|Home | About | Journals | Submit | Contact Us | Français|
In a cross-sectional study, we assessed effects of calcineurin inhibitor (CNI) or rapamycin on T-regulatory (Treg) cells from children with stable liver (n=53) or kidney (n=9) allografts several years post-transplant. We analyzed Treg number, phenotype, suppressive function, and methylation at the Treg-specific demethylation region (TSDR) using Tregs and peripheral blood mononuclear cells. 48 patients received CNI (39 as monotherapy) and 12 patients received rapamycin (9 as monotherapy). Treg numbers diminished over time on either regimen, but reached significance only with CNI (r=−0.424, p=0.017). CNI levels inversely correlated with Treg number (r=−0.371, p=0.026), and positively correlated with CD127+ expression by Tregs (r=0.437, p=0.023). Patients with CNI levels >3.6 ng/ml had weaker Treg function than those with levels <3.6 ng/ml, whereas rapamycin therapy positively correlated with Treg numbers (r=0.628, p=0.029) and their expression of CTLA4 (r=0.726, p=0.041). Overall, CTLA4 expression, TSDR demethylation and an absence of CD127 were important for Treg suppressive function. We conclude that rapamycin has beneficial effects on Treg biology, whereas long-term and high dose CNI use may impair Treg number, function and phenotype, potentially acting as a barrier to attaining host hyporesponsiveness to an allograft.
Though essential for allograft survival in most recipients, immunosuppression can cause opportunistic infections, malignancies, cardiovascular diseases and nephrotoxicity (1). Despite excellent one-year graft survival rates, chronic allograft dysfunction persists, perhaps reflecting, at least in part, impairment of FOXP3+ Tregs (2). CNI use, central to most immunosuppressive regimens (3), blocks IL-2 production and may adversely affect Tregs (4–6). Experimentally, rapamycin enhances de novo conversion, proliferation and suppressive function of Tregs (5, 6), and may increase Treg numbers clinically (7, 8), though ongoing deterioration in renal function occurred in some (9) but not other series (10). Given this uncertainty, we developed a composite approach to simultaneously analyze 4 key Treg parameters: numbers, phenotype, suppressive function under highly standardized conditions, and FOXP3 epigenetic status. We show the clinical utility of our approach by determining the effects of CNI or rapamycin therapy in pediatric transplant (Tx) recipients.
Healthy volunteer donors (n=13, 7 males) provided peripheral blood mononuclear cells (PBMC) via the University of Pennsylvania Human Immunology Core.
We studied 62 patients with stable liver (n=53) or kidney (n=9) allografts (Tables 1–2). CNI (tacrolimus) blood levels were 0.7–7.4 ng/ml (mean=3.61±0.21 ng/ml), and rapamycin levels were 1–9.1 ng/ml (mean=4.1±0.69 ng/ml). Patients were similarly distributed for gender, race, time post-Tx, donor type (living or deceased), proportions receiving liver or kidney allografts, primary diseases leading to Tx, and concomitant medications, but varied in ages (CNI=13.1±0.5 and Rapa=10.7±0.9 years, p=0.03). As ages were not correlated with data except time post-Tx (r=0.561, p<0.001), rapamycin and CNI groups were suitable for comparisons.
PBMC were isolated from 32 ml of blood and shipped overnight at 4 °C to Children's Hospital of Philadelphia (CHOP). For standardization, CHOP samples were also kept at 4°C overnight. Cell numbers and viability were determined, and Tregs isolated using magnetic beads (CD4+CD25+Treg isolation kit, Miltenyi-Biotec). Aliquots of Tregs were immediately tested in suppression assays (11), and frozen for flow cytometry and DNA isolation. To ensure suppression assays reflected effects of immunosuppression on Tregs and not conventional T effector (Teff) cells, the same healthy donor CFSE-labeled T-cells were used in each assay. In preliminary studies, these cells were tested with different healthy donor Tregs to ensure absence of variability due to HLA differences (data not shown). In parallel, we assessed proliferation of patient CFSE-labeled CD3-stimulated CD4+CD25− Teff cells. Aliquots of isolated Tregs were stained for CD4, CD25, CD127, CTLA4, FOXP3 (PCH101, eBioscience) and LIVE/DEAD Kit (Invitrogen). PBMC stimulated overnight with CD3/CD28 mAb-coated beads 1:1 (Invitrogen) and 100 U/ml IL-2 (Roche) were stained with CD4, CD25, CD31, CD39, Helios and FOXP3 mAbs. Staining for FOXP3, CTLA4 and Helios were performed using FOXP3 Fix/Perm kit (eBioscience). FOXP3 TSDR-methylation assays used DNA from 60–100×103 Tregs or 100–400×103 PBMC (12). TSDR-methylation was evaluated separately by gender, but also percentages of demethylated Tregs in girls were calculated as %Tregs = (100-HM)*2, where HM is %TSDR-hypermethylation.
