JC Polyomavirus Infection Reveals Delayed Progression of the Infectious Cycle in Normal Human Astrocytes

Animal models are crucial in advancing biomedical research and defining the pathogenesis of human disease. Unfortunately, not all diseases can be easily modeled in a nonhuman host or such models are cost prohibitive to generate, including models for the human-specific virus JC polyomavirus (JCPyV). JCPyV infects most of the population but can cause a rare, fatal disease, progressive multifocal leukoencephalopathy (PML). There have been considerable advancements in understanding the molecular mechanisms of JCPyV infection, but this has mostly been limited to immortalized cell culture models. In contrast, PML pathogenesis research has been greatly hindered because of the lack of an animal model. We have further characterized JCPyV infection in primary human astrocytes to better define the infectious process in a primary cell type. Albeit a cell culture model, primary astrocytes may better recapitulate human disease, are easier to maintain than other primary cells, and are less expensive than using an animal model.

ABSTRACT

JC polyomavirus (JCPyV) infects 50 to 80% of the population and is the causative agent of a fatal demyelinating disease of the central nervous system (CNS). JCPyV presents initially as a persistent infection in the kidneys of healthy people, but during immunosuppression, the virus can reactivate and cause progressive multifocal leukoencephalopathy (PML). Within the CNS, JCPyV predominately targets two cell types, oligodendrocytes and astrocytes. Until recently, the role of astrocytes has been masked by the pathology in the myelin-producing oligodendrocytes, which are lytically destroyed by the virus. To better understand how astrocytes are impacted during JCPyV infection, the temporal regulation and infectious cycle of JCPyV were analyzed in primary normal human astrocytes (NHAs). Previous research to define the molecular mechanisms underlying JCPyV infection has mostly relied on the use of cell culture models, such as SVG-A cells (SVGAs), an immortalized, mixed population of glial cells transformed with simian virus 40 (SV40) T antigen. However, SVGAs present several limitations due to their immortalized characteristics, and NHAs represent an innovative approach to study JCPyV infection in vitro. Using infectivity assays, quantitative PCR, and immunofluorescence assay approaches, we have further characterized JCPyV infectivity in NHAs. The JCPyV infectious cycle is significantly delayed in NHAs, and the expression of SV40 T antigen alters the cellular environment, which impacts viral infection in immortalized cells. This research establishes a foundation for the use of primary NHAs in future studies and will help unravel the role of astrocytes in PML pathogenesis.

IMPORTANCE Animal models are crucial in advancing biomedical research and defining the pathogenesis of human disease. Unfortunately, not all diseases can be easily modeled in a nonhuman host or such models are cost prohibitive to generate, including models for the human-specific virus JC polyomavirus (JCPyV). JCPyV infects most of the population but can cause a rare, fatal disease, progressive multifocal leukoencephalopathy (PML). There have been considerable advancements in understanding the molecular mechanisms of JCPyV infection, but this has mostly been limited to immortalized cell culture models. In contrast, PML pathogenesis research has been greatly hindered because of the lack of an animal model. We have further characterized JCPyV infection in primary human astrocytes to better define the infectious process in a primary cell type. Albeit a cell culture model, primary astrocytes may better recapitulate human disease, are easier to maintain than other primary cells, and are less expensive than using an animal model.

INTRODUCTION

Progressive multifocal leukoencephalopathy (PML) is a fatal demyelinating disease caused by the neuroinvasive pathogen JC polyomavirus (JCPyV) (1, 2). JCPyV resides in the kidneys in 50% to 80% of the healthy adult population, causing an asymptomatic infection (3–7). During immunosuppression, JCPyV can reactivate and lead to the development of PML in the central nervous system (CNS) (6, 8, 9). Although it is rare, individuals who are HIV-1 positive (1, 10) and those using immunomodulatory therapies for immune-mediated diseases (11–13), such as natalizumab for multiple sclerosis, are at increased risk for PML development (12, 14). While experimental treatments in development have led to some clinical improvements in patients, the current standard is treatment of the underlying immunosuppression and supportive care (15–17). Successful treatment of the immunosuppression has improved the 1-year PML survival rates, which range from ∼20% in natalizumab-associated PML to 50 to 80% in HIV-associated PML (15, 17). Although there is no cure for PML, patients can survive for up to 15 years with the disease, albeit with significant morbidity, including neurological symptoms that range from mild to severe dysfunction (15, 18).

In the CNS, the virus primarily targets two types of glial cells, astrocytes and oligodendrocytes (19–21). The study of clinical patient samples has indicated that myelin-producing oligodendrocytes are targeted by the virus, leading to demyelination of the CNS and onset of PML. However, the role of astrocytes in JCPyV infection and PML remains unclear (19–21). Questions underlying astrocytic infection, including whether astrocytes provide protection during PML pathogenesis or whether astrocytes support a favorable environment for viral replication, are not clearly understood (22). While cell culture models have provided significant advances to the field, limitations include differences in primary versus immortalized cell types that could confound comparisons to disease pathogenesis in vivo. Thus, understanding the JCPyV-host cell interactions and the viral infectious cycle in biologically relevant primary astrocytes is critical to better understand JCPyV pathogenesis and to develop therapies for PML.

JCPyV pathogenesis is the result of the virus hijacking host cell machinery, and the temporal regulation of the infectious cycle in the nucleus is carefully orchestrated by numerous host cell transcription factors; this, in part, contributes significantly to the unique host and cellular tropism of the virus. JCPyV is a small, nonenveloped, double-stranded DNA virus that is ∼5,100 bp in length (23, 24). The genome is organized into three regions: the regulatory region or noncoding control region (NCCR), containing the origin of replication, TATA box, and enhancer elements and two bidirectional coding regions that contain the early genes and late genes (25). Isolates of JCPyV from PML patients include naturally occurring variants, such as Mad-1 and Mad-4, that all have this general genome composition, with most of the differences displayed in the NCCR. For example, the NCCR of the Mad-4 variant contains a 19-bp deletion and only one late-proximal TATA box compared to the NCCR of the Mad-1 variant, which contains duplicate TATA boxes and an enhancer element (26). There are modified variants as well, including the Mad-1/SVEΔ, a laboratory-derived variant with improved infectivity in cell culture due to insertion of the simian virus 40 (SV40) enhancer (27).

For productive viral replication to occur in the nucleus, many host transcription factors, including nuclear factor kappa-light-chain enhancer of activated B cells (NF-κB) (28, 29), Spi-B transcription factor (SpiB) (30), and in glial cells, nuclear factor of activated T cells 4 (NFAT4) (31) and subtypes of the nuclear factor 1 (NF-1) family (32, 33), bind to the regulatory region to initiate early gene transcription. This results in the production of the virus origin binding protein, large tumor antigen (T Ag), and other proteins needed to transform the cell into a virus factory. These early gene products orchestrate this transformation by binding to retinoblastoma (Rb) (34), consequently driving the cell into S phase, sequestering p53, and thus, blocking apoptotic pathways (35). T Ag also functions as a DNA helicase to unwind the viral DNA for replication (36). With increasing viral early gene transcript concentration within the nucleus, the infectious cycle undergoes a switch toward viral DNA replication and late gene transcription, producing agnoprotein and viral protein 1 (VP1), VP2, and VP3 (26). These viral proteins encapsidate the newly replicated viral DNA. Understanding the infectious cycle in human cells that contribute to PML pathogenesis is necessary to design therapeutics and prevent disease, but unfortunately, there are not many biologically representative cell culture models or in vivo models that outline JCPyV infection in astrocytes or oligodendrocytes.

