Understanding at a molecular level, the immunologic response of polyomavirus nephropathy (PVN), a critical cause of kidney graft loss, could lead to new targets for treatment and diagnosis. We undertook a transcriptional evaluation of kidney allograft biopsies from recipients with PVN or acute rejection (AR), as well as from recipients with stable allograft function (SF). In both the PVN and AR groups, Banff histologic scores and immunohistochemical analysis of inflammatory infiltrates were similar. Despite their different etiologies, the transcriptional profiles of PVN and AR were remarkably similar. However, transcription of genes previously linked to AR including CD8 (65.9 ± 18.8) and related molecules IFN-γ(55.1 ± 17.0), CXCR3 (49.9 ± 12.8) and perforin (153.8 ± 50.4) were significantly higher in PVN compared to AR (30.9 ± 2.0, 14.0 ± 7.3, 12.1 ± 7.3 and 15.6 ± 3.8-fold, respectively; p < 0.01). Importantly, transcription of molecules associated with graft fibrosis including matrix collagens, TGFβ, MMP2 and 9, as well as markers of epithelial-mesenchymal transformation (EMT) were significantly higher in PVN than AR. Thus, renal allografts with PVN transcribe proinflammatory genes equal in character and larger in magnitude to that seen during acute cellular rejection. BK infection creates a transcriptional microenvironment that promotes graft fibrosis. These findings provide new insights into the intrarenal inflammation of BK infection that promotes graft loss.
Over the past decade, infection of kidney transplants with BK polyomavirus has become increasingly appreciated (1,2). The infection is manifested by both an inflammatory response and subsequent fibrotic response, leading to renal dysfunction and eventual irreversible graft loss in 40–70% of infected grafts (3,4). Strongly implicated in the pathogenesis of this disease is over-immunosuppression with concomitant tubular injury (5). It remains unclear whether the tubular damage is solely due to viral effects or if it is also augmented by the virally-directed (and perhaps by-stander) immune response. Regardless, the short-term benefits of reduced acute rejection rates over the past decade have not resulted in significant improvements in long-term graft loss, in part due to a reciprocal increase in BK infection (6). Thus, understanding the character of the BK immune response stands to benefit the detection, management and treatment of recipients with BK polyomavirus nephropathy (BK PVN).
Renal biopsy is the gold standard of diagnosis of BK PVN. The histologic hallmarks of this disease are viral cytopathic changes in renal tubular epithelial cells, which occur in medullary and distal tubules in early disease (7), followed by proximal tubules in more advanced stages (8). Other sites of infection reported include vascular endothelium (9) and parietal glomerular epithelium (10). Accompanying these changes is an interstitial inflammatory cell infiltrate with tubulitis. Some clinicians contend that this infiltrative process is an appropriate viral-specific immune response, and that reduction in immunosuppression is warranted. On the other hand, changes seen in PVN are in many cases indistinguishable from the accepted hallmarks of acute cellular rejection, and thus a similar inflammatory response could in another setting be most responsive to enhanced immunosuppressive therapy. Given that diametrically opposed treatments both, in their appropriate settings, result in restoration of renal function suggests that viral-specific infiltrates may have unique characteristics compared to allospecific infiltrates. Understanding distinctions and similarities could elucidate the cellular and molecular events of polyomavirus infection and identify new therapeutic targets and biomarkers of disease. This could also more completely define the pathophysiology of intra-renal inflammation and its relationship to clinical renal dysfunction.
To this end, we evaluated and characterized the events and outcomes of BK polyomavirus infection in our transplant program since its inception in 1999, aided by serial viral monitoring and intra-graft transcriptional analysis. We find that patients with PVN demonstrate significant elevation of transcripts for inflammatory cytokines and CD8+ T-cell cytotoxic molecules like that seen in acute rejection and functionally significant PVN is associated with large viral loads in biopsies. Moreover, viral infection induces a series of gene transcripts associated with graft fibrosis prior to the clinical diagnosis of chronic graft nephropathy. Thus, BK infection of a renal allograft results in substantial intra-graft cellular immune response, similar in character but greater in magnitude to acute rejection that uniquely predisposes toward pro-fibrotic response but is nevertheless responsive to reduced immunosuppression.
