Polyomavirus BK with rearranged noncoding control region emerge in vivo in renal transplant patients and increase viral replication and cytopathology

rr-NCCRs are associated with higher plasma BKV load

The NCCR architecture was analyzed by PCR in 129 KT patients, including 70 with PVAN-positive biopsy results, providing a total of 286 same day–matched plasma and urine samples (Table I). The majority of patients had ww-NCCR detectable in plasma (n = 85; 66%) and in urine (n = 119; 92%; Table I, Patients column). rr-NCCRs consisted of larger and shorter amplicons representing insertions (ins-)NCCR in 14%, deletions (del-)NCCR in 42%, and mixtures of ins/ww- and del/ww-NCCR in 6 and 38%, respectively (Table I, rr-NCCR architecture in samples column). The number of patients with rr-NCCR BKV in plasma represented a significant fraction (P < 0.0005, Kolmogorov-Smirnov; see footnote a in Table I, Patients column) compared with the fraction of patients with rr-NCCR in urine (P = 0.45, Kolmogorov-Smirnov). The higher frequency of patients with rr-NCCR in plasma compared with urine was statistically significant (P < 0.0001, χ² test; see footnote b in Table I, Patients column). The same was true when the analysis was restricted to patients with the histological diagnosis of PVAN in biopsies (n = 70; rr-NCCR in plasma 35, 50%; ww-NCCR in urine 67, 96%; P < 0.0001, χ² test). When we compared the NCCR architecture in 129 patients with (n = 70) and without histologically defined PVAN (n = 59; Table I, Patients column), the detection of rr-NCCR in plasma was significantly associated with a positive biopsy (35 of 70 vs. 9 of 59; P < 0.0001, χ² test).

When expressed per sample, ww-NCCR BKVs were detected in the majority of plasma and urine, but rr-NCCRs were significantly more frequent in plasma compared with urine samples (64 and 12 of 286, respectively; P < 0.0001, χ² test; Table I, Samples column). For the difference of 52 samples (18%), we found discordance between rr-NCCR BKV in plasma and ww-NCCR in urine (illustrated for 5 patients in Fig. 1 A). BKV rr-NCCR variants were associated with higher plasma BKV loads compared with ww-NCCR BKV (median 4.36 × 10⁵ vs. 1.97 × 10⁴ geq/ml; P < 0.000002, Mann-Whitney U test; Table I, Samples column). In contrast, no association with ww- or rr-NCCR architecture was apparent for urine BKV loads (median 2.89 × 10⁸ vs. 2.56 × 10⁷ geq/ml; P = 0.098, Mann-Whitney U test). However, urine BKV loads were on average 3,000-fold (range 100–10,000-fold) higher than plasma BKV loads.

To corroborate the association of rr-NCCR and plasma BKV loads, we prospectively analyzed 733 serial BKV DNA–positive plasma samples from 227 KT patients, including 57 cases with documented PVAN (Table I). The results identified the overall rr-NCCR prevalence rate, ranging from 15% per sample to 24% per patient, and confirmed the association of rr-NCCR with higher median BKV loads (rr-NCCR 4.59 × 10⁵ geq/ml and ww-NCCR 1.94 × 10⁴ geq/ml; P < 0.00002; Table I). BKV rr-NCCRs were detected in 113/733 plasma samples (15%) consisting of ins-NCCR in 11% and del-NCCR in 35%, as well as mixtures of ins/ww and del/ww NCCRs in 13 and 41%, respectively (Table I, rr-NCCR architecture in samples column). Again, rr-NCCR in plasma was significantly associated with a PVAN-positive biopsy (26 of 57 vs. 19 of 141; P < 0.0001, χ² test).

