The plasmacytoma variant translocation 1 gene (PVT1) is an lncRNA that has been designated as an oncogene due to its contribution to the phenotype of multiple cancers. Although the mechanism by which PVT1 influences disease processes has been studied in multiple cancer types, its role in cervical tumorigenesis remains unknown. Thus, the present study was designed to investigate the role of PVT1 in cervical cancer in vitro and in vivo. PVT1 expression was measured by quantitative PCR (qPCR) in 121 invasive cervical carcinoma (ICC) samples, 30 normal cervix samples, and cervical cell lines. Functional assays were carried out using both siRNA and LNA-mediated knockdown to examine PVT1’s effects on cervical cancer cell proliferation, migration and invasion, apoptosis, and cisplatin resistance. Our results demonstrate that PVT1 expression is significantly increased in ICC tissue versus normal cervix and that higher expression of PVT1 correlates with poorer overall survival. In cervical cancer cell lines, PVT1 knockdown resulted in significantly decreased cell proliferation, migration and invasion, while apoptosis and cisplatin cytotoxicity were significantly increased in these cells. Finally, we show that PVT1 expression is augmented in response to hypoxia and immune response stimulation and that this lncRNA associates with the multifunctional and stress-responsive protein, Nucleolin. Collectively, our results provide strong evidence for an oncogenic role of PVT1 in cervical cancer and lend insight into potential mechanisms by which PVT1 overexpression helps drive cervical carcinogenesis.
Citation: Iden M, Fye S, Li K, Chowdhury T, Ramchandran R, Rader JS (2016) The lncRNA PVT1 Contributes to the Cervical Cancer Phenotype and Associates with Poor Patient Prognosis. PLoS ONE 11(5): e0156274. https://doi.org/10.1371/journal.pone.0156274
Editor: Bibekanand Mallick, National Institute of Technology, Rourkela, INDIA
Received: August 15, 2015; Accepted: May 11, 2016; Published: May 27, 2016
Copyright: © 2016 Iden et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This work was supported by the Women's Health Research Program at the Medical College of Wisconsin (http://obgyn.mcw.edu/research/whrp/). The funders had no role in any part of preparation of this manuscript.
Competing interests: The authors have declared that no competing interests exist.
Long noncoding RNAs (lncRNAs) provide important targets for cancer diagnostics and therapeutics due to their critical role in numerous cellular processes such as epigenetic changes, gene enhancer and tumor suppressor activity, and miRNA sequestering. LncRNAs are pervasive in the genome, frequently show cell type- and temporal-specific regulation of gene expression, and can influence many cellular processes via multiple disparate mechanisms . Plasmacytoma variant translocation 1 (PVT1) is a highly conserved lncRNA ~50kb downstream of MYC that has attracted significant attention from the cancer field due to its frequent co-amplification with MYC in several solid tumors . The first studies providing evidence that PVT1 may contribute to carcinogenesis demonstrated frequent translocations in mouse plasmacytomas [3,4] and human Burkitt’s lymphomas [5–7]. The oncogenic effects of PVT1 have been further highlighted by more recent studies demonstrating its overexpression and amplification in multiple cancer types [8–17]. More importantly, PVT1 expression has been significantly correlated with clinical features such as risk, recurrence, and survival in various cancers [8,11–13,18].
Despite the wealth of knowledge regarding the oncogenic properties of PVT1 in multiple cancers, very little is known about its precise biologic function. In fact, the handful of studies providing PVT1 mechanistic data exceedingly suggest that it exerts its effects in a cell-type and/or disease-specific manner. For example, PVT1 function has been attributed to its binding and stabilization of the Myc  and Nop2  proteins in breast and hepatocellular carcinoma, respectively. In gastric cancer cells, PVT1 acts to repress the expression of p15 and p16 via its physical interaction with the polycomb group protein, EZH2 . Also via EZH2 recruitment and regulation of thyroid-stimulating hormone receptor, PVT1 induces proliferation of thyroid cancer cells . Finally, computational analysis of PVT1 suggests that it may act via binding and sequestration of mir-200 family members in breast cancer tissue . Due to their complexity and breadth of results, these studies emphasize the importance of disease-specific investigation of PVT1 mechanisms.
Our interest in PVT1 originated from our work and others’ demonstrating that it is a frequent site of HPV integration in cervical cancer, further suggesting an important biological function [22–24]. Thus, we sought to better define the role of PVT1 in cervical carcinogenesis by examining PVT1 expression patterns in cervical cancer tumors and cell lines and the functional effects of this lncRNA on cancer-related processes such as proliferation, cell motility, apoptosis, and chemoresistance. Finally, because many lncRNAs rely on protein partners to carry out their effects, we employed RNA affinity chromatography and mass spectrometry sequencing to elucidate cervical cancer cell-specific PVT1 binding partners.
The study protocol was approved by the institutional review board of the Medical College of Wisconsin IRB (PRO00011466; Molecular Characterization of Cervical Cancer).
SiHa (ATCC® HTB35™), HeLa (ATCC® CCL2™), and DoTc2 4510 (ATCC® CRL-7920™) human cervical cancer cell lines and HPV 16 E6/E7-transformed, normal ectocervical cells (Ect1/E6E7; ATCC® CRL-2614™) were obtained from the American Type Culture Collection (ATCC, Manassas, VA). Cervical cancer cells were maintained in either Eagle's Minimum Essential Medium (EMEM; SiHa and HeLa) or Dulbecco’s Modified Eagle’s Medium (MEM; DoTc2) supplemented with 10% Fetal Bovine Serum (FBS; ATCC). Ect1/E6E7 cells were maintained in Keratinocyte-Serum Free medium (GIBCO) with 0.1 ng/ml human recombinant EGF, 0.05 mg/ml bovine pituitary extract, and additional calcium chloride 44.1 mg/L. The ATCC authenticates all cell lines by examining their short tandem repeat (STR) DNA profiles. Cells were kept at 37°C in high humidity with an atmosphere of 95% air and 5% CO2. Cells were harvested using Trypsin-EDTA (0.25% Trypsin, 0.53 mM EDTA; ATCC) and counted using Trypan Blue 0.4% Solution (AMRESCO, Solon, OH) and hemacytometer counting chamber. For some experiments, cisplatin (1 mM; Sigma, St. Louis, MO) was reconstituted in distilled water and stored at -20°C until use. Cells were exposed to cisplatin at a concentration of 10 μM, 50 μM, and 100 μM in growth media for 4 h in culture media. Cisplatin-containing culture media was removed and replaced with fresh growth media 4 h later and cells incubated at 37°C overnight. For experiments investigating PVT1 expression following hypoxia and immune response stimulation, SiHa cells were respectively treated with cobalt chloride (150 μM; Sigma) or interferon alpha (IFN-α; 10 μM; Sigma) for 48 h prior to RNA extraction (methods below). Finally, other SiHa cells were treated with cyclohexamide (10 μM; Sigma) for different time points before being subjected to lysis for Myc Western blotting.