While experimental animal models to study JCPyV pathogenesis have been attempted, the most tractable model systems have not been able to recapitulate the clinical symptoms of PML. Early animal models, including Syrian golden hamsters (37, 38), owl monkeys, and squirrel monkeys, resulted in tumorigenesis upon JCPyV infection due to the oncogenic potential of the JCPyV protein T Ag (39–42). These studies reinforced the fact that nonhuman cells lacked the appropriate host factors for the virus to initiate transcription of the late genes in order to complete the infectious cycle (43), thus resulting in tumor formation. To overcome this challenge, recently developed animal models have included engrafted human cells and humanized or weakened immune systems (21, 44). In the most-recently reported animal model for PML pathogenesis, Kondo et al. (21) developed a humanized mouse model with engrafted glial progenitor cells (GPCs). Their results, unlike other models, highlighted that the primary cells targeted by JCPyV were GPCs and astrocytes, demonstrating that astrocytes are the main target in PML pathogenesis (21). In contrast, oligodendrocytes were infected in a delayed manner and were not required for viral propagation and spread (21), which represents a significant paradigm shift in the understanding of PML development within the field. This research illuminated the importance of astrocytic infection in PML, which is currently understudied in the field.

There are a few reports of JCPyV infection of primary astrocytes in the literature. In 2004, progenitor cell-derived astrocytes (PDAs) were used to understand their capacity to support JCPyV infection, with the researchers concluding that cell death was the result of necrosis and not induction of apoptotic pathways (45). Further research validated the susceptibility of astrocytes to JCPyV infection, in contrast to progenitor cells, in which infection was lower (46). A 2003 microarray study revealed 355 genes upregulated and 130 downregulated during infection of primary human astrocytes, leading to further examination of specific proteins, such as Grb-2, cyclin A, cyclin E, PAK2, and transforming growth factor β receptor 1 (TGFβ-R1), in JCPyV infection (22). Another microarray analysis, in 2013, examined the genes influenced by JCPyV infection during the differentiation of brain-derived multipotential CNS progenitor cells (neural progenitor cells [NPCs]) into PDAs. Their findings highlighted transcription factors, including nuclear factor I-X (NFI-X), NFI-A, c-Jun, and c-Fos, that promoted JCPyV infection during the differentiation to PDAs (47). A recent study examined JCPyV DNA replication in primary astrocytes, SVG-A cells (referred to herein as SVGAs; an immortalized, mixed population of glial cells transformed with simian virus 40 [SV40] T antigen), and primary human choroid plexus cells (48). Erickson and Garcea (48) demonstrated that replication in the nucleus of primary astrocytes was like that of other polyomaviruses, recruiting similar host DNA damage response proteins to sites of replication. The authors concluded that there was either a delay or cessation in viral DNA replication in infected astrocytes (48).

The purpose of this study was to expand on previously published research to improve our understanding of JCPyV infectivity in primary human astrocytes, while comparing this to infection in SVGAs, a mixed-glial cell model widely used to study JCPyV. SVGAs, which constitutively express SV40 T Ag, were developed to study JCPyV infection in vitro (49). Due to the challenges of generating an animal model, SVGAs have been an important model cell line in the field, being implemented in numerous studies and significantly advancing JCPyV research, but due to their transformed characteristics, it is difficult to make direct comparisons to JCPyV infection in primary cell types. This study establishes a foundation for future JCPyV research by using primary human astrocytes, which may reflect a more accurate approach to in vitro infection studies. We demonstrate that normal human astrocytes (NHAs) can be readily infected with JCPyV and serve as an effective model to study JCPyV infection. NHAs demonstrate a delayed infectious cycle in comparison to the progression of the infectious cycle in SVGAs, suggesting that infection may proceed in a delayed manner in primary cells. Overexpression of SV40 T Ag in NHAs revealed that T Ag can drive the infectious cycle and induce expression of cyclins, which serve as a marker of cell cycle progression. Finally, with advancements in culturing primary cells, the experiments and methodologies used in this study can be extrapolated to other models used to understand viral infections. By performing single-cell analyses, examining multiple proteins in one cell, and performing statistical procedures like dimensionality reduction, one can strengthen in vitro studies, especially where animal models are lacking or are difficult to understand and develop.

RESULTS

NHAs display established characteristics of astrocytes.

Astrocytes, one of the most abundant glial cells in the brain, with many key roles in the CNS, including support of myelination (50–52), have been implicated in PML (21) and yet are challenging to study due to limited markers. The most commonly used astrocyte marker, glial fibrillary acid protein (GFAP), is a reliable marker for reactive astrocytes, and yet, it has limitations (53), since as much as 40% of astrocytes can be found negative (54). Aldehyde dehydrogenase 1 family, member L1 (ALDH1L1), is another astrocyte marker, and unlike GFAP, has a broader pattern of expression in astrocytes (55). To characterize NHAs before JCPyV infection, cells were subcultured to determine variability in the expression of both astrocyte markers, GFAP and ALDH1L1, over time. NHAs (passage 2 [P2] to P8), SVGAs (P78 to P84), and human embryonic kidney 293A cells (HEKs) (P66), a non-glial cell control that expresses low levels of GFAP (56), were plated, fixed, and analyzed by indirect immunofluorescence using antibodies targeting both astrocyte markers (Fig. 1A). NHA morphology is characterized as protoplasmic because they were isolated from gray matter (57). At early passages (P2 to P4), NHAs displayed many major branches from the astrocyte soma and thousands of branchlets (Fig. 1A), similar to the findings in other published research (58). As subculturing continued, however, these protoplasmic traits diminished (Fig. 1A). This is in comparison to ALDH1L1, however, of which expression remained diffuse through continued passages, corroborating the findings in other research (55). In-Cell Western assay (ICW) was used to quantify GFAP and ALDH1L1 expression in all three cell types (Fig. 1B and C). GFAP expression decreased in NHAs from P2 to P3 but then remained consistent through P8. Levels of GFAP expression were higher in NHAs and SVGAs than in HEKs (Fig. 1B and C). ALDH1L1 expression in NHAs also decreased from P2 to P3, but at less appreciable levels. Both glial cell types, however, expressed higher levels of ALDH1L1 than did the non-glial cell type (Fig. 1C). These data corroborate a previous report that GFAP expression in NHAs decreases over time (59) but ALDH1L1 expression remains present (55).

FIG 1.

NHAs display established characteristics of astrocytes. NHAs and SVGAs (at indicated passages) were plated and fixed for staining using antibodies targeting glial fibrillary acidic protein (GFAP) and aldehyde dehydrogenase family 1, member L1 (ALDH1L1). HEK293A cells (HEKs) were used as a non-glial cell-type control. (A) Representative images of NHAs, SVGAs, and HEKs stained with antibodies specific for GFAP (green) and ALDH1L1 (red); nuclei were counterstained with DAPI (blue). Cells were analyzed by indirect immunofluorescence using an epifluorescence microscope at ×40 magnification. Bars = 20 μm. (B) Cells were plated in 96-well plates, stained using antibodies specific for GFAP or ALDH1L1 (green) or CellTag (red), and analyzed by ICW. (C) Percentage of GFAP or ALDH1L1 for each passage was quantitated by ICW signal intensities (GFAP or ALDH1L1/CellTag × 100 = % response) for NHAs (P2 to P8), SVGAs (P78 to P84), and HEKs (P66). Box and whisker plots represent the distribution of data from 6 samples, with median values denoted by the black lines and whiskers representing values 1.5 times the distance between the first and third quartiles (interquartile range). Colored diamonds represent individual points for NHAs and SVGAs (6 replicates), and HEKs represent the experimental control, comprising box-and-whisker plots. Outliers are represented by black points. The Kruskal-Wallis test was used to compare all three cell types within each passage analysis. *, P < 0.01. Data are representative of three independent experiments.

Multiple variants of JCPyV infect NHAs.

