Background
Simian virus 40 (SV40) was first recognized and isolated during the late 1950's [
1] and recently achieved fame because it was carried over inadvertently as live virus into poliovirus vaccine preparations from 1955-1963 in the U.S. and elsewhere [
2]. Approximately 60% of the population in the U.S. and abroad was exposed to SV40. Initially this caused little alarm, but the virus was later found to induce mesotheliomas in hamsters [
3,
4] and afterwards was found in a high percentage of certain types of human cancers, especially mesotheliomas, but not in surrounding tissues [reviewed in [
2]]. Discussions and investigations regarding the molecular identity of the SV40 isolates, revealed the sequences found in cancers were 'wild type', not laboratory strains [
5], ruling out artifacts. Retrospective studies on human cohorts inadvertently exposed to SV40 via poliovirus vaccine increased the level of concern. A two-fold elevation in the risk of neural cancers was noted in the children of 50,000 individuals exposed to SV40 during pregnancy [
5], though study design criticisms were registered. A three-fold elevation in the incidence of mesothelioma was reported in infants and children in an exposed cohort [
2], and other studies reviewed therein also indicated an elevated risk of brain tumors. SV40 seroprevalence in children born in Texas from 1980-95 indicates that endemic levels of infection are 5.9% [
6], or, as reviewed in Butel and Lednicky, from 3 to 13% of the number of persons not exposed to vaccine [
5]. A recent investigation points to an actual prevalence of 2% [
7] after correcting for cross-reactivity to JC and BK viruses. In Finland, where SV40 was not a contaminant in poliovirus vaccine, the seroprevalence is zero. A vaccine against SV40 [
8] is being developed.
There is a wealth of information about the mechanisms of action of SV40 in rodent and human cells
in vitro and
in vivo[
9‐
13]. SV40 Tag was found to bind and inactivate p53 and pRB, abrogating apoptotic mechanisms and control of cell proliferation, allowing cellular overgrowth and escape from senescence, and interestingly Tag-p53 complexes bind and activate the IGF-1 promoter, resulting in increased malignant cell growth [
14,
15]. Tag also binds the co-activators of IRF transcription, p300 and CBP [
12]. Small t antigen (tag) immortalizes cells in conjunction with Tag [
9], through binding and inhibition of protein phosphatase 2A. Information concerning the permissivity of human cells for SV40 has emerged [
6,
16] whereby p53 binding to Tag partially inactivated viral replicase activity, and then cells could support an active (but low-level) infection without a lytic component, together with a level of Tag presence which failed to activate a full-blown immune response [
6]. Episomal and DNA-integrated viral replication are possible during such infection. For a more recent and extremely thorough treatment of SV40 infection, latency, and transformation of human mesothelial cells, see [
17]. Testa et al. [
10] suggest that asbestos and SV40 are co-carcinogenic, which may contribute to the long latency period between asbestos exposure and the development of mesothelioma (which is most often associated with the presence of SV40 virus in humans, and also with asbestos exposure). Definitive recent work by Kroczynska, et al. [
18] demonstrated that crocidolite asbestos and SV40 are co-carcinogens in human mesothelial cells, and in causing mesothelioma in hamsters. In that study, SV40 did not cause malignant mesothelioma
per se, but increased the incidence from 20% (crocidolite alone) to 90% (crocidolite + SV40).
3-MC is a well-known, thoroughly investigated, potent human and animal hepatocarcinogen of the type known as polycyclic aromatic hydrocarbons (PAH), which bind to the cytosolic Ah receptor, translocate to the nucleus via association with ARNT, and in association with ARNT bind DNA, activating transcription of genes containing XRE
's or ARE's [
19] (such as
CYP1A1) and eliciting an AP-1 antioxidant response. The gene expression of 3-MC has been investigated in exposed rat kidney/liver, and mouse liver [
20‐
22] and
in vitro in rat hepatocytes [
23], and compared to other hepatotoxicants. Gene expression changes included the induction of
GSTμ, CYP1A1 and
A2, and several acute phase proteins in the liver, and
CYP1A1 and
A2 in the kidney [
21]. PAH also form direct protein-DNA adducts [
19].