Cells (2.0 x 105) cells were plated into a 6-well dish and allowed to adhere overnight. Cells were then transfected with 2 small interfering RNA (siRNA) designed against PVT1 combined in an equimolar ratio (20 or 40 nM; FlexiTube Hs_PVT1_5 and Hs_PVT1_6; Qiagen, Valencia, CA) or a negative control siRNA (AllStar Negative Control; Qiagen) conjugated with Alexa Fluor 488. Transfections were carried out using DharmaFECT1 (GE Dharmacon, Lafayette, CO) per the manufacturer’s instructions. Transfection efficiency was monitored via fluorescent microscopy and confirmed via quantitative PCR (qPCR) 48 h after transfection. siRNA transfections were performed in both SiHa and HeLa cervical cell lines and utilized in proliferation, apoptosis and cell death, and migration and invasion assays.
For locked nucleic acid (LNA) oligonucleotide transfections, cells (2.0 x 105) were plated into a 6-well dish and the following day transfected with 10 nM PVT1_10 or PVT1_15 LNA (Exiqon, Woburn, MA) or a scrambled control LNA (Negative Control A, Exiqon) using DharmaFECT 1 (GE Dharmacon) at a concentration of 0.2 μl per 100 μl growth medium. The sequences of PVT1_10 LNA, PVT1_15 LNA and Negative Control A are 5'-AACACGTCTATACGC-3', 5’-AGATCACTGTAAATCC-3’ and 5'-GGCAGGATCTATGGCA-3', respectively. Transfection media was replaced with fresh growth media 24 h following transfection and downstream assays performed 48 h later. LNA-transfected SiHa cells were used to confirm the effects of siRNA transfection on cell proliferation, for determining the effects of PVT1 knockdown on cisplatin responsiveness, Myc degradation experiments, and effects on nucleolin-dependent mRNA expression.
RNA isolation, cDNA synthesis, and quantitative PCR (qPCR)
RNA was isolated using mirVana miRNA Isolation Kit (Ambion, Austin, TX) and cDNA synthesized using High-Capacity RNA-to-cDNA Kit (Applied Biosystems, Foster City, CA). For some experiments, live cultured cells were washed with 1x PBS and separated into cytoplasmic and nuclear RNA portions by use of nonionic detergents following the PARIS kit instructions (Life Technologies). RNA was isolated from each portion and DNase treated using DNA-free Kit (Life Technologies). Expression of PVT1, MYC and control RPS18 was assessed using EvaGreen qPCR Mastermix (MidSci, St. Louis, MO) on a StepOnePlus Real-Time PCR System (Life Technologies). The thermal cycling program included an initial enzyme activation step at 95°C for 10min, followed by 40 cycles consisting of a 10sec denaturing step at 95°C, and a 1min annealing/extension step at 60°C for all primers. Fluorescent intensity was measured at 62°C at the end of each cycle. Forward and reverse primers for PVT1 were 5’-CCGACTCTTCCTGGTGAAGC-3’ and 5’-GTATGGTCAGCTCAAGCCCA-3’, 5’-TACCCTCTCAACGACAGAG-3’ and 5’- TCTTGACATTCTCCTCGGTG-3’ for MYC, and 5’-CTTCAGTCGCTCCAGGTCTT-3’ for RPS18 (for data normalization).s for PVT1 were 5’-CCGACTCTTCCTGGTGAAGC-3’ and 5’-GTATGGTCAGCTCAAGCCCA-3’, 5’-TACCCTCTCAACGACAGAG-3’ and 5’- TCTTGACATTCTCCTCGGTG-3’ for MYC, and 5’-CTTCAGTCGCTCCAGGTCTT-3’ for RPS18 (for data normalization).
Expression of miRNAs was conducted using TaqMan® MicroRNA assays (Life Technologies) for hsa-miR-1204 (002872), hsa-miR-1205 (002778), hsa-miR-1206 (002878), hsa-miR-1207-3P (002826), hsa-miR-1207-5P (241060_mat), hsa-miR-1208 (002880), and RNU48 (001006; for data normalization) according to manufacturer’s instructions. Briefly, RNA was isolated using mirVana miRNA Isolation Kit and cDNA synthesized using the TaqMan MicroRNA Reverse Transcription Kit (Life Technologies) with provided primers for each miRNA. qPCR amplification was then performed on a StepOnePlus Real-Time PCR System using the manufacturer’s parameters. All samples were analyzed in triplicate using 2− ΔΔCt method for relative gene expression, expressed as 2−ΔΔCt.
RNA fluorescent in situ hybridization (FISH) and immunocytochemistry
ViewRNA TYPE 6 probe sets (Affymetrix, Santa Clara, CA) designed against PVT1 (NR_003367.2) were used to visualize PVT1 expression in vitro. Experiments were conducted according to manufacturer’s protocol using reagents provided in the QuantiGene ViewRNA ISH Cell Assay Kit (Affymetrix). Briefly, SiHa cells were seeded onto 12 mm circular coverslips and grown to 70–90% confluence before being fixed with 4% formaldehyde. Cells were then washed with 1X PBS and incubated for 5 min in Detergent Solution for permeabilization. Cells were washed once more and incubated with Working Protease Solution for 25 min. Probes were diluted (1:100) in pre-warmed Probe Set Diluent and incubated with the cells for 3 h at 40°C. Cells were washed, hybridized with Working PreAmplifier Mix Solution for 30 min at 40°C, washed again, and hybridized with Working Amplifier Mix Solution for 30 min at 40°C. After more washing, cells were hybridized with the labeled probes for 30 min at 40°C. Following this incubation, coverslips were washed, stained with DAPI, and mounted on glass slides. Cells were imaged with a Zeiss LSM510 confocal microscope located in the Children’s Research Institute Imaging Core at the Medical College of Wisconsin. Images were processed in Adobe Photoshop CS5.1.
