Pathological and Clinical Correlates of FOXP3+ Cells in Renal Allografts during Acute Rejection


* Corresponding author: F. Veronese,


The localization and significance of regulatory T cells (Treg) in allograft rejection is of considerable clinical and immunological interest. We analyzed 80 human renal transplant biopsies (including seven donor biopsies) with a double immunohistochemical marker for the Treg transcription factor FOXP3, combined with a second marker for CD4 or CD8. Quantitative FOXP3 cell counts were performed and analyzed for clinical and pathologic correlates. FOXP3+ cells were present in the interstitium in acute cellular rejection (ACR) type I and II, at a greater density than in acute humoral rejection or CNI toxicity (p < 0.01). Most FOXP3+ cells were CD4+ (96%); a minority expressed CD8. FOXP3+CD4+ cells were concentrated in the tubules (p < 0.001), suggesting a selective attraction or generation at that site. Considering only patients with ACR, a higher density of FOXP3+ correlated with HLA class II match (p = 0.03), but paradoxically with worse graft survival. We conclude that infiltration of FOXP3+ cells occurs in ACR to a greater degree than in humoral rejection, however, within the ACR group, no beneficial effect on outcome was evident. Tregs concentrate in tubules, probably contributing to FOXP3 mRNA in urine; the significance and pathogenesis of ‘Treg tubulitis’ remains to be determined.


In recent years intensive research has been devoted to the control of allograft rejection and the acquisition of long-term allograft tolerance. Regulatory T cells, which promote a state of peripheral antigen-specific self-tolerance by suppressing the activation and expansion of reactive effector cells, have been implicated in these processes (1–4).

Naturally occurring CD4+CD25+ T cells (Tregs) represent approximately 10% of CD4+ T cells in the peripheral blood of normal rodents and humans (5–7). First described by Sakaguchi et al. (8), Tregs are hyporesponsive to T-cell receptor stimulation and suppress the proliferation and activation of both CD4+ and CD8+ T cells (9,10). The suppressive mechanisms of Tregs are not yet fully understood but include engagement of the Treg TCR by antigen, direct cell contact, local secretion and cytokine signaling, such as TGF-β, IL-2 and IL-10, and inhibition of transcription of genes central to effector functions (11–13). Suppression has both an antigen-specific component (TCR) for the Treg and a nonspecific component for the cell suppressed, via cytokines. Studies in mice have shown that Tregs can prevent or ameliorate allograft rejection (14) and certain autoimmune diseases (15–17). As Tregs actively participate in prevention of organ transplant rejection and induction of transplantation tolerance, they also have an increasing potential for immunotherapy in these settings (17–21).

Detection of putative Tregs in tissues by immunohistochemical techniques has been facilitated with the discovery of the transcription factor FOXP3, a forkhead-winged helix transcription factor gene that is expressed in CD4+CD25+ T suppressor cells (22,23). The specificity of FOXP3 expression (or the murine form Foxp3) for Treg function has been shown in mice and humans, although FOXP3 can be expressed in some activated human CD4 or CD8 T-cell clones that do not express CD25 constitutively (23). In human peripheral blood, 95.7% of CD4+CD25high T cells expressed FOXP3, whereas no staining was detected in the CD25 population or in resting CD8+ T cells by flow cytometry (24). Ex vivo retroviral gene transfer of FOXP3 converted peripheral CD25CD45ROCD4+ naïve T cells into a regulatory T-cell phenotype, with an impaired proliferation and production of cytokines such as IL-2 and IL-10 upon T-cell receptor stimulation. These acquired peripheral Tregs suppressed proliferation of other T cells in a cell–cell contact-dependent manner (22,25,26).

Tregs producing FOXP3 infiltrated murine cardiac allografts in recipients treated with CD154 monoclonal antibody plus donor-specific transfusion (27) and functional Tregs were detected in murine skin allografts by their ability to passively transfer tolerance to naive recipients (28). No definitive pathologic studies have been done on FOXP3 cell infiltrates in human grafts. Higher levels of FOXP3 mRNA was reported in the urine of graft recipients with acute rejection, which predicted reversal of rejection and graft survival (29).