Materials and Methods
Patients and biopsy acquisition
All patients enrolled in our Institutional Review Board-approved protocols following informed consent were available for evaluation and inclusion. All immunosuppressive therapy in our program is protocol guided and varies widely based on protocol. Within our program, recipients received either tacrolimus (target trough 8—12 ng/mL) and/or mycophenolate mofetil (2 gm/day) and/or sirolimus (target trough 8—15 ng/mL) and/or prednisone (0—30 mg/day) (Table 1). Some patients also received induction therapy with daclizumab (n = 28) or rabbit anti-thymocyte globulin (ATG) at a dose of 10 mg/kg (n = 35) or 20 mg/kg in the absence of maintenance steroids (n = 18) (11) or Alemtuzumab induction in the absence of maintenance steroids (n = 28) (12). Some patients received prolonged daclizumab induction followed by tacrolimus and mycophenolate mofetil (n = 9) as previously described (13).
Table 1. Clinical characteristics of BK PVN recipients and comparison groups
1p = 0.04 compared to PVN.
2Excludes patients with graft loss.
3ATG = antithymocyte globulin.
Recipient age (years)
44 ± 15
43 ± 4
Donor age (years)
42 ± 10
44 ± 18
40 ± 5
2 ± 1
4 ± 5.0
Time of diagnosis (days posttransplant)
Serum creatinine at time of biopsy (mg/dL)
Serum creatinine 1 yr post biopsy
Daclizumab/prednisone/MMF/tacrolimus (n = 5)
Daclizumab/prednisone/MMF/tacrolimus (n = 5)
Rabbit ATG3/prednisone/MMF/tacrolimus (n = 2)
Daclizumab/prednisone/rapa/cyclosporine (n = 1)
Rabbit ATG/tacrolimus/sirolimus (n = 3)
Solumedrol/prednisone/MMF/tacrolimus (n = 3)
Rabbit ATG/prednisone/MMF/tacrolimus (n = 6)
Rabbit ATG/tacrolimus/sirolimus (n = 2)
Biopsies (16-gauge) were obtained based on protocol surveillance criteria (14) or as clinically indicated. Normal kidney tissue (NK; n = 15) was obtained from living donor kidneys under direct visualization prior to organ procurement by open nephrectomy and was also used for calibration and direct comparison. Patients were assigned to the polyomavirus nephropathy group (PVN; n = 10) when the biopsy demonstrated viral cytopathic changes with intranuclear inclusion bodies, associated renal tubular epithelial cell injury including tubular epithelial cell necrosis and denudation of basement membranes as well as positive immunohistochemical staining for the BK T antigen. Biopsy and transcriptional findings were compared to those derived from a group of patients with clinical acute rejection (AR; n = 17). This group consisted of patients on standard triple immunosuppression therapy who were biopsied to evaluate a rise in serum creatinine of ≥15% from baseline, had at least 1a Banff grade acute cellular rejection without other demonstrable pathology and lacked viral cytopathic changes and T antigen immunostaining, as we have previously described (15). The findings in these two groups were also compared to a group of biopsies from recipients with stable allograft function (SF; n = 10), defined as having a protocol biopsy at least 1 month post-transplantation without a change in serum creatinine (<10% above baseline) and in the absence of any histologic abnormality including drug toxicity, infection, or acute or chronic rejection (15). While there were no significant differences in demographics or time of biopsy between recipient groups (Table 1), serum creatinine at the time of biopsy was lowest in the SF group and highest in AR (p = 0.01).
