Detection of JC virus DNA and proteins in bone marrow of HIV-positive and HIV-negative patients

Abstract

Background

We sought to determine the prevalence of JCV in bone marrow samples from HIV-positive and HIV-negative patients and whether the bone marrow is a site of latency and neurotropic transformation of JC virus, the agent of progressive multifocal leukoencephalopathy(PML).

Methods

We collected bone marrow aspirates, archival bone marrow, blood, and urine samples from 75 HIV-negative and 47 HIV-positive patients without PML, as well as bone marrow, and urine or kidney samples from 8 HIV-negative and 15 HIV-positive patients with PML. Samples were tested for JCV DNA by quantitative PCR and JCV protein expression by immunohistochemistry. JCV regulatory regions (RR) were characterized by sequencing.

Results

JCV DNA was detected in bone marrow specimens of 10/75(13%) of HIV-negative, and 22/47 (47%) of HIV-positive patients without PML, but in 3/8(38%) HIV-negative and 4/15(27%) HIV-positive patients with PML. JCV DNA(2-1081 cps/μg cellular DNA) was detected in multiple leukocyte subpopulations of blood and bone marrow samples. JCV large T antigen, but not VP1 capsid protein, was expressed in bone marrow plasma cells. Bone marrow JCV RR sequences were similar to those usually found in brain of PML patients.

Conclusions

The bone marrow is an important reservoir and possible site of neurotropic transformation for JCV.

Introduction

JC virus(JCV) [1] causes progressive multifocal leukoencephalopathy(PML), a deadly demyelinating disease of the central nervous system(CNS), in immunocompromised patients[2, 3]. Primary infection is asymptomatic and occurs in late childhood [4]. Approximately 86% of healthy adults are seropositive for JCV[5] and the virus remains quiescent in healthy individuals. In immunocompromised hosts, however, JCV can reactivate and cause a lytic infection of oligodendrocytes, resulting in PML.

The events leading to JCV reactivation and pathogenicity remain incompletely understood. While JCV coding region is very well conserved, its non-coding regulatory region(RR) is hypervariable, and has been associated with neurovirulence. Indeed, a stable, non-pathogenic RR is usually found in urine samples from healthy and immunosuppressed individuals and has been called the “archetype” RR. Conversely, RR isolated from CNS of PML patients contain deletions and duplications, as seen in the RR of JCV prototype Mad-1[6].

Hematogenous spread is the likely mode of JCV dissemination into the central nervous system. Indeed, although healthy individuals usually do not carry JCV in their blood, several studies have shown JCV to be associated with peripheral blood leukocytes and cell-free plasma in immunosuppressed people [7-10]. However, the phenotype of cells carrying JCV in the bloodstream remains undetermined, as JCV has been found in association with B cells as well as other leukocyte subpopulations[11]. This controversy calls for further examination of the interactions between JCV and the hematopoietic system.

Little is known about JCV prevalence in the bone marrow of healthy individuals as well as immunocompromised patients. In 1988, Houff et al first demonstrated JCV DNA and capsid protein in bone marrow mononuclear cells from two PML patients [12]. Thereafter, JCV DNA detection has been reported in 21/54(39%) of tonsils[13], as well as in twelve cases of bone marrow samples, from PML patients[12, 14], leukemia patients[15], and bone marrow transplant recipients[16]. More recently, the potential role of bone marrow as a reservoir of JCV has been highlighted by the occurrence of PML in four patients with multiple sclerosis and one with Crohn’s disease treated with the immunomodulatory drug natalizumab (Tysabri)[17-20]. Indeed natalizumab blocks α4 integrins, and promotes release of immature leukocytes from bone marrow into the bloodstream. It has been hypothesized that these cells may be infected by JCV[21, 22], leading to viral dissemination to the CNS. We have previously described one patient with PML who had JCV RR sequences in bone marrow usually found in CNS isolates of the virus[23]. In an attempt to resolve a central question in JCV pathogenesis, we therefore sought to characterize the determinants of JCV latency and reactivation in human bone marrow.

Materials and Methods

Specimen collection

This study was approved by our Institutional Review Board. Altogether we have collected specimens from 145 individuals. These included bone marrow aspirates, archival bone marrow, blood, and urine samples from 75 HIV-negative and 47 HIV-positive patients without PML, as well as bone marrow, and urine or kidney samples from 8 HIV-negative and 15 HIV-positive patients with PML.