The purpose of this study was to determine the site and phenotype of FOXP3 cells in human renal allografts, and the clinical correlates of their presence.

Materials and Methods

Selection of cases

The study group consists of 73 renal transplant biopsies done at the Massachusetts General Hospital (MGH), including 70 biopsies taken for graft dysfunction from 1997 to 2004 and three biopsies from prior years. These cases were selected only for the diagnosis of acute cellular rejection (ACR) type I (n = 35) or type II (n = 10), acute humoral rejection (AHR, n = 14) or calcineurin inhibitor toxicity (CNI, n = 14) and the availability of sufficient paraffin-embedded tissue. For those cases with more than one biopsy, only the first biopsy was used. Seven intraoperative donor biopsies were used as controls. Median of time post transplant to biopsy in months was 1.0 (0.3–17.0). This study was approved by the Institutional Review Board, number 2006-P-000704/1, MGH.

Clinical data were collected from the MGH renal transplant and pathology databases (Table 1). Immunosuppression usually consisted of CNI, mycophenolate mofetil and prednisone. ACR was generally treated with glucocorticoids, muromonab-CD3 (OKT3) or thymoglobulin, AHR plasmapheresis and intravenous immunoglobulin were added. Graft function was determined by serum creatinine measurements at the time of biopsy, at 2 and 12 months, and at last follow-up. Graft failure was defined as return to dialysis.

Table 1.  Demographic and clinical data of transplant patients at the time of biopsy
  1. 1Mean ± SD.

  2. 2DD = deceased donor; LUR = living unrelated donor; LRD = living related donor.

  3. 3HLA mismatches in loci A, B: 0–2:3–4.

  4. 4HLA mismatches in loci DR: 0–1:2.

  5. 5Median and interquartile range.

Case no.73
Age (year)42±131
Male sex (%)58
Caucasian (%)92
Type of donor: DD:LUR:LRD2 (%)52:11:37
HLA class I MM3 (%)58:42
HLA class II MM4 (%)61:39
Immunosuppression regimen (%)
 Calcineurin inhibitor93
Time to biopsy (months)1.0(0.3–17.0)5
Time to last follow-up (months)34 (14–60)
Graft loss (%)
 CNI toxicity50


Immunohistochemistry was performed on formalin-fixed, paraffin-embedded tissue with a novel double marker technique optimized for the simultaneous identification of FOXP3 and a surface differentiation molecule (either CD4 or CD8). Sections were baked for 30 min in an oven, deparaffinized in xylene, rehydrated in absolute and 95% ethanol and incubated for 5 min in 3% H2O2 in methanol to block endogenous peroxidase. Antigen retrieval was done with Borg Decloaker™(Biocare Medical, Walnut Creek, CA) pH 9.5 in a pressure cooker and blocking by normal goat serum 1:50 and avidin D 1:10 anti-CD4 (clone BC/1F6, IgG1) or anti-CD8 (clone BC/1A5) monoclonal antibodies (Biocare Medical) diluted at 1:25 in Van Gogh yellow diluent (Biocare Medical), were incubated overnight at 4°C. After biotin blockade, slides were incubated with Universal Link biotinylated goat purified anti-mouse IgG (Biocare) for 20 min, followed by Streptavidin-horseradish peroxidase (Biocare) for 20 min, and developed with 3,3'-diaminobenzidine. All steps included washing with TBS/Tween 20. Slides were then incubated in a sequence for blocking steps with TBS/BSA 1% for 20 min, 5% skim milk in TBS for 20 min and normal goat serum 1:50 and avidin D 1:10 for 20 min. A rabbit polyclonal anti-FOXP3 antibody (ab10563, Abcam, Cambridge, MA) diluted at 1:800 in Van Gogh yellow was incubated overnight at 4°C. After biotin blockade, a biotinylated goat anti-rabbit IgG secondary antibody was used for 35 min followed by avidin-biotinylated-alkaline phosphatase complex (ABC-AP, Vector Laboratories, Burlingame, CA, USA) for 60 min. All steps included washing with TBS/Tween 20 and with TBS only before developing FOXP3 staining. Tissue sections were developed with Vector Blue™(Vector) alkaline phosphatase substrate and mounted in Faramount (DAKO, Carpinteria, CA).

