A functional variant at the miR-885-5p binding site of CASP3 confers risk of both index and second primary malignancies in patients with head and neck cancer

Abstract

Caspases are important regulators and executioners in the apoptosis pathways and play crucial roles in carcinogenesis. We tested the hypothesis that functional variants of CASP genes are associated with risk of squamous cell carcinoma of the head and neck (SCCHN) and second primary malignancy (SPM). We genotyped 7 selected, potentially functional single nucleotide polymorphisms (SNPs) located in the microRNA binding sites of the 3′ untranslational region (UTR; 2 in CASP3, 1 in CASP6, and 4 in CASP7) and evaluated their associations first with risk of SCCHN in 1066 patients with SCCHN and 1074 cancer-free control subjects and then with SPM in 846 patients in the same non-Hispanic white study population. We found that compared with the CASP3 TT genotype of rs1049253, the variant TC/CC genotypes were associated with significantly increased risk of SCCHN (adjusted odds ratio=1.29 and 95% confidence interval=1.07-1.56) and SPM (adjusted hazard ratio=1.79 and 95% CI=1.02–3.16) and worse SPM-free survival (log-rank P = 0.020), but no associations were found for the other 6 SNPs. We then performed additional experiments to seek functional relevance of the rs1049253 SNP. First, the luciferase activity and miR-885-5p mimic transfection tests suggested that CASP3 was the target of miR-885-5p and that rs1049253T>C resulted in altered regulation of the CASP3 expression. Second, the rs1049253 CC genotype was associated with reduced levels of CASP3 mRNA in peripheral blood mononuclear cells from 118 SCCHN patients and 103 cancer-free control subjects and lower levels of CASP3 protein expression in 11 head and neck cancer cell lines, compared with the TT genotype. Taken together, our data suggest that the miR-885-5p binding site rs1049253T>C SNP in the 3′-UTR of CASP3 modulates CASP3 expression at both mRNA and protein levels and thus contributes to SCCHN susceptibility. Guan, X., Liu, Z., Liu, H., Yu, H., Wang, L.-E., Sturgis, E. M., Li, G., and Wei, Q. A functional variant at the miR-885-5p binding site of CASP3 confers risk of both index and second primary malignancies in patients with head and neck cancer.

Cancers in the oral cavity, pharynx, and larynx account for >90% of squamous cell carcinoma of the head and neck (SCCHN), which is one of the 6 most common cancers worldwide (1). In the United States, ∼52,140 new SCCHN cases occur annually, accounting for 3.2% of all newly diagnosed noncutaneous malignancies, with 11,460 deaths in 2011 (2). Furthermore, ∼15% of patients with SCCHN develop a second primary malignancy (SPM), which is a significant cause of post-treatment morbidity and mortality of this disease (3). The use of tobacco and alcohol (4, 5) are the known risk factors for both primary and secondary SCCHN, and some cancer treatments (6, 7) are also known to play a role in SPM development. However, these factors alone do not fully explain the risk of both primary and secondary SCCHN.

Apoptosis, or programmed cell death, is an important mechanism for maintaining internal homeostasis by removing irreparably damaged cells (8–10). Therefore, deregulation of apoptosis can disrupt the delicate balance between cell proliferation and cell death, thus leading to cancer development (8). Cells have two main signaling pathways that lead to apoptosis, an extrinsic (death receptor) pathway and an intrinsic (mitochondrial) pathway (11–13); however, both pathways converge to activate the effector caspases-3, -6, and -7, which are important mediators for apoptotic process by cleaving numerous intracellular substrates in the initiation of cell dissolution (14). Therefore, caspases play a central role in the execution phase of cellular apoptosis, and many types of human cancers and cell lines contain somatic mutations and genetic variants in these caspase genes (15–20), which are known to be associated with cancer risk, including SCCHN (21–24).

Recent studies have indicated that microRNAs (miRNAs), which bind to the 3′ untranslated region (UTR) of mRNAs of their target genes, play an important role in regulating apoptosis related to carcinogenesis (25, 26). Elevated or decreased expression of miRNAs has been found in various tumor types, and miRNAs may function as tumor suppressor genes and/or oncogenes in many cancers, including SCCHN (27–29). Furthermore, single-nucleotide polymorphisms(SNPs) located in the miRNA binding sites (i.e., miRNA-binding SNPs) of the target genes may affect the regulation and function of miRNA-mediated genes and are, thus, associated with cancer susceptibility (30).