Biopsy preparation and histology
Biopsies were obtained under ultrasound guidance. A portion of the cortex was macro-dissected and placed into liquid nitrogen at the bedside divided at the bedside as previously described (15). The remaining portion or additional portions for histology were fixed in formalin, sectioned, stained with hematoxylin and eosin, PAS and Masson's trichrome stains. Tissue sufficient for Banff classification was available for all biopsies and was evaluated in a masked fashion by a transplant pathologist (DK).
Immunohistology of kidney grafts
Additional sections of the graft were immunostained as previously described (15) with antibodies specific for relevant lymphocyte markers: LCA (Dako, Carpinteria, CA), CD3 (Dako), CD4 (Novocastra, Burlingame, CA), CD8 (Dako), CD68 (Dako), CD56 (Dako), L26 (Dakocytomation), perforin (Kamiya, Seattle, WA) and granzyme B (Monosan, Burlingame, CA). Polyomavirus infection was evaluated using the anti-SV40 large T-antigen antibody PAB-416 (Oncogene Research Products) which cross reacts with SV40, BK and JC T proteins and C4d staining was performed using the antibody Bi-RC4D (Bionet Inc, Southbridge, MA). The degree of infiltration for each cell phenotype was semi-quantitatively scored from 0 to 6 based on immunohistochemical staining in the renal cortex as previously published (15). Scoring of infiltrating cells was based on evaluation of largest collection and most severe involvement of the biopsy, except for scores of 1 and 2, in which there was an average of 2 or fewer positive cells per 20X field or more than 2 positive cells per 20X field without clustering of cells, respectively. Staining of the tubular epithelium by anti-HLA DR was scored from 0–4 with 0 equivalent to no staining, 1 < 5% of cortical tubules stained, 2 5–25% of cortical tubules stained, 3 with 25–50% staining and 4 equivalent to more than 50% of tubules staining positively. In patients with serial biopsies, only the index biopsy was included in the semi-quantitative analysis.
Quantification of transcripts by real-time polymerase chain reaction (RT-PCR)
Real-time PCR allows for precise relative quantitation of gene transcripts (16,17) directly from renal allograft biopsies (14,15,18). Samples for RNA were snap frozen in liquid nitrogen and processed for cDNA as previously described (14). cDNA (100 ng) from each biopsy was used for RT-PCR. Gene targets were chosen based on potential relevance to acute allograft injury as previously described and included proinflammatory, cytotoxic and T-cell cytokines, costimulatory molecules and growth factors (15), using a 384-well format, Low Density Array (Applied Biosystems Inc., Foster City, CA). Additionally, biopsies were also evaluated for expression of 24 genes that have either been linked to the generation of kidney fibrosis in either transplantation or native renal disease (Figure 6) using a similar low density microarray format (Applied Biosystems). Each target was analyzed in quadruplicate. In addition, primers for 18s ribosomal RNA (an internal control for template input that we have shown does not change based on the clinical conditions under evaluation), were analyzed for each reaction. Reactions were amplified: 50°C for 2 min, 99°C for 10 min, 35 cycles at 99°C for 15 sec and 60°C for 1 min. Quadruplicate reactions for each target gene were normalized to 18s ribosomal RNA, an internal control for template input that we have shown does not change based on the clinical conditions under evaluation (15).
Individual samples were compared to a pooled sample of cDNA from normal kidney (NK) prepared by combining equal amounts of cDNA from each individual within the group to establish a homogenous reference for normal kidney. Data were analyzed using Sequence Detector version 1.7.1 software (Applied Biosystems). Quantification was derived using the comparative threshold cycle method as previously described (14) and reported as an n-fold difference of the experimental sample to the pool of biopsies from NK group. No transcript associations segregated by patient demographics or immunosuppressants.
