Background
Lung cancer is responsible for the highest cancer-associated mortality rate worldwide. Only 16% of patients affected with Non-small cell lung cancer (NSCLC), which is the most common subtype, are alive 5 years after diagnosis, and this number has hardly improved over several decades [
1]. One reason for this poor prognosis is that only 15% of lung cancers are diagnosed at an early stage. Till recently the standard of care for NSCLC at stages I-IIIA was surgery, resulting in patient survival rates of 23% in stage IIIA, 33% in stage IIB, and up to 89% in stages IA [
2]. Adjuvant chemotherapy after radical resection of localized NSCLC improves survival at 5 years by about 5% [
3]. However, there is still a relatively high risk of relapse, and up to 40% of all stage IB and 60% of stage II patients die from their disease despite receiving adjuvant chemotherapy [
4]. The integration of prognostic and predictive biomarkers has the potential of identifying patients who are at a low-risk of relapse following surgery and do not need further therapy, and conversely, patients who are at a high risk of relapse and who potentially may derive the greatest benefit from adjuvant treatment, including chemotherapy or personalized treatment based on individual tumor profiling. Therefore, an effort to identify more robust prognostic and predictive biomarkers is needed [
5].
The Helicase-like Transcription Factor (HLTF) is a member of the yeast mating SWItch/Sucrose Non Fermenting (SWI/SNF) family of proteins involved in chromatin remodeling. Several studies demonstrated its function in gene transcription [
6], cell cycle [
7], DNA repair [
8,
9], and genome stability maintenance [
10], supporting its tumor suppressor role. In cancer, two different alterations in
HLTF expression were reported: (i) an epigenetic silencing by hypermethylation of its promoter and (ii) an alternative splicing of its mRNA, leading to the production of several shorter forms of the protein lacking DNA repair domains. The hypermethylation of
HLTF promoter was first identified in colon cancer [
11] and was reported in other types of cancers, including gastric cancers [
12‐
16]. It was shown in HeLa cells that
HLTF mRNA was alternatively spliced in the exons 19 to 22 region, resulting in the expression of shorter truncated protein forms. The distinctive character of the
HLTF spliced mRNA variants (I21R) is that they contain the intron 21 between exons 21 and 22. To date, the expression of HLTF protein forms was reported in head and neck, cervix and thyroid [
17‐
20] cancers and associated with a poor prognosis [
16].
In lung cancer, one study assessed the hypermethylation of
HLTF in a cohort of 54 patients with NSCLC [
21]. Promoter hypermethylation was found in 21 patients (39.6%), including 9/20 squamous cell carcinoma (SCC) and 12/33 adenocarinoma (ADC). Patients whose tumors harboured
HLTF hypermethylation had shorter survival, in comparison with patients whose tumors had a hypomethylated
HLTF promoter (log-rank,
p = 0.035). So far, to our knowledge, there are no published data about the expression of HLTF (wild-type and its truncated forms) in lung cancer.
The purpose of this study is to assess the expression of wild-type (WT) and spliced variants (I21R) of HLTF mRNAs in NSCLC and evaluate their clinical relevance. We analyzed publicly available databases for HLTF in lung cancer and assessed its expression in NSCLC cell lines and in a clinically annotated cohort of 171 patients with resected stage I-II NSCLC.
Methods
In silico analyses
Available genomic profiling data (mutation, copy number, DNA methylation [correlation only], and mRNA expression) for
HLTF were downloaded from cBioPortal, an online portal for accessing data from The Cancer Genome Atlas (TCGA) project and other cancer genome profiling initiatives (
http://www.cbioportal.org/public-portal). Additional cancer genome profiling data were obtained from the Catalogue of Somatic Mutations in Cancer (
http://cancer.sanger.ac.uk/cosmic; [
22]).
Patient characteristics
A total of 171 patients with resected stage I-II NSCLC collected at University Health Network (Toronto, Canada) were included in this study. These patients had surgery between 1996 and 2005. The length of follow-up: median 5.4 years, range 0.1–12 years. As these patients all underwent surgery before 2005, none of them received adjuvant chemotherapy as it did only become standard after 2005. The clinical and demographic characteristics of the patient cohort are listed in Table
2.
