Antigen-specific T cell responses to BK polyomavirus antigens identify functional anti-viral immunity and may help to guide immunosuppression following renal transplantation

Abstract

Infection with the polyoma virus BK (BKV) is a major cause of morbidity following renal transplantation. Limited understanding of the anti-viral immune response has prevented the design of a strategy that balances treatment with the preservation of graft function. The proven utility of interferon-gamma enzyme-linked immunospot (ELISPOT) assays to measure T cell responses in immunocompetent hosts was the basis for trying to develop a rational approach to the management of BKV following renal transplantation. In a sample of transplant recipients and healthy controls, comparisons were made between T cell responses to the complete panel of BKV antigens, the Epstein–Barr virus (EBV) antigens, BZLF1 and EBNA1, and the mitogen phytohaemagglutinin (PHA). Correlations between responses to individual antigens and immunosuppressive regimens were also analysed. Antigen-specific T cell responses were a specific indicator of recent or ongoing recovery from BKV infection (P < 0·05), with responses to different BKV antigens being highly heterogeneous. Significant BKV immunity was undetectable in transplant patients with persistent viral replication or no history of BKV reactivation. Responses to EBV antigens and mitogen were reduced in patients with BKV reactivation, but these differences were not statistically significant. The T cell response to BKV antigens is a useful and specific guide to recovery from BKV reactivation in renal transplant recipients, provided that the full range of antigenic responses is measured.

Introduction

Infection with the polyoma virus BK (BKV) is a major cause of morbidity following renal transplantation and a challenging clinical problem. In particular, our limited understanding of anti-viral immune responses has prevented the design of rational strategies that balance treatment with the preservation of graft function [1,2]. Initial exposure to BKV usually occurs in childhood, after which it becomes latent in urothelial tissues [3,4]; however, reactivation of the virus leading to tissue damage and ascending urinary tract infection can occur with immunosuppression [5]. BKV replication is detectable in up to 30% of patients following renal transplantation [6], and if unchecked these infections can lead to allograft dysfunction or even loss [5,7–9].

The current management of BKV in most centres is cautious reduction of immunosuppression, with the aim of restoring virus-specific immunity without precipitating graft rejection [10]. Previous studies have shown that recovery from polyoma virus infections is associated with the development of virus-specific T cells to BKV antigens [11–13], which remain detectable months after the resolution of infection [14]. However, as no routine clinical tests of host anti-viral responsiveness are available, surrogate measures of the effect of immunosuppression reduction are used, including serial measurements of viral nucleic acid and renal function.

Interferon gamma (IFN)-γ–enzyme-linked immunospot (ELISPOT) assays are a rapid, sensitive technique that has become the standard clinical laboratory method for detecting antigen-specific T cells in response to infectious diseases or vaccination in immunocompetent individuals [15–19]. Therefore, we questioned to what extent a similar analysis of the host immune response to BKV in immunosuppressed recipients of renal transplants might be used to guide management of patients with viral reactivation. Using overlapping peptide pools covering all immunogenic early and late BKV antigens, two immunodominant Epstein–Barr virus (EBV) antigens (BZLF-1 and EBNA-1) and the mitogen phytohaemagglutinin (PHA), we analysed the responses of transplant recipients with or without evidence of BKV reactivation and healthy controls. The aim of this study was to begin to explore the temporal relationship between immunity, viral replication and immunosuppression in the acute situation. Transplant patients with active BKV replication had evidence of impaired immune function, with viral clearance associated with the development of strong T cell responses against individual BKV antigens; however, the dominant antigen response varied markedly between patients. We conclude that an IFN-γ–ELISPOT assay is a simple and reliable way to measure the spectrum of T cell responses to BKV and may be a useful tool to guide immunosuppression reduction in patients with evidence of BKV replication.