Data are shown as mean ± SEM. For continuous variables with normal distribution, statistical significance was assessed by Student's t-test to compare two groups, and Pearson correlation assay used to test correlations, unless specified. In some cases, we performed partial correlation assays with FOXP3 as a controlling variable. For testing association in two-way tables, we used chi-square or Fisher exact test. To test correlations between variables not normally distributed, we used nonparametric Spearman test. To standardize results of each suppression assay, we calculated area-under-curve as described (11) and shown (Figure S1). A two-tailed p value of <0.05 was considered statistically significant. For all calculations, all available data without any type of variable reductions (outliers etc.) or adjustments were used.
Patient Tregs isolated using CD4+CD25+ beads had consistently high purity (85.8±5%), but unlike in healthy donors, had variable FOXP3+ expression (Figure S2A). To assess contamination by CD4+CD25+CD127+ activated Teffs, we analyzed FOXP3+ and FOXP3− cells within CD4+CD25+ isolates. About 50% of patients had FOXP3+ cells mostly within the CD25highCD127− subset, and FOXP3− cells occurred mostly within the CD25−/lowCD127+ subset (Figure S2B, patients a, d), like in normal donors. However, additional unusual distributions of FOXP3+ and FOXP3− cells were noted, including increases in CD25+CD127+FOXP3+ cells (Figure S2B, patients b & c) or CD25+/highCD127−/lowFOXP3− cells (Figure S2B, patients e & f). In healthy donor Tregs, CD25high or CD25highCD127− gating strategies showed 90–96% FOXP3+ cells, but in Tx recipients with unusual FOXP3 distributions this led to only 75–80% FOXP3+ cells and extremely low yields (20%–50% of yields with magnetic beads, data not shown). Given limited blood volumes available from pediatric recipients, sorting of CD4+CD25+CD127− cells would not generate enough Tregs for functional and methylation studies, and would provide the same variable FOXP3 purity as achieved by bead isolation. We therefore stained aliquots of each isolated Treg sample for CD4, CD25, FOXP3, CD127 and CTLA4, and adjust our data according to FOXP3 expression, as shown in figure legends and text. Thus, we calculated numbers of CD4+CD25+ isolated Tregs and CD4+CD25+FOXP3+ Tregs and described both statistics. When two groups of Tregs were compared (e.g. CNI & rapamycin), we tested differences in FOXP3 expression. We also evaluated correlations of various markers in isolated Tregs with suppressive function, FOXP3 epigenetic status and clinical variables, and used CD4+CD25+FOXP3+ phenotype to exclude effects of varying FOXP3 expression. During Treg isolation, calculation of viable CD4+ cells and CD4+CD25+ Tregs collected from 1×106 PBMC allowed us to evaluate Treg numbers with regard to number of isolated CD4+CD25+ Tregs, number of CD4+CD25+FOXP3+ Tregs, and number of TSDR-demethylated Tregs. TSDR-demethylation data was adjusted by gender as described in Methods, but also analyzed separately for boys and girls.
The number of CD4+CD25+ Tregs and CD4+CD25+FOXP3+ Tregs inversely correlated with years post-Tx (r=–0.357, p=0.02 for CD4+CD25+ Tregs in liver recipients). If rapamycin-treated patients were excluded, this correlation increased (r=−0.424, p=0.017). There was no significant correlation for rapamycin (r=−0.30, p=0.43), and no correlation between patient age and Treg number, indicating an effect specific for chronic CNI use. These data suggest continuous CNI therapy is linked with progressive decline in Treg numbers.