Quantitation of infiltrating cells

Sections stained for FOXP3 and either CD4 or CD8 were scored in coded slides by one observer (RBC). The number of positive cells for one or both markers in the cortex was recorded, as well as cells located at perivascular areas and at corticomedullary junction aggregates, generating the scores for analysis (see below). The cells in the interstitium were counted separately from those in the tubules. The area of the cortex was measured with a loupe (included glomeruli and vessels) and the data were expressed as the number of cells/mm2 for each of the markers. For wedge biopsies, five 20× fields or 10 40× fields were scored and cells/mm2 calculated using the area per field. In 12 instances, a satisfactory double stain could not be achieved for one of the markers (9 CD4 and 3 CD8), due to high background, low intensity or insufficient remaining tissue. The reproducibility from slide to slide from the same case was assessed (see Results). Interstitial and intratubular cells were scored as CD4+FOXP3, CD4+FOXP3+ or CD4FOXP3+ and CD8+FOXP3, CD8+FOXP3+ or CD8FOXP3+, respectively.

Statistical analysis

Data were expressed as mean ± SD or median and interquartile ranges. For continuous variables, statistical significance was assessed by Student's t-test or, if the variable deviated from a normal distribution, as is the case of FOXP3 expression, by nonparametric tests (Mann–Whitney, Wilcoxon or Kruskall Wallis). Binary variables were analyzed by Fisher exact test. Spearman coefficient was used to correlate FOXP3 counts per mm2 in slides double stained for CD4/FOXP3 and CD8/FOXP3.

FOXP3 was correlated with the functional graft outcome at 2 and 12 months and at last follow-up. When duplicate stains were available for FOXP3, the mean FOXP3/mm2 was used. Graft outcome was evaluated by level of serum creatinine at last follow-up or graft failure, defined as return to dialysis. The association between FOXP3 expression and graft survival (time between transplant and allograft failure or censor, including death with functioning kidney in the latter) was evaluated by the Kaplan–Meier survival analysis, and differences in survival were measured by the log-rank test.

The influence of several variables on FOXP3 expression was analyzed by a multiple linear regression model: age, gender, HLA class II mismatch, immunosuppressive regimen, type of acute rejection and time posttransplant. FOXP3 measurements were log transformed prior to multivariate analysis to reduce skewness. A p value of less than 0.05 was considered significant.


Demographics and clinical data for the 73 patients with indication biopsies are shown in Table 1. Ninety-three percent of the patients were maintained on a calcineurin inhibitor-based regimen. Biopsies were done from 3–5338 days posttransplant (median 45 days). Serum creatinine at biopsy was 3.5 ± 1.9 mg/dl, and at 2 and 12 months were 1.7 ± 0.8 and 1.6 ± 0.6 mg/dl, respectively. Time from biopsy to last follow-up was 34 months (median), when average creatinine was 1.8 ± 0.8 mg/dl. Graft loss was 11%, 71% and 50% in patients with ACR, AHR and CNI toxicity, respectively, over the time of follow-up.