Real-time PCR assessment of viral load
Urine and plasma were obtained prospectively at all protocol time points and at all medically indicated biopsies. Real time PCR for detecting BK virus using the Light-Cycler (Roche Molecular Biochemicals, Indianapolis, IN) has been previously described (19). In this assay, BKV primers for the partial sequences of VP1, VP2 and VP3 genes are used and are distinct in detecting JC and SV40 viruses. Standards for quantifying BK virus were generated as previously described (19). In each assay, standards of 5, 50, 500 and 5000 copies/reaction mixture are included to generate a standard curve for quantification of positive samples. Each assay run includes both negative control and positive controls (DNA extracted brom BK virus ATCC VR-837). The minimum level of detection is five plasmid copies per reaction mixture. In individuals with detectable viruria, urine and serum screening was increased to weekly for 3 months, then every 3 months. All specimens were tested in duplicate with one reaction containing the positive control. A negative BKV result had to have a positive result for the positive control to be considered valid.
In situ hybridization and PCR analysis of BK PVN kidney biopsies
Renal allograft biopsies were placed in routine formalin fixation and paraffin embedding. For each patient sample, a total of 3—5 μm sections were cut onto silanated slides (Histoserve, Rockville, MD). For each biopsy a total of 8 slides were prepared. Viral DNA was analyzed in infected cells using in situ hybridization as previously described (20). Intrarenal viral concentration was measured by real time-PCR as previously described (21) on a total number of 4 sections. Viral copy number was then expressed as number of copies obtained by PCR per positive cell. AR and SF biopsies (n = 10) were negative for BK virus using this technique.
Significance of differences between immunohistochemical staining scores and Banff histology subscores was determined using the Mann-Whitney U-test. Gene transcript data were normalized by log transformation and compared between the three biopsy groups (AR, PVN, SF) with one-way analysis of variance (ANOVA). Post-hoc inter-group comparisons were made using a Bonferroni correction to appropriately account for multiple comparisons. Significance was defined as a two-sided p < 0.05.
Clinical features of BK PVN recipients
Of all recipients transplanted in our center since the program's inception in June 1999, 10 (8%) were diagnosed with BK PVN based on characteristic histologic features of the graft biopsy. These features included interstitial inflammation as well as viral cytopathic changes including nuclear inclusions in tubular epithelial cells, enlarged irregular nuclei and chromatin smudging, as well as detached tubular cells with denuded patches of basement membrane (Figure 1A). Additionally, all biopsies had positive immunostaining for polyomavirus T antigen. The clinical characteristics of these 10 patients are shown in Table 1. Mean serum creatinine was elevated in these recipients, but was significantly lower than the level seen in AR recipients (p = 0.04). During the time period that these cases were identified, there was no routine surveillance after transplantation to monitor viral replication in urine or serum. Viral DNA measurements in all these cases were made at the time of biopsy diagnosis.
Immunosuppression plays a vital role in the development of BK PVN and it is unclear whether specific agents rather than total immunosuppressive burden is critical. Therapy in these patients included daclizumab induction followed by mycophenolate mofetil, prednisone and tacrolimus (n = 5) or rabbit anti-thymocyte globulin induction followed by triple therapy with mycophenolate, prednisone and tacrolimus (n = 2) or followed by sirolimus, tacrolimus and rapid prednisone taper (n = 3). There was one recipient of simultaneous kidney and pancreas transplant. Trough levels of tacrolimus were 8–12 ng/mL and 8–10 ng/mL for sirolimus. These levels were not substantially different from the target levels in recipients with AR. No cases of PVN were diagnosed in patients treated with alemtuzumab or high dose (20 mg/kg) rabbit ATG induction. These induction strategies were combined with minimized maintenance immunosuppression approaches (11,12).
