Merkel Cell Polyomavirus Downregulates N-myc Downstream-Regulated Gene 1, Leading to Cellular Proliferation and Migration

Merkel cell carcinoma was first described in 1972 as a neuroendocrine tumor of skin, most cases of which were reported in 2008 to be caused by a PyV named Merkel cell polyomavirus (MCPyV), the first PyV linked to human cancer. Thereafter, numerous studies have been conducted to understand the etiology of this virus-induced carcinogenesis. However, it is still a new field, and much work is needed to understand the molecular pathogenesis of MCC. In the current work, we sought to identify the host genes specifically deregulated by MCPyV, as opposed to other PyVs, in order to better understand the relevance of the genes analyzed on the biological impact and progression of the disease. These findings open newer avenues for targeted drug therapies, thereby providing hope for the management of patients suffering from this highly aggressive cancer.

ABSTRACT

Merkel cell polyomavirus (MCPyV) is the first human polyomavirus etiologically associated with Merkel cell carcinoma (MCC), a rare and aggressive form of skin cancer. Similar to other polyomaviruses, MCPyV encodes early T antigen genes, viral oncogenes required for MCC tumor growth. To identify the unique oncogenic properties of MCPyV, we analyzed the gene expression profiles in human spontaneously immortalized keratinocytes (NIKs) expressing the early genes from six distinct human polyomaviruses (PyVs), including MCPyV. A comparison of the gene expression profiles revealed 28 genes specifically deregulated by MCPyV. In particular, the MCPyV early gene downregulated the expression of the tumor suppressor gene N-myc downstream-regulated gene 1 (NDRG1) in MCPyV gene-expressing NIKs and hTERT-MCPyV gene-expressing human keratinocytes (HK) compared to their expression in the controls. In MCPyV-positive MCC cells, the expression of NDRG1 was downregulated by the MCPyV early gene, as T antigen knockdown rescued the level of NDRG1. In addition, NDRG1 overexpression in hTERT-MCPyV gene-expressing HK or MCC cells resulted in a decrease in the number of cells in S phase and cell proliferation inhibition. Moreover, a decrease in wound healing capacity in hTERT-MCPyV gene-expressing HK was observed. Further analysis revealed that NDRG1 exerts its biological effect in Merkel cell lines by regulating the expression of the cyclin-dependent kinase 2 (CDK2) and cyclin D1 proteins. Overall, NDRG1 plays an important role in MCPyV-induced cellular proliferation.

IMPORTANCE Merkel cell carcinoma was first described in 1972 as a neuroendocrine tumor of skin, most cases of which were reported in 2008 to be caused by a PyV named Merkel cell polyomavirus (MCPyV), the first PyV linked to human cancer. Thereafter, numerous studies have been conducted to understand the etiology of this virus-induced carcinogenesis. However, it is still a new field, and much work is needed to understand the molecular pathogenesis of MCC. In the current work, we sought to identify the host genes specifically deregulated by MCPyV, as opposed to other PyVs, in order to better understand the relevance of the genes analyzed on the biological impact and progression of the disease. These findings open newer avenues for targeted drug therapies, thereby providing hope for the management of patients suffering from this highly aggressive cancer.

INTRODUCTION

For nearly 40 years, BK polyomavirus (BKPyV) and JC polyomavirus (JCPyV) have been the only known human polyomaviruses (PyVs). During the last decade, 11 new human PyVs, including KI polyomavirus (KIPyV), WU polyomavirus (WUPyV), Merkel cell polyomavirus (MCPyV), human polyomavirus 6 (HPyV6), human polyomavirus 7 (HPyV7), human polyomavirus 9 (HPyV9), New Jersey polyomavirus (NJPyV), trichodysplasia spinuolsa-associated polyomavirus (TSPyV), Malawi polyomavirus (MWPyV), HPyV12, and St. Louis polyomavirus (STLPyV), have been discovered (1). To add to this list, a putative human PyV named Lyon-IARC PyV (LIPyV) was recently isolated from human skin (2). Although little information is known about the pathogenesis of these novel human PyVs, some of them have been linked to human diseases: BKPyV-associated nephropathy, JCPyV-associated progressive multifocal leukoencephalopathy (PML), WUPyV-associated bronchitis, HPyV6/HPyV7-associated dermatosis, and TSPyV-associated trichodysplasia spinulosa. MCPyV is responsible for an aggressive type of skin cancer called Merkel cell carcinoma (MCC) (3, 4), as its presence has been confirmed in 80% of MCC cases (5). MCPyV is clonally integrated into the host cancer cell genome and harbors mutations that impair viral replication activity, which are important events in MCPyV-mediated MCC (4).

PyVs are small (40 to 50 nm in diameter), nonenveloped, double-stranded DNA (dsDNA) viruses that have a capsid with icosahedral symmetry (6, 7). All PyVs harbor a circular genome with a size ranging from 4.5 to 5.4 kbp, which can be divided into two oppositely oriented protein-encoding regions, the early T antigen gene and the late VP gene, separated by a noncoding control segment (8). Alternative splicing of all PyV early genes gives rise to viral protein large T antigen (LT) and small T antigen (ST), while the late region encodes viral capsid-forming proteins VP1, VP2, and VP3 (9, 10). MCPyV-positive MCCs express the early T antigen gene and require the early gene products for tumor cell survival and proliferation (11).

Although the LT and ST of almost all known human PyVs have displayed transforming properties in various experimental models, the oncogenic potential varies among these proteins (12). Unlike other human PyVs, where LT acts as a major transforming protein, MCPyV ST has shown strong transforming activities (13), indicating that in some aspects MCPyV behaves differently from other human PyVs (14). In addition, the proven implication of only one PyV (MCPyV) in human carcinogenesis raises the question of why MCPyV alone is carcinogenic but the other human PyVs are not. Even though some efforts were made to explain the biological differences between MCPyV and other human PyVs (15, 16), the mechanistic aspect which renders MCPyV carcinogenic is not very well defined.