FOXP3+ cells in ACR. Antibody to FOXP3 stained the nuclei of scattered mononuclear cells in the interstitial infiltrate and tubules in all grafts with ACR (Figures 1 and 2). FOXP3+ cells were often prominent in aggregates around vessels (Figure 1A) and at the corticomedullary junction. FOXP3+ cells were very infrequent in glomeruli and no FOXP3+ cells were found in the occasional endarteritis lesions identified. FOXP3+ cells averaged 10.2 ± 11.2 cells/mm2, with 90% of the FOXP3+ cells in the interstitium. The vast majority of the FOXP3+ cells expressed CD4 (96 ± 8%) (Figures 1B and 2A). In contrast, in FOXP3/CD8 double stains, few FOXP3+ cells were found that expressed CD8 (mean of 6 ± 2%, p < 0.001 vs. CD4) (Figure 2B). We also sought evidence for CD4/CD8 double negative FOXP3+ cells by a triple stain (CD4, CD8, FOXP3) in a limited survey: only extremely rare FOXP3+ cells could be identified that lacked both CD4 and CD8.

Figure 1.

Figure 1.

(A) Renal biopsy with ACR: double immunohistochemical stain with antibody to FOXP3 (blue) and anti-CD4 (brown) as described in the Materials and Methods. An aggregate of lymphocytes in the cortex is shown with numerous double positive cells (CD4+FOXP3+) among the CD4+FOXP3 cells (100 × original magnification). (B) Higher magnification of the same area reveals the nuclear staining for FOXP3 and surface staining for CD4 (brown) in several cells (e.g., cells with arrows). All of the FOXP3 cells in this field are CD4 positive (400 × original magnification).

Figure 1.

Figure 1.

(A) Renal biopsy with ACR: double immunohistochemical stain with antibody to FOXP3 (blue) and anti-CD4 (brown) as described in the Materials and Methods. An aggregate of lymphocytes in the cortex is shown with numerous double positive cells (CD4+FOXP3+) among the CD4+FOXP3 cells (100 × original magnification). (B) Higher magnification of the same area reveals the nuclear staining for FOXP3 and surface staining for CD4 (brown) in several cells (e.g., cells with arrows). All of the FOXP3 cells in this field are CD4 positive (400 × original magnification).

Figure 2.

Figure 2.

(A) In a biopsy with ACR, an intratubular CD4+ cell expresses FOXP3 (arrow). Double stain for FOXP3 (blue) and CD4 (brown) (400× original magnification). Insert shows another intratubular CD4+FOXP3+ cell (600× original magnification). (B) ACR biopsy stained for FOXP3 (blue) and CD8 (brown). Most intratubular FOXP3 are CD8 negative (arrowheads). One cell is CD8+FOXP3+ (arrow; 400× original magnification).

Figure 2.

Figure 2.

(A) In a biopsy with ACR, an intratubular CD4+ cell expresses FOXP3 (arrow). Double stain for FOXP3 (blue) and CD4 (brown) (400× original magnification). Insert shows another intratubular CD4+FOXP3+ cell (600× original magnification). (B) ACR biopsy stained for FOXP3 (blue) and CD8 (brown). Most intratubular FOXP3 are CD8 negative (arrowheads). One cell is CD8+FOXP3+ (arrow; 400× original magnification).

We tested the reproducibility of the counts by comparing the FOXP3+ cells/mm2 in different stains from the same graft biopsy (FOXP3/CD4 vs. FOXP3/CD8 double stains) and found a high correlation (r = 0.903, p < 0.0001) (Figure 3).

Figure 3.

Correlation of FOXP3+ scoring (cells/mm2) from biopsies stained with CD4/FOXP3 and CD8/FOXP3; r = 0.903, p < 0.0001 (n = 61).

We sought evidence for concentration of FOXP3+ cells in tubules by calculating the percentage of CD4+ cells that expressed FOXP3 in the tubules versus the CD4+ cells in the interstitium. We found that 15.6 ± 17.8% CD4+ T cells in tubules express FOXP3 versus 3.2 ± 2.3% of CD4+ cells in the interstitium (p < 0.001). A relative concentration of FOXP3+ expressing CD8 cells in tubules was also found, although the percentages were much lower.