We also analyzed 56 BKV-positive urine samples from KT patients without concurrent BKV viremia (median, 14,700 geq/ml; range, 1,000–8.9 × 10⁸) and found ww-NCCR BKV in all cases. When examining 264 healthy blood donors, BKV was only detected in urine, but not in plasma in 26 (9.8%) cases, all of which were of archetype ww-NCCR configuration (median viral load, 2,600 geq/ml; range, 1,000–64,900). We concluded that rr-NCCR BKV variants appeared more frequently in plasma than in urine in viremic KT patients, and were then associated with significantly higher plasma BKV loads. In contrast, the day-matched urine BKV loads were generally 3,000-fold higher than the corresponding plasma BKV loads, regardless of the NCCR architecture. This observation, together with the discordance of rr- and ww-NCCR architecture in day-matched plasma and urine samples provided quantitative and molecular evidence that renal tubular epithelial cells and urothelial cells in ureters and bladder, indeed, reflected different viral replication compartments in immunosuppressed KT patients, thereby extending previous histopathology data (6).

rr-NCCR BKV replace ww-NCCR BKV with persisting viremia

In preceding plasma samples from patients with rr-NCCR variants, we detected ww-NCCR BKV in all but one case. A detailed analysis of 35 PVAN patients with rr-NCCR (Table I) revealed that plasma rr-NCCR BKV variants, indeed, emerged from initial ww-NCCR BKV strains after prolonged viremia (Fig. 1 B). Regression analysis revealed that the time to first rr-NCCR emergence correlated with peak plasma viral load (P < 0.000001; Fig. 1 C). The linear and the quadratic regression models largely overlapped between BKV loads of 5.5–7.5 log10 geq/ml. However, the convex shape of the quadratic regression model provided a better fit of the overall data, suggesting that higher peak plasma BKV loads overproportionally increased the risk of rr-NCCR emergence. Accordingly, rr-NCCRs were detectable in patients with high peak plasma BKV load of 6.5 log10 geq/ml after <4 mo, whereas 100-fold lower viremia (4.5 log10) required >7 times as long. The data indicated that duration and peak replication favored the change from ww- to rr-NCCR BKV variants as the majority species. The only case where rr-NCCR BKV was already present in the first BKV-positive plasma sample was a patient retransplanted after allograft loss and nephrectomy after PVAN. This observation was intriguing, and suggested that impaired BKV-specific cellular immunity is not only critical for curtailing BKV replication (23), but possibly also for recurrence of BKV replication and direct emergence of rr-NCCR variants (24, 32).

rr-NCCR BKV are associated with histological disease progression

Comparing PVAN histology in needle biopsies from 39 patients with either rr-NCCR (n = 21) or ww-NCCR (n = 18) in plasma, we found that the percentage of surface with tubules showing cytopathic changes staining positive for large T-antigen (LTag) was significantly higher in cases with rr-NCCR BKV variants in plasma compared with ww-NCCR BKV (median 15%, mean 26.6 ± 29.4 vs. median 3.5%, mean 11.3 ± 17.1; P = 0.03; Mann-Whitney U test). In addition, cytopathically altered areas from patients with rr-NCCR BKV were more likely to have lymphocytic inflammatory infiltrates compared with ww-NCCR BKV (60 vs. 23.07%, respectively; P = 0.04, Mann-Whitney U test). Finally, detection of rr-NCCR in plasma was associated with PVAN pattern B (Fig. 2 B), whereas ww-NCCR more readily presented as PVAN pattern A (Fig. 2 A; 15/21 vs. 6/18, respectively; P = 0.03, Fischer's exact test).

rr-NCCR are diverse and without corresponding changes in VP1

To characterize the architecture of rr-NCCR in more detail, we sequenced 5 ins-NCCR and 5 del-NCCR from 10 different patients with PVAN. Insertions consisted of partial duplication of the P-Q region, whereas deletions preferentially affected the Q-R-S region (Fig. 3). Thus, NCCR rearrangements changed transcription factor-binding sites and enhancer elements, but no common feature was apparent (Fig. S1, available at http://www.jem.org/cgi/content/full/jem.20072097/DC1). Sequencing of the antigenic determinants of VP1 revealed BKV serotype I and IV in 8 and 2 rr-NCCR variants, respectively. In all cases, the VP1 sequences were identical with the one detected when the majority species were of archetype ww-NCCR (Fig. 3 B and Fig. S2). Thus, NCCR rearrangements were not linked to changes in VP1 serotype.