For co-staining of Nucleolin with PVT1 RNA FISH, immunocytochemistry was performed prior to the staining with DAPI step above. Cells were washed 3 times with PBS +1% BSA and 0.1% Tween-20 and then blocked at room temperature for 2 h in blocking buffer (PBS + 10% BSA, 0% normal donkey serum and 0.1% Tween-20). Coverslips were then incubated at 4°C overnight in blocking buffer containing a polyclonal antibody directed against Nucleolin (1:1000; Cell Signaling Technology, Danvers, MA). Cell were again washed 3 times and incubated for 1 h at room temperature in block solution containing a FITC-conjugated secondary antibody for visualization of Nucleolin. After a final wash, coverslips were mounted using Vectashield Antifade Mounting Medium with DAPI (Vector Laboratories, Burlingame, CA) and imaged as described above.
Cell proliferation assay
Cell proliferation was measured using the Cell Proliferation ELISA BrdU (colorimetric) kit (Hoffmann-La Roche, Nutley, NJ) according to the manufacturer’s protocol. Briefly, 48 h after transfection, 1.0 x 104 cells were plated in a 96-well dish and allowed to adhere overnight. Cells were then labeled with BrdU for 2 h, fixed, and incubated with anti-BrdU-POD. Unbound peroxidase conjugates were removed by washing. Substrate was then added and absorbance values (370 nM–492 nM) were measured 15 min later.
The PrestoBlue® Cell Viability assay was also used according to manufacturer’s instructions. SiHa cells were transfected with LNA as described above. Forty-eight h later, cells were plated onto a 96-well plate (1.0 x 104 cells/well) and exposed to cisplatin (Sigma) at a range of concentrations between 10–200 μM for 4 h. Cisplatin culture media was removed and replaced with growth media after the 4 h exposure and then incubated at 37°C overnight. The following day, fluorescence values were measured after incubation of cells with the PrestoBlue® reagent for 1 h at 37°C.
Cell death and apoptosis assays
Cell death was assayed using the Cell Death Detection ELISA Kit (Hoffmann-La Roche) as per manufacture directions. Briefly, 48 h after transfection, 0.5 x 105 cells were harvested and lysed. Cellular nucleosomes were bound to the prepared ELISA plate through histone components. Anti-DNA-peroxidase was added, and unbound peroxidase conjugates were removed by wash. Substrate was then added and absorbance values were measured 15 min later (405 nM–490 nM). Apoptosis was characterized by measuring active caspase-3 (ng/mg of total cell lysate) using the Human Caspase-3 (Active) ELISA Kit (Life Technologies, Grand Island, NY). Seventy-two h after transfection, cells (1.0 x 106) were lysed in Cell Extraction Buffer (Life Technologies) supplemented with Protease Inhibitor Cocktail (Sigma) and phenylmethanesulfonyl fluoride (Sigma).
Cell migration and invasion assays
Forty-eight h after transfection, media containing serum was removed and cells were washed with 1x PBS then cultured for 24 h in serum-free EMEM. After the starvation period, cells were harvested using HyQTase and plated (1 x 105 cells) and onto Corning Transwell Permeable Supports (Corning, Tewksbury, MA). For migration assays, Transwell membranes were equalized in growth media 24 h before cell seeding. For invasion assays, Transwell membranes were coated with Matrigel Matrix (Corning; 1:6 in EMEM) and dried for 6 h at 37°C. Cells were cultured in the presence and absence of chemoattractant (EMEM + 20% FBS) and harvested 6 h or 48 h later for analysis of migration and invasion, respectively. Non-migrant cells were removed from the upper face of the Transwell membrane with a cotton swab, while migrant cells were fixed and stained with Crystal Violet (EMD Millipore, Billerica, MA) and visualized/counted using light microscopy.
Fresh frozen cervical tissue or 3.6 x 106 cells were washed with 1x PBS and lysed for 10 min on ice using Cell Extraction Buffer (Life Technologies) supplemented with 1 mM PMSF (Sigma) and a Protease Inhibitor Cocktail (Sigma). Undigested cellular debris was pelleted by centrifugation at 10,000 RPM and remaining lysate was quantified using the Bradford Assay (Sigma). Lysate (20 μg) was run out on a Criterion Tris-HCl polyacrylamide gel (Bio-Rad Laboratories, Hercules, CA), transferred to PVDF membranes, blocked with 1X TBS + 10% nonfat dry milk at room temperature for 1 h, and incubated with primary antibody (Myc or Nucleolin, 1:1000; Cell Signaling Technology) overnight at 4°C. The following day, membranes were washed, incubated for 1 h at room temperature with HRP-conjugated goat anti-rabbit IgG (1:5,000; Cell Signaling Technology), and developed using ECL (Life Technologies). Chemiluminescence was measured using Molecular Imager ChemiDoc XRS+ (Bio-Rad Laboratories) and visualized protein bands were quantified using Image Lab Software (Bio-Rad Laboratories).
RNA affinity chromatography and mass spectrometry sequencing
Several overlapping sense and antisense single-stranded DNA oligonucleotides (ssDNA oligos; 100 nt long) were generated against the complete sequences of exons 2 and 9 of PVT1 (NR_003367.3; Eurofins MWG Operon, Huntsville, AL). Exons 2 and 9 were chosen based on their increased propensity for protein binding as determined in silico. ssDNA were hybridized to obtain double-stranded DNA (dsDNA), all of which had a T7 promoter sequence at the 5’end. The T7 promoter-containing dsDNA was in vitro-transcribed using the MaxiScript T7 kit (Ambion) and resultant RNA was then 3’-biotinylated with the Pierce Desthiobiotinylation kit (ThermoScientific). Desthiobiotinylated RNAs were incubated with streptavidin-bound magnetic beads from Pierce Magnetic RNA-Protein Pull-Down Kit (Thermo Scientific) as per the manufacturer’s protocol. Following washing, 200 μg of SiHa whole cell lysate was added to the RNA-bound streptavidin magnetic beads and incubated at 4°C with gentle agitation for 90 min. After 3 washes with Wash Buffer (Thermo Scientific 20164), the RNA bound proteins were eluted in SDS sample buffer and resolved by SDS-PAGE on a 12% acrylamide gel. Following silver staining, protein bands of interest were Mass Spectrometry sequenced at Taplin Mass Spectrometry Facility (Harvard Medical School, Boston, MA). Mass spectrometry “hits” were analyzed for the number of unique and total peptides obtained and area under the curve. Each hit was in silico trypsin digested and the number of peptides obtained with in silico digestion was compared with the number of peptides obtained in the mass spectrometry results. Only those proteins that had 20% or more of the trypsin-digestible fragments present were considered to be true hits. Western blotting (as described above) following RNA affinity chromatography was further used to validate true hits.