FOXP3 expression in grafts by diagnosis. The number of FOXP3+ cells/mm2 was significantly higher in ACR type I and type II, compared with AHR and CNI toxicity (Table 2). In AHR, the expression of FOXP3 was a rare event (20% of the level in type I ACR). This pattern of FOXP3 expression was observed in both CD4 and CD8 stains. The differences could be explained in part by the difference in CD4+ cell infiltration in AHR, which was 30% of that in type I ACR (98 ± 112 cells/mm2 vs. 328 ± 210 cells/mm2, respectively; p < 0.05). However, some of the AHR cases also met the Banff criteria for ACR.

Table 2.  Expression of FOXP3 (cells/mm2) according to histological diagnosis
DiagnosisCD4 stainsCD8 stains
NTotal CD4+Interst FOXP3+Tub FOXP3+% FOXP3nTotal CD8+Interst FOXP3+Tub FOXP3+% FOXP3
  1. Total = absolute number of CD4+ or CD8+ T cells/mm2 in the cortical area; Interst = absolute number of FOXP3+ cells/mm2 in interstitium; Tub = absolute number of FOXP3+ cells/mm2 in tubules.

  2. % FOXP3 = percentage of CD4+ or CD8+ T cells with FOXP3.

  3. Twelve cases (9 CD4 and 3 CD8) were excluded because of inadequate sampling.

  4. 1ACR I versus AHR, CNI toxicity, Donor: p < 0.01 (CD4 and CD8 stains: total, interstitial and tubular).

  5. 2ACR I versus II: p = 0.001.

ACR I132328 ± 21012 ± 131.2 ± 1.63.6 ± 2.732752 ± 53728.2 ± 10.60.9 ± 1.20.1 ± 0.4
ACR II9304 ± 2088.6 ± 6.50.9 ± 0.72.8 ± 0.810360 ± 1895.5 ± 5.60.8 ± 2.20.06 ± 0.1
AHR1398 ± 1122.2 ± 3.80.3 ± 1.02.0 ± 2.214329 ± 3791.9 ± 3.00.3 ± 0.40.0 ± 0.0
CNI toxicity1076 ± 581.4 ± 1.80.4 ± 0.54.4 ± 6.214122 ± 810.7 ± 1.10.1 ± 0.20.0 ± 0.0
Donor77.2 ± 8.60.3 ± 0.50.0 ± 0.05.0 ± 10.0711.9 ± 12.20.2 ± 0.40.0 ± 0.00.0 ± 0.0

Although no significant difference was found between ACR type I and II in the FOXP3 cells/mm2, FOXP3+/CD8+ ratios were higher in type I than in type II ACR (median and interquartile range: 24.8 (8.4–61.7) vs. 5.6 (3.4–9.2), p = 0.004). Three of five patients with ACR who lost the graft had a high FOXP3+/CD8+ ratio. FOXP3+/CD4+ ratios were similar.

In donor biopsies, FOXP3 was expressed only in CD4+ T cells (Table 2). There were no FOXP3+CD4 or FOXP3+CD8+ cells. FOXP3+ cells were rare and located in the interstitium, preferentially in cortical peritubular capillaries.

Clinical correlations. Taking the median value of FOXP3+ cells/mm2 (7.0 cells/mm2) from the average of CD4 and CD8 stains as the threshold for high or low FOXP3 expression, graft outcome was analyzed at different time points posttransplant in patients with ACR according to the magnitude of FOXP3 expression. The length of follow-up (time from biopsy to last creatinine or graft loss) did not differ between patients with high or low FOXP3. Kaplan–Meier analysis of 2-year graft survival in patients with ACR type I or type II is shown in Figure 4. Graft survival was lower in patients with higher FOXP3 scores in comparison to the other group, 81.8% versus 100% respectively (p = 0.02). Repeating this analysis with FOXP3+/CD4+ or FOXP3+/CD8+ ratios, no difference in graft survival was found (data not shown). No difference in creatinine was found between patients with higher or lower FOXP3 in the time of biopsy (Table 3). At last follow-up, patients with higher FOXP3 scores had a lower creatinine (1.4 ± 0.4 mg/dl vs. 1.9 ± 0.8 mg/dl; p = 0.05), but grafts that were lost were necessarily excluded from this analysis.