Cell lines and cell culture
NSCLC cell lines were purchased from the American Type Cell Collection (ATCC,
http://www.atcc.org), and cultured according to ATCC recommendation. Among these, there were 33 of the adenocarcinoma (ADC) subtype (H1693, H2122, H2228, H2279, H1573, H1395, H522, H1792, H838, H1819, H4011, H2291, H2073, H1568, H920, H1993, H4006, HCC827, H3255, H23, H4019, H2126, H1437, H1944, H2009, H2405, H1373, H1355, H1975, HCC2935, A549, H650, H1650), two large cell carcinoma (H661, H4017), one mixed adenosquamous carcinoma (H647) and two of undefined histology (DFC1032, DFC1024). MGH7 cells (squamous cell carcinoma, SCC) were cultured as described [
23].
mRNA expression
Total RNA was extracted from cell lines with RNeasy Mini Kit (QIAGEN) according to the manufacturer’s instructions. RNA purity and concentration were assessed with Nanodrop (Thermo Scientific). Total RNA (input 150 ng) extracted from cell lines and patient tumours were reverse transcribed into cDNA (SuperScript III, Invitrogen). Droplet Digital polymerase chain reaction (ddPCR) was performed based on the manufacturer’s recommendations (QX200, Bio-Rad). ddPCR is a highly sensitive qPCR due to a step of sample fractioning (limiting dilutions) by generation of droplets (water:oil emulsion). It allows retrieving an absolute count of RNA copies for each sample, and is particularly indicated for low-expressed targets. Each ddPCR was performed with 22.5 ng cDNA in triplicates. Reaction conditions were as follow: ddPCR cycle was set up at 95 °C for 5 min, 40 cycles of [30 s at 95 °C and 1 min at 58 °C], 5 min at 4 °C, and finally 5 min at 90 °C. Results were analyzed with QuantaSoft (Bio-Rad), and the cut-off to define positive and negative droplets was set up at 10,000 arbitrary units of fluorescence amplitude. This signal is then used in the calculation of HTLF copy number by a Poisson regression (QuantaSoft, Bio-Rad).
Primers to detect either WT HLTF mRNA (F: 5’-GTTCAAAGATTAATGCGCT-3′ and R: 5’-AAAGACAGGAATGTTGTAAACTGAGA-3′) or HLTF mRNA variants I21R (F: 5’-TCCAGTTTCAAAGGTAAAGTACTC-3′ and R: 5’-GCCAGTGGTCAACAACAGAA-3′) by ddPCR were designed with Primer3 and purchased from Eurogentec. Primers were tested for nonspecific amplicons and primer dimers by visualizing PCR products on 1% agarose gels and droplet distribution profile (QuantaSoft, Bio-Rad).
Statistical analyses
Expression levels of different variants of HLTF were measured in triplicates. The reliability was assessed by calculating the intra-class correlation coefficient (ICC) based on the within and between variances estimated using the variance component analysis. For the outcome analysis, the three replicates were averaged for each sample. Two outcome variables were assessed: overall survival (OS) and disease-free survival (DFS). Both were measured from surgery date. For OS, the time was calculated up to the date of death or last follow-up with death of any cause as an event; for DFS, the time was calculated up to the date of relapse, death or last follow-up with death or relapse as events. There were 71 deaths (number of events for OS) and 81 events for DFS in the cohort. The averages of the three replicates of WT and I21R HLTF expressions were tested for their associations with OS and DFS using the Cox proportional hazards regression. Both variants of HLTF were also dichotomized at their respective medians and were tested as categorical variables using the log-rank test. The percentages for OS and DFS for the high and low values of each of these covariates were calculated using the Kaplan-Meier method. A composite covariate was created by combining WT and I21R HLTF expression levels and a data-driven covariate was defined as “Low WT HLTF and High I21R HLTF” vs. the rest. This new covariate was also tested for its association with OS and DFS by employing the log-rank test. These covariates were tested for their association with outcome, adjusting the model for age (≤65, > 65), sex, stage (I vs. II), and histology (ADC vs. the rest) using Cox regression. All p-values were based on the Wald test. HLTF expressions (continuous) were also tested for their associations with the clinical factors (age, sex, histology, and stage) using the Mann-Whitney test. A cut-off of p ≤ 0.05 was used for statistical significance.
Discussion
The purpose of this study was to assess the expression of WT and variant forms of
HLTF mRNAs in NSCLC and evaluate their clinical relevance. Our hypothesis was that the expression of
HLTF mRNA variant I21R has a poor prognosis on patients with NSCLC. In head and neck, cervix and thyroid cancers, the expression of HLTF truncated protein has been associated with poor outcome [
16‐
20]. The present study showed that in a cohort of 171 patients, the combination of low expression of WT
HLTF transcript and high expression of I21R
HLTF transcript was associated with poor prognosis in early stage NSCLC.
Overall, in silico analysis showed that
HLTF alterations, including gene amplifications, high expression, and methylation occurred more frequently in SCC than in ADC. Mutations in
HLTF were rare in both ADC and SCC; however, the mutations observed in ADC were different from those found in SCC. In ADC, mutations occur in DNA binding domain and DNA repair domains (Fig.
