrow) (400×, original magnification, immunoperoxidase).

Discussion

We assessed the significance of FOXP3 expression in renal transplant biopsies for cause, focussing on its association with histologic lesions, transcriptome changes and time posttransplant. In kidney transplant biopsies for cause, FOXP3 mRNA expression was associated with rejection (both ABMR and TCMR), younger donor age, longer time posttransplant, and PBTs reflecting inflammation and injury. Thus FOXP3 is a feature of renal inflammation, but differs from many other features on inflammation in being time dependent. FOXP3 mRNA was not associated with renal function within 6 months after biopsy, although there was a trend toward more impaired renal function with longer follow-up, as expected given the association of FOXP3 with inflammation. These data suggest that the occurrence of FOXP3 mRNA reflects time dependent entry of regulatory T cells into sites of chronic inflammation.

FOXP3 is associated with inflammation, which is associated with poorer outcomes, but FOXP3 does not independently predict outcome. Our study confirms previous histopathologic studies that found that FOXP3 was associated with inflammation, including studies of nonhuman primate kidneys (26), cardiac allograft biopsies (27), human kidney transplant urine (16) and human kidney allograft biopsies (12,17). Some renal tissue from nephrectomy specimens had FOXP3 mRNA, probably attributable to foci of chronic inflammation, suggesting that FOXP3 is not a reflection of alloimmunity but of chronic inflammation per se.

These data present a novel view of FOXP3 as a time-dependent feature of injured and inflamed compartments. Time emerged as a significant independent factor in FOXP3 expression in univariate and multivariate analyses, distinguishing FOXP3 from other inflammation-associated factors such as the ‘i’ score, the QCAT and GRIT burden, which were not time dependent. We were able to detect the time dependency because our inclusion of unselected biopsies for cause, even many years after transplantation. We cannot exclude that the increased FOXP3 expression in late allografts is related to a decrease in immunosuppression with time posttransplant; given the multiple combinations of immunosuppressives in our patient cohort and the fact that all patients received less immunosuppression over time, this issue cannot be solved in this study.

FOXP3 joins some other known time-dependent features of renal biopsies, including interstitial fibrosis and tubular atrophy with persistent inflammation. Protocol biopsies have confirmed that early renal transplants display more inflammation, whereas biopsies after several months reveal more atrophy and fibrosis (28,29), associated with persistent inflammation. These changes are time dependent but not relentlessly progressive: for example, atrophy and fibrosis emerges after the first few months but shows only slow progression after the first year. Other time-dependent variables include B cells, plasma cells and their related transcripts (manuscript in press). Thus we believe that time is better modeled as a dichotomous variable than a linear relationship, that is before 6 months and after 6 months, but determining and validating the optimal way of correcting for time will require larger data sets.

The fact that ABMR cases have FOXP3 expression similar to TCMR cases is compatible with our previous findings that TCMR and ABMR share many molecular disturbances (19) related to inflammation. ABMR and TCMR share expression of cytotoxic T-cell transcripts (granzyme B, perforin) and transcripts associated with IFN-γ effects. Presumably ABMR or TCMR creates an inflammatory compartment in the graft, which then recruits FOXP3+ T cells (either Tregs or effector T cells transiently expressing FOXP3) (15). Veronese and colleagues found that the number of FOXP3+ cells is lower in ABMR than in TCMR (12), but their ABMR cases were earlier than ours. Most ABMR in our study were chronic active ABMR based on current Banff criteria (25). All cases with ABMR occurred 14 months or more after transplants, compared to a median of 1 month (0.3–17 months) in the Veronese study.

We suggest that FOXP3 positive cells or their precursors accumulate in inflamed sites as a strategy to control the potential for autoimmunity in such sites. This is an extension of the theory of Stockinger et al., that regulatory cells control ‘spaces’ in the immune-inflammatory response (30), and accounts for the major features of FOXP3 expression: time dependency, association with inflammation and expression in some control kidneys. FOXP3+ T cells could either home to inflamed sites from blood or develop from precursors in such sites, possibly under the influence of TGF-β1(31), which drives CD4+ naïve T cell to express FOXP3. The importance of time and location of CD25+CD4+ T cells as mechanisms to prevent rejection was demonstrated in a mice skin grafts model (32).

Kidney inflammation is associated with adverse transplant outcomes, even if FOXP3 positive cells in that site limit the risks. In a protocol biopsy study, donor-specific hyporesponsiveness of the patient was associated with higher numbers of FOXP3+CD4+CD25+ and less chronic Banff lesions in the biopsy and better renal function compared to patients with nonhyporesponsiveness (18). Thus the benefit of FOXP3 positive cell may be masked in biopsies for cause by the effect of inflammation. We propose that the FOXP3 positive T cells in renal allografts do not reflect a cognate immune response, but are part of the naturally occurring Treg population whose purpose is to stabilize inflamed sites that cannot be restored to normal. Thus TCMR, ABMR and other types of renal inflammation induce differentiation of FOXP3+ T cells from precursors in situ in a time-dependent manner, or develop the ability to recruit such cells over time. This is likely to be as a general feature of injured and inflamed sites, rather than a feature of allorecognition per se. The best transplants would be those in which there is no injury or inflammation and thus little or no FOXP3, but chronically inflamed tissue with FOXP3 may be more stable than such tissue with no FOXP3. No independent effect of FOXP3 emerged in our multivariate analysis, but the sample size is not sufficient to exclude such an effect.

Acknowledgments

Special thanks and appreciation to Dr. Zija Jacaj and Debra Lieberman for help with collection of the clinical data; and to Vido Ramassar, Anna Hutton, Stacey Lacoste and Sujatha Guttikonda for technical support. This research has been supported by funding and/or resources from Genome Canada, Genome Alberta, the University of Alberta, the University of Alberta Hospital Foundation, Roche Molecular Systems, Hoffmann-La Roche Canada Ltd., Alberta Advanced Education and Technology, the Roche Organ Transplant Research Foundation, the Kidney Foundation of Canada, and Astellas Canada. Dr. Halloran also holds a Canada Research Chair in Transplant Immunology and the Muttart Chair in Clinical Immunology. Michael Mengel received a research stipend from the Dr. Werner Jackstädt Kidney Foundation.

Ancillary