Fresh bone marrow aspirates were obtained from 34 adult patients without PML (28 HIV-negative, 6 HIV-positive), who required bone marrow examination as part of their clinical care. Available peripheral blood samples (n=24) and urine samples (n=17) were also collected. Bone marrow aspirate, peripheral blood and urine were collected from 3 HIV-negative patients with PML. Peripheral blood leukocytes (PBL) were isolated from blood samples using Ficoll gradient centrifugation.

Formalin-fixed, paraffin embedded archival bone marrow samples from 2003 to 2007 were retrieved on 93 patients, including 47 HIV-negative and 41 HIV-positive patients without PML, and 5 HIV-negative patients with PML.

Formalin-fixed paraffin embedded autopsy samples of vertebral bone, brain, and kidney from 15 HIV-positive patients with PML were obtained from NeuroAIDS Consortium.

Cell sorting of bone marrow and peripheral blood leukocytes

From the 34 subjects without PML, 25 of the 34 bone marrow samples and 17 of the 24 peripheral blood samples were stained with fluorescent antibodies targeting cell surface markers: CD3, CD19, CD20, CD16, CD56, and CD34. These cells were sorted using 5-color flow cytometry apparatus (FACSVantage, Becton-Dickinson) and bone marrow aspirate and peripheral blood from 2 patients were sorted using immunomagnetic beads by AutoMACS(Miltenyi Biotec). Sorted cell subpopulations included: B(CD19⁺, CD20⁺), T(CD3⁺), natural killer(NK)(CD3^-, CD16⁺, CD56⁺), natural killer T(NKT)(CD3⁺, CD56⁺), polymorphonuclear (PMN)(forward and side scatter), monocytes(forward and side scatter and granularity), and pluripotent stem cells(CD34⁺).

DNA extraction from fresh clinical samples

Sorted bone marrow and peripheral blood cell fraction were resuspended in 200μl of PBS. DNA extraction was performed using QIAgen DNA blood mini kit (QIAgen). For urine, 200μl of fluid was taken from the samples and used for direct extraction with the QIAgen minElute kit (QIAgen).

DNA extraction from archival bone marrow samples

Ten slices, each 5μm thick, from the formalin-fixed, paraffin-embedded bone marrow blocks were collected into an eppendorf tube and a new microtome blade was used for each block. DNA was extracted after deparaffination in 100% xylene for 10 minutes, and then in 50% xylene and 50% ethanol for 10 minutes, followed by 100% ethanol for 10 minutes twice, all performed at 56°C. The dried sample was dissolved in 200 μl of Tissue Lysis solution, part of the QIAgen DNeasy blood and tissue kit(QIAgen). DNA was extracted following the kit instructions.

Quantitative JCV PCR (QPCR)

We used QPCR to detect and quantify JCV DNA in all samples as previously described[24]. The limit of detection of the assay was 100 copies of JCV DNA/ml of urine, or 2 copies of JCV/μg of PBL DNA.

Cloning and sequencing of JCV RR

We used primers JC 5′- ATT CAT TCT CTT CAT CTT GTC TTC GTC CCC ACC TTT AT-3′(nt 4846-4883) and JCR 422 5′-TTT TTC CCG TCT ACA CTG TCT TCA CCT G-3′(nt 425-398). For each PCR reaction 20 pmol of primers were used in 30 cycles of amplification with an annealing temperature of 63°C. The PCR products corresponding to the expected size were cloned with a TOPO-TA Cloning Kit (Invitrogen, Carlsbad, CA). 10 clones for each PCR product were analyzed by restriction enzyme digestion and electrophoresis. We sequenced ten clones per each amplified fragment length. The DNA sequence was obtained on an ABI 3730xl sequencer (Applied Biosystem). Sequence analyses were performed using the Lasergene Software MegAlign 7.1(DNA STAR, Madison, WI).

Immunohistochemistry staining (IHC)

IHC staining used the following antibodies, VP1(PAB597, a generous gift from Dr Walter Atwood), large T-antigen (SV40 Tag v-300, Santa Cruz Biotechnology), CD20, CD8, CD4, CD3, CD138(Dako) as previously described[25]. Antigen retrieval was performed in boric acid (0.2M) at 60 °C overnight.

Statistical analysis

Correlation between JCV DNA detection and CD4 counts, HAART usage, or HIV viral load was performed using two-tailed student-t test. Comparison between JCV DNA detection in the bone marrow samples of HIV-positive and HIV-negative patients was performed using Fischer’s exact test, two tailed.