Histologic evaluation of kidney graft biopsies
All biopsies underwent routine light microscopic evaluation using Banff criteria as well as with immunostaining for identification of lymphocyte phenotypes and cytotoxic markers. Representative photomicrographs of PVN and AR biopsies are shown in Figure 1. Mean Banff subscores from PVN, AR and SF biopsies are compared in Figure 2. Both PVN and AR have evidence of tubulitis and interstitial inflammation of similar magnitude and significantly more severe than in SF biopsies (p < 0.001 and p = 0.004, respectively). While PVN and AR demonstrated similar levels of chronic interstitial inflammation, tubular atrophy was more severe in PVN compared to the AR biopsies (1.5 ± 0.6 vs 0.8 ± 0.7, respectively; p = 0.005) and SF (p = 0.004). Other chronic changes such as glomerulosclerosis and interstitial fibrosis were not significantly different between the groups. Thus, PVN biopsies by Banff criteria have similar levels of inflammation as AR, a feature not useful for distinguishing between these processes as already noted (22)
Cellular infiltrates in PVN and AR are similar in composition and quantity
All biopsies were assessed by immunohistochemical staining for major cellular immune cell subsets as well as cytotoxic markers and HLA-DR and representative sections of immunostaining in PVN and AR grafts are shown in Figures 3A and 3B, respectively. Biopsies were evaluated using a semi-quantitative scoring system as described in the methods (Figure 4). In both AR and PVN groups, T-cell infiltrates were substantial and of similar magnitude and were significantly greater than in SF biopsies. These infiltrates consisted of both CD4+ and CD8+ T cells, with CD4+ cells present in greater frequency than CD8+ cells in both groups. Additionally, there was a moderate component of macrophages within the infiltrate, but not substantially different in intensity between AR and PVN. In contrast to AR biopsies, however, PVN biopsies demonstrated more contiguous and intensely staining clusters of B cells (mean score 4.4 ± 0.2) compared to AR (3.4 ± 0.2; p = 0.003) and which were nearly absent in SF (0.9 ± 0.5; p = 0.004). While there were more NK T cells (3.3 ± 1.2) and less substantial HLA DR staining (1.1 ± 0.3) in PVN biopsies, this was not statistically significant compared to AR (2.0 ± 0.0 and 2.0 ± 0.3, respectively). Finally, the levels of expression of both cytotoxic molecules granzyme B and perforin were also of similar intensity between groups and not significantly elevated compared to SF biopsies. Thus, biopsies with PVN had remarkable qualitative similarities to biopsies in AR recipients. Moreover, the inflammatory cell infiltrate included both B, T and NK T cells as well as macrophages, with a modest increase in B cells compared to the inflammatory cell infiltrates of AR.
In situ hybridization quantitates viral load in PVN biopsies
The extent of viral infection can vary in recipients with PVN and the extent of disease can be difficult to quantitate. Thus, we analyzed viral load within the biopsies by in situ hybridization as described in the methods. Sixteen biopsies from 10 recipients with PVN were evaluated for the expression of BK viral DNA using in situ hybridization and quantitated by in situ PCR as described in the methods. In 9 of 10 recipients with disease, BK polyomavirus was detected by in situ hybridization with a mean of 36.4 ± 19.8 positive cells per section of kidney and a mean of 1.29 × 105± 2.8 × 104 viral copies per cell. Thus, substantial viral replication occurs within individual cells, although there is no clear correlation between viral load and biopsy profile in this limited number of samples.
PVN is associated with marked gene transcript activity
Transcript levels for relevant immune and fibrosis gene targets were assessed in each of the biopsy groups. Summarized in Figure 5 are representative factors with significant deviations from normal, non-transplanted kidneys. Expression in stable function allografts is shown for comparison as previously reported (15). SF biopsies demonstrated relative transcriptional quiescence compared to PVN and AR. There were insignificant levels of T cell and macrophage transcripts including CD3, CD8, IFN-γ, IL-2, IL-10, CD28 (not shown), CD25 (not shown) and RANTES consistent with the modest infiltrates detected by immunostaining (data not shown) (15) and detectable although modest elevations of granulysin and CD154 (Figure 5A). In contrast, PVN and AR biopsies showed marked transcriptional activity (Figure 5A). Nearly all transcripts studied were significantly elevated (p < 0.05) in the extent of expression compared to stable functioning grafts with the exception of IL-2, IL-4 and CD86 in PVN kidney biopsies, and IL-2, IL-4, HLA-DR, granulysin, and TNFα and β, in AR grafts.