Virus-mediated host genetic alteration is one of the major contributing factors in virus-induced cancers. The aim of this study was to investigate the effect of human PyV early genes on host global gene expression, with a focus on the identification of the genes uniquely altered by the expression of MCPyV early genes that may confer the transforming abilities to MCPyV. Of the numerous genes identified in this study, N-myc downstream-regulated gene 1 (NDRG1) was found to be downregulated by MCPyV T antigen. Interestingly, previous studies have shown that NDRG1 functions as a metastatic suppressor (17 – 20) and a transcriptional repressor and is involved in cell cycle inhibition (21, 22), cellular differentiation (20), and apoptosis (23). However, nothing is known about the role of this protein and the underlying mechanism by which it affects the cell-transforming ability of MCPyV. Taken together, our work identifies significant changes in gene expression upon MCPyV oncogene expression which may explain the difference in the carcinogenic potential of MCPyV from that of other human PyVs. In addition, altered expression of NDRG1 by MCPyV provides mechanistic insight into host cell signaling deregulation in MCPyV-mediated cellular proliferation.

RESULTS

Early gene products from different human PyVs differently affect cellular gene expression.

To determine whether some human PyVs differ in their ability to alter cellular gene expression, we generated normal immortalized keratinocyte (NIK) lines expressing the entire early region of five human PyVs, namely, BKPyV, JCPyV, KIPyV, MCPyV, and WUPyV. In addition, simian virus 40 (SV40) was included in the study due to its well-characterized transforming activities in experimental models. NIKs were transduced with recombinant retroviruses containing the early regions of the six PyVs, and the relative and absolute expression levels of ST and LT were found to be nearly the same as those shown in our previously published study (24). The expression profiles of 24,000 annotated genes were determined by use of an Illumina microarray (Illumina HumanHT-12 BeadChip array, v4). The microarray data were first subjected to quality control, where 3 different methods, namely, unsupervised hierarchical clustering of the duplicates, scatterplots, and boxplots of gene expression data, were applied. Sample analysis was performed in duplicate for all viruses except for MCPyV with its corresponding negative control, for which quadruplicate analyses were performed. Unsupervised hierarchical clustering was performed on genes that passed the filtering criteria, which revealed that all the samples clustered separately, with their duplicates and quadruplicates clustering next to each other (Fig. 1A).

FIG 1.

The ability to deregulate cellular genome expression varies among different PyVs. (A) Schematic presentation of the unsupervised clustering of replicates after removing the background. (Top) The dendrogram shows the clustering of duplicates and quadruplicates, determined using centered correlation and average linkages. (Bottom) The heat map shows the differential expression of the genes in all samples. Each row indicates the expression of a specific gene across all the samples, while each column represents the sample in which gene expression was measured. The color scale at the bottom reveals the relative expression level of the genes among all the samples. Blue and red represent down- and upregulation, respectively. (B) The histogram shows the total number of differentially expressed (either downregulated or upregulated) genes upon each class comparison. Each PyV, representing one class, was compared with the negative control (pLXSN), and the resulting deregulated genes with a 1.5-fold change in expression with a P value and FDR of <0.001 for each class are represented in the graph. The numbers on the top of each bar show the total number of up- and downregulated genes by early genes of each PyV. (C) The Venn diagram represents the common and differentially expressed genes for the MCPyV (MCV) data set from this study and the studies of Berrios et al. (25), Masterson et al. (26), and Daily et al. (27). The number 1 in the middle indicates the gene (HIST1C1) that was commonly deregulated in the 4 data sets. (D) Cluster analysis of differentially expressed genes involved in cell cycle regulation. The heat maps obtained from BioCarta show the differential expression of 28 genes involved in the cell cycle at the G₁/S checkpoint (left) or the 23 genes related to cyclins and cell cycle regulation (right) between MCPyV and pLXSN. Color intensities reflect the fold change in expression relative to that in the control cells. Blue and brown show down- and upregulation, respectively.

Subsequently, we compared the expression profile data for each PyV with the expression profile data for the negative control, i.e., NIKs transduced with an empty retrovirus (pLXSN). The expression of genes is provided as the ratios of the values obtained relative to the values obtained under the control condition after normalization of the data. For comparison between these classes, genes were considered differentially expressed when they displayed a difference of at least a 1.5-fold increase or decrease in expression pattern in both replicates with a P value and a false discovery rate (FDR) of <0.001. Using these selection criteria, we identified numerous genes deregulated by each PyV upon comparison with the negative control (Fig. 1B). Notably, most of the genes were downregulated in each class comparison. The exception was the WUPyV genes, for which the number of upregulated genes was higher than the number of downregulated ones. However, SV40 scored a maximum for the deregulation of genes (n = 967), while MCPyV scored 325 genes on the list.

A comparison of the genes deregulated by MCPyV in this study with the published data sets from 3 different studies (25 – 27) revealed a total of 73 genes to be commonly deregulated in our study and at least one of the previous studies (Fig. 1C). Only 1 gene, HIST1H1C, was found to be deregulated unanimously in all studies included in the comparative analysis. The encoded protein is involved in cell senescence, DNA repair, and the cell cycle. A comprehensive list of the 73 genes was prepared and is included in Table S1 in the supplemental material. Strikingly, pathway analysis of the 73 genes showed genes regulating mostly cell senescence, DNA repair, the cell cycle, and signal transduction pathways, and these are included in Table S2.

Next, we focused on the expression profile induced by MCPyV, since it is the only PyV clearly associated with human carcinogenesis. The MCPyV-deregulated genes (n = 325) were subjected to BioCarta pathway analysis. Although a significant number of genes whose expression was altered by MCPyV are functionally related to many pathways, genes involved in cell cycle regulation and mitogen-activated protein kinase (MAPK) pathways ranked the highest, with a significant number of genes showing variable expression (Table S3). Our observation revealed that essential genes involved in cell cycle regulation, particularly at the G₁/S phase, are modulated by the expression of MCPyV early genes (Fig. 1D). These genes included cyclins, cyclin-dependent kinases (CDKs), and cyclin-dependent kinase inhibitors (28). These results further strengthen the notion that cell cycle deregulation may be one of the major driving factors in MCPyV-mediated carcinogenesis.

Comparative analyses of deregulated cellular gene expression mediated by the 6 PyVs.