Results

Detection of JCV DNA in fresh bone marrow aspirates

We tested fresh bone marrow aspirates from 34 patients without PML(Table 1). Using QPCR analysis we found that 10/34(29%) of these bone marrow aspirates had at least one cell subpopulation that was positive for JCV. Furthermore, 6/24(25%) matching peripheral blood samples obtained at the time of bone marrow aspirate had at least one cell subpopulation containing JCV DNA. Finally 7/17(41%) urine samples from these patients had detectable JCV DNA. Every patient with detectable JCV DNA in PBL also had detectable JCV in bone marrow. However, 4 patients with JCV-positive bone marrow aspirate did not have detectable JCV DNA in PBL. JCV presence in urine was not correlated to detection in bone marrow and peripheral blood. Only 2 of the 7 positive urine samples were from patients with detectable JCV DNA in bone marrow or PBL. Among these 34 patients, 6(18%) were HIV-positive. Three of these 6 patients had detectable JCV in bone marrow (50%) compared to 7/28(25%) HIV-negative patients. Two of these three HIV-positive patients with detectable JCV DNA in bone marrow also had detectable JCV in PBL.

Table 1.

Detection of JCV DNA by PCR in fresh clinical samples

No. positive/total No. tested
Diagnosis	Bone marrow aspirates	Peripheral blood leukocytes	Urine
Lymphoma	4/13	4/11	6/10
MM	1/4	0/3	NA
Leukemia	0/4	0/2	1/1
MGUS	2/5	1/3	0/2
Anemia	1/4	1/3	0/3
Infections	1/2	0/1	0/1
Thrombocytopenia	1/2	0/1	NA
Total	10/34 (29%)	6/24 (25%)	7/17 (41%)

MM, multiple myeloma; MGUS, monoclonal gammapathy of unknown significance, NA, not available

Of the 10 bone marrow aspirates with detectable JCV(Table 2), the first 8 were sorted by flow cytometry and samples #9 and #10 were sorted by AutoMacs. When available, peripheral blood leukocytes were also sorted by the same set of surface markers and using the same methods. JCV DNA was detected in multiple cell subpopulations including PMN, monocytes, B, T, NK, NKT cells, as well as the unsorted fraction and the fraction remaining after sorting. JCV DNA detection was not associated with any particular subpopulation. Overall, the JC viral load in both bone marrow and PBL were low. The highest JC viral load in bone marrow was 250 copies/μg cellular DNA in NK cells of patient #3. The lowest JC viral load was 2 copies/μg cellular DNA in the unsorted samples of both patients #2 and 5. The highest JC viral load in peripheral blood was 1081 copies/μg in the remaining fraction of patient #6. Detection of JCV DNA in one cell type from the bone marrow did not always correspond to its presence in the same cell type in the peripheral blood of the same patient. HIV-positive patient #5, with anemia had five sorted cell subpopulations that contained JCV DNA.

Table 2.

Quantification of JC viral load by QPCR in sorted cell subpopulations from bone marrow aspirates and peripheral blood leukocytes

Patient # Diagnosis	1 MGUS		2 MGUS		3 Infection		4 MM		5 Anemia (HIV+)		6 Thrombo (HIV+)		7 Lymphoma		8 Lymphoma		9 Lymphoma		10 Lymphoma (HIV+)
Copies of JCV/μg
Sample	BM	PBL	BM	PBL	BM	PBL	BM	PBL	BM	PBL	BM	PBL	BM	PBL	BM	PBL	BM	PBL	BM	PBL
Unsorted	-	NA	2	-	48	NA	-	NA	2	13	-	265	-	14	-	-	7	NA	-	NA
PMN	-	NA	8	-	-	NA	47	NA	-	-	NA	NA	-	-	18	-	NA	NA	NA	NA
Mono	-	NA	-	-	69	NA	NA	NA	-	-	NA	NA	178	-	-	-	NA	NA	NA	NA
B	172	NA	-	30	-	NA	72	NA	-	-	36	-	-	-	-	88	-	NA	-	NA
T	-	NA	-	29	219	NA	-	NA	37	32	NA	NA	-	-	43	69	NA	NA	NA	NA
NK	-	NA	38	-	250	NA	NA	NA	44	-	NA	NA	-	97	-	105	NA	NA	NA	NA
NKT	-	NA	-	56	-	NA	NA	NA	84	96	NA	NA	-	-	-	-	NA	NA	NA	NA
CD34+	-	NA	-	-	-	NA	NA	NA	-	-	-	NA	-	-	-	-	NA	NA	NA	NA
Remaining	-	NA	-	-	-	NA	-	NA	121	-	-	1081	-	-	-	-	-	NA	10	NA