Immune response gene transcripts in PVN and AR
Figure 5A demonstrates expression of T-cell related transcripts for selected markers of activation (CD3, IL-2, RANTES, T-bet, IL-4), costimulation (CD154, CD80, CD86), inflammation (TNFα, TNFβ, IL-10) and cell death and cytotoxicity (Granzyme B, LTβ, FasL, Bax, Granulysin). In these broad categories, we saw marked similarity in the extent of transcription between PVN and AR. The T-cell responses in both biopsy groups consisted of a predominant Th1 expression profile with little evidence of Th2 subtypes. Similar to AR biopsies, PVN biopsies demonstrated marked levels of costimulatory markers CD154, CD80 and CD86 and while the levels of proinflammatory transcripts TNFα and TNFβ were more apparent in PVN (20.7 ± 7.5-fold and 10.8 ± 3.2-fold, respectively) than in AR (4.4 ± 1.2-fold and 9.1 ± 3.8-fold), they were not significantly different.
Programmed cell death as depicted by FasL transcription was not significantly different between PVN (35.7 ± 9.8-fold) and AR (165.8 ± 55.6-fold) biopsies but markedly elevated compared to SF (p < 0.001). There was also evidence of functional cytotoxic activity within the infiltrate with 15–30-fold levels of expression of cytotoxic proteins granzyme B and granulysin. Thus, despite differences in the events instigating an immune response within the kidney graft, the transcriptional profile of BK PVN shares many common features to that of AR, with the presence of T-cell activation, as well as dramatic cytotoxic graft destruction and programmed cell death.
However, comparison of these two groups did demonstrate some features that are quantitatively different (Figure 5B). Despite similar levels of immunostaining within the groups, CD8 gene expression was significantly more intense in PVN compared to AR (65.9 ± 18.8 vs 30.9 ± 2.0-fold, p < 0.01). Functional markers of cytotoxic T-cell function IFN-γ and perforin had higher relative levels of expression in PVN (55.1 ± 17.0-fold and 153.8 ± 50.4-fold, respectively) compared to AR (14.0 ± 7.3-fold and 15.6 ± 3.8-fold; p < 0.01). Expression of CXCR3 was also elevated in PVN (49.9 ± 12.8-fold) compared to AR (12.1 ± 7.3-fold; p < 0.001), supporting the presence of an even more intense CD8 functional response within the BK-infected graft than in AR. Finally, transcripts of HLA-DR in PVN biopsies (19.6 ± 7.6-fold) were substantially higher than in AR (2.8 ± 1.0-fold; p = 0.02) and was in contrast to the lower level of protein expression seen by immunohistochemistry in PVN compared to AR.
PVN induces a pro-fibrotic response within the graft of greater magnitude than AR
The prominence of tubular atrophy within the PVN grafts suggests a relationship between graft injury and graft loss due to infection. We thus examined a series of gene transcripts that have been associated with fibrosis generation and regulation both in transplanted grafts and native kidney disease. The results are shown in Figures 6A and 6B. Within both AR and PVN grafts, there were increased levels of transcripts of genes related to matrix proteins collagen I, IV, fibronectin and vimentin compared to SF grafts (Figure 6A). Moreover, in PVN, the levels of collagens I, IV and fibronectin were significantly elevated compared to AR (p = 0.03, 0.04 and <0.01, respectively). Expression of growth factors associated with fibrosis were noted for a significant increase in TGFβ in both PVN and AR (9.4 ± 4.5 and 3.2 ± 1.7–fold) compared to SF (p = 0.001), with a small but significant increase in transcription of CTGF, a downstream effector of TGFβ in PVN compared to SF (p = 0.02). However, there were no significant differences in FGF2, PDGFβ, IGF1 and VEGF between AR, PVN and SF.