Next, we determined whether MCPyV displays unique features in deregulating cellular gene expression in comparison to the other PyVs. For this purpose, single class comparisons between the early genes of the 6 PyVs expressed in NIKs and the pLXSN control expressed in NIKs were followed by the evaluation of the Venn intersections of the 6 data sets using R scripts. This led to the identification of a total of 23 genes, namely, C12orf24, C1orf116, C9orf41, CCNA1, CDR2L, CTSH, DLK2, ECM1, FOXQ1, INPP4B, KIAA0101, KIF13B, KLF6, LIPG, MXRA5, NDRG1, PTPRE, PYGB, S100A16, SH3KBP1, SLC1A3, TRIB1, and UGT1A6, whose expression was specifically altered by MCPyV and 60 other deregulated genes that were common to all PyVs (Fig. 2A). In addition, 97, 44, 25, 285, and 398 genes were specifically deregulated by BKPyV, JCPyV, KIPyV, WUPyV, and SV40, respectively. To further elucidate the MCPyV signature genes, the BRB-ArrayTools package was used, whereby a single class comparison between MCPyV and pLXSN was made after subtracting the background genes and the genes resulting from the class comparisons of BKPyV, JCPyV, KIPyV, WUPyV, SV40, and the negative control, pLXSN. Interestingly, this method also showed that MCPyV exclusively and significantly altered the expression of 28 genes in comparison to their expression in the other PyVs (Fig. 2B). Notably, 23 of 28 MCPyV-deregulated genes identified in this analysis were also found in the Venn diagram intersections. However, an increase of 5 genes whose expression was significantly different, namely, SPRR2E, CTSC, ANXA2, PTGS1, and DUSP10, was seen in the latter approach. Further, to understand whether these 23 genes are specific to MCPyV-mediated deregulation, we did a comparative analysis between the published data sets for SV40 (29, 30), BKPyV (31, 32), and JCPyV (33, 34) and found 2 genes, CCNA1 and LIPG, commonly deregulated for SV40 and 3 genes, KIAA0101, MXRA5, and SLC1A3, commonly deregulated for BKPyV (Fig. 2C and D). However, no gene was found to be commonly deregulated for JCPyV.

FIG 2.

MCPyV specifically deregulates certain cellular genes. (A) The Venn diagram represents the genes common to and differentially expressed by 6 PyVs: MCPyV (MCV), JCPyV (JCV), KIPyV (KIV), WUPyV (WUV), BKPyV (BKV), and SV40. Single class comparisons were made between all 6 PyVs and the pLXSN control, and then R scripts were used to obtain Venn diagram intersections of the 6 data sets. The numbers at the extremities represent the genes specifically deregulated by each PyV, while the number 60 in the middle refers to the genes that were commonly deregulated by the 6 PyVs mentioned above. The 23 genes showing the MCPyV-specific signature are highlighted in red. (B) The heat map shows the relative expression of 28 genes in the other 5 human PyVs which were uniquely and specifically deregulated by MCPyV early genes. The 28 genes listed are uniquely and significantly deregulated by MCPyV, with 15 genes being downregulated and 13 genes being upregulated. The numbers on the y axis show the number of genes, while the numbers on the x axis represent the number of samples. The color bar at the bottom represents the fold change scale, varying from −2.4 (blue, downregulated) to 2.3 (red, upregulated). (C and D) The 23 genes of the MCPyV-specific signature compared to the SV40-specific (C) and BKPyV-specific (D) signatures. (E) The bar diagram shows the number of genes involved in biological (left) and molecular (right) functions. Using Gene Ontology software, the 28 genes representing the specific signature of MCPyV were analyzed for their involvement in various biological processes. Each bar represents one biological category, and the numbers on the top of each bar show the number of genes out of 28 involved in the respective functional category. The number of genes is reported on the y axis, while the x axis represents the categories of biological functions. (F) Reactome Pathway analysis showing the top 5 pathways regulated by the 23 genes. FDR, false discovery rate.

We also evaluated whether the products of these MCPyV-deregulated genes are involved in crucial cellular pathways by using Gene Ontology (GO) software. The analysis showed that 19 of these genes were involved in the response to stimuli and in the regulation of processes linked to cellular transformation (Fig. 2E). Importantly, as shown in Table S4, some of these genes have been found to be deregulated in different types of human cancers. Reactome Pathway analysis using the 23 genes revealed glucuronidation, biological oxidation, Tp53- and G₁/S-mediated transcription, and phase II conjugation of compounds to be the top 5 pathways regulated by them (Fig. 2F). Together, these results show that the products of the MCPyV early gene have a unique property of deregulating cellular gene expression when its properties are compared to those of the other PyVs.

Validation of the role of MCPyV early proteins in altering cellular gene expression.

In order to confirm the microarray data, we performed quantitative real-time PCR (qRT-PCR). Of the 28 MCPyV-deregulated genes, we selected 5 (NDRG1, KLF6, TRIB1, INPP4B, and ANX2A), based on their biological functions as major tumor suppressors, as described in Table S4. qRT-PCR confirmed that all 5 genes were downregulated in NIKs in the presence of the viral genes (Fig. 3A). Similar findings were obtained when MCPyV early genes were expressed in hTERT-expressing human keratinocytes (hTERT-HK). NDRG1, KLF6, TRIB1, and INPP4B were significantly downregulated in the presence of viral early genes, but ANXA2 was not (Fig. 3B). To corroborate our findings, we also determined whether silencing of the expression of MCPyV LT and/or ST in an MCPyV-positive MCC cell line (MKL-1 cells) influences the expression of the 5 selected genes. Figure 3C shows that silencing the expression of ST alone or both ST and LT (PAN) resulted in a significant increase in the expression of NDRG1, KLF6, and INPP4B. There were no observed changes in their transcript levels in the MCPyV-negative cell line UISO when it was transduced with ST or PAN short hairpin RNAs (shRNAs) (Fig. 3D).

FIG 3.