rr-NCCRs are sufficient to increase early reporter gene expression

To investigate functional effects of rr-NCCR in regard to early and late gene expression, we constructed a bidirectional reporter vector phRG, having the red fluorescent protein (RFP) and GFP in the same positions as the early and late genes in the circular BKV DNA genome (Fig. 3 A). Human embryonic kidney (HEK) 293 cells were transfected and assayed for red and green fluorescence. The expression patterns became detectable as early as 1 d after transfection (dat) and remained stable for at least 10 d of observation (all scored at 2 dat; Fig. 3 B). Archetype ww-NCCR conferred a strong late gene expression (green signal) and a relatively weak early gene expression (red signal, at ∼1/10 of the green signal). In contrast, all rr-NCCRs increased early gene expression by two- to sevenfold, and decreased late gene expression, irrespective of insertion or deletion architectures. Simple deletions or duplications had effects similar to more complex ones, suggesting that the architectural context and not the mere number of transcription factor binding sites might be relevant. We therefore generated two NCCR variants by in vitro mutagenesis that reflected small insertions or deletions of the natural NCCR sequences. Both ins(P24-37)NCCR and del(R8-18)NCCR affected early and late gene expression in a manner similar to in vivo–occurring variants ins(23.1)NCCR or del(15.10)NCCR (Fig. 3 B). As NCCR expression might be affected by host cell type and differentiation, we tested two prototype rr-NCCRs (del 5.3, ins 7.3) in addition to the ww-NCCR (ww 1.4) in cells known to support BKV replication in vitro (Vero monkey kidney cells and human umbilical cord vein endothelial cells) and in vivo (primary human renal proximal tubular epithelial cells [RPTECs]). We found that the expression patterns for all three NCCR prototypes were similar in all tested cell types (Fig. 3 C). Northern blot analysis and RNA protection experiments at 2 dat indicated that NCCR-driven RNA expression correlated with RFP and GFP signals (unpublished data). To exclude vector-mediated positional effects on gene expression, the del(5.3)NCCR was inserted in reverse orientation relative to RFP and GFP (phRG-del[5.3R]). Transfection phRG-del(5.3R) into the different cell types revealed strong early but weak late gene expression, as observed for del(5.3)NCCR, but in opposite colors (Fig. 3 C, compare column 2 and 4). We concluded that rr-NCCR BKV variants were of diverse architecture, but all increased early gene expression in vitro as a common hallmark.

To investigate whether the genomic NCCR rearrangement still allowed for modulatory effects in trans, we examined the responses of the bidirectional reporter vector phRG to BKV LTag, steroids, and agnoprotein expression, which have been reported to positively and negatively regulate NCCR expression (33). As shown for the prototype ins(7.3)NCCR (Fig. 4), increasing LTag expression as detected by Western blot raised late gene expression by 5-fold within 24 h (green, Fig. 4 A). Because glucocorticoid response elements are present in BKV NCCR, we examined whether steroid treatment modulated gene expression in Vero monkey kidney cells that, unlike HEK293, expressed functional glucocorticoid receptors. Treatment with dexamethasone at 10⁻⁶ M increased early and late gene expression from the ww-NCCR (3-fold), and to lesser degrees, also from the del(5.3)- (2.5-fold) and ins(7.3)-NCCR (2-fold; Fig. 4 B). When BKV agnoprotein was cotransfected, both early and late reporter gene expression signals were ∼10-fold reduced for all 3 NCCRs (Fig. 4 C). Collectively, these results indicated that rr-NCCR consistently increased early gene expression relative to archetype ww-NCCR, but still remained susceptible to modulatory effects in trans conveyed by LTag, steroids, or agnoprotein.