However, the gene expression patterns induced by 3-MC in conjunction with SV40 used as an immortalizing principle have not been described. Human uroepithelial cells immortalized with SV40 (hereafter referred to as HUC) were compared to the descendant MC-SV-HUC-T-2 (hereafter referred to as HUC-TC) line which was immortalized by SV40 and subsequently transformed to tumorigenicity using 3-MC, in order to observe specific gene expression changes induced by the transforming agent. Previously, Reznikoff et al. [
24] developed these cell lines and showed that treatment of HUC with SV40 followed by 3-MC, but not with either treatment individually, produced tumors in athymic mice. In the present experiment, we expected to see up-regulation of oncogenes, down-regulation of tumor suppressor genes, and other evidence of activation typical of cancer cell lines. In actuality, many were clearly virally-related when compared to the already virally-immortalized HUC, indicating a possible new interaction between viral elements and 3-MC during cellular transformation to full tumorigenicity. Here we explore those changes and discuss their possible biological significance.
Discussion
The morphology, growth characteristics, phenotype, karyotype, and ultrastructure of these cell lines were extensively described previously [
24,
28‐
35]. The parent HUC non-transformed cell line did not produce tumors after inoculation
in vivo up through at least passage 80 (over 3 years) in culture. However, the parent cell line was highly unstable chromosomally. Wu et al. [
33] demonstrated that marker chromosomes of 3 tumor cell lines (HUC-TC) were stabilized relative to the parent non-transformed cell line, by malignant transformation. HUC-TC were transformed at passages 12-15, and we obtained cells from the repository that were passage 14. We used these cells at passage 19. We obtained the parent HUC non-transformed cell line at passage 32 and used it at passage 38. We inoculated these HUC-TC (passage 19) into athymic mice and tumors were produced in the same manner as the original experiments [
24], (data not shown). Given the previous extensive characterization of these cells and the limited number of passages that elapsed between the time we obtained and used the cells for experimentation, the likelihood of significant alterations in the genome is limited, but cannot be completely ruled out.
It was expected that the gene expression results would strongly reflect the 3-MC treatment. We chose to use the human cancer array and therefore changes in other metabolic genes such as CYP1A1, which is also known to occur upon 3-MC treatment, were not measured. The gene expression changes seen upon comparing HUC with HUC-TC were surprising in that they were highly related to SV40 treatment although both cell types had been SV40-treated. It appeared that a non-transient (i.e., present through several cell passages) expression and enhancement of anti-viral responses occurred in HUC-TC as a result of the treatment with 3-MC. Below we discuss how this activity might result in carcinogenesis.
Cellular antiviral responses typically begin with host cell recognition of the internal presence of SV40 double-stranded RNA, an indicator of viral replication [
25]. The response includes up-regulation of
IFN's α/β/γ, with multiple effects such as up regulation of the expression of
2',5'-OAS 1 and 2, seen here, activating the RNase L homodimer. Active RNase L then cleaves double-stranded viral RNA and stimulates apoptosis. But clearly apoptosis was not activated (Fig.
1, which is a representative image, shows no apoptotic cells or bodies and Fig.
2 shows that both cell types had a healthy growth rate in culture). The activation of PKR by type I interferons would then typically result in binding of eIF2a to GDP and eIF2b, a recycling factor for eIF2a, inactivating eIF2a and blocking the initiation of protein translation. (The down-regulation of the related
eIF4E 25-kDa subunit, in this experiment, may have been another mechanism of decreased protein translation.) PKR then typically activates NFκB, which translocates to the nucleus, binds DNA in the promoter regions of NFκB-responsive genes, and initiates transcription of proliferation-related or stress-responsive genes, the latter of which lead to apoptosis. PKR activation blocks viral transcription and translation, as does the up regulation of
MxA and
MxAB (up regulated here) in response to interferons. Here, PKR may have stimulated pro-proliferative genes but pro-apoptotic genes may have been incompletely or improperly activated, or such activation may have been ineffective due to the up regulation of opposing signals (see below). Waring, et al. [
20] have identified a gene expression profile that is similar to that of 3-MC and mediates hepatic toxicity through the AhR either directly or through the effects on NFκB, resulting in the inhibition of cell adhesion protein expression. If such a pathway acts through NFκB, it may be similar to the PKR-mediated NFκB activation pattern seen here, producing a tumorigenic phenotype. Additional pro-apoptotic elements were up regulated (
endonuclease III, caspase 4, caspase 10, death receptor 5 (TRICK2A), TNFRSF25 (WSL + TRAMP + Apo-3 + DDR3)) however these cells were not apoptotic. The reason for unchecked proliferation may be related to the up regulation of multiple blockers of apoptosis (
DR1, HSIAH2, IAP3, DAP1), known to act either as decoys that bind and inactivate apoptotic ligands, or act upstream of the caspases. In addition, pRB is known to be bound by Tag, nullifying cell cycle checkpoint control. p53 protein was at least partly functional in these cells, as we noted several
p53-inducible gene expression increases, as well as
mdm2 up regulation. However Tag is known to bind p53 and render it incapable of initiating apoptosis. Although p53 and pRB binding by Tag can account for both loss of apoptosis signaling and checkpoint control, there were many other changes at the mRNA level related to these important functions and indicative of cellular dysregulation.