To determine whether NHAs are permissive to multiple variants of JCPyV, NHAs and SVGAs were infected with Mad-1, Mad-4, and Mad-1/SVEΔ, fixed at 48 and 72 h postinfection (hpi), and stained with antibodies specific for the early viral gene product T Ag and the late viral gene product viral protein 1 (VP1) (Fig. 2). The time points correspond to T Ag and VP1 production, respectively, in SVGAs and served as a starting point to understand JCPyV infection in NHAs. Mad-1 and Mad-1/SVEΔ resulted in equivalent percentages of cells expressing T Ag in NHAs and SVGAs at 48 hpi (Fig. 2A and B). However, the expression of VP1 was significantly lower in NHAs than in SVGAs at both 48 and 72 hpi. The infectivity of Mad-4 was lower, but a similar trend in the percentages of NHAs expressing VP1 was observed (Fig. 2C). This suggests that JCPyV, regardless of variant, can establish infection in NHAs but the production of late viral proteins is delayed, suggesting a delay in the infectious cycle. Based on similarities with multiple JCPyV variants, Mad-1/SVEΔ was utilized for subsequent experiments due to enhanced infectivity and ease of production.

FIG 2.

Multiple JCPyV strains demonstrate robust T Ag production but delayed production of VP1 in NHAs. NHAs and SVGAs were infected with the Mad-1/SVEΔ (A), Mad-1 (B), and Mad-4 (C) strains of JCPyV at MOIs of 0.5, 0.5, and 0.05 FFU/cell, respectively, and fixed at 48 or 72 hpi. Cells were stained with antibodies specific for the early gene product T Ag and the late gene product VP1 and analyzed by indirect immunofluorescence. Percent infection was determined by counting the number of T Ag- or VP1-positive nuclei divided by the number of DAPI-positive nuclei for five ×20 fields of view for triplicate samples. Data are representative of three individual experiments. Error bars indicate standard deviations (SD). Student’s t test was used to determine statistical significance comparing data for NHAs and SVGAs within each time point and viral protein. *, P < 0.01.

JCPyV infection of NHAs occurs in a delayed manner independent of MOI.

To determine if the delay in infection in NHAs occurs when they are inoculated with increased virus concentrations, NHAs and SVGAs were infected at various multiplicities of infection (MOIs), fixed at 48 and 72 hpi, and stained with antibodies specific for JCPyV T Ag and VP1 (Fig. 3A and B). At 48 hpi, the percentages of JCPyV-infected NHAs expressing T Ag were comparable to the results for SVGAs, regardless of the MOI utilized. In contrast, VP1 expression was not detectable in NHAs at 48 hpi (Fig. 3B). At 72 hpi, the percentages of NHAs expressing T Ag continued to increase (Fig. 3B), and yet, the numbers of VP1-positive cells were significantly reduced in comparison to the numbers in SVGAs (Fig. 3B). Infecting the NHAs with an MOI of ∼6 resulted in ∼50% of the cells expressing T Ag at 72 hpi, but this still resulted in less than 3% of cells expressing VP1 (data not shown). This suggests that JCPyV can establish infection in NHAs but the production of late viral proteins is prolonged, suggesting a delay in the infectious cycle.

FIG 3.

Late JCPyV gene products are delayed across a range of MOIs. (A) Representative images of NHAs and SVGAs infected with JCPyV (MOI = 0.1 FFU/cell), fixed at 48 hpi, and stained with antibodies specific for JCPyV T Ag (red) or glial fibrillary acidic protein (GFAP) (green); nuclei were counterstained with DAPI (blue). Cells were analyzed by indirect immunofluorescence using an epifluorescence microscope at ×20 magnification. Bars = 50 μm. (B) NHAs and SVGAs were infected with JCPyV at the indicated MOIs (FFU/cell) and fixed at 48 or 72 hpi. Cells were stained with antibodies specific for the early gene product T Ag and the late gene product VP1. Percent infection was determined by counting the number of T Ag- or VP1-positive nuclei divided by the number of DAPI-positive nuclei for 5 ×20 fields of view for triplicate samples. Data are representative of three individual experiments. Error bars indicate SD. Student’s t test was used to determine statistical significance comparing data for NHAs and SVGAs at each respective MOI, within each time point and viral protein. *, P < 0.01.

To determine the extent of the delay in VP1 production in infected NHAs, both cell types were infected and analyzed for the production of viral proteins T Ag and VP1 over a long-term time course (12 days) (Fig. 4A). VP1 expression in NHAs was significantly lower than in SVGAs (Fig. 4B), despite the increased expression of T Ag in NHAs. However, appreciable amounts of VP1 were detected in NHAs at 12 days postinfection (dpi), suggesting that VP1 production is delayed rather than absent (Fig. 4A). In SVGAs, at 3 and 7 dpi, 42 to 53% of infected cells expressed VP1, and at 12 dpi, 63% of cells expressed VP1. In NHAs, however, only 4 to 15% of infected cells were expressing VP1 at 3 and 7 dpi, and only 40% expressed VP1 at 12 dpi (Fig. 4B). Finally, to determine the efficiency of viral assembly and release, supernatant was collected at each time point from the infected cells and was then used to inoculate naive SVGAs. Cells were fixed at 3 dpi and stained using a VP1-specific antibody to determine whether the viral particles released into the supernatant were infectious. The results shown in Fig. 4C illustrate that supernatant collected from SVGAs resulted in a significantly higher percentage of infected cells than supernatant collected from NHAs. By 12 dpi, however, the infectious virus in the supernatant collected from NHAs increased noticeably, resembling the trend observed in the infectivity analysis (Fig. 3). These data suggest that VP1 production in NHAs is significantly delayed before 7 dpi compared to the VP1 production in SVGAs; however, after 7 dpi, levels of VP1 increase, releasing infectious virus particles from the cell.

FIG 4.

The JCPyV infectious cycle is significantly prolonged in primary astrocytes. (A) NHAs were infected with JCPyV (MOI = 0.1 FFU/cell) and fixed at 3, 7, and 12 dpi. Cells were stained with antibodies specific for T Ag and VP1, and percent infection was quantified by indirect immunofluorescence, counting the number of T Ag- or VP1-positive nuclei and dividing by the number of DAPI-positive nuclei for 5 fields of view at ×20 magnification for triplicate samples. (B) The ratios of VP1- to T Ag-positive cells from NHAs in the experiment (results are shown in panel A) and from SVGAs were measured. (C) Naive SVGAs were inoculated with viral supernatant collected from the specific cell types on the days indicated, incubated for 72 h, and then fixed and stained for VP1. Data represent percentages of JCPyV-infected cells for five ×20 fields of view for triplicate samples. All data are representative of at least 3 independent experiments. Error bars indicate SD. Student’s t test was used to determine statistical significance. *, P < 0.05.

SV40 large T Ag restores VP1 production in NHAs.

SVGAs were developed specifically to support JCPyV infection, by using the origin-defective mutant of SV40 (49). Constitutive expression of the SV40 large T Ag in the nucleus allows efficient transcription and replication of the JCPyV genome during JCPyV infection (60). It is thought that the enhanced efficiency is due to SV40 large T Ag binding more effectively to the JCPyV origin of replication, ultimately leading to enhanced transcription of the JCPyV early gene products, including JCPyV T Ag (61, 62). The expression of T Ag drives the infectious cycle forward, leading to the switch from producing T Ag to replication and then production of the late gene products, including VP1. To determine if the delay in the JCPyV infectious cycle in NHAs could be overcome by the constitutive expression of SV40 large T Ag, NHAs stably expressing the origin-defective mutant of SV40, referred to herein as NHA-Ts, were generated (Fig. 5A). NHAs, SVGAs, and NHA-Ts were infected with JCPyV and fixed at 48, 72, and 96 hpi, and JCPyV T Ag and VP1 protein production were measured using indirect immunofluorescence. T Ag production was similar across the three cell types at 48 hpi, but there was a significant difference in VP1 production (Fig. 5B). VP1 expression was significantly higher in NHA-Ts than in NHAs, and the level of VP1 production was comparable to that in SVGAs at 48 hpi (Fig. 5B). At 96 hpi, NHAs demonstrated a significantly higher level of T Ag expression but a significantly lower level of VP1 than in NHA-Ts and SVGAs. In contrast, NHA-Ts continued to express higher percentages of both T Ag and VP1 (Fig. 5B). These data suggest that SV40 large T Ag enhances JCPyV infection in NHAs and allows efficient production of VP1.