The regulation of this fibrotic response was also studied (Figure 6B). There were marked increases in MMP-2 and MMP-9 in PVN (4.6 ± 2.5 and 87.8 ± 65.1-fold) compared to AR grafts (0.3 ± 0.1 and 11.8 ± 6.5-fold; p = <0.001 and p = 0.002, respectively). While expression of BMP-7, an antagonist to TGFβ-induced fibrosis within the kidney (23), was increased significantly in PVN (1.1 ± 0.8-fold) compared to AR (0.4 ± 0.4-fold), this level was similar to SF and normal kidney. This is in spite of the relatively marked activation of profibrotic markers in PVN, suggesting dysregulation of matrix metabolism within the graft.
Recently, epithelial-mesenchymal transformation (EMT) has been appreciated as a process that mediates late graft deterioration following transplantation (24). We evaluated whether the extent of tubular atrophy in PVN could be related to EMT via the expression of E-cadherin, alpha-smooth muscle actin and S100A4. Transcript levels for E-cadherin were not significantly different in all three groups. While transcript levels for α-SMA, a marker for myofibroblasts, were significantly higher in PVN compared to AR (p = 0.04), they were not substantially different than in SF. However, S100A4, a mesenchymal marker associated with deteriorating kidney graft function (24,25), was significantly higher in PVN grafts (8.5 ± 3.5-fold; p < 0.001) than AR (2.4 ± 0.9) and SF (2.4 ± 0.9), suggesting an association between EMT induction in PVN grafts and long-term graft failure.
Despite the critical role that PVN may play in long-term graft loss, there is surprisingly little known about the immunology of this infection or its acute or chronic effects on the allograft. Recognizing this entity early is crucial to initiate appropriate treatment (26). There is limited effective anti-viral therapy although a number of centers have adopted the use of cidofovir or leflunomide, but recent in vitro studies suggest that these agents have limited therapeutic efficacy (27) while having significant and potential clinical toxicities. The current standard of care is to reduce total immunosuppressive load while monitoring for acute rejection. Often the inflammation of PVN is substantial and distinguishing it histologically from AR may be difficult. Reduction in immunosuppression may result in rising serum creatinine and repeat biopsy showing ongoing or worsening inflammation. Treatment with immunosuppressants may aggravate disease by supporting viral replication. Thus, a clinical conundrum exists in managing these patients. Identifying the overriding disease and potential treatment targets in the case of PVN are critical questions that remain in transplantation. Thus, part of the goal of these studies was to characterize the transcriptional profile of PVN to better identify the cellular and molecular events of BK polyomavirus infection and this is the first such description undertaken in the literature. Our intent was also to identify novel markers and thus identify therapeutic strategies and potentially unique targets of intervention.
There is remarkable similarity of transcriptional expression between PVN and AR, and in some ways this is predictable. During infection, host cell receptors, as yet unidentified, facilitate viral attachment and cellular infection. Natural killer cells of the innate immune system provide initial antiviral cytotoxic activity and subsequent adaptive immune responses follow with CD8+ cytotoxic T-cells mediating MHC class I restricted antiviral responses (28). While immune staining did not detect significant differences by histology between AR and PVN, a strong presence of CD8 function and infiltration via IFNγ and perforin transcripts were clearly elevated in PVN, even more so than in AR, with a strong CXCR3 expression related to IFN-γ. The differences not only point out a critical aspect of PVN infection, but also note the limitations to immunostaining as a sole marker of disease. Moreover, the use of non-invasive monitoring diagnostic testing for AR using urinary mRNA expression for cytotoxic markers granzyme B and perforin (29) or CXCR3 (30) may not adequately distinguish AR from PVN, and could lead to misdiagnosis in recipients in the absence of biopsy or viral quantification. While prior studies suggest that such markers can distinguish AR from a urinary tract infection, these were specifically bacterial in orgin and not viral (31). What is not clear is whether other viral instigators of inflammation and immune activation may result in a similar molecular response and that any particular perturbation within the graft may result in a similar pattern of expression.