-, JCV DNA not detected; PMN, polymorphonuclear cells; Mono, monocytes; MGUS, Monoclonal Gammapathy of Unknown Significance; MM, multiple myeloma; Thrombo, Thrombocyptopenia; BM, Bone marrow ; PBL, Peripheral blood leukocytes; NA, not available

JCV DNA detection in archival bone marrow samples from HIV-positive and HIV-negative patients

We sought to further explore the impact of HIV infection on JCV detection in bone marrow samples. We identified 88 patients (41 HIV-positive, 47 HIV-negative controls) with archival bone marrow samples, performed at Beth Israel Deaconess Medical Center from 2003 to 2007. The indications for bone marrow biopsies in HIV-positive patients included: lymphoma(n=20), pancytopenia(n=8), anemia(n=6), polycythemia(n=2), MGUS(n=2), leukopenia(n=1), splenomegaly after solid organ transplant(n=1), and these were matched with selected HIV-negative controls(Table 3). The median age of HIV-positive patients was 42(range 31-69) compared to 55(range 24-89) for HIV-negative individuals. HIV-positive patients had a median CD4 count of 157/μl(range 2-928), 75% were on treatment with highly active antiretroviral therapy (HAART) and 27% had undetectable HIV plasma viral load. In the HIV-positive group, 19/41(46%) bone marrow samples had detectable JCV DNA by QPCR, compared to 3/47 samples (6%) in the HIV-negative group(p<0.01). Interestingly, among patients with lymphoma, 10/20(50%) were JCV DNA positive in the HIV-positive cohort compared to only 1/22(5%) in the HIV-negative cohort(p=0.01). Furthermore, CD4 count, HAART usage, and peripheral HIV viral load did not significantly correlate with positive JCV detection in the bone marrow.

Table 3.

PCR detection of JCV DNA in archival bone marrow samples from HIV⁺ and HIV^- patients

	HIV+	HIV-
Indication for BM	no. positive/ total no. tested		p
Anemia	2/6	1/8	ns
Splenomegaly	1/1	0/1	ns
Polycythemia	0/2	0/2	ns
Thrombocytopenia	0/1
MGUS	0/2	0/2	ns
Leukopenia	1/1	0/5	ns
Pancytopenia	5/8	1/7	ns
Lymphoma	10/20	1/22	0.001
Total	19/41 (46%)	3/47 (6%)	<0.01

BM, bone marrow biopsy; MGUS, monoclonal gammapathy of unknown significance; ns, not significant; p value indicate significance using Fischer’s Exact Test, two tailed.

Detection of JCV in PML patients

To determine whether JCV was present in bone marrow of patients with active JCV replication in their brain, we tested 8 HIV-negative and 15 HIV-positive cases with PML. Bone marrow samples from the 8 HIV-negative PML patients (Table 4), including 3 fresh bone marrow aspirates and 5 archival specimens were obtained 7-139 days post PML diagnosis. Of 8 bone marrow samples, 3 (38%) had detectable JCV DNA. We obtained matching peripheral blood samples from 7 of these HIV-negative PML patients and urine samples from 5 of them. Of the 3 patients with detectable JCV DNA in bone marrow, 2 had also detectable JCV DNA in PBL, and only one had detectable JCV DNA in urine. In the 15 HIV-positive PML patients, archival vertebral bone and kidney specimens obtained at autopsy were tested. Of those, 4 (27%) patients had detectable JCV DNA in bone specimens and only one of them had detectable JCV DNA in the kidney. Separately, 4 others had detectable JCV DNA in kidney specimens only.

Table 4.

Detection of JCV DNA by QPCR in clinical samples from HIV⁺ and HIV^- PML patients

	Bone marrow	PBL	Urine	Kidney
No positive/total no tested
HIV-/PML	3/8	2/7	4/5	NA
HIV+/PML	4/15	NA	NA	5/11
Total	7/23 (30%)	2/7 (29%)	4/5 (80%)	5/11 (45%)

PBL, peripheral blood leukocytes; NA, not available;

Grouping all of our samples by HIV and PML status, we detected JCV DNA in the bone marrow specimens of 10/75(13%) of HIV-negative and 22/47(47%) of HIV-positive patients without PML (p<0.001), and 3/8 (38%) HIV-negative and 4/15 (27%) of HIV-positive patients with PML (p=0.65).