FIG 5.

SV40 T Ag reestablishes VP1 production in NHAs. (A) Representative epifluorescence images of NHAs, SVGAs, and NHAs stably expressing SV40 T Ag (NHA-Ts) as determined by staining cells with antibodies specific for SV40 T Ag (red) and astrocyte marker GFAP (green); nuclei were counterstained with DAPI (blue). Bars = 20 μm. (B) NHAs, SVGAs, and NHA-Ts were infected with JCPyV (MOI = 0.1 FFU/cell), incubated at 37°C, fixed at 48, 72, and 96 hpi, stained with antibodies specific for JCPyV T Ag and VP1, and analyzed by indirect immunofluorescence. Data represent percentages of JCPyV-infected cells for five ×20 fields of view for triplicate samples. Error bars indicate SD. One-way ANOVA was used to determine statistical significance comparing NHAs and NHA-Ts to SVGAs for each time point and viral protein. *, P < 0.01.

Production of VP1 transcript is delayed in NHAs.

To further understand the delay in JCPyV infection in NHAs and define whether there are cell type-dependent differences in viral transcription, viral transcripts were quantified at 24 to 96 hpi by quantitative PCR (qPCR) in NHAs, SVGAs, and NHA-Ts. The 24-hpi time point was included to serve as an initial measure of viral transcripts in each cell type, which demonstrated no significant difference in T Ag transcript levels between NHAs and SVGAs (Fig. 6A). At 48 hpi, both SVGAs and NHA-Ts had significantly more T Ag transcripts than NHAs (Fig. 6A), and this trend was retained over the course of time points analyzed. Not surprisingly, there was a significant difference in VP1 transcript levels from 48 to 96 hpi in NHAs in comparison to the levels in SVGAs, but the transcript levels in NHA-Ts were restored to levels comparable to those in SVGAs (Fig. 6B). Interestingly, VP1 transcript levels increased in NHAs to be almost comparable to the levels in SVGAs and NHA-Ts at 72 and 96 hpi. These data correspond with the long-term-infection results (Fig. 4), as after 7 dpi, there was a significant increase in VP1-expressing cells in NHAs, further corroborated by qPCR analyses. Additionally, as T Ag accumulates in the infected cell, viral DNA replication is occurring and the switch to transcription of the late viral genes ensues, including the production of VP1. Thus, it is hypothesized that without the presence of SV40 large T Ag, effective late gene transcription in NHAs does not occur until 48 h after it occurs in SVGAs and NHA-Ts.

FIG 6.

Reduced production of T Ag protein delays transcription of VP1 in NHAs. (A) NHAs, SVGAs, and NHA-Ts were infected with JCPyV (MOI = 0.1 FFU/cell) and incubated for 24, 48, 72, and 96 h. At each time point, RNA was extracted, and viral transcript levels were determined by qPCR. Data represent the absolute quantification of viral transcripts for triplicate samples. Error bars indicate SD. Data are representative of 3 independent experiments. One-way ANOVA was used to determine statistical significance comparing data for NHAs and NHA-Ts to data for SVGAs for each time point and viral protein. *, P < 0.01. (B, C) Cells infected in parallel were also fixed and counterstained for viral proteins, and the relative fluorescence intensity units (RFUs) for T Ag (∼70 cells) (B) and VP1 (∼30 cells) (C) were quantified by indirect immunofluorescence for five ×20 fields of view. (B) The Kruskal-Wallis test was used to compare the relative intensities of T Ag and VP1, comparing all cell types within each time point. *, P < 0.01.

To further define the transcriptional and translational productivity of JCPyV infection, the relative intensities of T Ag and VP1 protein expression in infected cells were quantified using ImageJ (63). Approximately 70 T Ag- and ∼30 VP1-expressing cells were quantified from 48 to 96 hpi (Fig. 6B). There were significant differences in T Ag expression in infected NHAs compared to expression in SVGAs and NHA-Ts at all time points (Fig. 6B). As T Ag expression is needed to drive the transcriptional and translational productivity of VP1, these data help to explain the low levels of VP1-positive cells in NHAs compared to the levels in SVGAs. Additionally, NHA-Ts serve as a suitable positive control for this experiment, as there was no significant difference in T Ag expression in comparison to that in SVGAs (Fig. 6B), confirming that cell type-dependent differences did not affect the relative intensities. Finally, when the infectious cycle was able to progress to produce VP1 in infected NHAs, there were no differences in the relative intensities compared to the intensities in SVGAs and NHA-Ts (Fig. 6C). These data suggest that after adequate production of JCPyV T Ag, the infectious cycle is driven forward, ultimately producing VP1 transcript and quickly being translated to VP1 protein. These data reinforce the idea that the delay in the infectious cycle in NHAs is caused by insufficient levels of JCPyV T Ag protein being produced in the cell.

Polyomavirus proteins dysregulate the cell cycle in primary and immortalized cells.

All polyomaviruses produce T Ag during the infectious cycle to transform the cell into a viral factory that allows the effective replication of viral DNA and, ultimately, virus production (34). Viral T Ag is a multifunctional protein, serving as a DNA helicase (36), binding to p53 and blocking apoptosis, and inducing the infected cell into S phase (35). To observe the broad view of cell cycle progression during JCPyV infection of NHAs, cells were stained for cell cycle markers cyclin E and cyclin B1. Cyclin E accumulates in the cell during S phase, and cyclin B1 accumulates during the progression to G₂/M phase (64–66). Cyclin B1 has been used to investigate the cell cycle status of JCPyV-infected cells, as well as that of other viruses (21, 67–69). Thus, analyzing the expression of cyclin E and viral proteins T Ag and VP1 provides a better understanding of the infectious cycle over time in NHAs and SVGAs in relation to the cell cycle status. Using confocal microscopy, fields of view were analyzed for nuclear (N) and cytoplasmic (C) cyclin B1 and cyclin E and nuclear T Ag or VP1 expression in 30 cells for NHAs, SVGAs, and NHA-Ts at 48 to 96 hpi (representative images are presented in Fig. 7A).

FIG 7.

JCPyV T Ag drives nuclear expression of cyclin E in all cell types, while cyclin B1 expression is variable. (A to C) NHAs, SVGAs, and NHA-Ts were infected with JCPyV (MOI = 5 FFU/cell), incubated for 48, 72, and 96 h, and then fixed and stained with antibodies specific for T Ag (red), cyclin E (magenta), and cyclin B1 (green) and complementary secondary antibodies. Bars = 20 μm. (A) Representative confocal micrograph images (60× objective) of samples at 96 hpi. (B) Image analysis was performed for mock-infected and infected cells (∼30 cells per time point and sample), quantifying the RFUs for nuclear and cytoplasmic expression (N/C) of cyclin E (B) and cyclin B1 (C) using ImageJ. One outlier was removed from the cyclin E analysis (B), and two outliers were removed from the cyclin B1 analysis (C). The dashed lines represent an N/C ratio of 1. Solid lines represent the median values. The Kruskal-Wallis test was used to compare the relative intensities of T Ag, comparing all cell types within each time point. *, P < 0.01.