Figure 4.

Kaplan–Meier estimates of graft survival in patients with acute rejection based on the level of FOXP3 expression at the time of biopsy (high > 7.0 cells/mm2; low ≤ 7.0 cells/mm2).

Table 3.  Graft functional outcome according to level of FOXP3 expression in patients with ACR
Time pointsHigh FOXP31 (>7.0 cells/mm2)nLow FOXP3 (≤7.0 cells/mm2)np2
  1. 1Median (7.0) of FOXP3+ cells/mm2 from the average of CD4 and CD8 stains as the cut off for high or low FOXP3 expression.

  2. 2High versus low FOXP3 expression.

  3. 3Means ± SD.

  4. 4Last sCr = serum creatinine at last follow-up.

  5. 5Time from biopsy to last creatinine or graft loss.

At biopsy2.9 ± 1.33192.8 ± 0.9200.80
2 months1.6 ± 0.5201.6 ± 0.3220.99
12 months1.5 ± 0.5181.7 ± 0.7210.31
Last sCr41.4 ± 0.4181.9 ± 0.8220.05
Time to outcome (mo)525 ± 212230 ± 22220.52

For AHR graft, survival at 2 years was poor. In these patients, FOXP3 expression was scored as high (>2.5 cells/mm2) or low (≤2.5 cells/mm2), with a graft survival of 50% versus 17%, respectively (p = 0.71).

FOXP3 expression correlated with HLA class II mismatch founding patients with ACR (Table 4). Patients with ACR and 0 or 1 DR mismatch had a higher FOXP3 expression compared to patients with 2 DR mismatches (median and interquartile range: 10.1(3.9–14.4) vs. 2.8(1.5–8.3), p = 0.03). Furthermore, in the first year posttransplant the magnitude of FOXP3 expression was greater than it was found to be after 12 months (p = 0.01). FOXP3 levels in patients not taking calcineurin inhibitors (usually taking rapamycin) also tended to be higher (p = 0.07). There was no correlation between FOXP3 expression and age, gender, type of donor and HLA class I mismatch.

Table 4.  Comparison of FOXP3 expression by demographic data, HLA match, immunosuppressive regimen and time posttransplant in patients with ACR
 NFOXP3+ cells/mm2 1p
  1. 1FOXP3+ cells/mm2= sum of interstitium and intratubular FOXP3+ cells from average of CD4 and CD8 stains.

  2. 2Median and interquartile range.

  3. 3Calcineurin inhibitor (tacrolimus or cyclosporine).

>50 years137.6(1.5–14.4)20.47
≤50 years3211.3(2.9–13.9) 
Type of donor
HLA class I (A, B)
 0-2 MM267.9(2.5–13.9)0.94
 3-4 MM176.1(2.5–15.1) 
HLA class II (DR)
 0 or 1 MM3010.1(3.9–14.4)0.03
 2 MM142.8(1.5–8.3) 
CNI3 based regimen
Time posttransplant
 ≤12 months3614.5(9.9–16.2)0.01
 >12 months86.1(2.4–11.5) 

Multivariate analysis using log-transformed FOXP3 values yielded a more normal distribution. After adjustment for the effects of age, gender, mismatch HLA class II, type of immunosuppression and time posttransplant, there remained a strong difference in the levels of log FOXP3 between patients with ACR type I and AHR (Table 5). A decrease of approximately 77% in log FOXP3 levels was observed in AHR when compared to ACR type I (p < 0.001). We also found that a higher HLA class II mismatch was associated with a decrease in log FOXP3 levels (p < 0.03), in relation to HLA class I mismatch, confirming the univariate analysis with nontransformed values (Table 4). In addition, we found a trend of increased FOXP3 related to calcineurin inhibitor-free drug regimens (usually with rapamycin).