Expression of differentially expressed genes across the different Merkel cell carcinoma cell lines. Total RNA was extracted from NIKs (A) or hTERT-HK (B) stably expressing the early genes of MCPyV and converted into cDNA, as described in Materials and Methods. Expression of the indicated genes was analyzed in these samples. (C and D) The knockdown of both LT and ST (PAN) or ST alone in MCPyV-positive MCC cells and MKL-1 cells (C) and MCPyV-negative cells (UISO cells) (D) was achieved by transduction with lentivirus-based shRNA, as described in Materials and Methods. Scrambled shRNA (Scr) was used as a negative control. Cells were collected and processed for total RNA. After reverse transcription, the mRNA levels of the indicated genes were determined by qRT-PCR and normalized to the levels of the GAPDH (glyceraldehyde-3-phosphate dehydrogenase) housekeeping gene. (E) Total protein lysates isolated from the Merkel cell carcinoma-positive cell line MKL-1 transduced with PAN shRNA or NIKs or hTERT-HK stably expressing MCPyV early genes were subjected to immunoblotting, and the expression of NDRG1 was checked in these samples. The results (±SD) are representative of those from at least two independent experiments performed in duplicate. *, P < 0.05 and P > 0.01; **, P < 0.01 and P > 0.001.

Comparison of the data obtained with the different cell lines indicated that NDRG1 was the most consistent MCPyV-deregulated gene in keratinocytes and in the Merkel cell carcinoma-derived cell line. Moreover, NDRG1 participates in the second most important pathway, which is the transcriptional regulation of cell death genes by Tp53, as revealed by the Reactome Pathway analysis (Fig. 2F). To further confirm these observations, we determined the NDRG1 protein levels in the three experimental models described above. Silencing of early gene expression by PAN shRNA in MKL-1 cells or expression of MCPyV early genes in NIKs as well as in hTERT-HK resulted in the rescue of or decrease in NDRG1 protein levels, respectively (Fig. 3E). Together, these data show that MCPyV ST and LT can downregulate NDRG1 mRNA and protein levels.

Ectopic expression of NDRG1 in cells expressing early genes of MCPyV or MCC cell lines inhibits cellular proliferation and migration.

Next, we aimed to understand the biological significance of MCPyV-mediated NDRG1 downregulation. hTERT-HK, which had previously been transduced with recombinant retroviruses containing the MCPyV early gene (MCPyV-hTERT-HK), were transfected with the NDRG1-expressing vector pBABE (Fig. 4A). Ectopic expression of NDRG1 decreased the number of colonies by approximately 50% (Fig. 4B and C). To corroborate these findings, the MCC cell lines MKL-1 and MKL-2 were transduced with lentiviruses expressing NDRG1 under the control of a doxycycline-inducible promoter. Induction of NDRG1 expression by doxycycline resulted in a decrease in cellular proliferation of the MCC cell lines. At day 12, MKL-1 and MKL-2 cells showed 31.9% and 34.4% decreases in proliferation compared to the doxycycline-treated controls, respectively (P < 0.05) (Fig. 4D and E). Interestingly, the cellular morphology of the two MCC cell lines showed distinct features. Upon NDRG1 induction, the MKL-1 cells formed smaller cellular aggregates than the controls, whereas the MKL-2 cells showed larger clumps (Fig. 4F). Even though NDRG1 decreased the overall cellular proliferation in the MCC cell lines, these results indicate that NDRG1 differentially impacts the cellular physiology in the two cell lines.

FIG 4.

Effect of NDRG1 overexpression on cellular activities in hTERT-HK expressing early genes of MCPyV or MCC cell lines. hTERT-HK expressing the early region of MCPyV were transiently transfected with the pBABE empty vector or the pBABE vector expressing NDRG1. (A) Expression of NDRG1 in the transfected cells. (B and C) Cells transfected with NDRG1 or not were plated in 6-well plates at a ratio of 1:10, 1:100, or 1:1,000 after selection with hygromycin for 7 to 8 days, as described in Materials and Methods. Representative images (B) and the percentage of colonies (C) are shown. (D to F) MCC cell lines MKL-1 and MKL-2 were transduced with the empty vector or NDRG1-expressing doxycycline-inducible lentiviral constructs in the presence of 0.5 μg/ml of doxycycline. (D) NDRG1 expression in MKL-1 and MKL-2 cells was confirmed by immunoblotting. (E) Cell proliferation was monitored by use of the WST-8 cell proliferation assay reagent (Dojindo). First, the fold increase in cell proliferation for the assay data point (day 12) was determined by dividing the OD value of the data point by that obtained on day 1. To calculate the relative cell proliferation activity in the presence of NDRG1 expression, the value of the fold increase for NDRG1-induced cells was divided by that for empty vector control-transfected cells. (F) Representative image of MCC cell lines expressing or not expressing NDRG1 at 5 days after doxycycline treatment. (G and H) MCPyV was seeded in 6-well plates and transiently transfected with the pBABE empty vector or the pBABE vector expressing NDRG1. After 48 h of transfection, a scratch was introduced using a pipette tip and the cells were imaged every 24 h, as described in Materials and Methods. The migration of hTERT-HK was observed for 48 h, and representative images (G) and a bar graph showing wound healing or closure of the wound expressed as a ratio over that for the control at 0 h (H) are shown. The results obtained from three independent experiments are shown. Error bars indicate the standard deviation. Statistical significance was determined by Student's t test. *, P < 0.05 and P > 0.01; **, P < 0.01 and P > 0.001; ***, P < 0.0001.

As NDRG1 is also implicated as a metastasis suppressor (35), we evaluated whether NDRG1 influences cell migration in our experimental models. MCPyV-hTERT-HK were transiently transfected with the NDRG1 overexpression vector, and a wound healing assay was performed every 24 h for a period of 2 days. We observed nearly complete wound closure (94.9%) in control MCPyV-hTERT-HK after 48 h, whereas we observed only 26.1% wound closure in NDRG1-overexpressing MCPyV-hTERT-HK (Fig. 4G and H). Together, these results show that NDRG1 plays an important role in cellular proliferation and migration.

NDRG1 overexpression differentially regulates the cell cycle in MCC cell lines.