rr-NCCR increase BKV replication and cytopathology in vitro

To directly examine the impact of rr-NCCR on BKV gene expression and replication, we generated recombinant BKV strains by cloning each of the indicated 10 rr-NCCR, as well as the archetype ww-NCCR, into the genomic backbone of the BKV Dunlop strain. At 3 dat of recombinant BKV DNA into Vero cells, DNase-protected BKV DNA loads in supernatants were 1–3 log10 higher viral loads for all 10 recombinant rr-NCCR BKV strains compared with ww(1.4)NCCR (Fig. 5 A). More detailed analysis revealed that supernatants of del(5.3)NCCR BKV–transfected cells had 3 and 1.5 log10 higher viral loads at day 3 and 7, respectively, and a pronounced cytopathic effect at day 21 compared with ww(1.4)NCCR BKV (Fig. 5 B). Western blot analysis of the transfected Vero cells demonstrated that LTag expression occurred ∼12–24 h earlier in del(5.3)NCCR BKV, with a corresponding acceleration of VP1 expression compared with ww(1.4)NCCR BKV strains (Fig. 5 C). When Vero cell supernatants were harvested 7 dat and seeded for infection of primary RPTECs, we observed that more cells were infected by the recombinant del(5.3)NCCR BKV than by the ww(1.4)NCCR BKV, with corresponding supernatant viral loads being 100-fold higher (unpublished data). Immunofluorescence staining of RPTECs at day 4 after infection identified a correspondingly higher number of infected cells in del(5.3)NCCR BKV– than in the ww(1.4)NCCR BKV–infected cells (Fig. 5 D, green stain; 10× magnification). Moreover, LTag preceding expression of the late agno gene could be observed at the single-cell level (Fig. 5 D, LTag red nuclear stain, arrows in 40× magnification; agno green cytoplasmic stain) and VP1 expression (not depicted). Thus, we concluded that rr-NCCR were sufficient to confer increased early gene expression, replication, and cytopathology to recombinant BKV in vitro.

Patients.

We characterized BKV NCCR in 1,019 plasma and 342 urine samples from 412 KT patients. Day-matched plasma and urine samples were obtained from KT patients as part of the prospective BKV study protocol approved by the University of Maryland Transplant Center. BKV loads had been determined by real-time PCR in same day–matched plasma and urine samples (n = 286 from 129 patients, including 70 patients with PVAN). The NCCR architecture was determined retrospectively on specimens stored at −20°C. PVAN was graded based on the Banff classification, as recommended by Hirsch et al. (17), as follows: PVAN A, minimal to mild viral cytopathic changes (cy1+–cy2+), insignificant inflammatory infiltrates (<i1+), tubular atrophy (<ct1+), or fibrosis (<ci1+); PVAN B, viral cytopathic changes scoring from mild to severe (cy2+–cy4+) and significant inflammatory infiltrates scoring from moderate to severe (i1+–i3+), but only mild tubular atrophy (≤ci2+) and mild fibrosis (≤ci2+); PVAN C, moderate to severe tubular atrophy (ct3+) and interstitial fibrosis (ci2+–ci3+), variable scores for cytopathic changes (cy1+–cy4+), and for inflammatory infiltrates (i1+–i3+) (17). During the years 2005 and 2006, an additional 733 consecutive BKV DNA–positive plasma samples from 227 patients, including 57 with PVAN, had been submitted for BKV testing to the Molecular Diagnostics Laboratory of the Division of Diagnostics, which is an accredited quality controlled laboratory (EN ISO) of the University of Basel (www.zid.ch). If the samples had detectable BKV loads, BKV NCCR sequences were amplified as an independent confirmatory assay to confirm BKV specificity. Patients with sufficient sampling density defined as >6 consecutive plasma samples collected over >12 wk were identified (n = 35, PVAN in 28) to analyze the dependence of rr-NCCR emergence on time and plasma viral load. NCCR architecture was examined in 56 KT patients (median age 44 yr; range 17–66 yr; 52% female) with urinary BKV shedding, but without concurrent BKV viremia. As controls, we examined BKV in urine and plasma of 264 healthy blood donors (median age 46 yr; range 18–67 yr; 49% female; IRB approval 267/06).

BKV load and NCCR analysis.

BKV DNA was quantified after DNA extraction from 200 μl plasma, urine, or cell culture supernatant using a commercial kit (QIAmp; QIAGEN) and a previously described real-time PCR protocol (1). NCCR amplification used a standard nested PCR approach with the outer primer pairs BKTT5 5′-GAGCTCCATGGATTCTTC-3′ and BKTT6 5′-CCAGTCCAGGTTTTACCA-3′ and the inner primer pair BKTT7 5′-CCCTGTTAAGAACTTTATCCATTT-3′ and BKTT8 5′-AACTTTCACTGAAGCTTGTCGT-3′. The PCR products were compared after 2%TAE-agarose gel electrophoresis with ww-NCCR controls, and excised for sequence analysis, if indicated. Dilution series of increasing ratios of rr-NCCR versus ww-NCCR and cloning into plasmids indicated that the sensitivity of detecting minority species in NCCR mixtures by the PCR protocol was as low as 5%. Patients were scored as bearing rr-NCCR when one or multiple samples showed rr-NCCR. Samples containing rr-NCCR or mixtures of rr- and ww-NCCR (Table I, rr-NCCR architecturein samples column) were scored as rr-NCCR.