RNA immunoprecipitation (RIP) assays were performed using the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (EMD Millipore) following manufacturer’s instructions. Total cell lysate from 1.7 x 107 SiHa cells was used for immunoprecipitation with Myc and Nucleolin antibodies (same as for Western blotting). As a nonspecific IgG control, purified Rabbit IgG was run in parallel. Following protein digestion, RNA was extracted and DNase treated using DNA-free Kit (Life Technologies). Levels of PVT1 were quantified in input and immunoprecipitated RNA.
Data Analysis and Statistics
Experiments were performed in triplicates and differences between mean data were analyzed using Student’s t tests (two-tailed). Results have been graphed as means ± standard error of the mean. Cisplatin dose-response curves were generated using a variable slope model. All statistical analyses and graphing of data were performed using GraphPad Prism 6.0 (La Jolla, CA), except for survival analysis which was performed using SPSS11.0 software (Chicago, IL). Kaplan-Meier survival curves were plotted and compared using log-rank (Mantel-Cox) tests. P values less than 0.05 were considered significant.
PVT1 expression is upregulated in tumors from cervical cancer patients and correlated with poorer survival
PVT1 is a multi-exon gene with its contiguous genomic region containing a cluster of 6 annotated microRNAs (miRNAs). PVT1 and local miRNA expression levels were determined by qPCR for cancerous and adjacent normal tissues from cervical cancer patients. PVT1 expression was significantly higher in tumors from cervical cancer patients versus adjacent normal tissue (n = 127 tumor; n = 30 adjacent normal; p<0.001; Fig 1A). Expression of miRNAs 1204 and 1206 (but not 1205, 1207-3p, 1207-5p, or 1208; S1A–S1D Fig) was significantly higher in tumor versus normal adjacent cervical tissue (Fig 1B and 1C). Of note, because PVT1 and MYC are frequently co-amplified in cancer, we additionally examined expression of MYC in cervical cancer and adjacent normal tissues. Although we observed a trend toward increased MYC in cancer compared to adjacent normal, the difference in average expression of MYC was not statistically significant between these two groups (S1E Fig).
(A) PVT1 expression was significantly higher in primary cervical tumors (n = 127) compared to adjacent normal cervical tissue (n = 30; p<0.001). Expression of miRNAs 1204 (B; p<0.05) and 1206 (C; p<0.001), but not other local miRNAs, was also significantly higher in cervical tumor tissue (n = 38) compared to adjacent normal tissue (n = 18). (D) Kaplan-Meier plots revealed an association of higher tumor PVT1 levels with significantly poorer survival times compared to lower PVT1 levels (p = 0.03).
The 127 cervical cancer patients were divided into high and low PVT1-expressing groups, using the median PVT1 expression level, to investigate their correlation with prognosis. Using Kaplan-Meier survival analysis and log-rank tests, we found that high PVT1 expression was significantly correlated with shorter cumulative survival time for cervical cancer patients (p = 0.03; Fig 1D). Together, these data suggest that, in general, PVT1 expression is elevated in cervical cancer tumors and that the higher the expression of PVT1, the poorer the prognosis.
PVT1 expression in cervical cancer cell lines
To begin our in vitro examination into the role of PVT1 in cervical cancer cells, we determined PVT1 expression levels in various commercially available cervix-derived cell lines via qPCR (Fig 2A). Out of 4 cervical lines, HPV 16-positive SiHa cells exhibited the highest PVT1 expression, while the lowest PVT1 expression was observed in HPV 16 E6/E7-transformed cells derived from normal ectocervix (Ect1/E6E7). Next, we sought to identify specific oncogenic stressors that may drive enhanced PVT1 expression in cervical cancer. To this end, we exposed SiHa cells (used for the remainder of our studies) to various transformative stimuli prior to measuring the effects on PVT1 expression via qPCR. Interestingly, we found that PVT1 expression in SiHa cells was significantly increased in response to 48 h incubation with INF-α or the hypoxia mimetic, cobalt chloride (CoCl2; Fig 2B). Other stimuli, such as epidermal growth factor (EGF), insulin-like growth factor (IGF), phorbol 12-myristate 13-acetate (protein kinase C activator), and gamma-irradiation did not significantly affect PVT1 expression (data not shown). Finally, we investigated subcellular localization of PVT1 using RNA FISH and found expression of PVT1 in both the cytoplasm and nucleus of SiHa cells. Further, control antisense-transfected SiHa (Fig 2C) exhibited similar nuclear and cytoplasmic PVT1 staining, of which was absent in PVT1 knockdown cells (Fig 2D). In summary, PVT1 is expressed in various cell lines derived from human cervix, can be augmented by immune and hypoxic stimuli, and is localized throughout the nucleus and cytoplasm suggesting that this lncRNA may participate in various cervical cancer cell processes, possibly via more than one unique mechanism.
(A) PVT1 expression in commercially available cervical cell lines. Lowest PVT1 expression was observed in HPV 16 E6/E7-transformed cells derived from normal ectocervix (E6/E7-Ecto), while SiHa cervical cancer cells displayed the highest PVT1 expression compared to 2 other cervical cancer-derived lines (HeLa and DoTc2). (B) Expression of PVT1 in SiHa cells was further increased upon 48h treatment with INF-α (10 μM) or the hypoxia mimetic cobalt chloride (CoCl2; 150 μM). Representative images of RNA FISH experiments in SiHa cells transfected with either control (C) or PVT1 (D) LNAs. Control LNA-transfected cells exhibited punctate signals for PVT1 (red) in both the nucleus (stained with DAPI in blue) and the cytoplasm. Both nuclear and cytoplasmic PVT1 staining was absent in PVT1 LNA-transfected cells. Phase image (bottom left panel) depicts cell morphology. Scale bar (white) = 10 μm. *p<0.05, **p<0.01.