Analysis of the cell cycle profile by flow cytometry showed that NDRG1 overexpression resulted in a modest but reproducible decrease in cells in S/G₂ phase (22.4% decrease; Fig. 5A and B). In addition, an increase in the sub-G₀ population was observed in the same cells in the presence of ectopic NDRG1 levels (2.1-fold compared to that for the control [P < 0.01]; Fig. 5A and C). Interestingly, not many differences in the cell cycle profiles of the MCC cell lines were observed (Fig. 5D). This can be attributed to the fact that they are slow-cycling cells with doubling times of nearly 3 days. However, the bromodeoxyuridine (BrdU) incorporation assay also did not show many changes in MKL-1 cell lines overexpressing NDRG1 (Fig. 5D, left, and Fig. 5E), whereas MKL-2 cells showed a significant decrease (49.1% decrease over that for the control [P < 0.05]; Fig. 5D, right, and Fig. 5E). These data highlight different mechanisms in the NDRG1-mediated inhibition of cellular proliferation in hTERT-HK and Merkel cancer-derived cell lines.

FIG 5.

Cells engaged in DNA synthesis are reduced in MCPyV-hTERT-HK and MKL-2 cells but not MKL-1 cells expressing NDRG1. (A to C) hTERT-HK expressing the early region of MCPyV were transiently transfected with the pBABE empty vector or the pBABE vector expressing NDRG1. (A) The cell cycle profiles of cells overexpressing NDRG1 or not were fixed, and the cells were stained with PI. (B and C) The percentage of cells in S/G₂ phase (B) and the percentage of cells in the sub-G₀ population (C) are represented as bar graphs. (D) Representative results depicting the cell cycle profile (top) and BrdU incorporation in S phase of the cell cycle (bottom) of cells analyzed by flow cytometry. (E) MKL-1 or MKL-2 cells treated with doxycycline for 8 days were labeled with 10 mM BrdU for 1 h. The incorporated BrdU and cellular DNA were stained by anti-BrdU antibody and propidium iodide, and BrdU incorporation was quantitated. The results (±SD) are representative of those from three independent experiments. Statistical analysis was performed using Student's t test. *, P < 0.05 and P > 0.01; **, P < 0.01 and P > 0.001; ***, P < 0.0001.

NDRG1 regulates expression of key cell cycle regulators, CDK2 and cyclin D1.

After studying the role of NDRG1 in cell cycle regulation, we aimed to investigate the expression of cell cycle regulators in the presence of NDRG1. First, we determined by immunoblotting (IB) whether the expression of MCPyV early genes in hTERT-HK influences the protein levels of positive regulators of the cell cycle, namely, cyclin D1 and cyclin-dependent kinase 2 (CDK2). Figure 6A shows that the expression of viral genes resulted in a significant increase in cyclin D1 and CDK2 protein levels (6.6- and 3.5-fold, respectively, compared to that for the control; P < 0.001 and P < 0.01, respectively). Overexpression of NDRG1 in the same cells partially reduced the levels of these cellular proteins (43.6% and 49.6%, respectively; Fig. 6B). However, as seen in the BrdU incorporation assay, the MCC cell lines MKL-1 and MKL-2 transduced with lentiviruses expressing NDRG1 in the presence of doxycycline showed differential regulation of cyclin D1 and CDK2 expression. A significant decrease in cyclin D1 expression along with a modest decrease in CDK2 expression was observed for MKL-2 cells, whereas no changes in expression of the two proteins were observed for MKL-1 cells (Fig. 6C). This justifies the observed discrepancies in BrdU incorporation observed for the MCC cell lines.

FIG 6.

Interrelation between the products of early genes of MCPyV, NDRG1 and cell cycle regulatory proteins CDK2 and cyclin D1 (A) Total protein lysates from cells stably expressing the early genes of MCPyV (MCPyV) or not (pLXSN) were prepared and immunoblotted for the indicated proteins mostly known to be involved in cell cycle regulation. (B) Protein lysates from cells expressing the early genes of MCPyV and overexpressing NDRG1 were subjected to Western blotting and probed with the indicated antibodies. (C) Lysates from MKL-1 and MKL-2 cells overexpressing NDRG1 were probed for the indicated proteins. (D to G) The knockdown of both LT and ST (PAN) MCC-positive cells (MKL-1, MKL-2, MS-1, and CVG-1 cells) was achieved by transduction with lentivirus-based shRNA, as described in Materials and Methods. Scrambled shRNA (Scr) was used as a negative control. Immunoblot analysis for NDRG1, β-catenin, CDK2, cyclin D1, and LT was performed in MKL-1 (D), MKL-2 (E), CVG-1 (F), and MS-1 (G) cells. β-Actin was used as a loading control. (H) Total protein lysates from cells stably expressing the early genes of MCPyV transfected with cyclin D1 siRNA (si cyclinD1) or control siRNA (si control) were prepared and immunoblotted for expression of cyclin D1. (I and J) Cells transfected with NDRG1 or not were plated in 6-well plates at a 1:100 ratio for 7 to 8 days, as described in Materials and Methods. Representative images show the colonies (I), and the number of cells per colony is shown in the bar graph (J). **, P <0.01 and P > 0.001.

Furthermore, to understand whether MCPyV early gene expression has any impact on the regulation of NDRG1, CDK2, and cyclin D1 in different MCC cell lines, we knocked down the early genes in MKL-1, MKL-2, MS-1, and CVG-1 cells and checked their protein expression by Western blotting. The immunoblot showed an increase in NDRG1 expression in all four cell lines upon transduction with PAN shRNA compared to that for the controls (Fig. 6D to G). However, MKL-2, MS-1, and CVG-1 cells showed decreases in cyclin D1 and CDK2 levels upon ST and LT knockdown (Fig. 6D to G). Thus, while in the less transformed hTERT-HK NDRG1 overexpression resulted in a reduction in the levels of positive cell cycle regulators, this phenomenon was conserved in only some Merkel cancer cell-derived cell lines, possibly due to a different status of their cellular transformation. In any case, silencing the expression of the viral oncogenes rescued NDRG1 protein levels in all analyzed Merkel cancer cell-derived cell lines.

To further characterize the role of cyclin D1 in the proliferation of cells expressing the MCPyV early genes, we silenced cyclin D1 expression in MCPyV-hTERT-HK by small interfering RNA (siRNA) (Fig. 6H). Although we did not observe significant changes in the total number of colonies of the mock-transfected and cyclin D1 siRNA-transfected cells, silencing of cyclin D1 expression resulted in a strong reduction of the colony size (Fig. 6I and J), highlighting the inhibition of cellular proliferation. These results further indicate that NDRG1 might involve multiple downstream effector molecules, validating our previous observation (Fig. 5D).