JCV protein detection in archival bone marrow samples

To determine whether JCV remained quiescent in bone marrow or underwent a replicate cycle and expressed protein, we then performed IHC staining on all archival samples that had detectable JCV DNA by QPCR. None of the samples had detectable JCV VP1 capsid protein. However, JCV large-T antigen (TAg) was found in 6/19(32%) of HIV-positive samples and 2/3(67%) of the HIV-negative samples that had detectable JCV by QPCR, respectively. QPCR negative samples were used as controls for TAg staining. To characterize the specific cell type supporting JCV TAg expression, we performed double IHC with cell markers on those samples that were positive for TAg. These experiments showed that TAg was mostly present in cells that also stained for CD 138, a plasma cell marker (Figure 1).

Figure 1.

Double immunostaining for JCV T Ag and CD138⁺ plasma cells in archival bone marrow sample from an HIV-POSITIVE patient without PML. Three CD138⁺ plasma cells (red) express JCV T Ag (brown, arrows) next to one uninfected plasma cell (red only, asterisk), and one T Ag+ CD138-negative cell (brown only, arrowhead). Bar=50μm

Characterization of JCV regulatory region sequences in bone marrow

To determine the type of the JCV RR present in archival bone marrow samples, we sequenced the RR after PCR amplification and cloning. We were able to amplify 6 of the 19 QPCR positive HIV-positive, and 1 of the 3 QPCR positive HIV-negative formalin-fixed paraffin-embedded samples. All 7 RR had tandem repeats of the 98 base pair(bp) element similar to the Mad-1 JCV prototype. However, each patient sample contained unique point mutations in either the first and/or second 98 bp repeat element or in the agno gene. We were not able to obtain sequences from JCV RR in fresh bone marrow aspirates or peripheral blood samples.

Discussion

In our quest to better understand the mechanisms that lead to JCV pathogenesis and dissemination, our goals were to determine the prevalence of JCV in bone marrow and to characterize the phenotype of JCV-infected cells. We therefore combined a highly specific cell sorting method with a very sensitive QPCR assay, which demonstrated that 29% fresh bone marrow aspirates contained JCV DNA. The bone marrow is an ideal site for long term viral latency as seen with the herpes viruses, Epstein-Barr virus and Cytomegalovirus [26, 27]. Furthermore, the bone marrow contains progenitor and mature cells that are known to be susceptible to JCV infection. Indeed, JCV DNA has been found in B lymphocytes in peripheral blood [7]. In vitro, JCV can infect both primary CD34⁺ cells and primary tonsillar B lymphocytes[9]. Lastly, nuclear protein extracted from B lymphocytes can bind to the nuclear factor-1(NF-1) sites in the JCV regulatory region to promote JCV replication[10]. Our demonstrations of a high prevalence of JCV in bone marrow ascertain that the bone marrow compartment harbors JCV. Moreover, since all subjects with JC viremia also had JCV in bone marrow, the virus may potentially spread throughout the body via the hematogenous route.

HIV infection is the major risk factor for PML, and approximately 80% PML patients have AIDS [28, 29]. Our data indicate that JCV is significantly more prevalent in bone marrow of HIV-positive than HIV-negative patients with similar hematological conditions. One possible explanation is that HIV-positive patients experience a prolonged and more severe immunosuppression that may promote JCV reactivation. Others have shown evidence of molecular interactions, either directly or indirectly, between HIV and JCV [30-34]. Since HIV presence in bone marrow has been demonstrated [35, 36], it is therefore possible that JCV and HIV interplay in this compartment is one important factor of JCV neurotropic transformation.

In the bone marrow, JCV was detected in sorted cell groups including, PMN, monocytes, B, T, NK, NKT cells, and what remained after sorting, which include progenitor cells prior to expression of mature cell markers, and hematological abnormal cells. This is consistent with our previous observation that JCV DNA can be associated with multiple leukocyte subpopulations in peripheral blood[11]. Interestingly, sorted CD34⁺ cell group did not have any detectable JCV. However, our ability to detect JCV DNA in the CD34⁺ cell subpopulation may have been limited due to the small percentage of CD34⁺ cells in the human bone marrow [37]. The sorted CD34⁺ cell group always contained ten to one hundred times fewer cells than the other groups.