Materials and methods

Patient recruitment and clinical samples

Patients attending the Oxford Transplant Unit and healthy controls identified through advertisement within the department were recruited to donate whole blood following written informed consent. Ethical approval for the study was granted by the Berkshire Research Ethics Committee (REC reference 08/H0607/50). Samples were collected for ELISPOT analysis from 26 transplant recipients and 12 healthy controls. Transplant patients were treated according to local protocols with induction therapy including basiliximab (14 patients) or alemtuzumab (12 patients) and maintenance with combinations of tacrolimus and mycophenolate or azathioprine and glucocorticoids. Two patients in the study had biopsy-proven acute rejection, one patient in the BKV-positive group and one in the cleared BKV group. In both cases the rejection was steroid-responsive, and no patients in the study received additional anti-rejection agents, such as anti-thymocyte globulin. Anti-viral prophylaxis against cytomegalovirus (CMV) was prescribed routinely to all patients, except where donors and recipients were both CMV seronegative. Serial samples were obtained from four patients with active BKV viraemia at the time of initial recruitment. As the Oxford Transplant Unit uses decoy cell detection (urothelial cells containing BKV viral inclusion bodies) to screen for BKV reactivation following transplantation – patients with two or more urine samples positive for decoy cells then have blood quantitative polymerase chain reaction (qPCR) performed – patients were divided into three groups at the time of donation, as follows: (i) BKV-positive, currently decoy cell-positive, nine patients; (ii) cleared BKV, currently decoy cell- and qPCR-negative but previously decoy cell-positive, nine patients; and (iii) BKV-negative, patients who had never been decoy cell- or qPCR-positive, eight patients. All patients in the BKV-positive group had sustained decoy cell positivity (more than two positive samples > 2 weeks apart) and seven patients were viraemic (as assessed by qPCR) at the time samples were collected. Decoy cell status is used to identify BKV reactivation in Oxford because of the rapid turnaround and sensitivity for detecting clinically significant viral replication [6]. Although the detection of decoy cells has a lower positive predictive value for the development of polyomavirus nephropathy than plasma PCR [20], a recent review of our decoy screening programme has shown that 75% of patients with sustained decoy cell positivity develop viraemia an average of 45 days after the onset of decoy cell positivity. BKV qPCR was performed by the HPA Regional Laboratory, Bristol, UK. All patients were EBV-seropositive at the time of transplantation. Demographic and clinical details are shown in Supplementary Fig. S1.

Isolation of peripheral blood mononuclear cells (PBMCs)

Cell separation tubes (Sigma, Gillingham, UK) were used to isolate PBMCs from whole blood by density gradient centrifugation. The PBMC layer was removed, washed twice with sterile phosphate-buffered saline (PBS; Fisher Scientific, Loughborough, UK) and rested overnight in RPMI-1640 media, supplemented with 10% fetal calf serum, 1% l-glutamine, 1% penicillin–streptomycin solution, 50 µM β-mercaptoethanol (all from Sigma) and 1% HEPES buffer (Gibc oBRL, Scotland, UK) (referred to herein as complete media).

Preparation of antigens

Peptide pools (15 amino acids in length with overlaps of 11 amino acids) of BKV early (large T and small t) and late (VP1–VP3) antigens, the immunodominant EBV antigens BZLF1 and EBNA1 were purchased from JPT (Berlin, Germany) and stored at −20°C as a lyophilized powder until used. Peptide pools were dissolved in dimethylsulphoxide (DMSO; Sigma) and diluted with Dulbecco's PBS (Gibco BRL) to create stock solutions (0·15 mg/ml). Lymphocyte mitogens, concanavalin A (ConA), 1 mg/ml stock solution and PHA 1 mg/ml stock solution (both from Sigma) were used as positive controls.