Nuclear cyclin E expression was significantly higher in NHAs expressing JCPyV T Ag than in mock-infected cells at all time points measured (Fig. 7B). In comparison, SVGAs infected with JCPyV had higher levels of cyclin E in the nucleus than did mock-infected cells at 48 and 96 hpi, and this trend was only seen in NHA-Ts at 48 hpi (Fig. 7B). Conversely, cyclin B1 expression was significantly higher in the cytoplasm of NHAs than in that of mock-infected cells and higher in the nucleus of SVGAs than in that of mock-infected cells at 96 hpi (Fig. 7C). A principal-component analysis (PCA) was performed examining the relationship of T Ag expression and the nuclear and cytoplasmic expression of cyclin B1 and cyclin E within each cell population. A loading plot of the first two, most-explained principal components illustrating the relationship of each of the 5 variables was established (Fig. 8A). The first component (x axis) separates cyclin B (nuclear and cytoplasmic) and T Ag expression (positive) and cyclin E expression (negative). The second component (y axis) is influenced by all vectors (positive) (Fig. 8A). Applying the same loading plot to each cell type and time point, NHAs were defined as having low expression of T Ag and cyclin B1, distinctly different than in SVGAs and NHA-Ts, at 48 hpi (Fig. 8B). However, at later time points, the NHA population began to resemble the SVGA and NHA-T populations (Fig. 8B).

FIG 8.

PCA separates the NHA population early during JCPyV infection. (A) A loading plot of the first two components. The first component (x axis) separates cyclin B1 and T Ag expression (positive) and cyclin E expression (negative). The second component (y axis) is influenced by all vectors (positive). (B) Principal-component analysis was performed on samples from the experiment whose results are shown in Fig. 7, using the loading plot to demonstrate correlations between nuclear and cytoplasmic expression of cyclin E and cyclin B and nuclear T Ag expression, separated by cell type and faceted by time point.

Cyclin E and cyclin B1 expression were also examined in VP1-positive cells for each cell type at 96 hpi (Fig. 9A). Only the 96-h time point was quantified for NHAs, as the numbers of VP1-positive cells for this cell type at the earlier time points were insufficient for analysis. As shown by the results in Fig. 7B and C, NHAs expressed higher levels of cyclin E but lower levels of cyclin B1 in the nucleus than did mock-infected cells (Fig. 9B). In SVGAs, cyclin E and cyclin B1 levels were slightly higher in the nucleus and there was no difference in cyclin expression in NHA-Ts compared to the results for the mock-infected populations (Fig. 9B). A PCA was also performed on VP1-positive cells, and a loading plot illustrates the principal components used to define each of the five variables (Fig. 10A). The first component (x axis) is influenced by all vectors (positive). The second component (y axis) separates cyclin E and VP1 expression (negative) and cyclin B expression (positive) at 96 hpi (Fig. 10A). Applying the loading plot to each cell type, the expression levels of cyclins in VP1-positive NHAs more closely resemble those of SVGAs and NHA-Ts in comparison to the distinct populations observed in T Ag-positive cells at earlier time points. However, VP1-positive NHAs still express lower levels of cyclin B1 (Fig. 10B). These data, along with the previous PCA, illustrate that at earlier times during infection, NHAs are uniquely defined by the level of viral protein expression and cyclin expression, but as infection progresses, infected NHAs become more variable and the population begins to resemble those of SVGAs and NHA-Ts.

FIG 9.

Nuclear cyclin E expression is enhanced in NHAs expressing VP1. NHAs, SVGAs, and NHA-Ts were infected with JCPyV (MOI = 5 FFU/cell), incubated for 96 h, and then fixed and stained with antibodies specific for VP1 (red), cyclin E (magenta), and cyclin B1 (green) and with complementary secondary antibodies. Bars = 20 μm. (A) Representative confocal micrograph images (60× objective) of samples at 96 hpi. (B) Image analysis was performed for mock-infected and infected cells (∼30 cells per time point and sample), quantifying the RFUs for nuclear and cytoplasmic expression (N/C) of cyclin E (B) and cyclin B1 (C) using ImageJ. One outlier was removed from the cyclin B1 analysis (C). The dashed lines represent an N/C ratio of 1. Solid lines represent the median values. The Kruskal-Wallis test was used to compare the relative intensities of VP1, comparing all cell types within each time point. *, P < 0.01.

FIG 10.

VP1 expression correlates with cyclin expression in JCPyV-infected cells. (A) A loading plot of the first two components in PCA. The first component (x axis) is influenced by all vectors (positive). The second component (y axis) separates cyclin E and VP1 expression (negative) and cyclin E expression (positive) at 96 hpi. (B) Principal-component analysis was performed on samples from the experiment whose results are shown in Fig. 9, using the loading plot to demonstrate correlations between nuclear and cytoplasmic expression of cyclin E and cyclin B1 and nuclear VP1 expression, separated by cell type.

DISCUSSION

SVGAs, transformed with SV40 T Ag, serve as a robust model to study JCPyV infection and have been especially important due to the challenges of growing the virus in commonly used cell culture models. Studies in SVGAs have significantly improved our understanding of JCPyV infection, as these cells are cost effective and easy to maintain and utilize in the laboratory. Unfortunately, the immortalized characteristics of these cells present challenges for comparisons to viral infection that results in disease in human hosts, and their use may not reflect the progression of infection in a natural host cell (70). First, SVGAs are a heterogenous population of cells, which makes it difficult to generate conclusions from a single cell type. Furthermore, cells transformed with SV40 T Ag, like SVGAs, are susceptible to chromosomal changes through repeated passages, demonstrating alterations in both karyotype and phenotype in SVGAs (71). As a result of these challenges, we further characterized JCPyV infection in primary astrocytes, specifically analyzing the temporal regulation of the virus. We have demonstrated herein that NHAs exhibited established characteristics of protoplasmic astrocytes based on the expression of astrocytic markers GFAP and ALDH1L1 (Fig. 1A). The expression of astrocytic markers decreases in initial passaging but remains constant thereafter and, thus, independent of the initial events of the infectious cycle (Fig. 1A). The infectious cycle in NHAs is comparable to that in SVGAs, but subsequent steps are delayed, resulting in a significant reduction of the late viral gene product VP1. As it is well understood that SV40 T Ag can bind more effectively than JCPyV T Ag to the early region of JCPyV to initiate JCPyV replication (61, 62), we sought to characterize the differences in JCPyV infectivity in primary cells compared to its infectivity in cells transformed with SV40 T Ag, NHA-Ts. The expression of SV40 T Ag conferred an enhanced capacity of NHAs to support JCPyV infection to the level of SVGAs. These findings illustrate that primary astrocytes support JCPyV infection but have a delayed infectious cycle and that the expression of SV40 T Ag alters the course of infection in NHAs through cellular reprogramming that is not attributable to JCPyV infection.

Other cell culture models have been used to study JCPyV infection in vitro. This includes other immortalized cells, such as COS-7 cells (CV-1 cells transformed with SV40 T Ag) (72) and 293TT cells (human embryonic kidney [HEK] cells expressing high levels of SV40 T Ag) (60), as well as primary cells like RPTECs (renal proximal tubule epithelial cells) (73), neural progenitor cells (NPCs) (47), choroid plexus epithelial cells (48, 86), and human embryonic stem cell (hESC)-derived oligodendrocytes (74). RPTECs represent a potential primary model for studying the low-level persistent infection in the kidney. NPCs and hESC-derived oligodendrocytes are useful models to study glial cell-specific infection but are challenging to differentiate and maintain in cell culture (75). On the other hand, primary astrocytes are easy to culture and maintain and, most recently, were discovered to play a critical role in PML development in a human-chimeric-glial cell mouse model (21). Previously, PML development has been defined by the lytic infection of oligodendrocytes, which were thought to be the main target of PML (1, 76). Kondo et al. suggested that astrocytes are readily targeted by the virus and produce higher levels of JCPyV infection than oligodendrocytes and, interestingly, that astrocytes may act as a reservoir for JCPyV infection (21). The delayed characteristics noted in that study align with our results demonstrating that JCPyV can infect NHAs, producing significantly more JCPyV T Ag-positive cells and fewer infected cells expressing VP1 (Fig. 2 and 4). Infecting NHAs with multiple JCPyV variants, which resulted in the same trends in T Ag and VP1 production, demonstrated that the NCCR region may not influence the temporal regulation of the virus in primary cells (Fig. 2); however, future studies would be necessary to further validate this finding.