Table 5.  Multivariate analysis of log FOXP3 correlations
ModelEffect on log FOXP31Beta coefficientp value
  1. 1Beta coefficient expressing effect on log FOXP3 were obtained in a multiple linear regression model.

  2. 2Time from transplant to last follow-up.

Acute rejection
 ACR type I 
 ACR type II−0.06−0.040.73
Female gender0.110.110.36
HLA class II mismatch−0.31−0.280.03
Rapamycin use0.400.210.07
Time posttransplant2<


T-cell-mediated immunoregulation is believed to be critical for the acquisition of tolerance to allograft antigens. Intensive research has been focused on the induction, development and action of T regulatory cells, specifically CD25+CD4+ cells, to mediate graft acceptance (1–4,18–21). The inoculation of Tregs from normal syngeneic mice, together with naïve T cells, in T-cell-deficient mice with allografts resulted in a significant increase in graft survival (30). In other experimental models, CD25+CD4+ Tregs prevented skin allograft rejection in mice pretreated under certain conditions (31,32).

The present study is the first immunohistochemical demonstration of FOXP3 cells in human allograft rejection, although murine studies have been reported (27,33). Using a two-color technique, we showed in human kidney grafts that the vast majority of FOXP3+ cells are CD4+ T cells, with few CD8+ and rare CD4CD8 cells. CD4+FOXP3+ cells were commonly localized in dense interstitial infiltrates of ACR type I and II. While we did not assess the function of these cells, the common coexpression of CD4 is more consistent with a Treg phenotype than as simply a low level related to activation of CD4 and CD8 cells (24). Judging from the low frequency of positive cells in the histochemical stains (3.4% of CD4+ cells and 0.1% of CD8+ cells), we probably detected primarily the cells with high FOXP3 expression.

Zheng and colleagues (34) hypothesized that the immune response directed at an allograft can have two components—a graft destructive one by cytotoxic T cells and associated molecules, and a protective process mediated by regulatory T cells through the action of FOXP3 transcription factor and other molecules. The balance between these two opposing forces can influence graft outcomes. In allograft kidneys in mice, greater FOXP3+ infiltrates correlated with acceptance of the graft (33); some of the cells were found in tubules mimicking rejection. In a study evaluating the effect of immunosuppressive drugs on the induction of FOXP3 in human T cells in vitro and FOXP3 mRNA expression in biopsies of cardiac allograft recipients, Baan and colleagues (35) reported that cyclosporine, tacrolimus and daclizumab inhibited FOXP3 gene transcription, whereas rapamycin did not. Bestard et al. (36) induced donor-specific cellular hyporesponsiveness in recipients of kidney grafts by a CNI-free protocol that included anti-thymocyte globulin, mycophenolate mofetil and rapamycin. Patients with acquired donor-specific hyporesponsiveness showed FOXP3+ lymphocyte infiltrates in 6-month protocol biopsies associated with excellent renal function.

We observed that the proportion of CD4+ cells in tubules expressing FOXP3 is significantly higher than in the interstitium, irrespective of the severity of ACR. The reasons for the selective localization are not known. Among the possibilities are selective attraction (chemokines), retention or proliferation. Alternatively, induction of FOXP3 may be promoted by the tubular cells, which are known to produce TGFβ under certain conditions (37), but two studies did not show a correlation between latent TGFβ or active TGFβ and FOXP3 mRNA expression by PCR in kidney allografts (27, 38). Whether TGFβ is responsible for the conversion of Treg in tubulitis is an issue to be investigated. The significance of the localization is also not established, but we hypothesize that ‘Treg tubulitis’ might be beneficial to graft acceptance. ‘Treg tubulitis’ may down-regulate the immune response where effector CD4+ helper Th1 and cytotoxic CD8+ T cells are active (34). Selective intratubular accumulation of FOXP3+ cells may help to explain the increased levels of urinary FOXP3 mRNA in acute rejection described by Muthukumar and colleagues (29).