To determine whether bone marrow harbors quiescent JCV or sustains a productive infection by this virus. We performed IHC staining of the archival bone marrow samples of HIV-positive and HIV-negative patients showing expression of JCV TAg in a third of those with detectable JCV DNA, but not the VP1 protein. TAg is a regulatory protein that is expressed early in the viral life cycle and is crucial for JCV DNA replication [38, 39] and has been associated with cellular oncogenic transformation[40]. Conversely, the VP1 protein is the major component of the viral capsid, and is expressed at the time of viral assembly. These data suggest that JCV may remain quiescent in bone marrow, or undergo an abortive, rather than a productive infectious cycle.

Among patients without PML, JCV DNA was significantly more prevalent in the bone marrow of HIV-positive patients (22/47; 47%) versus HIV-negative patients (10/75; 13%) (p<0.001). Interestingly, this difference was accentuated when looking at the subset of HIV-positive and HIV-negative patients with lymphoma (10/20 (50%) vs 1/22 (5%), respectively (p=0.001)). The role, if any that JCV might play in HIV associated lymphomas deserves further studies. HIV seropositivity increases the risk of developing lymphoma by 60-165 folds prior to the HAART era[41-43], which has not markedly changed since the availability of HAART[44]. Hence, the pathogenesis of HIV associated lymphomas is likely multifactorial[45], including induction of cytokines resulting in B cell proliferation, prolonged immunosuppression, and inability to control oncogenic viruses such as the herpesviruses EBV and HHV8[46, 47], which may in turn cause genetic alterations in B cells. A previous study showed that central nervous system lymphomas contained significant quantity of JCV genome, and that JCV TAg can co-express with EBV in the same cells[48]. In this context, the finding of JCV TAg, an oncogenic protein, in bone marrow cells from the B cell lineage is tantalizing.

Having ascertained the presence of JCV in bone marrow, we sought to characterize JCV RR in this compartment, since this non-coding region contains determinants of neurotropism and neurovirulence. However, this task was rendered difficult by the low JC viral load found in bone marrow. All of our cloned JCV RR from bone marrow specimens were similar to the JCV Mad-1 prototype[49], but multiple clones from each specimen contained unique point mutations in either the first or second 98bp repeats, ruling out a contamination with Mad-1 plasmid and cross-contaminations between patients samples. Unlike the archetype RR, which is present in most urine samples, the Mad-1-like RR containing tandem repeats of a 98 bp element are usually detected in most CNS samples of PML patients[6], and also in blood of PML patients who had a poor outcome[50]. This finding suggests that the bone marrow may be the site of JCV neurotropic transformation. These results confirm our initial observation that both archetype and rearranged RR with tandem repeat pattern may coexist in the bone marrow of an HIV-negative PML patient with rheumatoid arthritis[23].

There are several limitations to our study. First, due to the difficulties in collecting large numbers of bone marrow samples, our specimens are heterogeneous. Indeed, efficiency of JCV DNA extraction and detection by PCR may vary in fresh bone marrow aspirates or formalin-fixed paraffin-embedded biopsies. Nevertheless, we chose to process specimens and analyze data separately according to the type of samples obtained (Tables 1 and 3). One exception involved bone marrow specimens from PML patients, which are extremely rare, and only archival vertebral bone, rather than iliac crest biopsy was available from HIV-positive PML cases (Table 4). This may explain in part why we found JCV in samples of only one third of PML patients tested. Nevertheless, these results suggest that viral persistence in bone marrow may not be necessary at time of active viral replication in the brain. Second, we may have underestimated JCV prevalence in fresh bone marrow aspirates. Indeed of 34 samples, 25 were sorted and the remaining 9 samples were extracted as a whole. While 10 of the sorted samples contained detectable JCV DNA, none of the 9 unsorted samples were positive. Due to the low JC viral load in bone marrow, it is possible that in unsorted samples, the larger quantity of chromosomal DNA may have diluted the small amount of JCV DNA present. Lastly, we were only able to clone and sequence JCV RR from formalin-fixed paraffin-embedded bone marrow biopsy samples, probably because of the low JC viral load in fresh bone marrow aspirates and peripheral blood. This limited our ability to compare the JCV RR in the bone marrow and the peripheral blood of the same individuals, which could aid in clarifying JCV latency and reactivation.

In conclusion, this is the first study to show prevalence of JCV in bone marrow from both HIV-positive and HIV-negative patients, with and without PML. The presence of JCV DNA and TAg, and the neurotropic RR implicate bone marrow as an important site in JCV pathogenesis. Further understanding of JCV primary infection, latency and reactivation are urgently needed to prevent the occurrence of PML a growing number of patients treated with novel immunomodulatory therapies for cancer and autoimmune diseases.