CNI blood levels correlated inversely with the number of Tregs (and FOXP3+ Tregs) in PBMC, numbers of Tregs in the CD4+ subset, and numbers of TSDR-demethylated Tregs in PBMC (Table 3, Figure 1A). Additionally, CNI blood levels correlated with CD127+ expression in CD25+FOXP3+ Tregs (Figure 1A). Two groups of CNI patients defined by mean trough CNI levels (< or > 3.6 ng/ml) had differential Treg function, being impaired in those with CNI >3.6 ng/ml. This was observed in all patients, in liver recipients alone, and in CNI monotherapy patients (Figure 1A and not shown), excluding possible additional effects of combined therapies. CNI “low” and “high” groups had no differences in other clinical and demographic data (not shown). In contrast to CNI, rapamycin blood levels were positively correlated with Treg numbers (Figure 1B) and high CTLA4++ expression in isolated Treg cells (Figure 1B). Patients on CNI had significantly higher CD127+ expression in CD4+CD25+FOXP3+ Tregs (8.31±0.85% vs. 4.95±0.47%, p=0.046 for all patients, 7.80±0.74% vs. 4.99±0.54%, p=0.048 for patients on monotherapies). To test effects of high CNI or rapamycin concentrations, we compared patients with CNI blood levels >3.6 ng/ml with patients having rapamycin levels >4.1 ng/ml and found that CNI patients had fewer Tregs in the CD4+ subset (1.73±0.15% in CNI vs. 2.96±0.71% in Rapamycin, p=0.011). These data indicate differential effects of long-term CNI and rapamycin therapy on Treg numbers and suppressive function.
In liver Tx patients, tacrolimus blood levels correlated with number of biopsy-proven acute rejection episodes (r=0.327, p=0.042 for all CNI patients; r=0.351, p=0.033 for CNI monotherapy, Spearman tests), but neither rapamycin blood level nor dosing had such correlations. Given CNI-rejection data, we tested if patients with multiple rejections differed from others in any Treg-associated markers. We found CNI patients with multiple rejections (>=2) had more CD127 and tended to have less CTLA4 expression in Tregs compared with 0–1 rejections group (Figure 2A for all patients and not shown for liver recipients). The number of acute rejections correlated with in vitro proliferative ability of patient CD4+CD25− Teffs (r=0,798, p=0.03, Spearman), suggesting a persisting greater capacity for T cell proliferation in patients with histories of multiple rejections.
To assess whether CTLA4 expression is important for Treg suppression in Tx recipients, we divided patients into 2 groups based upon mean value of CTLA4++ expression (< or > 9.5% CTLA4++ Treg). Tregs in the group with higher CTLA4++ expression had better suppression (Figure 2B), lower CD127 expression, and lower rates of Teff proliferation in vitro (Figure 2C). The same differences were observed in CNI-treated patients, indicating this was not due to rapamycin-CNI differences. The suppressive function of patient Tregs correlated with CTLA4++ expression (r=0.462, p=0.03 for CD4+ responders). Low CD127 expression appeared important for Treg suppression since CD25+FOXP3+ Tregs with higher CD127+ expression were weaker suppressors (Figure 2D), and this was true when CNI-treated liver recipients were analyzed separately (not shown). While FOXP3 expression correlated with CTLA4 and absence of CD127 expression in all tested groups (not shown), and despite variable FOXP3+ expression in Tregs after isolation, FOXP3+ “high” and “low” Tregs (< or > mean 47.7%) had comparable suppressive function (Fig. 2E), despite differences in CTLA4 and CD127 expression (Figure 2F). FOXP3 expression in Tregs did not correlate with suppressive function for any group (all patients, liver grafts, boys, girls, patients on CNI) (Figure S3 and not shown). Collectively, these data suggest that CD127− and CTLA4+ are likely surrogate markers of functional FOXP3 protein and therefore should be used in conjunction with FOXP3 evaluation by flow cytometry.