IFN-γ–ELISPOT assay

Polyvinylidene (PVDF) 96-well plates (Millipore, Watford, UK) were coated with 100 µl of IFN-γ monoclonal antibody at 15 µg/ml (clone 1-D1K) (MAbTech, Stockholm, Sweden) and incubated at 2–8°C overnight. The following day plates were washed five times with PBS (Fisher Scientific) and incubated for 2 h at 37°C with 200 µl/well of complete media. BKV and EBV peptide pools were added in triplicate to each plate at a final concentration of 3 µg/ml. PHA and Con A were used at final concentrations of 5 µg/ml, while DMSO at the same concentration used to resuspend the peptides in complete media was used as a negative control (all in triplicate). PBMCs (2 × 10⁵) were added to each well and incubated at 37°C 5% CO₂ for 18–20 h. At the end of the incubation, plates were washed five times with PBS/0·05% Tween 20 (Sigma), then incubated at room temperature for 2 h with 100 µl of 1 µg/ml biotinylated IFN-γ antibody (clone 7-B6-1) (MAbTech). The plates were then washed again before being coated with 100 µl streptavidin–alkaline phosphatase (ALP) (MAbTech) at 1 µg/ml and incubated at room temperature for 1 h. After further washing, the plates were developed with 100 µl/well of 0·22 µm filtered 5-bromo-4-chloro-3-indolyl phosphate nitroblue/tetrazoliun chloride (BCIP/NBT) (Pierce, Cramlington, UK). Development was continued until spots appeared in the positive control wells. Spots were counted using an automated plate reader, AID ELISPOT (AID, Strassberg, Germany), with AID ELISPOT software version 3·5 (Cadama Medical Ltd, Stourbridge, UK).

Data analysis

Statistical analyses were performed with GraphPad prism version 5 (GraphPad Software, Inc., CA, USA). Differences between groups were analysed by one-way analysis of variance (anova) or independent t-tests. Statistical significance was defined as P < 0·05.

Results

Resolution of BKV infectivity is associated with measurable antigen-specific T cell immunity

We set out to test the hypothesis that host responsiveness to BKV antigens may be a useful guide to treatment in the clinical setting of renal transplantation. We wanted to know if these assays would be sufficiently sensitive to be used reliably on an individual basis, and therefore we recruited a small but well-defined sample of patients from the Oxford transplant programme and controls. Recruited subjects included healthy controls (average age 43 years, range 30–62), transplant patients with no evidence of BKV replication (average age 48 years, range 30–68, time after transplant 74–225 days), transplant patients with ongoing BKV replication (average age 53 years, range 31–71, time after transplant 48–263 days) and transplant patients with a history of BKV reactivation (average age 48 years, range 31–63, time after transplant 59–211 days), but who no longer had evidence of active replication.

In each case we determined the range of T cell responses to BKV antigens by incubating PBMCs with a series of overlapping peptide pools from all five immunogenic BKV antigens, as well as two immunodominant EBV antigens and the mitogen PHA as controls (Fig. 1). Patients who no longer had evidence of viral replication (currently plasma qPCR and urinary decoy cell-negative), which we have termed cleared BKV, demonstrated significantly increased responses to BKV antigens (large T P < 0·05; small t P < 0·01; VP1 P < 0·002; VP2 P < 0·005; VP3 P < 0·002) compared to patients who were currently BKV-positive, transplant recipients without evidence of BKV reactivation or healthy controls. Using 50 antigen-specific IFN-γ spot-forming units (SFU)/million PBMCs as a robust measure of a positive response, we found that all the patients who had recently cleared infection were responsive to at least one BKV antigen (Fig. 1c–g). In contrast, patients with evidence of current BKV replication on the basis of decoy cell positivity showed much lower responses to BKV antigens, which were below the 50 SFU/million PBMCs threshold except in the case of one patient, who became BKV-negative in the week following assessment (Fig. 1c–g). Interestingly, transplant patients with ongoing or recent BKV replication both showed a trend towards reduced T cell responses to EBV antigens and PHA compared to healthy and other transplant controls, suggesting a global impairment in T cell function due to the level of immunosuppression (Fig. 1a,b,f). However, these responses were variable and most importantly did not distinguish clearly between patients with ongoing BKV replication or recovery. Healthy controls and transplant patients without evidence of clinical BKV reactivation after transplantation showed good responses to PHA and EBV antigens, but generally lower responses to BKV antigens. Two healthy immune-competent controls had evidence of BKV immunity (> 50 SFU/million PBMCs), consistent with prior exposure to BKV and persisting cellular immunity (Fig. 1c–g).

Fig. 1.

Enzyme-linked immunospot (ELISPOT) responses to polyomavirus BK (BKV) and control antigens following renal transplantation. Responses to overlapping peptide pools of Epstein–Barr virus (EBV) antigens (a,b), early BKV antigens (c,d), late BKV antigens (e–g) and mitogen (h) were assessed by interferon (IFN)-γ–ELISPOT assay. Responses are shown for transplant recipients who are BKV^- (never decoy cell-positive), BKV⁺ (decoy cell-positive) or cleared BKV [previously positive but now decoy cell- and quantitative polymerase chain reaction (qPCR)-negative, in red] and healthy controls. Points represent individual patients, bars are means ± standard error of the mean (*P < 0·05; **P < 0·01-Kruskal–Wallis).