PVT1 silencing decreases cervical cancer cell proliferation, migration and invasion
Next, SiHa cells transfected with PVT1-targeted siRNA (siPVT1) were utilized to examine the effects of this lncRNA on cervical cancer cell proliferation and motility. Transfection with PVT1-targeting siRNAs resulted in, on average, 70% knockdown of PVT1 compared to control-transfected cells (Fig 3A). Further, there was no significant difference in local miRNA (1204–1208) or MYC mRNA expression in siPVT1 versus scrambled control-transfected cells (siCONT; S2 Fig). siPVT1-transfected SiHa cells exhibited a significant decrease in BrdU incorporation, a measure of replicating cells, at 72 h post-transfection compared to siCONT, suggesting that PVT1 plays a role in driving cell proliferation (Fig 3B). These effects of PVT1 knockdown on SiHa proliferation were also confirmed by transfection with PVT1-targeted LNA. Finally, in comparison to siCONT cells, siPVT1 cells displayed a significant decrease in migration (Fig 3C) and invasion (Fig 3D) potential. Of note, these effects of PVT1 knockdown are mimicked in another cervical cancer cell line, HeLa (S3 Fig), providing further support for a role of this lncRNA in cervical carcinogenesis.
(A) Transfection of SiHa cervical cancer cells with siRNAs targeting PVT1 (siPVT1) resulted in an approximate 70% knockdown in PVT1 lncRNA expression as compared to cells transfected with a scrambled control siRNA (siCONT). (B) SiHa cells transfected with siPVT1 exhibited a significant decrease in proliferation compared to siCONT cells. Transfected SiHa cells were also assessed for changes in (C) migration and (D) invasion 6 h or 48 h following introduction of chemoattractant (FBS), respectively. siPVT1 cells showed a significant decrease in both cell migration and invasion compared to siCONT cells. Quantitative results are graphed on the left, while representative images are on the right. ***p<0.001, ****p<0.0001
PVT1 silencing increases apoptosis and response to cisplatin in cervical cancer cells
To date, siRNA-mediated knockdown strategies targeting PVT1 have demonstrated its role in cisplatin sensitivity in malignant pleural mesothelioma  and gastric cancer cells . However, the anti-apoptotic signature induced by PVT1 in other cancer cell types [8,18,20,26] may indicate that its contribution to cisplatin resistance is more far-reaching. Thus, our next experiments were designed to investigate the role of PVT1 in apoptosis and cisplatin sensitivity in cervical cancer cells. Extending the previously mentioned findings, our results show that PVT1 knockdown in SiHa cells significantly increased levels of cytoplasmic histone-associated DNA fragments (Fig 4A) and active caspase-3 (Fig 4B), suggesting that PVT1 can also contribute to cervical carcinogenesis via inhibition of cell death and apoptosis. Further, PVT1 knockdown by either siRNA (Fig 4C) or LNA oligonucleotides (Fig 4D) leads to increased responsiveness of SiHa cells to cisplatin. Collectively, the cell phenotypic (loss of PVT1) functional data suggests that PVT1 is required for proliferation, maintenance and cisplatin resistance of cervical cancer cells.
siPVT1 cells exhibited a significant increase in cell death (A) and apoptosis (B) compared to siCONT cells. (C) Proliferation of siPVT1 SiHa cells was significantly decreased in response to multiple doses of cisplatin. (D) Dose-response curve for control and PVT1 LNA-transfected SiHa cells following 4 h treatment with cisplatin. The IC50 value of the PVT1 knockdown cells was significantly decreased compared to control knockdown cells (p<0.0001), suggesting a role for this lncRNA in cisplatin resistance. **p<0.01, ***p<0.001, ****p<0.0001
PVT1 interacts with Myc and Nucleolin proteins in cervical cancer
Many lncRNAs exert their functions via binding specific protein partners. PVT1 has been shown to bind and stabilize the Myc protein by preventing its phosphorylation at threonine 58 to promote carcinogenesis in breast cancer cells . However, although RNA immunoprecipitation results suggest that Myc and PVT1 associate in SiHa cells (S4A Fig), western blotting of protein lysate from cells exposed to siRNA-mediated knockdown of PVT1 did not reveal significant differences in total or T58-phosphorylated Myc levels (S4B Fig), nor in the rate of Myc degradation (S4C and S4D Fig). Thus, it appears that PVT1 likely exerts its effects through an alternative mechanism in cervical cancer cells.
To this end, using RNA affinity chromatography in conjunction with mass spectrometry sequencing, we identified potential protein binding partners of PVT1 in vitro using whole-cell lysates from SiHa cervical cancer cells. Of the several proteins identified, we chose to further examine the multifaceted DNA and RNA binding protein, Nucleolin, as it could account for the wide-reaching effects of PVT1 knockdown. We confirmed the interaction of PVT1 and Nucleolin in SiHa cells via their colocalization using combined RNA FISH and immunocytochemistry (Fig 5A). Next, we performed reciprocal capture of Nucleolin or PVT1 with complementary approaches. In the first approach, we identified Nucleolin by western blotting with anti-Nucleolin when PVT1 RNA was captured by RNA affinity chromatography (Fig 5B). In the second approach, significant PVT1 enrichment was observed in anti-Nucleolin (Fig 5C) immunoprecipitates. Because PVT1 has previously been shown to bind to specific proteins and mediate their stability [17,19], we next investigated the effect of PVT1 knockdown on Nucleolin protein expression. Somewhat surprisingly, LNA-mediated knockdown of PVT1 had no significant effect on total Nucleolin protein levels (S5 Fig), suggesting a disparate function of PVT1-Nucleolin binding. Work is ongoing to elucidate the mechanistic properties of the PVT1-Nucleolin interaction in cervical cancer cells.