DISCUSSION

In this study, we performed comparative gene expression profiling of cells expressing the early genes of 6 PyVs, namely, BKPyV, JCPyV, KIPyV, MCPyV, SV40, and WUPyV, with the aim of identifying the unique features of MCPyV that endow it with oncogenic characteristics. Our results showed that, in comparison to the other 5 PyVs, MCPyV uniquely deregulated 28 genes, with 13 genes displaying upregulation and 15 genes displaying downregulation. We hypothesized that the specific deregulation of these 28 genes may be due to the direct oncogenic potential of MCPyV early protein activities, as the transforming nature of these oncoproteins has been described previously (13). Also, comparisons between our data sets and data sets from different publications revealed differences in the deregulated gene lists. This may be due to the experimental models, cell types, types of transfection, or technologies used to appreciate the changes in gene expression, which were not exactly the same as ours, leading to a not so perfect overlap. The same can also be considered true for the gene list comparisons that we found for SV40, BKPyV, and JCPyV. In fact, MCPyV distinctly downregulated certain genes, including tumor suppressor genes, such as NDRG1, INPP4B, KLF6, TRIB1, and ANXA2. The reduced expression of these tumor suppressor genes has been reported in several cancer types, including lung, prostate, and breast cancer (36 – 38). Strikingly, the genes which were specifically upregulated by MCPyV also included oncogenes, such as FOXQ1, DUSP10, and CTSH. The unique ability of MCPyV to suppress tumor suppressor genes and activate oncogenes may explain its high oncogenic potential in comparison to that of other human PyVs, as virus-mediated cancers are thought to be the result of the accumulation of genetic alterations induced by their oncoproteins.

Genetic alterations disturbing cell cycle regulation are one of the leading causes of cancer development. In fact, cell cycle progression is a firmly controlled process where cyclins, CDKs, and CDK-interacting protein/kinase-inhibitory proteins (cip/kip family) coordinate to ensure the proper transition of the cell cycle across cell cycle checkpoints (28). Certain oncoviruses have evolved diverse strategies to deregulate cell cycle progression because the loss of proper cell cycle control is one of the major driving forces in cellular transformation (39). In fact, dsDNA viruses, such as human papillomaviruses (HPVs) and PyVs, depend on the cell cycle machinery for replication. Therefore, they need to push the cell into S phase of the cell cycle. Although the PyV early region gene products have been reported to target cellular proteins implicated in cell cycle regulation, little is known about cell cycle regulation for cells infected with MCPyV (1).

Our results showed a significant modulation in the expression of genes related to particular cell cycle genes involved in the G₁/S checkpoint, showing that cell cycle regulation is a highly influenced pathway in MCPyV-infected cells. Several CDK inhibitors, including CDK inhibitor 1A (CDKN1A), CDK inhibitor 2B (CDKN2B), CDK inhibitor 2C (CDKN2C), and CDK inhibitor 2D (CDKN2D), are known to prevent the progression of the cell cycle and are strongly downregulated by MCPyV (Fig. 1D; see also Table S3 in the supplemental material). Interestingly, the regulation of CDKN1A by KLF6 has been reported (37), and we also found CDKN1A to be strongly downregulated by MCPyV. In addition to some other CDKs, CDK4 is strongly upregulated in the presence of MCPyV early genes (Fig. 1D). Furthermore, there was an observable slight upregulation of some cyclins, including CCNE1 (cyclin E1), CCND3 (cyclin D3), CCND2 (cyclin D2), and core cell cycle regulation gene CDC25, with a marginal suppression of retinoblastoma 1 (RB1) in the presence of MCPyV early genes. All these indicators of cell cycle modulation suggest that MCPyV favors S-phase progression. However, the impact of MCPyV on cell cycle deregulation needs to be further elucidated. There is evidence that MCPyV early proteins promote cell growth, as illustrated by the reported robust cell death and cell cycle arrest associated with the inhibition of early gene expression, using PAN shRNA in MCC cell lines (13). The requirement of an intact pRb binding site by MCPyV to induce cellular growth shed light on the fact that MCPyV targets the cell cycle to increase its replication (40). Moreover, a MCPyV-infected MCC cell line showed impaired cell cycle arrest following exposure to UV radiation. Subsequently linked to LT, these results suggest that the presence of the virus affects the normal cell cycle in cells exposed to UV (41).

Our results further indicated that among the genes deregulated specifically by MCPyV, NDRG1 was reproducibly downregulated in all model systems. NDRG1 is known to be a tumor suppressor and metastasis suppressor in a variety of cancer cells, like those of the brain, breast, colon and rectum, pancreas, prostate, and esophagus, whereas it is also known to promote tumorigenesis in some other forms of cancer of the kidney, liver, mouth, skin, and uterine cervix (42). The mechanism by which NDRG1 exerts its effect is largely dependent on the cellular context (42). In MCF-7 breast and EJ bladder cancer cell lines, NDRG1 overexpression was shown to decrease cellular proliferation (21). Similarly, in pancreatic cancers, NDRG1 overexpression led to an inhibition of tumor growth and an increase in apoptosis (43). Keeping in line with these observations, our results also showed that NDRG1 overexpression served to induce cell arrest in MCPyV early gene-transduced cells, as was evident from the reduced number of cells in S/G₂ phase compared to that for the controls. These results further imply that MCPyV downregulates NDRG1 to aid cell cycle progression, thereby promoting cell survival.

Our investigation further revealed the wound healing capacity to be severely compromised in MCPyV-positive cells overexpressing NDRG1. However, the majority of the work with NDRG1 failed to unravel the underlying mechanism responsible for these diverse biological effects. One of the reports suggested that NDRG1 expression results in ATF3-mediated expression of the KAI gene to inhibit metastasis in prostate cancer (44). Another study highlighted the upregulation of PTEN, SMAD4, and NEDD4L to be important contributors to the antitumor effect of NDRG1 (45). However, a number of studies support the fact that NDRG1, via its interaction with either the Wnt receptor (46) or glycogen synthase kinase 3β (47), leads to the regulation of the β-catenin distribution and activity, affecting cell proliferation and migration. We show here that NDRG1 overexpression has important effects on the expression of CDK2 and cyclin D1, which are important regulators of the cell cycle checkpoint.