Developing antigen-specific responses are restricted to BKV

To test further the hypothesis that viral clearance is associated with a measurable change in antigen-specific immunity, we used serial IFN-γ–ELISPOT assays to monitor the development of BKV immunity in patients who were decoy cell- and qPCR-positive at the time of recruitment. In each case, the development of antigen-specific immune responses was associated with patients becoming decoy cell- and qPCR-negative, with intervals ranging from 1 week to 1 month. Conversely, persistent viraemia, despite immunosuppression reduction, was associated with a failure to develop significant antigen-specific responses. Antigen-specific T cell responses were widely variable in range and specificity between individual patients (Fig. 2), but were specific to BKV antigens, with no significant increases in responsiveness to EBV antigens or mitogen detected (data not shown). Similar findings were made in longitudinal studies of two patients with biopsy-proven BKV nephropathy, as illustrated in Fig. 2c, which shows the clinical course of one patient who had persistent viraemia until complete cessation of immunosuppression, which was associated with the development of antigen-specific T cell responses to BKV.

Fig. 2.

Serial enzyme-linked immunospot (ELISPOT) responses to Epstein–Barr virus (EBV) and polyomavirus BK (BKV) peptides. (a,b) Serial ELISPOT responses to EBV and BKV antigens are shown for two patients (a and b represent separate patients) who were BKV-positive at the time of their initial donation (black circles) and then retested once quantitative polymerase chain reaction (qPCR) and decoy cell-negative (red squares). BKV-specific T cell responses developed in each patient, but to different antigens. No significant increases in the responses to EBV antigens [or to phytohaemagglutinin (PHA), data not shown] were detected. Bars represent mean ± standard error of the mean. (c) The development of antigen-specific responses to BKV in a single patient with biopsy-proven BKVN showing the relationship between immunosuppression, viral titre and renal function (upper panel) and general and antigen specific immunity (lower panel) at various times after transplantation. Upper panel shows viraemia by qPCR (blue triangles), serum creatinine (black circles) and immunosuppression (red squares) calculated using the index of Vasudev et al., where one immunosuppressive unit equals 5 mg prednisolone, 100 mg azathioprine, 2 mg tacrolimus or 500 mg mycophenolate mofetil [21]. The lower panel shows spot-forming units/million peripheral blood mononuclear cells (PBMCs) for interferon (IFN)-γ–ELISPOT assays performed at various times after transplantation shown as means ± standard error of the mean (light green = PHA, brown = VP1, purple = small t, blue = VP3, dark green = VP2, light brown = large T and black = BZLF1).

Heterogeneity in the response to BKV antigens

Previous studies investigating the antigen-specific responses to BKV following renal transplantation have typically looked at one or two antigens, usually VP1 and/or large T. Therefore, we wondered if knowledge of the full range of antigenic responses would improve sensitivity and our ability to detect anti-viral immunity in the acute setting. Given the variation in responses seen between patients, which included robust responses to VP2, VP3 or small t but not VP1 or large T, we next investigated how the antigenic responses to different antigens correlated in individual patients. These data are presented as correlation curves (Fig. 3). In patients who had cleared BKV clinically, there was a weak correlation between the responses to VP1 and large T antigens (r² = 0·435, P = 0·038), but no correlation with responses to any other antigens (r² < 0·2 and P > 0·05). Restricting analyses to only one or two antigens (e.g. large T) could result in a failure to detect up to 60% of people with significant antigen-specific responses [> 50 plaque-forming units (PFU)/million PBMCs] to BKV (Supplementary Table S1). Therefore, we conclude that maximum sensitivity to detect meaningful T cell responses to BKV can be achieved only through assessment of the full range of antigens.

Fig. 3.

Heterogeneity in the polyomavirus BK (BKV) response. (a) The correlation between responses to VP1, small t, VP2, VP3 and phytohaemagglutinin (PHA) (ordinate) and large T (abscissa) are shown, along with a linear regression and 95% confidence band (dotted lines) for patients who cleared BKV. Data points represent individual patients. (b) Similar analysis to (a), but with comparison to VP1 (abscissa).