(A) SiHa cells stained for PVT1 (green) and nucleolin (red), and DAPI. Boxed cells (top panel) are at higher magnification (bottom panel). White arrows show PVT1-nucleolin colocalization. Scale bars = 20μM. (B) Western blot for Nucleolin of SiHa protein captured by PVT1 RNA affinity chromatography. (C) PVT1 RNA was immunoprecipitated with anti-Nucleolin, but not the negative control rabbit IgG (Cont). **p<0.01
The current study revealed that PVT1 expression levels in cervical cancer tumors were significantly higher than in adjacent noncancerous tissue and that high PVT1 expression levels correlated with patients’ poorer overall survival. PVT1 expression was also examined in vitro, with the SiHa cervical cancer cell line expressing the highest levels of the lncRNA compared to others. Moreover, PVT1 expression in SiHa cells was further augmented under hypoxic conditions or following immune response stimulation by INF-α. Functionally, PVT1 knockdown in cervical cancer cells was associated with decreased proliferation, migration and invasion and increased apoptosis and cisplatin sensitivity, suggesting that this lncRNA likely plays a pivotal and multifaceted role in cervical carcinogenesis. Finally, the effects of PVT1 on cervical cancer cells may be facilitated by its association with the protein Nucleolin.
The human PVT1 gene is a long intergenic noncoding RNA that gives rise to multiple splice variants. Ensembl data suggests the presence of at least 25 unique human PVT1 transcripts  and exon 2 of PVT1 can also form a circular RNA (circRNA) that is highly expressed in HeLa cervical cancer cells , thus making the study of PVT1 function particularly challenging. Here, we have used siRNA and LNA-mediated knockdown of PVT1 to study its role in cervical carcinogenesis. While both methods of knockdown were particularly efficient in knocking down multiple regions along the full-length PVT1 transcript (with exception of primers designed to span exons 1 and 2), only LNA-transfected cells were effective in knocking down the exon 2-derived circRNA (S1 Table). Further, our RNA FISH images (Fig 2C and 2D) clearly demonstrate that knockdown of PVT1 results in complete loss of staining for both nuclear and cytoplasmic PVT1. Due to the fact that the probes used for this assay cover 8 (NR_003367.2) out of 9 (NR_003367.3) exons of PVT1 and that the signal from each individual probe can be amplified up to 96,000-fold, we conclude that most, if not all, PVT1 isoforms are being affected by our knockdown. Thus, while the current manuscript indicates that PVT1 plays a pivotal role in cervical carcinogenesis, future experiments are required to definitively pinpoint the precise splice variant(s) responsible for the oncogenic effects of PVT1 in cervical cancer.
The 8q24 region of the human genome, which contains MYC and PVT1, is one of the most common sites of cancer-related amplifications [29,30]. The MYC oncogene encodes a well-characterized transcription factor that plays a role in various cancer-related processes such as cell cycle progression, cellular transformation, and apoptosis  making it essential that studies investigating PVT1 and its cellular mechanisms include concurrent examination of MYC. Co-amplification and expression of MYC and PVT1 has been shown to play a role in the promotion of the malignant phenotype in malignant pleural mesothelioma , ovarian and breast cancer [8,19], and neuroblastoma . Here, we show that PVT1 expression is significantly higher in cervical cancer tissue compared to normal controls (Fig 1A). Unlike in other cancers, however, we did not observe a concurrent increase in MYC mRNA levels in cervical tumors (S1E Fig). Further, using publicly available TCGA data from cBioPortal [33,34], we found that 13% (40 cases) of cervical cancer tumors have amplification and/or overexpression of PVT1, yet co-amplification of PVT1 and MYC only occurs in 27.5% (11/40 cases) and PVT1 amplification or mRNA upregulation with no alteration in MYC occurs in 65% (26/40 cases). Although we also found evidence of physical interaction between PVT1 and Myc protein in SiHa cells (S4A Fig), we did not observe any significant effects of PVT1 knockdown on Myc protein levels or stabilization (S4B and S4C Fig). Collectively, these data suggest that co-regulation of PVT1 and MYC expression may not be as common in cervical cancer as it is in other cancer types and that the result of their interaction is unlike that described previously . Future work is necessary to elucidate the biological function of the PVT1-Myc association in cervical cancer.
We have provided three lines of evidence confirming an interaction between PVT1 and Nucleolin in SiHa cells (Fig 5). Nucleolin is a multifunctional protein that is well known to be dysregulated in disease [35,36]. Although abundantly expressed in the nucleolus, Nucleolin also localizes to cytoplasmic and plasma membrane compartments [37–39], and its subcellular localization can indicate its functional role in different cell types . Our RNA FISH/immunocytochemistry data suggest that, in SiHa cells, colocalization of Nucleolin and PVT1 is chiefly nuclear/nucleolar (Fig 5A). Nucleolin in these compartments regulates ribosomal biogenesis, facilitates rDNA transcription, and governs oncogene expression . Interestingly, these regulatory functions of Nucleolin are intimately tied to cellular stress responses under conditions such as cancer-related hypoxia and immune activation via viral infection [40–43]; stimuli we demonstrate here to also significantly augment PVT1 expression in SiHa cells (Fig 2B) and that are especially relevant in cervical disease [44–48]. Thus, our future research aims to examine the importance of the PVT1-Nucleolin interaction in the context of these processes, and whether they underlie the oncogenic effects of PVT1 in cervical cancer in vivo.
In summary, we propose that high expression levels of PVT1 contribute to the cervical cancer phenotype via modulation of cell proliferation, apoptosis, and cell motility. Perhaps most importantly, high PVT1 expression correlates with poorer survival outcome in cervical cancer patients and may play a central role in resistance of cervical cancer cells to cisplatin. In vitro, PVT1 expression is significantly augmented by cancer-related hypoxic and immune-stimulatory treatments, suggesting possible mechanisms underlying its heightened expression in cervical cancer. Finally, we have identified Nucleolin as a protein binding partner of PVT1 in cervical cancer cells. Future work in our laboratory is aimed to fully elucidate the functional consequences of PVT1-Nucleolin binding as well as to determine what role, if any, the PVT1-Myc interaction plays in cervical carcinogenesis.
S1 Fig. Local miRNA and MYC expression in cervical cancer.
(A-D) miR1205, 1207, and 1208 expression was not significantly different between adjacent normal (n = 18) and cancer tissue (n = 38). (E) MYC mRNA levels were also not significantly different between the two groups. (F) PVT1 expression in commercially available cervical cell lines. Lowest PVT1 expression was observed in HPV 16 E6/E7-transformed cells derived from normal ectocervix (E6/E7-Ecto), while SiHa cervical cancer cells displayed the highest PVT1 expression compared to 2 other cervical cancer-derived lines (HeLa and DoTc2).