Previous research has also demonstrated that the overexpression of SV40 T Ag enhances the replication of JCPyV, which naturally replicates poorly in cell culture. For example, 293TTs, HEK cells with an SV40 T Ag expression plasmid and an integrated SV40 genome, can support infection because of their capacity to produce high levels of SV40 T Ag (60). Similarly, results reported herein illustrate that JCPyV infection is enhanced in NHAs that express SV40 T Ag in trans (Fig. 5B). Although viral DNA replication was not measured directly, previous research has shown that SV40 T Ag enhances JCPyV replication (60). In line with previous results, NHA-Ts had equivalent levels of both early and late JCPyV transcripts and proteins, like those of SVGAs (Fig. 6). A possible explanation lies in the flanking sequences around the core origin of JCPyV T Ag not being able to stimulate DNA replication as well as SV40 T Ag (62). This dependency results in JCPyV T Ag binding less efficiently to the viral genome (61), leading to less virus replication and, ultimately, lower expression of the late gene products, including VP1.

Polyomavirus T Ag proteins can impact the cell cycle, allowing the forced entrance of quiescent cells into S phase and binding to p53 and Rb (35, 77), which is necessary for viral replication. SV40 T Ag can also induce the expression of cyclins A, B1, and E but not of cyclin D (78, 79). Cyclins are commonly used as markers for the cell cycle (21, 67–69) and have previously been used as cell cycle markers during JCPyV infection (21, 22). Cyclin E is activated during the end of G₁ and accumulates in the cell during S phase, and it has previously been demonstrated to be upregulated around 5 dpi during JCPyV infection but to decrease later (22). Cyclin B1 on the other hand, accumulates in the cell during late stages of the G₂ phase and becomes activated in the nucleus during the initiation of the M phase. Furthermore, previous research has demonstrated that cyclin B1 expression increased in infected astrocytes from initial infection up to 15 dpi (22). Unfortunately, previous immunofluorescence analyses were subject to a qualitative approach that could lead to inherent biases of defining a cell expressing a given protein and not quantitatively illustrating the range of expression in infected cells. Also, cyclins are dynamic proteins within the cell, translocating to the cytoplasm and nucleus, and thus, other techniques used may not be able to accurately quantitate these translocation events (63).

Quantitative fluorescence microscopy methods allow single-cell analyses to determine the amount of a protein within a cell, information that could be lost by biochemical fractionation from an analysis such as Western blotting (63). Examining the nuclear (N) to cytoplasmic (C) ratios (N/C), it was determined that cyclin E expression was higher in the nucleus of infected cells than in the nucleus of mock-infected cells in both NHAs and SVGAs, which was shown previously in cells infected with JCPyV (22). When examining cyclin B expression, however, infected NHAs had a lower N-to-C ratio than mock-infected cells, which contrasted with the results for SVGAs, especially at 96 hpi (Fig. 7B). These findings illustrate the advantages of using this technique and highlight the differences of JCPyV infection in NHAs. As JCPyV infection continues, the cell progresses into the G₂ phase and is arrested at G₂/M, shown by the increase in cyclin B1 expression in the cytoplasm, which allows virus assembly and subsequent virus release (Fig. 11). In NHAs, cyclin B1 is activated but is sequestered in the cytoplasm, while in SVGAs, cyclin B1 expression is greater in the nucleus of infected cells than in the nucleus of mock-infected cells (Fig. 7A and C). This difference is most likely due to the expression of SV40 T Ag, as the infected cells of the newly transformed NHA-T population resemble SVGAs (Fig. 8B and 10B), resulting in the infected cell not being able to progress to later stages of the cell cycle.

FIG 11.

Model of the temporal regulation of JCPyV infection and cell cycle progression, comparing NHAs to SVGAs. JCPyV translocates to the nucleus after arrival in the endoplasmic reticulum (ER) (∼6 to 12 h); comparable levels of T Ag transcript are produced at 24 hpi in NHAs and SVGAs. T Ag transcript is translated to T antigen protein, but at decreased levels in NHAs compared to the levels in SVGAs. Subsequently, less VP1 transcript is produced, delaying the translation of VP1 protein. Cyclin E continues to accumulate in the nucleus of NHAs from 48 to 96 hpi as T Ag transforms the cell to support viral replication. VP1 transcript begins to accumulate in the nucleus at 72 hpi and cyclin B accumulates in the cytoplasm, and at 96 hpi, VP1 is produced. In comparison, JCPyV infection in SVGAs occurs in a more rapid fashion, where T Ag transcript is translated to T Ag protein and cyclin E accumulates in the nucleus at 48 hpi. VP1 transcripts drastically increase at 48 hpi as well, and at 72 hpi, nucleus expression of cyclin E is reduced, and VP1 production occurs soon after.

Image analysis also allows the generation of a unique data set, defining selected cells by the expression of multiple proteins. In this study, the nuclear and cytoplasmic expression of host proteins, as well as nuclear viral protein levels, was analyzed by performing a principal-component analysis across various time points, revealing interesting trends developed by the correlations among the variables chosen. Compared to SVGAs and NHA-Ts, NHAs were defined as a small, unified population of cells expressing low levels of cyclin B and T Ag. As infection progressed, however, the NHA population began to overlap SVGAs and NHA-Ts, most likely the result of continued JCPyV T Ag production, further inducing expression of cyclin E and potentially driving the cells into S phase (Fig. 8B and 10B).

This research further characterizes primary astrocytes as a useful model to study early events in JCPyV infection, as NHAs support the production of abundant levels of JCPyV T Ag. Furthermore, the techniques and methodological approaches explained here can be extrapolated to other primary cells, including oligodendrocyte progenitor cells and mature oligodendrocytes to determine the temporal regulation of JCPyV infection in this other critical cell type in the pathogenesis of PML. The differences between infection in primary astrocytes and in the transformed model SVGA cell line provide insights into the mechanism by which SV40 T Ag enhances JCPyV infection. Taken together, these analyses illustrate a timeline of the progression of JCPyV infection in a primary cell type, NHAs. Initially delayed (compared to that in cells expressing SV40 T Ag), the infectious cycle is prolonged due to the low levels of JCPyV T Ag (Fig. 11). However, the prolonged expression of JCPyV T Ag may lead to T Ag binding to the viral DNA, as well as p53 and Rb, inducing the cell into S phase and ultimately driving the transcription of the late genes, such as VP1. In a field where the development of an animal model is hampered due to the narrow host range of JCPyV, this research may serve to establish a foundation for future studies to examine and discover molecular mechanisms of astrocytic viral infection that ultimately lead to human disease.

MATERIALS AND METHODS

Cells, plasmids, and viruses.

Normal human astrocytes (NHAs) (passage 1 [P1]) were obtained from Lonza (product no. CC-2565) and maintained in astrocyte growth medium (AGM) supplemented with AGM SingleQuots supplements (product no. CC-3186) and 1% penicillin-streptomycin (P-S) (Mediatech, Inc.). NHAs were provided with Lonza’s Certificate of Analysis. The donor was a 19-week-gestation female with no detected levels of HIV, hepatitis B virus (HBV), or hepatitis C virus (HCV). Cells were most likely isolated from the cerebrum and structures associated with the cerebrum and, thus, most likely represent primarily type I astrocytes. NHAs were used at low passage numbers (P3 to P8) and subcultured according to Lonza instructions. SVGAs and HEKs, generously provided by the Atwood Laboratory (Brown University) (49), were cultured in complete minimum essential medium (MEM) (Corning) and complete Dulbecco’s modified Eagle medium (DMEM) (Corning), respectively, with 10% fetal bovine serum (FBS), 1% P-S, and 0.2% Plasmocin prophylactic (InvivoGen).