2), which might alter HLTF transcriptional and DNA repair abilities. Conversely, in SCC, mutations did not occur in functional domains but there are 2 nonsense mutations leading to the expression of a shorter protein containing only the DBD. This suggests that these shorter proteins would only have transcriptional activity. Further investigations are required to assess the functional consequence and potential clinical impact of these mutations in cancer. Copy number alterations were also found to be different between ADC and SCC; high amplifications were rare in ADC, but 83% of SCC have either a gain (57%) or an amplification (26%) of
HLTF. These observations are consistent with the fact that
HLTF is located on chromosome 3q, which is frequently amplified in SCC. We also analyzed the association of
HLTF expression with its methylation status in both NSCLC types. In both ADC and SCC, there was a negative correlation between methylation and
HLTF expression, but a high expression was more frequently seen in SCC, which might be related to the higher frequency of gene copy number. Intriguingly, we did not notice any difference in
HLTF expression levels (WT and I21R) between ADC and SCC by RT-ddPCR. This discrepancy may be possibly explained by the fact that we assessed
HLTF mRNA variants separately, while data reported in cBioportal considered only WT
HLTF expression without distinguishing the variants.
In the available online data, only WT
HLTF expression was assessed. To our knowledge, to date the expression of the
HLTF spliced variants with intron 21 retention (I21R) has not been assessed. Using the RT-ddPCR with specific primers that we constructed, we were able to evaluate the expression of WT
HLTF mRNA and its spliced variants I21R. Spliced variants I21R lead to the expression of shorter protein forms, which are thought to disturb WT HLTF function and act as oncogene proteins [
16]. Studies in head and neck, cervix and thyroid cancers showed that the expression of such shorter proteins was associated with poor prognosis [
17‐
20]. They replace WT HLTF progressively and accumulate along the carcinogenic process, most likely due to their higher stability compared with WT HLTF. It was reported that the I21R transcripts have a lower abundance than the WT
HLTF transcript in mouse heart and brain transcriptomes [
7,
24]. We analyzed RNA-seq data from TCGA for the presence of the intron 21 sequence and found that its expression was a rare event in NSCLC. In both the NSCLC cell lines and the 171 resected NSCLC from patients, WT
HLTF levels were significantly higher than I21R
HLTF.
Castro et al. studied the methylation for several genes including
HLTF in NSCLC and reported that patients with
HLTF methylation have shorter survival [
21]; this study represents the only study of
HLTF in lung cancer. They reported
HLTF methylation frequency for NSCLC and did not observe any significant difference for
HLTF methylation between ADC and SCC (12/33 vs. 9/20, respectively;
p = 0.57, Fisher exact test). cBioportal does not provide gene methylation frequency but only correlations with the expression of a given gene. In both ADC and SCC, we observed a negative correlation between
HLTF expression and methylation. Interestingly,
HLTF expression was affected more by the variation in
HLTF copy number than its promoter methylation status.
Conclusion
So far to our knowledge, our study is the first to assess the clinical impact of WT and variant forms of
HLTF expression in patients with NSCLC. TCGA in silico analysis of
HLTF alterations including mutations, amplification, and mRNA expression modifications were more frequent in SCC than in ADC. In NSCLC cell lines and patient samples, both the expressions of WT and spliced I21R
HLTF mRNAs were detected, but with the latter at lower levels. In a cohort of 171 patients with resected stage I-II NSCLC, the combination of a low WT
HLTF expression with a high I21R
HLTF expression was associated with shorter DFS both in univariate and multivariate analyses. Surgically resected early stage NSCLC are very heterogeneous and no prognostic factor has been clinically validated for the risk of relapse. Very likely, a panel of several biomarkers will be necessary to predict tumour with poor prognostic; that would therefore require more intensive follow-up and treatment. If validated in independant cohorts, the combination of low WT and high I21R
HLTF might belong to this biomarker panel for the prognostic of surgically resected NSCLC. As detailed in a review article we published recently [
16], the
HLTF gene could be involved in various ways during the stages of tumour initiation and progression, by its ability to alternatively express proteins of different sizes with distinct functions ranging from tumour suppressor to oncoprotein. The involvement of alternative RNA splicing in producing tumour promoting proteins is a process that does not require inactivating mutation of a tumour suppressor gene and might be an underestimated carcinogenic mechanism. Further studies should precisely investigate the functions of these HLTF protein forms and their role in cancer development.
Acknowledgements
We acknowledge Dr. Claude Lachance for his help in ddPCR design.