Cell-cycle arrest was signaled as well, since
p21waf 1/cip 1(up regulated) is a p53-inducible universal CDK inhibitor and its up regulation is known to inhibit cell proliferation. The response was clearly not successful, most likely due to pRB-Tag binding. Tag was present in these cell lines [
32], and there was evidence of an increase in the rate of proliferation in HUC-TC vs. HUC. Other cell-cycle genes up regulated include
CDK4/cyclin D2 and
CDK7. CDK7 together with cyclin H forms CAK, a kinase required for CDK activation. Although
p16ink 4was up regulated, it could not bind pRB, which would have been already bound by Tag, and so could not block cell cycle progression. Ultimately, apoptosis was blocked and cell-cycle control circumvented (Fig.
5,
6,
7).
These results imply stimulation of IFN-γ-related pathways by 3-MC. Treatment with exogenous IFN-γ blocked cell proliferation in tumor, but not non-tumor HUC. However metabolic activity (as assayed by MTS) was decreased in both cell lines treated with IFN-γ from day 4 onward. Since there was no elevation in the level of secreted IFN α or γ, and many IFN-γ-inducible transcripts were increased, we conclude that 3-MC treatment activated IFN pathways without affecting constitutive levels of IFN. An hypothesis is that activation of IFN-γ-related pathways by 3-MC rendered HUC-TC susceptible to growth-suppression by exogenous IFN-γ. These data support the idea that during immortalization (first step, SV40-mediated) cells become unresponsive to IFNγ-mechanisms of cell cycle control, but subsequently, during transformation (second step, 3-MC-mediated) cells are altered in such a way that they are rendered sensitive to IFNγ control of cell proliferation, but by then it is too late because other aspects of cellular function controlling growth have been irrevocably altered. The cell cannot 'retreat' along the pathway to which it has become immutably committed, i.e. immortality. The 'coup de grace', 3-MC transformation of the primed cell population, might then be facile.
Clearly the IFN-γ pathways activated by 3-MC were not
intrinsically growth-suppressive in nature, since HUC-TC exhibited more rapid growth than HUC in the absence of treatment with exogenous IFN-γ. Activation of IFN-γ-inducible gene expression may represent dysregulation of homeostatic IFN-γ pathways. This raises the question of how the altered pathways promote tumor growth and metastasis. We would remind the reader that it is known that a slight deviation in one or more components of a growth-suppressive pathway may alter the function of the entire pathway, achieving the opposite effect, e.g. TGFβ signalling either promoting or suppressing tumors [
36]. Demonstration of the suppressive effects of IFN-γ on cancer cell growth both
in vitro and
in vivo has been unequivocal [
37‐
39] and the production of IFN-γ in response to chemotherapy is one marker used to assess the success or failure of treatment
in vivo; it is considered an indicator of immune activation and anti-tumor activity [
40,
41]. In addition, studies of infectious diseases (scrapie, herpes simplex virus type 1, human cytomegalovirus, human influenza virus C, type 1 diabetes) have linked IFN-γ inducible gene expression with the presence of disease and/or anti-viral mechanisms [
42‐
45]. In a recent study employing HTLV-1 to transform Rat-1 fibroblast cells
in vitro, four of eight up regulated genes were IFN-stimulated genes, and the 2',5'-OAS promoter was activated by viral Tax indirectly through an NFκB-dependent pathway, linking IFN-signaling with Tax transformation [
46]. We observed evidence of PKR up-regulation, which can directly activate the NFκ-B pathway. The present study may be another example of transformation that occurs via IFN-γ pathways.