Even though the work with MCPyV-hTERT-HK showed clear roles of NDRG1 in limiting cellular proliferation via the inhibition of cell cycle progression, NDRG1 overexpression in MCC cell lines showed that differential mechanisms by which NDRG1 mediates its effects in the MKL-1 and MKL-2 cell lines may exist. This may be partly attributed to the fact that cell lines may have accumulated changes over time, owing to continuous passage, leading to differential outcomes. However, as seen with 3 of 4 MCC cell lines, MKL-2, MS-1, and CVG-1 cells, there was a decrease in expression of CDK2 and cyclin D1 upon early gene knockdown along with an increase in NDRG1 expression, indicating a redundant role of NDRG1 in regulating these cell cycle regulators. Further studies are needed to understand what features leading to the observed discrepancies in NDRG1 expression distinguish one MCC cell line from the others. This is in line with previous observations, where expression of a protein (survivin) was differentially regulated between MS-1 cells and the three other cell lines under study, including MKL-1 cells (48).

Recently, it was shown that NDRG1 is important to hepatitis C virus assembly by regulating the biogenesis of lipid droplets, considered to be the main site for virus assembly (49). In yet another study, it has been described that microRNAs encoded by Epstein-Barr virus (EBV) can downregulate NDRG1 to promote EBV-mediated epithelial carcinogenesis (50). This might be another aspect of the regulation of viral loads for which the viruses have developed strategies to downregulate this versatile tumor suppressor. While evaluation of this feature of NDRG1 was beyond the scope of the present investigation, it will be interesting to have a deeper understanding of the role of NDRG1 in MCPyV assembly, as little is known in this area. Moreover, the mechanism of NDRG1 downregulation by viral early genes needs to be investigated, as a clear decrease in the level of NDRG1 expression was seen at both the transcriptional and translational levels.

In summary, we showed that MCPyV deregulates the cell cycle by specifically modulating genes associated with cell cycle regulation and MAPK pathways and that NDRG1 is a key player in cell cycle arrest and migration by mediating its effect in downregulating cyclin D1 and CDK2 in Merkel cell carcinoma.

MATERIALS AND METHODS

Expression vectors.

All expression vectors for the early genes of BKPyV, JCPyV, KIPyV, SV40, and WUPyV were prepared as described before (24). Full-length MCPyV early genes were a kind gift from D. A. Galloway (Fred Hutchinson Cancer Research Center, Seattle, WA, USA). NDRG1 (length, 1,185 bp) was subcloned into retroviral vector pBABE-Hyg or lentiviral vector pLenti TRE empty EF-puro (51). To knock down ST alone or both LT and ST mRNA expression in MCC cells (Fig. 3C and D), pLKO sh sT1 (ST) and pLKO sh pan-T1 (PAN) were used together with a control pLKO shCtrl construct (scrambled shRNA [Scr]) (13). In Fig. 6, pan-T antigen (PAN) knockdown was performed by pLenti e7SK-shpanT-puro and pLenti e7SK-Ctrl-puro (Scr), in which shRNA sequences identical to those for pLKO sh pan-T1 and pLKO shCtrl, respectively, were cloned under the control of an e7SK promoter (51). Lentivirus was produced as described previously (13).

MCPyV-positive MCC cells (MKL-1, MKL-2, CVG-1, and MS-1 cells) were cultured in RPMI 1640 medium (Invitrogen Life Technologies, Cergy-Pontoise, France) supplemented with 10% fetal bovine serum (PAA, Pasching, Austria), 100 U/ml penicillin and 0.1 mg/ml streptomycin (Pen-Strep; Gibco, Invitrogen), 2 mM l-glutamine (PAA), and 1 mM sodium pyruvate (PAA) (11, 13). NIH 3T3, 293FT and Phoenix cells were cultured according to a previously described protocol (13, 52). Naturally immortalized keratinocytes (NIKs) and hTERT-HK were grown together with NIH 3T3 feeder cells in FAD medium, containing Ham’s F-12 medium (PAA), Dulbecco modified Eagle medium (Gibco), 2% fetal calf serum (PAA), and 100 U/ml penicillin and 0.1 mg/ml streptomycin (Pen-Strep; Gibco, Invitrogen), adenine (Sigma), 10 ng/ml human epidermal growth factor (R&D), 5 mg/ml insulin (Sigma-Aldrich), 400 μg/ml hydrocortisone (Sigma), 10 mg/ml ciprofloxacin hydrochloride (Euromedex), and 20 mg of cholera toxin (List Biological Laboratories). All cells were cultured at 37°C with 5% CO₂.

Retroviral and lentiviral infections.

Retroviral transduction of keratinocytes with the early genes from PyVs cloned in pLXSN was performed as previously described (24, 52). After viral transduction, keratinocytes were selected in medium containing 1 mg/ml G418 (PAA). Lentiviral transductions were performed as previously described (13). Briefly, MCC cells infected with the lentiviruses indicated above were selected for 6 days with 1 μg/ml puromycin. After puromycin selection of infected cells, fresh cell culture medium was added to recover the cells and the cells were harvested for analysis, as indicated above. For NDRG1 experiments, 0.5 μg/ml of doxycycline was added to the puromycin-selected cells.

Transfection.

For transfection, cells were plated in 6-well plates and transfected with control or cyclin D1 siRNA (Dharmacon) using the Lipofectamine 2000 reagent according to the manufacturer’s protocol.

Image acquisition and processing.

Images were acquired directly in 6-well plates using a Nikon Eclipse Ti wide-field inverted microscope. The images thus captured were analyzed using ImageJ software.

mRNA extraction and quality control.

For microarray analysis, total RNA was extracted from NIKs expressing the early genes of the 6 PyVs, by using an Absolutely RNA miniprep kit (Stratagene) according to the manufacturer’s protocol. RNA integrity was characterized by determining the RNA integrity number (RIN) and quantification was performed by measuring the 28S/18S rRNA ratio by using an Agilent 2100 bioanalyzer instrument and an RNA 6000 Nano kit.

Genome expression profiling.