TSDR-demethylation within PBMC or tissue samples is reportedly the most specific indicator of “true” Tregs (13). We therefore tested TSDR-demethylation correlated with the number of Tregs post-Tx, calculated as CD4+CD25+ cells, CD4+CD25+FOXP3+ cells, or as TSDR-demethylated CD4+CD25+ cells. We found no correlations between TSDR-demethylation and any Treg isolate in all patients, or in boys or in girls (Table 4). Moreover, TSDR-demethylation had no correlations with numbers of CD127− or CTLA4+ cells (not shown). To assess whether TSDR-demethylation in PBMC was affected by T cell activation conditions, we stimulated healthy donor PBMC with CD3/CD28 mAbs ± IL-2. CD4+ T cells significantly upregulated FOXP3 expression and increased TSDR-demethylation, and addition of IL-2 reduced TSDR-demethylation without affecting FOXP3 upregulation (Figure 3). Importantly, CD4+FOXP3+ stimulated cells were CD25+ and CD127−, and had comparable CD39, CD31 and Helios markers, despite different TSDR-demethylation. These data indicate monitoring TSDR-demethylation levels in PBMC is not an accurate means to assess Treg numbers post-Tx.
Hypermethylated-TSDR FOXP3 in boys negatively correlated with CTLA4 and with absence of CD127 expression in Tregs (Figure 4 A, B), i.e. with markers linked to enhanced suppressive function. TSDR-hypermethylated Tregs tended to have impaired suppressive function (Figure 4C), but the strongest correlation was found for FOXP3 protein (r=−0.732, p=0.002). In contrast to induced Tregs (iTregs) and FOXP3+ activated T cells, natural Tregs have TSDR-demethylation (13). To characterize patient Treg subsets, we calculated the ratio of %FOXP3+ cells to %TSDR-demethylated cells within isolated suppressive CD4+CD25+ Tregs (FOXP3+/TSDR ratio). In only 12.5% cases was the ratio ≈1, and CNI and rapamycin groups showed interesting differences. Most patients receiving CNI had FOXP3+/TSDR ratios <1, while in rapamycin-treated patients had ratios typically >1 (p=0.033, Fisher's exact test, Figure 4D). FOXP3+/TSDR ratios <1 indicate cells with TSDR-demethylation but not FOXP3 expression, consistent with CNI inhibition of the transcription factor, nuclear factor of activated T cells (NFAT) (14) (Figure 5), whereas FOXP3+/TSDR ratios >1 suggest FOXP3+ cells with methylated-TSDR, e.g. iTregs or activated Teffs. Since the group with FOXP3+/TSDR ratios >1 had more suppressive Tregs and higher CTLA4 expression than the group with FOXP3+/TSDR ratios <1 (Figure S4), it is likely the rapamycin group had increased numbers of FOXP3+ TSDR-methylated suppressive iTregs rather than Teff cells (Figure 5). Collectively, these data suggest TSDR-demethylation is a useful component of Treg (though not PBMC) characterization, and that the associated FOXP3+/TSDR ratio provides insight into the relative contributions of Treg subsets in clinical Tx recipients.
While murine studies showed increased numbers of FOXP3+ Tregs in tolerant allograft recipients (15), the induction of CD25 and FOXP3 upon activation of human Teffs (16) has led to far more variable clinical data. Thus, Treg numbers were increased in tolerant recipients (17, 18), whereas others found no differences (19, 20). Similarly, Treg numbers were decreased in patients with acute (21–23) or chronic (19, 24) rejection in some Centers but not others (25, 26). Despite an ever-expanding list of markers to identify “true” human Tregs, including CD127−/low and CD49d−, CD39+, CTLA4+, GARP+, CD120+, LAP+, CD27+ etc., none have proven appropriately Treg-specific using patient samples. E.g. in healthy donors CD4+CD25+CD127−/low gated cells are 90–95% FOXP3+, whereas that subset is enriched by FOXP3− cells in rheumatoid arthritis (27). In Tx recipients, we found variable FOXP3 expression by CD4+CD25+ cells, many FOXP3+ cells outside the CD25++CD127− gate, and contamination of that gate by FOXP3− cells. Others noted similar findings in renal Tx patients (28). Thus, currently flow cytometry alone cannot satisfactorily enumerate Treg cells post-Tx, and needs to be accompanied by assessment of Treg function.