Despite similar levels and types of acute inflammation mediating PVN and AR, the chronic effects of these diseases may differ considerably. Our studies are the first to prospectively highlight the transcriptional profile of developing fibrosis in PVN. By light microscopy, we detected a significant increase in the extent of tubular atrophy in PVN compared to AR with no substantial differences in the extent of fibrosis or chronic inflammation. At this early stage, prior to a diagnosis of chronic allograft nephropathy, it is notable that PVN biopsies demonstrated a ‘pro-fibrotic’ microenvironment with significantly increased expression of TGFβ, an important inducer of EMT, CTGF a downstream effector of TGFβ, MMP-2 and MMP-9 which regulate not only matrix metabolism but also inflammatory processes within the kidney (32), as well as marked expression of matrix proteins more so than compared to AR. Induction of EMT in kidney grafts has been associated with progressive graft failure manifested by tubular atrophy and interstitial fibrosis (24,25). Our studies demonstrate that a transcriptional gene associated with EMT is activated in PVN grafts and while we did not detect increases in αSMA, this is a mesenchymal cell marker which is associated with more established fibrosis. Reduction in E-cadherin protein expression in tubules is typically expected in EMT, but transcript levels here were not reduced. Understanding the induction of this mechanism or inappropriate dysregulation of repair mechanisms is critical to abrogate the negative effects of viral infection in promoting long-term graft loss.
The small number of samples studied here precludes our ability to determine whether these transcriptional differences may be useful in making the clinical discrimination between PVN and AR. Moreover, a possible role of NKT cells based on marked expression of NK-related transcripts such as granulysin, CD8 and perforin in viral biopsies could be explored further with novel discovery of other targets. With that being said, alternative approaches to studying PVN would require the characterization of a larger set of test samples with unambiguous diagnosis and then applied to samples where the diagnosis may be in question. The prospective use of transcriptional data to help guide clinical care can only be attempted after the technique has been defined in the context of established diagnostic techniques. A more global gene transcript evaluation using chip technology might provide additional candidate genes between PVN and AR. In order to obtain productive and consistent comparisons, stage of PVN disease based on histologic criteria would be required (26). Other mechanistic studies might include quantifying alloreactive T-cell populations simultaneously to anti-BK response using tetramer analysis. In vitro investigation has been limited by the inability to develop persistently infected cell lines (33). Finally, animal models of BK PVN have had limited acceptability as polyomaviruses maintain species specificity and results in primates may not be entirely applicable to human disease (34). Murine models have necessitated simultaneous ischemic renal injury with limited technical feasibility in a kidney transplant model (35).
In summary, recipient kidney grafts infected with BK polyomavirus transcriptionally express a wide variety of proinflammatory markers, with expression of T-cell activation and cytotoxicity transcripts. Associated with inflammation are the presence of programmed cell death and cytotoxicity, and the induction of gene transcripts associated with either overproduction or underutilization of matrix proteins. While this profile is not substantially different from that seen in cellular acute rejection of the graft, our studies provide new insights into the inflammatory response of a viral infection that promotes graft loss, similar to the inflammatory response of untreated and uncontrolled rejection.
These studies were supported by the Division of Intramural Research, NIDDK. We wish to thank the expert technical assistance of Gary Fahle and Steven Fisher of the Warren Magnuson Clinical Center Laboratory, as well as the nursing and research staff of the Organ Transplant Program of NIDDK. Presented in part in abstract form, American Transplant Congress, 2004.