S2 Fig. siRNA-mediated PVT1 knockdown did not affect expression of any local miRNAs nor MYC.
S3 Fig. PVT1 promotes cervical cancer cell proliferation, migration and invasion in HeLa cells.
(A) Transfection of HeLa cervical cancer cells with siRNAs targeting PVT1 (siPVT1) resulted in an approximate 60% knockdown in PVT1 lncRNA expression as compared to cells transfected with a scrambled control siRNA (siCONT). (B) HeLa cells transfected with siPVT1 exhibited a significant decrease in proliferation compared to siCONT cells. Transfected HeLa cells were also assessed for changes in (C) migration and (D) invasion 6 h or 48 h following introduction of chemoattractant (FBS), respectively. siPVT1 cells showed a significant decrease in both cell migration and invasion compared to siCONT cells. Quantitative results are graphed on the left, while representative images are on the right. **p<0.01, ***p<0.001, ****p<0.0001
S4 Fig. PVT1 binds, but does not stabilize, Myc in SiHa cells.
(A) PVT1 from SiHa total cell lysate immunoprecipitated with a MYC antibody, but not the negative control rabbit IgG (Cont). (B) PVT1 knockdown in SiHa cells did not significantly affect p58 phospho-MYC or total MYC protein. (C,D) Degradation of MYC protein was also not significantly affected by PVT1 knockdown. **p<0.01
S5 Fig. PVT1 knockdown does not affect Nucleolin protein levels.
S1 Table. Percent knockdown of different PVT1 regions achieved with siRNA or LNA.
* denotes primer set used throughout the manuscript.
We would like to thank Dr. Stephanie Cossette for her help on confocal image processing and Dr. Pippa Simpson for generation of the presented Kaplan-Meier survival curves.
Conceived and designed the experiments: MI KL TC RR JR. Performed the experiments: MI SF KL TC. Analyzed the data: MI SF KL TC. Contributed reagents/materials/analysis tools: RR JR. Wrote the paper: MI JR.
- 1. Prensner JR, Chinnaiyan AM (2011) The emergence of lncRNAs in cancer biology. Cancer Discov 1: 391–407. pmid:22096659
- 2. TCGA Research Network.
- 3. Cory S, Graham M, Webb E, Corcoran L, Adams JM (1985) Variant (6;15) translocations in murine plasmacytomas involve a chromosome 15 locus at least 72 kb from the c-myc oncogene. EMBO J 4: 675–681. pmid:3924592
- 4. Webb E, Adams JM, Cory S (1984) Variant (6; 15) translocation in a murine plasmacytoma occurs near an immunoglobulin kappa gene but far from the myc oncogene. Nature 312: 777–779. pmid:6440031
- 5. Graham M, Adams JM (1986) Chromosome 8 breakpoint far 3' of the c-myc oncogene in a Burkitt's lymphoma 2;8 variant translocation is equivalent to the murine pvt-1 locus. EMBO J 5: 2845–2851. pmid:3024964
- 6. Shtivelman E, Bishop JM (1990) Effects of translocations on transcription from PVT. Mol Cell Biol 10: 1835–1839. pmid:2181290
- 7. Huppi K, Siwarski D (1994) Chimeric transcripts with an open reading frame are generated as a result of translocation to the Pvt-1 region in mouse B-cell tumors. Int J Cancer 59: 848–851. pmid:7989128
- 8. Guan Y, Kuo WL, Stilwell JL, Takano H, Lapuk AV, et al. (2007) Amplification of PVT1 contributes to the pathophysiology of ovarian and breast cancer. Clin Cancer Res 13: 5745–5755. pmid:17908964
- 9. Barsotti AM, Beckerman R, Laptenko O, Huppi K, Caplen NJ, et al. (2012) p53-Dependent induction of PVT1 and miR-1204. J Biol Chem 287: 2509–2519. pmid:22110125
- 10. Ding J, Li D, Gong M, Wang J, Huang X, et al. (2014) Expression and clinical significance of the long non-coding RNA PVT1 in human gastric cancer. Onco Targets Ther 7: 1625–1630. pmid:25258543
- 11. Yang YR, Zang SZ, Zhong CL, Li YX, Zhao SS, et al. (2014) Increased expression of the lncRNA PVT1 promotes tumorigenesis in non-small cell lung cancer. Int J Clin Exp Pathol 7: 6929–6935. pmid:25400777
- 12. Ding C, Yang Z, Lv Z, Du C, Xiao H, et al. (2015) Long non-coding RNA PVT1 is associated with tumor progression and predicts recurrence in hepatocellular carcinoma patients. Oncol Lett 9: 955–963. pmid:25624916
- 13. Huang CS, Yu W, Cui H, Wang YJ, Zhang L, et al. (2015) Increased expression of the lncRNA PVT1 is associated with poor prognosis in pancreatic cancer patients. Minerva Med.
- 14. Paci P, Colombo T, Farina L (2014) Computational analysis identifies a sponge interaction network between long non-coding RNAs and messenger RNAs in human breast cancer. BMC Syst Biol 8: 83. pmid:25033876
- 15. Haverty PM, Hon LS, Kaminker JS, Chant J, Zhang Z (2009) High-resolution analysis of copy number alterations and associated expression changes in ovarian tumors. BMC Med Genomics 2: 21. pmid:19419571
- 16. Chapman MH, Tidswell R, Dooley JS, Sandanayake NS, Cerec V, et al. (2012) Whole genome RNA expression profiling of endoscopic biliary brushings provides data suitable for biomarker discovery in cholangiocarcinoma. J Hepatol 56: 877–885. pmid:22173169
- 17. Wang F, Yuan JH, Wang SB, Yang F, Yuan SX, et al. (2014) Oncofetal long noncoding RNA PVT1 promotes proliferation and stem cell-like property of hepatocellular carcinoma cells by stabilizing NOP2. Hepatology 60: 1278–1290. pmid:25043274
- 18. Takahashi Y, Sawada G, Kurashige J, Uchi R, Matsumura T, et al. (2014) Amplification of PVT-1 is involved in poor prognosis via apoptosis inhibition in colorectal cancers. Br J Cancer 110: 164–171. pmid:24196785
- 19. Tseng YY, Moriarity BS, Gong W, Akiyama R, Tiwari A, et al. (2014) PVT1 dependence in cancer with MYC copy-number increase. Nature 512: 82–86. pmid:25043044
- 20. Kong R, Zhang EB, Yin DD, You LH, Xu TP, et al. (2015) Long noncoding RNA PVT1 indicates a poor prognosis of gastric cancer and promotes cell proliferation through epigenetically regulating p15 and p16. Mol Cancer 14: 82. pmid:25890171
- 21. Zhou Q, Chen J, Feng J, Wang J (2015) Long noncoding RNA PVT1 modulates thyroid cancer cell proliferation by recruiting EZH2 and regulating thyroid-stimulating hormone receptor (TSHR). Tumour Biol.