The most highly up regulated gene was
lipocalin (
neutrophil gelatinase-associated lipocalin, NGAL). Lipocalin has been found in a high molecular weight complex associated with progelatinase B, one of several gelatinase isoforms commonly found in the urine and cancerous tissues of bladder cancer patients [
47]. Since 3-MC is a known human uroepithelial carcinogen which has previously tested positive for tumorigenicity in mice
in vivo[
48], this transcriptional up regulation appears to be a correlate for urothelial malignancy, and makes
lipocalin a biomarker of exposure as others have observed for bladder cancer [
49], as well as colorectal cancer [
50], and as reviewed in [
51].
These data provide evidence that SV40 and 3-MC may act synergistically to promote transformation to a tumorigenic phenotype. Inoculated HUC-TC produced tumors in athymic mice as in the original experiments and we found gene expression changes related to viral elements to be up regulated, along with many interferon-responsive genes. We hypothesize that first SV40 infection altered cellular pathways related to cell cycle control and apoptosis, then 3-MC exposure initiated changes in interferon response elements, matrix attachment proteins, DNA damage responses, and activation of oncogenes and/or inactivation of tsg's, resulting in transformation to the malignant phenotype. This may be relevant to human environmental exposures.
Methods
Cell Culture, Morphology and Histologic Staining
HUC and HUC-TC were obtained from The American Type Culture Collection, Rockville, MD at passage 32 (HUC) and 14 (HUC-TC). HUC (passage 38) and HUC-TC (passage 19) cells were plated onto 150 mm dishes (five per treatment) at a density of 1 × 105 cells/mL and allowed to reach 80 - 90% confluence, or 5 days, under standard culture conditions. Cells were fed three times per week (26 mL/dish MEM containing 10% FBS, 0.1 mM non-essential amino acids, 2 mM L-glutamine, 2.7 mg/mL D-glucose, 10 μg/mL insulin, 5 μg/mL transferrin, 0.1 μg/mL hydrocortisone). Glass coverslips were placed into the 150 mm dishes at the time of plating and carefully removed with forceps before RNA harvest was carried out on the remaining cells. Cells from glass coverslips were fixed by rinsing in room temperature sterile PBS and fixed in cold (-20°C) absolute ethanol for 24 h, stained using Hematoxylin and Eosin by standard histologic staining methods, and photographed.
MTS Assay for Cell Viability
The Promega Cell Titer 96 Aqueous One Solution™ assay was employed to measure the metabolic activity of IFN-γ-treated HUC and HUC-TC cells relative to control cells. This assay relies on the conversion of a tetrazolium compound (MTS) to a blue colored reduced formazan product, which requires cellular reducing capacity as NADH and NADPH. Cells that are not metabolically competent will not reduce MTS. Cells (HUC = passage 37, HUC-TC = passage 18) were plated at a density of 1.25 × 104 cells/mL into 96-well plates (Corning plastic, one plate/cell type/day for a total of 12 plates) and grown for 7 days. Cells were fed with fresh media, 1× [I] (where [I] = 8.3 ng/mL) or 100× [I], IFN-γ on days 2, 4 and 6. On days 2 - 7 one plate of each cell type was assayed using the MTS reagent. 20 μL of MTS reagent was added to each well and plates were incubated in the dark under standard tissue culture conditions for one hour. Optical density was measured on a Titertek Multiskan™ spectrophotometer at 490 nm. 8 wells were read per treatment condition, on each plate, and the readings averaged. Statistical analysis was carried out using an Excel™ spreadsheet and significance levels analyzed using a paired two-tailed t-test.