NHAs were transformed with SV40 large T Ag by transfecting them with the origin-defective mutant of SV40 inserted into the pMK16 vector (provided by the Atwood Laboratory, Brown University); the resulting cells are referred to as NHA-Ts. NHAs (P8) were seeded at 5,000 cells/cm² in a 25-cm² flask, transfected with 1 μg of pMK16 containing the SV40 Ori⁻ mutant, using Fugene (Roche) at a 1.5:1 ratio (Fugene/DNA), and incubated at 37°C. The medium was replaced with fresh AGM supplemented with the SingleQuots supplements every 2 days, and after 2 weeks, proliferation of cells in areas of the flask was clear. The cells were passaged, and the AGM medium was replaced with MEM with 1% P-S and 16% FBS. To validate that NHA-Ts express SV40 large T Ag, cells were fixed and stained by indirect immunofluorescence using the SV40 T Ag-specific antibody AB2 (1:50; Calbiochem). One hundred percent of the cells displayed nuclear staining for SV40 T Ag, which was validated before and after experiments were performed. NHAs, SVGAs, and NHA-Ts were grown in a humidified incubator at 37°C with 5% CO₂.

The generation and production of purified JCPyV strains Mad-1, Mad-4, and Mad-1/SVEΔ were described previously (80), and the strains were kindly provided by the Atwood laboratory (Brown University) or purchased through the ATCC (Mad-4).

JCPyV infection.

NHAs, NHA-Ts, and SVGAs were seeded into 96-well plates with 10,400 cells/well (∼70% confluence). Cells were infected (multiplicities of infection [MOIs] as indicated in figure legends) with 42 μl/well of the respective cell medium at 37°C for 1 h. After infection, the cells were fed with the respective cell medium at 100 μl/well and incubated at 37°C for the duration of the infection. At the time points indicated in Fig. 2 to 10, cells were fixed and stained for both viral proteins analyzed, T Ag and VP1. Additionally, for long-term experiments (Fig. 4), supernatant was collected from each well to be subsequently used to inoculate naive SVGAs with 42 μl/well of supernatant at 37°C for 1 h. After infection with supernatants, cells were incubated at 37°C for 72 h and then fixed and stained by indirect immunofluorescence using VP1-specific antibodies.

Indirect immunofluorescence staining and quantitation of JCPyV infection.

Following infection, cells were stained for both viral proteins and host cell proteins at room temperature (RT). NHAs, NHA-Ts, and SVGAs were washed with 1× phosphate-buffered saline (PBS) and then fixed in 4% paraformaldehyde (PFA) for 15 min and washed with PBS–0.01% Tween three times. Cells were permeabilized for 15 min using PBS–0.5% Triton X-100. Following permeabilization, NHAs, NHA-Ts, and SVGAs were blocked in PBS–10% goat serum for 45 min. Cells were then stained using antibodies specific for viral proteins and host cell proteins for 1 h at RT except for antibodies specific for GFAP and ALDH1L1, which were incubated at 4°C for ~16 h. Cells were subsequently washed three times in PBS–0.01% Tween (PBS-T) and counterstained with the appropriate secondary antibodies, indicated in Table 1. Following primary and secondary incubations, cells were counterstained using DAPI (4′,6-diamidino-2-phenylindole) at RT for 5 min and washed with 1× PBS.

TABLE 1.

Antibodies used in immunofluorescence and ICW assays

Protein	Antibody (dilution; source or manufacturer) (assay)
	Primary	Secondarya
JCPyV T Ag	PAB597 (1:5; hybridoma)	Anti-mouse Alexa Fluor 594 or anti-mouse Alexa Fluor 568 antibody (confocal) (1:1,000; Thermo Fisher Scientific)
JCPyV VP1	ab34756 (1:1000; abcam)	Anti-mouse Alexa Fluor 594 or anti-mouse Alexa Fluor 568 antibody (confocal) (1:1,000; Thermo Fisher Scientific)
SV40 T Ag	AB2 (1:50; Calbiochem)	Anti-mouse Alexa Fluor 594 antibody (1:1,000; Thermo Fisher Scientific)
Cyclin B1	4138 (1:300; Cell Signaling Technology)	Anti-rabbit Alexa Fluor 488 antibody (1:1,000; Thermo Fisher Scientific)
Cyclin E	50-9714-80 (2.5 μg/ml; Invitrogen)	—b
GFAP	ab4674 (1:2,000; abcam)	Anti-chicken Alexa Fluor 488 antibody (1:1,000; abcam) (IFA) or anti-chicken Li-Cor IRDye 800CW (1:12,000) (ICW)
ALDH1L1	19H14L20 (1:100; Invitrogen)	Anti-rabbit Alexa Fluor 594 antibody (1:1,000; Thermo Fisher Scientific) (IFA) or anti-rabbit Li-Cor IRDye 800CW (1:12,000) (ICW)

^aIFA, immunofluorescence assay; ICW, In-Cell Western assay.

^b—, cyclin E antibody is conjugated to eFluor 660.

To visualize the cells expressing nuclear VP1 or T Ag, a Nikon Eclipse Ti epifluorescence microscope (Micro Video Instruments, Inc.) equipped with a 20× objective was utilized. The percentage of cells infected was determined by counting the number of T Ag- or VP1-positive nuclei divided by the number of DAPI-positive nuclei for 5 fields of view/well. DAPI-positive nuclei were determined by using the Nikon NIS-Elements Basic Research software (version 4.50.00, 64 bit). A binary algorithm was generated using the software to separate cells based on intensity, diameter, and circularity in each field of view. At least 3 images/well, using a 20× objective with 1-s exposure and 2.2× gain, were taken to quantify the relative intensities of viral protein expression in the nucleus.

In-Cell Western assay and host protein quantification.

Cells were seeded in 96-well plates, and following fixation, NHAs (P2 to P8), SVGAs (P78 to P84), and HEKs (P66) were permeabilized with 1× PBS–1% Triton X-100 at RT for 15 min. Cells were incubated with Tris-buffered saline (TBS) Odyssey blocking buffer (Li-Cor) at RT for 1 h. Cells were stained with antibodies detecting either GFAP or ALDH1L1 (Table 1) in TBS Odyssey blocking buffer at 4°C for ∼16 h while rocking. After primary incubation, cells were washed with PBS-T and incubated with an anti-chicken or anti-rabbit Li-Cor IRDye 800CW (Table 1) and CellTag 700 stain (1:500; Li-Cor) at RT for 1 h. Cells were washed again with PBS-T and aspirated prior to scanning. Plates were imaged using the Li-Cor Odyssey CLx infrared imaging system to detect both the 700- and 800-nm intensities. Plates were read at 42-μm resolution, at medium quality, with a 3.0-mm focus offset. After scanning, the 700- and 800-nm channels were aligned using the Image Studio software (version 5.2), and In-Cell Western (ICW) analysis was performed in Image Studio. Subsequently, the ratio of the 800-nm channel (GFAP or ALDH1L1) signal to the 700-nm channel (CellTag) signal was determined (81).

Host-cell and viral protein expression by confocal microscopy.

Cells were seeded in 96-well, #1.5 glass-bottom plates (CellVis) to 40 to 50% confluence. NHAs were plated in AGM medium supplemented with AGM SingleQuots supplements, NHA-Ts were plated in MEM with 1% P-S and 16% FBS, and SVGAs were plated in MEM with 1% P-S and 2% FBS. Following infection (MOI = 1 focus-forming unit [FFU]/cell), cells were incubated for 48, 72, and 96 h. Cells were fixed and stained for host cell proteins cyclin B1 and cyclin E and viral proteins T Ag and VP1 as described above. Approximately 30 cells per cell type and time point were imaged for expression of no viral protein (mock infected) or of T Ag or VP1 (JCPyV infected) or for coexpression of cyclin B1 and cyclin E. Images were collected using an Olympus laser scanning confocal microscope (model IX81; Olympus America) with a 60× objective (oil immersion) with z-stacks for each image and 0.32-μm-slice spacing. Fluorescence images were collected using the 488-, 568-, and 647-nm multiline argon lasers.