Genome-wide gene expression profiling analysis was performed using Illumina HumanHT-12 (v4) expression bead chips, providing coverage of more than 24,000 annotated genes (47,231 probes corresponding to 1 to 3 probes per gene), including well-characterized genes and splice variants derived from the National Center for Biotechnology Information (NCBI) reference sequence. Using an Illumina TotalPrep RNA amplification kit (Ambion), 500 ng of the extracted RNAs was converted to cDNAs and subsequent biotin-labeled single-stranded cRNAs. The distribution of homogeneous in vitro transcription products (cRNAs) was checked with the Agilent bioanalyzer instrument and the RNA 6000 Nano kit. The biotin-labeled cRNAs (750 ng) were then hybridized overnight to HumanHT-12 expression bead chips. Subsequent steps included washing, streptavadin-Cy3 staining, and scanning of the arrays on an Illumina bead array reader. The fluorescence emission by Cy3 was quantitatively detected for downstream analysis. Illumina GenomeStudio software (v2010.2) was used to obtain the signal values (AVG Signal), with no normalization and no background subtraction being performed.

Data analysis.

The quality of the bead array data was verified using the internal controls present on the HumanHT-12 bead chip and was visualized as a control summary plot. Nonnormalized raw data were then imported into BRB-ArrayTools software (v4.3.0; developed by Richard Simon and the BRB-ArrayTools Development Team) for downstream analyses. Background subtraction, color correction, and Simple Scaling normalization were performed using the lumi R package plug-in (53). The quality of the data was further checked by generating boxplots of total gene expression data, principal-component analysis, unsupervised clustering using centered correlation and average linkage of replicates, and the consistency between duplicate probes. For class comparisons, genes were considered differentially expressed between groups when the mean expression was at least 1.5-fold different (up or down) with a corrected P value and false discovery rate (FDR) of less than 0.001. The test was based on comparison of the differences in the mean log intensities between classes relative to the expected variation in the mean differences. Technical replicates were averaged before class comparison.

Gene Ontology analyses were performed with the WebGestalt web-based gene set analysis tool kit using the whole human genome as a reference. In addition, gene set comparison was done in BRB-ArrayTools for Gene Ontology categories and biological pathways (BioCarta and KEGG). The gene set comparison tool analyzes predefined gene sets for differential expression among predefined classes. The significance values are based on testing of the null hypothesis that the list of genes that belong to a given GO category is a random selection from the project gene list against the alternative hypothesis that it contains more genes differentially expressed between the classes being compared. Pathway analysis was performed using the Reactome Pathway database (54).

RT-PCR and quantitative PCR.

Quantitative real-time PCR (qRT-PCR) was performed as previously described (55). Briefly, total RNA was extracted from cells using NucleoSpin RNA (Macherey-Nagel) and reverse transcribed to cDNA by using RevertAid H-Minus Moloney murine leukemia virus reverse transcriptase (MBI; Fermentas) according to the manufacturer’s protocol. The primer sequences used for qRT-PCR are listed in Table S5 in the supplemental material.

Immunoblotting and antibodies.

Whole-cell lysates were prepared, and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting (IB) were performed according to previously described protocols (55, 56). Protein concentrations were measured with the bicinchoninic acid (BCA) assay reagent; 30 to 40 μg of protein extracts was used for SDS-PAGE and immunoblot analyses, and IB was performed according to the previously described method (55, 56). The antibodies used for IB were those to β-actin (MP Biomedicals), NDRG1 (Cell Signaling), cyclin D1 (Cell Signaling), CDK2 (Pharmingen), and MMP-7 (Santa Cruz).

Cell cycle analysis.

Analysis of the cellular DNA content was ascertained by propidium iodide (PI) staining to determine the proportion of cells in different phases of the cell cycle. Cells were harvested by trypsinization and fixed using ice-cold 70% ethanol at 4°C for 30 min. The cells were pelleted and washed with phosphate-buffered saline (PBS) 3 times. The cells were resuspended in 500 μl of PBS with 20 μg/ml PI and 10 μg/ml RNase A and incubated at room temperature for 30 min. Analysis was performed using a FACSCanto flow cytometer.

BrdU (10 μM) was added to MCC cells expressing or not expressing NDRG1 for 1 h before harvesting. The cells were fixed in 10% buffered formalin for 10 min, denatured with 2 N HCl for 30 min, and permeabilized in 0.3% Triton X-100–PBS for 10 min at room temperature. The cells were neutralized and incubated with anti-BrdU antibody (1:2,000; Cell Signaling) in 1% bovine serum albumin (BSA)–PBS overnight at 4°C. The cells were washed with 1% BSA–PBS once and incubated with secondary anti-mouse IgG conjugated to Alexa Fluor 488 (1:1,000; Invitrogen) for 1 h at room temperature. The cells were washed; suspended in PBS containing RNase A (100 μg/ml), propidium iodide (50 μg/ml), and 0.05% Triton X-100; incubated for 30 min at 37°C in the dark; and then analyzed with a BD Accuri C6 flow cytometer (Becton Dickinson).

Colony formation assay.

After 24 h of transient transfection, hygromycin was added to the medium and cells were selected for 72 h. Thereafter, cells were split at 1:10, 1:100, or 1:1000 and allowed to grow for several days under hygromycin selection (57). The cells were washed in phosphate-buffered saline (PBS), fixed, and stained with crystal violet in 20% methanol, and the number of colonies was counted.

Wound healing assay.

Transfected cells were wounded in a continuous straight line using a pipette tip. The images were obtained at 0, 24, and 48 h postscratch under a light microscope using an objective with ×5 magnification (57).

Cell proliferation assay.

At day 6 postransduction, 2.5 × 10⁴ cells were seeded in a 96-well plate (day 0). Cell proliferation was measured using the WST-8 reagent (Wako) at days 1, 3, 5, 8, 10, and 12. Optical density (OD) values were normalized by the values from day 1. The WST-8 formazan product was measured at 440 nm with a reference filter at 600 nm.

Statistical analyses.

Student's t test was applied to check the statistical significance of the obtained data. The error bars in the graphs represent the standard deviation (SD).

Data availability.

The data set analyzed during the current study is available in the Gene Expression Omnibus database under accession number GSE137328.