Plasmids and primers.

In pTRELTag the cDNA coding for the LTag of BKV is under control of a tetracycline-controlled transactivator-dependent promoter. The intron of the LTag was removed by gene splicing by overlap extension (SOE) (49). Two DNA fragments were generated using primer pairs 1–2 (Fig. S3, available at http://www.jem.org/cgi/content/full/jem.200712097/DC1) and 3–4 and DNA extracted from a BKV-positive urine as template. The two DNA fragments were subsequently joined in a PCR reaction containing primer pair 1–4. The amplification product was digested with ClaI and NheI, followed by ligation into ClaI–NheI-digested pTRE2pur (BD Biosciences). The dual-reporter vector phRG contains hygromycin selectable marker, the enhanced GFP and RFP dsRed2 occurs in opposite directions with both mRNAs, followed by the same β-globin polyA sequence. Plasmid phRG was generated in a three-step procedure. First, the tetracycline response element/minimal CMV promoter cassette was deleted from pTRE2hyg by SOE. Two DNA fragments generated by PCR using primer pairs 5–6 and 7–8 and pTRE2hyg as template were joined in a PCR reaction containing primer pair 5–8. The amplification product was digested with AvrII and BamHI, followed by ligation into AvrII–BamHI-digested pTRE2hyg to yield phyg. Next, BamHI–MluI-cleaved dsRed2 (pdsRed2-C1; BD Biosciences) and ClaI- and SalI-digested EGFP (pEGFP-N1; BD Biosciences) were ligated into BamHI–SalI–restricted pTRE2hyg to obtain pTREdsRed-EGFP. Next, the BamHI–SalI dRed2-EGFP fragment was ligated into phyg cleaved with the same enzymes to yield phdsRed-EGFP. In the final step, the AseI–EcoRV β-globin fragment, blunt ended with T4 polymerase, was subcloned into pGEM-3Zf+ (Promega) to yield pGEMglobin. This plasmid was cleaved with SacI–XbaI, and the resulting fragment was ligated into phdsRed-EGFP digested with SacI–SpeI to obtain phRG. In phRG-NCCR plasmids, the expression of RFP (early) and GFP (late) is under control of the inserted BKV NCCR. NCCR sequences were amplified from patient material by nested PCR using primer pairs 9–10 and 11–12. The MluI–BssHII-cleaved amplicon was ligated into phRG digested with the same enzymes. The plasmids phRGinsP24-37 and phRGdelR8-18 were generated from the archetype ww-NCCR by a duplication of P24-37 or a deletion of R8-18, respectively, by SOE. The two DNA fragments were generated from phRGww by PCR using primer pairs 11–13 and 14–12 (P24-37) or 11–15 and 16–12 (R8-18), respectively, and joined in a PCR reaction containing primer pair 11–12. The amplification products were ligated into MluI–BssHII-digested phRG, as described. To construct recombinant BKV strains, the SacI fragment containing the NCCR was replaced in pBKV Dunlop plasmid (American Type Culture Collection [ATCC] 45025) bearing the full-length genome sequence (50). A synthetic linker (primer 17–18) was ligated into SacI-cleaved pGEM3-Zf+ to yield pGEMSacIlinker. The NCCR released from the indicated phRGww, ins, and del constructs by MluI–BssHII digest was ligated into pGEMSacIlinker cleaved with the same enzymes. Finally, the SacI NCCR fragment was ligated into SacI-restricted pBKV to generate the respective recombinant pBKVww-, ins-, and del-NCCR. pCMVAgno encodes the agnoprotein sequence in the pRC/CMV (Invitrogen). The entire agno sequence was amplified from patient material by PCR using primer pair 19–20. The amplification product was cleaved with BamHI and NotI, followed by ligation into pRC/CMV digested with the same enzymes. The integrity of all plasmid constructs was confirmed by sequencing.

Cell lines.