ImageJ quantification of host-cell and virus protein.

Images were analyzed using ImageJ. For confocal microscopy experiments, Fig. 12 illustrates the workflow performed using ImageJ. Briefly, a z projection taking the average intensity of each z slice (∼30 slices) was generated for each image. Each channel of the image was split, and a binary mask, referred to below as a mask, was generated, representing the nucleus of the cell (Fig. 12A). The image of the nucleus was dilated 10 times to encompass the cytoplasm (Fig. 12B). This region of the cell is an accurate representation of the cytoplasm because cyclin B1 and cyclin E expression is diffuse and is not localized to one specific region (Fig. 7A) (63). The image of the dilated mask had the nucleus subtracted to create a ring-shaped structure (Fig. 12C). To eliminate noncell background, a second mask was generated from both cyclin E and cyclin B expression and the previous mask subtracted, resulting in cytoplasmic regions of interest (ROIs), considering cellular boundaries (Fig. 12D to F). The mask created for both cyclin E and cyclin B1 expression was not used to quantify cytoplasmic intensity because there was no method to distinctly separate cells, and thus, it would not result in an accurate representation of the cells being selected. These ROIs were overlaid on the 488-nm channel (cyclin B1) and the 660-nm channel (cyclin E) to measure the relative intensities of cyclin expression in the cytoplasm (Fig. 12G and H). The relative intensities in the cytoplasm were derived using equation 1, below. ROIs were also created from the mask indicating the nucleus (Fig. 12A) to measure relative intensities of cyclin expression and viral protein expression in infected cells. Finally, the nucleus-to-cytoplasm (N/C) ratio was determined for each selected cell using equation 2 below.

$$\text{Cytoplasmic relative intensity}=\frac{\left[(\text{dilated nucleus}\times \text{area of dilated nucleus})-(\text{nucleus}\times \text{area of nucleus})\right]}{\text{whole cell area}-\text{nucleus area}}$$ (1)

$$N/C=\frac{\text{nuclear intensity}}{\text{cytoplasm relative intensity}}$$ (2)

FIG 12.

Workflow in ImageJ to quantify host cell and virus protein expression. Images from confocal microscopy experiments were imported into ImageJ and intensities determined by this general workflow. First, a binary mask representing the nucleus was generated (A). The nucleus was dilated 10 times to encompass the cytoplasm (B) and the image illustrated in panel A subtracted to generate an outline of the cytoplasm without the nucleus (C). A second mask was generated from both cyclin E and cyclin B expression (D). The mask illustrated in panel D was subtracted from the image illustrated in panel B to create cytoplasmic regions of interest (ROIs), considering cellular boundaries (E and F). These ROIs were overlaid on the 488-nm channel (cyclin B1) and the 660-nm channel (cyclin E) to measure the relative intensities of cyclin expression in the cytoplasm (G and H). Nuclear ROIs were generated from the mask illustrated in panel A, and the nuclear-to-cytoplasmic (N/C) ratio was determined. The image illustrated in panel I represents the merged channels of 488 and 660 nm, and the image illustrated in panel J represents the phase channel (not measured).

Absolute quantification of T Ag and VP1 transcript levels by qPCR.

NHAs, NHA-Ts, and SVGAs were plated to 70% confluence in 96-well plates. Medium was removed, and cells were infected with JCPyV at an MOI of 0.5 FFU/cell in 42 μl of the respective cell medium and incubated at 37°C for 1 h. After 1 h, cells were fed with 100 μl of the appropriate medium and incubated at 37°C for 24, 48, 72, and 96 h. At each time point, cells were washed with 1× PBS and then fixed with 4% paraformaldehyde and subsequently stained for JCPyV T Ag, SV40 T Ag, or JCPyV VP1, or separate wells were treated with 50 μl of TRIzol reagent (Invitrogen) and stored at −20°C. After validating similar levels of infectivity at 48 hpi by measuring immunofluorescence, cells suspended in TRIzol reagent were removed from the −20°C storage and RNA was extracted from each sample with the Direct-zol RNA kits according to the manufacturer’s protocol (Zymo Research). All 72 samples (3 cell types, mock and JCPyV infected, in triplicate for three time points) were randomized and prepped for RNA extraction in two separate isolation procedures, ensuring that each group had an even representation of cell type, treatment, and time point (36 samples/group). RNA was immediately converted to cDNA with the iScript cDNA synthesis kit (Bio-Rad) using 1 μg of RNA. The two groups generated from the RNA extraction were randomly divided into two further groups (ensuring that each group was represented equally) for qPCR analysis of viral transcripts using T Ag and VP1 primers (82); glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a housekeeping gene but was not used in quantification (16 samples per T Ag, VP1, and GAPDH primer set). Briefly, using Hard-Shell 96-well PCR plates (Bio-Rad), each sample contained 150 nM both forward and reverse primers, 100 ng of cDNA, 5 μl of iQ SYBR green supermix (Bio-Rad), and nuclease-free water to a total volume of 10 μl per well. Plates were sealed with Microseal B seals (Bio-Rad), and analysis was performed on a Bio-Rad CFX96 real-time system, using Bio-Rad CFX Manager software 3.1. qPCR results for T Ag and VP1 transcripts were calculated using absolute quantification. In brief, a standard curve was developed using the JCPyV-pUC19 plasmid, performing 10-fold dilutions to generate 1 to 10⁹ copies/μl of the plasmid. Quantification cycle (C_q) values for both T Ag and VP1 were determined at each dilution, creating a linear equation from the standard curve. This equation was used to extrapolate the copies/μl of T Ag and VP1 transcripts in infected NHAs, NHA-Ts, and SVGAs at each time point, determined from the raw C_q values in each sample. Absolute quantification of viral transcript levels of T Ag and VP1 was determined in triplicate samples from three independent experiments. The statistical significance of differences in the numbers of copies/μl of viral transcripts was determined between cell types at 24, 48, 72, and 96 hpi.

Statistical analysis and graphing in RStudio.

The two-sample Student’s t test assuming unequal variances was used to compare the mean values for at least triplicate samples that were normally distributed; a P value of <0.05 or <0.01 was considered statistically significant. One-way analysis of variance (ANOVA) was used to compare the values for two or more samples when the data were normally distributed. The Kruskal-Wallis test was used when comparing the median values of more than two samples. Finally, the Wilcoxon signed rank test was used when comparing data for two populations that were not normally distributed. All statistical analyses were performed in RStudio (version 1.2.1335), except the Student’s t test, which was determined in Microsoft Excel. The Shapiro-Wilk's test and a quantile-quantile plot (Q-Q plot), a plot to quickly visualize the normality of the data, were used to determine if populations were normally distributed in RStudio. Principal-component analyses (PCAs) of the correlations of either T Ag or VP1 expression and both cyclin B1 and cyclin E expression were determined from z-sectioned confocal micrograph images of ∼30 cells. Analyses were performed in RStudio using the library ggfortify (83) and the library ggplot2 (84) to interpret the PCA objects. Other plots were also generated using ggplot2, importing the raw values from Excel into RStudio using the library XLConnect (85).

Data availability.

Macros for creating masks and quantifying protein expression by relative fluorescence units (RFU) in ImageJ and code for PCA performed in RStudio are supplied upon request.

Note Added after Publication

In the originally published version, reference 86 was omitted. This reference was added to the References list and is now cited on page 14, line 6 from the bottom.