We observed FOXP3 cells only rarely in glomeruli and not in endarteritis lesions in these cases. We do not rule out localization to these sites, since we did not examine florid cases of transplant glomerulitis, nor did we have many samples of endarteritis. We have observed FOXP3+ cells in the intima in nephrectomy samples of grafts with severe cellular rejection, so that FOXP3 cells can traffic to that site (unpublished data).

The present study showed a better graft functional outcome in acute rejection, similar to the report of Muthukumar et al. (29). However, the effect was due to the paucity of FOXP3 cells in patients with humoral rejection, which was not ascertained in the prior study (29). Among our patients with cellular rejection alone and higher FOXP3 infiltrates survival was paradoxically decreased. Creatinine at last follow-up was lower in the group with higher FOXP3, but this was skewed by the exclusion of grafts that had been previously lost. No outcome correlation was found using FOXP3+/CD8+ or FOXP3+/CD4+ ratios. The lower expression of FOXP3 in humoral rejection could be largely attributed to a lower level of CD4+ T-cell infiltration in AHR; tubulitis is also not a feature of pure antibody-mediated rejection.

The proposed mechanisms by which Treg cells protect from immunological injury in experimental studies include local recruitment to inflammatory sites where they maintain unresponsiveness. In syngeneic islet grafts in nonobese diabetic mice, diabetes is prevented by treatment with an adenovirus vector encoding the active form of TGFβ1. Immunohistology showed a mononuclear peri-islet infiltrate primarily of CD4+ T cells, of which 10% expressed CD25 and FOXP3, suggesting that local Treg cells may block autoimmune islet destruction (32). In experimental models, molecules such as CCR4 chemokine receptor (27) and CD103 chain of αEβ7 integrin (39), respectively have been associated with attraction and retention of Tregs in tissues. Lee et al. (27) showed that FOXP3+ cells traffic to sites of active immunity, but higher levels of FOXP3 mRNA were found at the same sites and persist for a long term in cardiac allografts of recipient mice tolerized with CD154 mAb plus donor-specific transfusions. In clinical heart transplantation, the highest intragraft FOXP3 mRNA expression in endomyocardial biopsies was found during acute rejection, suggesting that the anti-donor response via TCR triggered the influx or generated FOXP3+ T cells into the graft as an intrinsic mechanism of the in vivo alloresponse (35).

Immunosuppressive drugs affect FOXP3+ Treg function. In experimental bone marrow transplantation, cyclosporine, but not rapamycin or mycophenolate mofetil, suppressed T regulatory function with increased T-cell proliferation and reduced graft survival (40). In line with these findings, we observed a trend towards a lower expression of FOXP3 in patients with CNI immunosupression, and the opposite with rapamycin.

The generation and regulatory functions of FOXP3+ Tregs require stimulation of the T-cell receptor (41). The requirement for antigen might favor the graft itself as the site for the initial signal. In addition, nonantigen-specific signals known to be important in Treg generation and activity, such as TGFβ (15,26,42–44), IL-2 (45), CTLA-4 (18,46) IFN-γ (47), are found in the graft. It is interesting that in both univariate and multivariate analysis we found that the higher the HLA class II DR matching the higher was the level of FOXP3 expression, suggesting that local Treg infiltration may be promoted by the direct recognition of graft antigens in the context of self HLA class II antigens. It is notable that in some tolerance models the graft must remain in the recipient to maintain tolerance, and kidneys, but not heart, allografts have this capacity (48). The functional capacity of intragraft Treg was demonstrated by their ability to transfer tolerance to naïve recipients of skin allografts (28). Further investigation is warranted of the events in the graft relevant to the maintenance of graft acceptance.


This study is supported by grants from the NIH and a fellowship from CAPES - Brazilian Research Agency (FV).