Correlation between immunosuppression and antigen-specific responses

BKV reactivation is widely considered to be a surrogate marker of excessive immunosuppression [22], with an increased incidence of BKV viruria and viraemia associated with the introduction of more potent immunosuppressive regimens [23,24]. Egli et al. recently demonstrated a weak but significant inverse correlation (r² = 0·28) between tacrolimus dose and responses to large T antigen [25]; therefore, we investigated whether a similar correlation existed in our cohort of patients for the large T and all other antigens. In our study there were no significant inverse correlations between large T or VP1 antigen responses and trough tacrolimus levels (r² = 0·138, P = 0·411 and r² = 0·000, P = 0·967, respectively) (Fig. 4), and no correlation between tacrolimus levels and responses to other BKV antigens (r² < 0·007, P > 0·8) or BZLF1 (P = 0·147). Furthermore, no significant correlations existed between any of the antigen-specific responses in patients who cleared BKV infection and the total burden of immunosuppression calculated using the immunosuppressive index defined by Vasudev et al. (where one immunosuppressive unit equals 5 mg prednisolone, 100 mg azathioprine, 2 mg tacrolimus or 500 mg mycophenolate mofetil) [21] (Fig. 4). Previous studies have not shown differential effects of basiliximab or alemtuzumab induction therapy on BKV reactivation [26,27] and there were no differences in our study in terms of likelihood of reactivation (six of 14 basiliximab versus two of 12 alemtuzumab, P = 0·216), although numbers were small.

Fig. 4.

Effects of immunosuppression on antigen-specific responses. (a–c) The correlation between individual patient responses to BZLF-1, large T and VP1 antigens [interferon (IFN)-γ–enzyme-linked immunospot (ELISPOT)/10⁶ peripheral blood mononuclear cells (PBMC)] and trough tacrolimus levels in patients who had cleared polyomavirus BK (BKV). Circles are individual patients. Linear regression (solid line) and 95% confidence bands (dotted lines) are shown. (d–f) Similar analysis to (a–c) comparing antigenic responses and overall immunosuppression using the immunosuppressive (IS) index [21].

Discussion

Serological evidence of previous polyoma virus exposure is found in up to 90% of the adult population [28–30]. Primary infection usually occurs in childhood, after which the virus remains resident in urothelial tissues resulting in intermittent, clinically asymptomatic shedding in the absence of immunosuppression [31]. Of the five polyoma viruses infecting humans, BK virus is associated most strongly with the development of disease following transplantation, where viral replication can lead to ureteric stenosis, tubulointerstitial nephritis (polyoma virus nephropathy) or haemorrhagic cystitis [1,20,32–34]. Although BKV reactivation has been reported after most types of solid organ transplantation [35–38], the incidence is disproportionately high following renal transplantation, due possibly to local tissue damage and ischaemia–reperfusion injury inducing viral replication [39,40].

As the early diagnosis of BKV reactivation may improve outcomes and reduce complications most renal transplant units have introduced surveillance programmes, with associated protocols for immunosuppression reduction or modulation [41]. BKV screening techniques include cytological evidence of viral replication (decoy cells in the urine) and/or qPCR for viral DNA. As viruria predates viraemia, the Oxford Transplant Unit has adopted urinary decoy cell screening to identify ‘at-risk’ patients with clinically relevant BKV replication [6]. Patients with sustained decoy cell positivity or evidence of graft dysfunction progress to have quantitative assessment of viraemia by qPCR. However, although the detection of viral nucleic acid or cytopathic effect can be used to diagnose infection accurately, it does not provide information on the host response to the virus. As clearance of BKV is primarily T cell-dependent [42,43], assays that can quantify the T cell response to the virus could be of value in guiding the titration of immunosuppression in patients who are BKV-positive.