- 22. Tang KW, Alaei-Mahabadi B, Samuelsson T, Lindh M, Larsson E (2013) The landscape of viral expression and host gene fusion and adaptation in human cancer. Nat Commun 4: 2513. pmid:24085110
- 23. Ojesina AI, Lichtenstein L, Freeman SS, Pedamallu CS, Imaz-Rosshandler I, et al. (2014) Landscape of genomic alterations in cervical carcinomas. Nature 506: 371–375. pmid:24390348
- 24. Iden M, Fye, S., Huang, Y., Liu, P, Rader, J.S. (2014) HPV integration and methylation patterns in cervical intraepithelial neoplasia and invasive cervical cancer via HPV capture and high-throughput sequencing [abstract]. Annual Meeting of the American Association for Cancer Research. San Diego, CA: AACR.
- 25. Riquelme E, Suraokar MB, Rodriguez J, Mino B, Lin HY, et al. (2014) Frequent coamplification and cooperation between C-MYC and PVT1 oncogenes promote malignant pleural mesothelioma. J Thorac Oncol 9: 998–1007. pmid:24926545
- 26. Zhang XW, Bu P, Liu L, Zhang XZ, Li J (2015) Overexpression of long non-coding RNA PVT1 in gastric cancer cells promotes the development of multidrug resistance. Biochem Biophys Res Commun.
- 27. Cunningham F, Amode MR, Barrell D, Beal K, Billis K, et al. (2015) Ensembl 2015. Nucleic Acids Res 43: D662–669. pmid:25352552
- 28. Salzman J, Chen RE, Olsen MN, Wang PL, Brown PO (2013) Cell-type specific features of circular RNA expression. PLoS Genet 9: e1003777. pmid:24039610
- 29. Beck-Engeser GB, Lum AM, Huppi K, Caplen NJ, Wang BB, et al. (2008) Pvt1-encoded microRNAs in oncogenesis. Retrovirology 5: 4. pmid:18194563
- 30. Beroukhim R, Mermel CH, Porter D, Wei G, Raychaudhuri S, et al. (2010) The landscape of somatic copy-number alteration across human cancers. Nature 463: 899–905. pmid:20164920
- 31. Nesbit CE, Tersak JM, Prochownik EV (1999) MYC oncogenes and human neoplastic disease. Oncogene 18: 3004–3016. pmid:10378696
- 32. Carramusa L, Contino F, Ferro A, Minafra L, Perconti G, et al. (2007) The PVT-1 oncogene is a Myc protein target that is overexpressed in transformed cells. J Cell Physiol 213: 511–518. pmid:17503467
- 33. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, et al. (2012) The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov 2: 401–404. pmid:22588877
- 34. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, et al. (2013) Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal 6: pl1. pmid:23550210
- 35. Berger CM, Gaume X, Bouvet P (2015) The roles of nucleolin subcellular localization in cancer. Biochimie 113: 78–85. pmid:25866190
- 36. Abdelmohsen K, Gorospe M (2012) RNA-binding protein nucleolin in disease. RNA Biol 9: 799–808. pmid:22617883
- 37. Borer RA, Lehner CF, Eppenberger HM, Nigg EA (1989) Major nucleolar proteins shuttle between nucleus and cytoplasm. Cell 56: 379–390. pmid:2914325
- 38. Ginisty H, Sicard H, Roger B, Bouvet P (1999) Structure and functions of nucleolin. J Cell Sci 112 (Pt 6): 761–772. pmid:10036227
- 39. Tuteja R, Tuteja N (1998) Nucleolin: a multifunctional major nucleolar phosphoprotein. Crit Rev Biochem Mol Biol 33: 407–436. pmid:9918513
- 40. Zhao Y, Chen XQ, Du JZ (2009) Cellular adaptation to hypoxia and p53 transcription regulation. J Zhejiang Univ Sci B 10: 404–410. pmid:19434769
- 41. Yang C, Maiguel DA, Carrier F (2002) Identification of nucleolin and nucleophosmin as genotoxic stress-responsive RNA-binding proteins. Nucleic Acids Res 30: 2251–2260. pmid:12000845
- 42. Daniely Y, Dimitrova DD, Borowiec JA (2002) Stress-dependent nucleolin mobilization mediated by p53-nucleolin complex formation. Mol Cell Biol 22: 6014–6022. pmid:12138209
- 43. Hiscox JA (2007) RNA viruses: hijacking the dynamic nucleolus. Nat Rev Microbiol 5: 119–127. pmid:17224921
- 44. Rofstad EK, Sundfor K, Lyng H, Trope CG (2000) Hypoxia-induced treatment failure in advanced squamous cell carcinoma of the uterine cervix is primarily due to hypoxia-induced radiation resistance rather than hypoxia-induced metastasis. Br J Cancer 83: 354–359. pmid:10917551
- 45. Sundfor K, Lyng H, Trope CG, Rofstad EK (2000) Treatment outcome in advanced squamous cell carcinoma of the uterine cervix: relationships to pretreatment tumor oxygenation and vascularization. Radiother Oncol 54: 101–107. pmid:10699471
- 46. Hockel M, Vorndran B, Schlenger K, Baussmann E, Knapstein PG (1993) Tumor oxygenation: a new predictive parameter in locally advanced cancer of the uterine cervix. Gynecol Oncol 51: 141–149. pmid:8276286
- 47. Fyles AW, Milosevic M, Wong R, Kavanagh MC, Pintilie M, et al. (1998) Oxygenation predicts radiation response and survival in patients with cervix cancer. Radiother Oncol 48: 149–156. pmid:9783886
- 48. Doorbar J (2006) Molecular biology of human papillomavirus infection and cervical cancer. Clin Sci (Lond) 110: 525–541.