HUV-EC-C (ATCC; CRL 1730) were grown in Kaighn's FK12 medium (51). HEK293 (ATCC CRL1573) and Vero76 cells (ATCC CRL1587) were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS. Primary RPTECs (Cambrex) were propagated in Renal Epithelial Cell Medium (Cambrex). Infection of RPTECs was performed as previously described (52). HEK293 cells stably expressing BKV LTag were generated by cotransfection of pTET-Off (BD Biosciences) and pTRELTag. Cells were selected and maintained in DME containing 10% FBS in the presence of 500 μg/ml Geneticin (G418; Invitrogen), 1.5 μg/ml Puromycin (Sigma-Aldrich), and 1 μg/ml Tetracycline (Sigma-Aldrich). Transfection of BKV genomes into Vero76 cells was initiated by cleavage of pBKV plasmid DNAs with BamHI, followed by relegation of the diluted DNA. Transfection of cells was performed at 60–80% confluency in 12-well plates and was achieved as specified by the manufacturer using Lipofectamine 2000 or LTX (RPTEC; Invitrogen) at a DNA/Lipofectamine ratio of 0.8:1. Transfection of HUV-EC-C was performed with Transpass D1 reagent (NEB) according to the manufacturer's instructions at a DNA/Transpass D1 ratio of 1.2:1. Medium was replaced after 3–6 h.

Western blot.

Cells were lysed in RIPA-buffer (150 mmol/liter NaCl, 50 mmol/liter Tris-HCl, pH 8.0, 1% NP-40, 0.5% deoxycholate, 0.1% SDS, and protease inhibitors [Roche]). Cell lysates were separated by SDS-PAGE and electrotransferred onto 0.2-μm nitrocellulose membranes (Whatman). Membranes were blocked with phosphate-buffered saline containing 5% nonfat dry milk, incubated with monoclonal mouse anti-SV40 T-antigen, Ab-2 (1:800; Calbiochem), polyclonal rabbit anti-LTag or anti-VP1 (both 1:2,500), or monoclonal anti-tubulin (1:1,000) and washed with phosphate-buffered saline containing 0.1% Tween 20. Anti–mouse or –rabbit horseradish peroxidase–labeled secondary antibody was used at 1:10,000 (Pierce Chemical Co.) or 1:20,000 (GE Healthcare) for detection of bound primary antibody by enhanced chemiluminescence.

Immunofluorescence.

4 d after infection, RPTEC cells were fixed in 4% PFA, and then fixed in methanol. Subsequently, cells were incubated at 37°C with primary and secondary antibodies for 30 min. We used the monoclonal mouse anti-SV40 T-antigen (Ab-2; 1:100) and an IgG-fraction of rabbit antiserum directed against BKV-agnoprotein (1:800) (51). The secondary antibodies were conjugated with Alexa Fluor 568 or 488 (1:500; Invitrogen). Nuclei were labeled in blue with DRAQ5 (Biostatus).

Microscopy and digital image processing.

Microscopy was performed using an epifluorescence microscope (model TE200; Nikon) equipped with suitable filters and a digital camera (Hamamatsu). Red and green cells were counted in a minimum of 3 10× microscopic fields. Images were taken at the times indicated in the figure legends. Digital images were recorded with Openlab 2.2 software and raw images were processed in Photoshop 6.0 (Adobe). Results are shown as the mean of 3 independent experiments yielding a SD of <10%. Confocal microscopy analyses and images were obtained using a microscope (Axiovert 200; Carl Zeiss, Inc.) equipped with a LSM510 confocal module and processed using LSM5 software version 3.2 (Carl Zeiss, Inc.).

Statistics.

The statistics were performed using SPSS-Version 14.0 (SPSS). For categorical values, Fisher's exact test or χ² test were used. Nonparametric tests were used to compare values lacking normal distribution (Mann-Whitney U test, Wilcoxon) where appropriate. All tests were two-sided, unless otherwise indicated.

Online supplemental material.

Fig. S1 shows NCCR sequences. Fig. S2 shows VP1 serotypes. Fig. S3 shows a list of primers used to amplify BKV sequences and generate plasmid DNAs for cloning. The online version of this article is available at http://www.jem.org/cgi/content/full/jem.20072097/DC1.