Treg suppression assays have pitfalls, since isolation of CD4+CD25+ (or CD4+CD25+CD127−) cells from Tx recipient results in a mix of suppressive (Tregs) and activated (Teffs) cells. We are unaware of any literature assessing at least FOXP3 (or better, FOXP3, CD127 and CTLA4) expression within isolated cells and corresponding assessment of Treg suppressive function post-Tx. Moreover, using autologous cells as responders in Treg assays can be misleading, since immunosuppression can decrease their proliferative capacity. Such use may explain why Tregs from patients with rejection had impaired suppressive function in some studies (24, 26) but not others (23, 29, 30). We therefore assessed CD25, CD127, CTLA4 and FOXP3 expression in isolated Tregs, and performed suppression assays under highly standardized conditions, with Tregs as the only variable. We found Treg function post-Tx correlated with CTLA4 and absence of CD127 expression, but not with FOXP3, despite its role as a master regulator in Tregs (2). This likely reflects how activated human T cells can induce FOXP3 without acquiring a Treg phenotype or suppressive function, and is consistent with dominance of CTLA4 over FOXP3 for human Treg suppressive function (11).
The epigenetic status of FOXP3 is useful given the TSDR is demethylated only in natural Tregs with stable FOXP3 expression (13). We found that TSDR-demethylation indeed correlated with a “true” Treg CD127−CTLA4+ phenotype and suppressive function, and described a new way to evaluate Treg subsets within isolated Tregs: the ratio of FOXP3+ cells to TSDR-demethylated cells (FOXP3/TSDR). Almost all Tregs from CNI patients had FOXP3/TSDR ratios <1, showing FOXP3 gene transcription was impaired, while TSDR FOXP3 demethylation was preserved. Conversely, most Tregs in rapamycin-treated patients had FOXP3/TSDR ratios >1, showing the Treg subset was enriched by suppressive, CTLA4+FOXP3+ cells with a methylated-TSDR, and indicative of iTregs (Figure 5). Rapamycin stimulates iTreg conversion through inhibition of the mTOR pathway and enhancing TGF-β production by conventional T cells (2).
Assessment of TSDR-demethylation is proposed as way to determine Treg numbers in PBMC or tissue samples (13), but requires at least two conditions. First, every TDSR-demethylated cell should have normal FOXP3 transcription and hence, FOXP3 protein. Second, cells with methylated-TSDR should be unable to produce FOXP3 or, if FOXP3 is produced, the cells should not be suppressive. This concept does not adequately account for iTregs. TSDR-demethylation in PBMC post-Tx showed no correlations with Treg numbers, calculated as CD4+CD25+, CD4+CD25+FOXP3+ or even as CD4+CD25+TSDR-demethylated cells. We are unaware of any previous comparison of Treg numbers calculated using TSDR-demethylation in PBMC and TSDR-demethylation in Tregs, and note that TSDR-demethylation in PBMC varies with activation.
Our study design has several points to note. We excluded patients undergoing acute or chronic rejection, or those graft failure, allowing us to evaluate Tregs from patients with stable graft function and receiving conventional immunosuppression. However, patients with impaired grafts may provide additional insights. Second, we excluded patients with conditions affecting Treg number or function, such as cancer, autoimmune diseases, hepatitis and serious infections, to minimize their possibly confounding effects. Lastly, our study had a small size and cross-sectional design. Despite this, our composite approach showed that even in patients with stable graft function, CNI use may adversely affect Tregs. Serial studies including collection of samples preand post-Tx will likely produce key additional insights, and are now underway in our lab.
Supported by grants from the National Institutes of Health: 1RC1 DK087270, UL1-RR026314 NCRR/NIH, and NIDDK KO8 1K08DK092282.
Ethical guidelines Clinical studies were approved by the respective IRB of each Center: CHOP-08-006311_CR1, CCHMC-2009-0018, and HSC-1000014523.
Disclosure The authors declare no conflict of interest.