The IFN-γ–ELISPOT assay is a relatively simple and highly reproducible assay that has become a standard clinical laboratory technique for monitoring the cellular immune responses to infections and vaccinations [44]. Although secretion of only a single cytokine is measured, IFN-γ production by T cells has been shown to correlate with functional immunity in a number of clinical settings [19,45–48]. ELISPOT assays are particularly suited to studying responses to polyoma viruses, as unlike other viral pathogens causing disease following transplantation such as CMV, polyoma viruses have a relative small genome (5·3 kb) and express only five antigenic proteins (Supplementary Fig. S2) [49].

While T cell responses to BKV in early transplant recipients have been reported previously using IFN-γ–ELISPOT assays [11,12], none of these studies analysed the full range of BKV antigens. By measuring immunity to a complete panel of BKV antigens, we have been able to demonstrate that the development of BKV-specific immune responses is associated temporally with the resolution of viraemia and viruria. Additionally, the development of these responses is specific to BKV and to individual antigens within the virus, with significant variation occurring between individuals (Figs 1 and 2 and Supplementary Table S1). This is relevant when considering previous studies that have focused on only one or two antigens [11,50], where a failure to detect relevant host immune responses may reflect the heterogeneity of the response. Our findings in the acute situation also complement the recent study by Muller et al., which showed long-term persistence of BKV responses in transplant recipients an average of 12 months after resolution of infection [14]. Additionally, relying on T cell responses to the large T antigen alone to guide clinical decisions may be problematic due to interference by BKV microRNA [51]. In SV40 (the prototypic polyoma virus), expression of this microRNA leads to specific inhibition of the large T protein with a resultant decrease in the cytotoxic host response to virus [52]. For these reasons, as well as the lack of correlation seen between antigen-specific responses to large T and all other BKV antigens except for VP1 (Fig. 3 and Supplementary Table S1), studies reliant upon the identification of responses to large T may fail to detect a clinically relevant host anti-viral immune response.

In agreement with the theory that BKV reactivation is the consequence of excessive immunosuppression, most of our patients with active BKV replication and those in recovery had reduced immune responses to EBV antigens and the mitogen PHA (Fig. 1). This reduction in EBV and PHA responses was not uniform, so differences did not reach statistical significance, but the overall result is suggestive of a widespread impairment in the T cell compartment in patients who develop BKV. While BKV nephropathy was relatively rare in the cyclosporin era, the increased use of tacrolimus has been associated particularly with a resurgence of this disease [23,53]. In our assays there was an inverse relationship between trough tacrolimus levels and responses to the large T antigen, but this was not statistically significant and the strength of this association was diminished further when the overall burden of immunosuppression was analysed (Fig. 4). As this assessment was conducted on patients who had already cleared BKV, who in many cases had already had their immunosuppression reduced, further prospective studies specifically examining the role of individual immunosuppressive agents and antigen-specific responses to BKV in larger groups of patients will help to define if tacrolimus is associated independently with an increased risk of viral reactivation.

In conclusion, our findings demonstrate that ELISPOT assays covering multiple BKV antigens are technically feasible, and can distinguish patients able to clear the virus. The development of significant anti-viral immunity as measured by IFN-γ–ELISPOT may prove valuable in guiding immunosuppression reduction in patients with active polyoma virus replication, with sensitivity in the assay, depending, however, upon the analysis of responses to all the BKV antigens. While further data from prospective studies in larger groups of patients will be required to explore the precise relationship between antigenic responses and the level and form of immunosuppression, our study suggests that an IFN-γ–ELISPOT assay for BKV is robust and discriminating.

Disclosure

All authors declare that they have no conflicts of interest.

Supporting information

Additional supporting information may be found in the online version of this article.

Fig. S1. Demographic data. No significant differences existed between any of the patient groups with respect to subject age (a), time post-transplantation (b), trough tacrolimus levels at the time of donation (c), graft function (d), lymphocyte count (e) or total white cell count (f). There was a trend towards an increased serum creatinine level in patients with current or previous polyomavirus BK (BKV) infections (d), but this was not statistically significant (P = 0·82). Bars represent mean ± standard error of the mean.

Fig. S2. Map of the polyoma virus BK showing the five antigen-coding genes. The genes for the early antigens (large T and small t) are shown in red and those for the late antigens (VP1-3) in blue. The product of the agnogene (black) the agnoprotein, is considered non-immunogenic.

Table S1. Antigen-specific responses in individual patients.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.