ELISA Assay for Interferon α and γ
Assays for quantitation of secreted interferons α and γ were performed in a 96-well format using commercially obtained assay kits. A Quantikine (DIF50, R & D Systems, Mpls., MN) kit was used for human IFN-γ including calibrated pure recombinant human interferon standards and a polyclonal antibody specific for human IFN-γ. A similar IFN-α kit was obtained from PBL Biomedical Laboratories, Inc. (product #41400, v.2.2). Standard curves for each were constructed and interferons were quantitated in pg/mL, according to manufacturers' instructions. HUC-TC cells (passage 38 or 39 for IFN-α or IFN-γ respectively) were plated at a density of 1.25 × 104 cells per mL into six dishes per cell type, and 100 μL of purified cellular supernatant per well was pipetted into the antibody-coated 96-well plate. The assay was carried out per the manufacturer's instructions, and results were read spectrophotometrically. Statistical analysis was carried out using an Excel™ spreadsheet.
In vitro IFN-γ Treatment of Cells
To assess the effect of IFN-γ on cell growth in culture, HUC (passage 40) and HUC-TC (passage 21) were treated with a known inhibitory concentration [I] of 8.3 ng/mL recombinant human IFN-γ (specific activity 2 × 108 U/mg, #I-3265, Sigma Chemical, St. Louis, MO) or control media 1 day post-plating, and grown for six days without media replacement. On day zero, cells were plated into 24 each 25 cm2 flasks at a density of 1.25 × 104 cells/mL (six control and six treated for HUC and HUC-TC). One dish from each treated and control dish was trypsinized using standard methods and counted each day beginning on day two post-plating. Counts were taken using a standard hemacytometer, in duplicate, and the results averaged. Significance was determined using an Excel™ spreadsheet and a paired two-tailed t-test.
RNA Preparation and Labeling of cDNA and Hybridization to Arrays
RNA was extracted by the addition of 14 mL TRIZOL™ reagent (Gibco BRL, Grand Island, NY) after triple rinsing with sterile room temperature PBS, according to the manufacturer's protocol. Six μg of total RNA per sample was reverse-transcribed and radioactively labeled using α
33P-dCTP in a previously-described PCR reaction [
52]. Labeled cDNA was hybridized overnight at 64°C and washed free of unhybridized cDNA in 0.5× SSC/1% SDS once, then twice in 2× SSC/1% SDS (20 min. per wash) at 64°C. Membranes (Atlas™ Human Cancer 1.2 Array [Cat. #7851-1], n = 5 per treatment) were exposed for 48 h to a rare-earth screen and read on a phosphorimager (Cyclone™ Scanner).
Data Manipulation & Statistical Analysis
The resulting intensities were uploaded into the Atlas Image 1.5™ software program (Clontech Inc., Palo Alto CA). Membranes were then aligned according to the manufacturer's instructions using the global normalization option and screened for bleed or other anomalies. The resulting reports were analyzed by group, for statistical significance (p < 0.05), using the NoSeCoLoR (NLR) software program, a normalization and local regression program [
53] as in previous studies [
52]. Statistically significant results were interpreted by use of current literature and diagrams constructed integrating experimental results with known biological pathways.
TaqMan™ Quantitative RT-PCR Confirmation of Selected Gene Changes
Using RNA from the same experiment as for gene expression, the expression changes of selected strong responding genes were confirmed using a Taqman™ real time quantitative RT-PCR assay (Perkin Elmer Prism ABI), as previously published [
52]. Primers were designed using Perkin Elmer Primer Express™, purchased from Keystone Biosource Inc. (Camarillo, CA) and prepared according to the manufacturer's instructions. The genes chosen for this assay were:
CDK4, DP2, p16ink 4,
β-actin (endogenous control),
FRA-1, GSH synthetase and p21waf 1/cip 1. These genes were altered on the array at p < 0.05, and were relevant to the mechanism of action, as observed by array results. The ΔΔCT method was used to calculate the fold-change in gene expression for the selected genes. β-actin was used as the endogenous control.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
LMC designed the study and carried out the in vitro experiments, analyzed the data and wrote the manuscript. TMM and MG carried out animal studies that were the basis for this study. LWY and HN performed Taq Man® real time RT-PCR and gene expression array experiments, respectively. KTM and ABD participated in planning the studies and analysis and interpretation of the data. All authors have read and approved the final manuscript.