Although some studies have investigated miRNA expression in SCCHN tissues and cancer cell lines (31, 32), few studies have investigated the associations between SCCHN risk and miRNA-binding SNPs and miRNA levels of caspase genes, particularly the apoptosis effector CASP3, CASP6, and CASP7, in both the intrinsic and extrinsic pathways. In the present study, we tested the hypothesis that genetic variants in the predicted miRNA-binding sites of CASP3, CASP6, and CASP7 may modulate expression of their target mRNAs and may, thus, be associated with risk of SCCHN and SPM. We first performed a case-control analysis of DNA samples from 1066 non-Hispanic white patients with SCCHN and 1074 control subjects frequency-matched for age, sex, and ethnicity. Among these 1066 cases, only 846 patients who were treated with curative intent at our institution were available for the analysis of SPM risk and SPM-free survival. To address the functional relevance of the selected miRNA-binding SNPs that are associated with SCCHN risk, we further evaluated the luciferase activity using cloning and reporter gene assays, as well as expression levels of their mRNA using real-time quantitative RT-PCR assay with peripheral blood mononuclear cells (PBMCs) from a subset of the subjects.

MATERIALS AND METHODS

Study population

Details of the recruitment of SCCHN case and control subjects have been reported elsewhere (22). Briefly, the 1066 non-Hispanic white subjects with newly diagnosed, untreated index primary tumors of the oral cavity (29.6%), oropharynx (50.7%), or larynx/hypopharynx (19.7%) were recruited at The University of Texas M. D. Anderson Cancer Center between October 1999 and October 2007. By using frequency matching for age (±5 yr), sex, and ethnicity, we identified 1074 cancer-free control subjects from among hospital visitors at M. D. Anderson Cancer Center during the same time period. We interviewed each eligible participant to obtain data on tobacco smoking and alcohol. Those participants who had smoked <100 cigarettes in their lifetime were defined as never smokers; otherwise, they were considered ever smokers. Those smokers who had quit for >1 yr were considered former smokers, and the remaining smokers were defined as current smokers. Similarly, participants who had consumed alcoholic beverages at least once a week for >1 yr previously were defined as ever drinkers. Ever drinkers who had quit drinking for >1 yr previously were defined as former drinkers, and the others were defined as current drinkers. Patients with SPM were recorded as described previously (33). Briefly, SPMs were considered if the second lesions had different histopathologic types, or if they developed >5 yr after treatment for the index tumor, and/or clearly separated by normal epithelium, according to clinical and radiographic assessment. Pulmonary lesions were included as a SPM, if they had a nonsquamous histology or if they were isolated squamous lesions over 5 yr from index SCCHN and considered by both thoracic oncologist and thoracic surgeon as a SPM. When there was a discrepancy or difference in opinions regarding recurrence or SPM, the second lesion was not considered a SPM but a local recurrence. Among the 1066 cases, 846 SCCHN patients were treated with curative intent and were available for analysis of SPM risk and SPM-free survival. Having given their written informed consent, each eligible subject provided additional information about risk factors for SCCHN. The University of Texas M.D. Anderson Cancer Center Institutional Review Board approved the research protocol.

miRNA-binding site SNP selection and genotyping

CASP3, CASP6, and CASP7 have been mapped to chromosome 4q33-q35.1, 4q25-q26, and 10q25.1-q25.2, respectively (34). It has been reported that CASP3 has 195 SNPs, CASP6 has 162 SNPs, and CASP7 has 263 SNPs (last search date: March 17, 2010; dbSNP database; http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?chooseRs=all&go=Go&locusId=836). Using the TargetScan online tool (34), we identified 22 SNPs in the 3′-UTR of CASP3, of which 10 are located in the predicted miRNA binding sites (35). However, only 2 of these 10 SNPs (rs1049253G>A [miR-885-5p], and rs1049216G>A [miR-181]) are reported to be common [i.e., minor allele frequency (MAF) > 0.05 in Caucasians]; for CASP6, 21 SNPs are located in the 3′-UTR, of which 8 are located in the predicted miRNA binding sites, but only 1 (rs1042891C>T [miR-944]) is reported to be common; for CASP7, 22 SNPs are in the 3′-UTR of CASP7, of which 10 are located in the predicted miRNA binding sites, and 4 (rs4353229T>C [miR-224], rs10787498T>G [miR-19a], rs1127687G>A [miR-136], and rs12247479A>G [miR-345]) are reported to be common; therefore, we selected and genotyped these 7 (i.e., 2 for CASP3, 1 for CASP6, and 4 for CASP7; Supplemental Table S1) for genotyping in all the subjects, using the TaqMan assay with the Sequence Detection Software on an ABI-Prism 7900 instrument, according to the manufacturer's instructions (Applied Biosystems, Foster City, CA, USA).

We extracted genomic DNA from the buffy coat fraction of the blood samples by using a blood DNA mini kit (Qiagen, Valencia, CA, USA), according to the manufacturer's instructions. The DNA purity and concentration were determined by spectrophotometer measurement of absorbance at 260 and 280 nm. Primers and probes were supplied by Applied Biosystems (Foster City, CA, USA). Each plate included 4 negative controls (no DNA), duplicated positive controls, and 8 repeat samples. Amplification was done under the following conditions: 50°C for 2 min, 95°C for 10 min, 40 cycles of 95°C for 15 s, and 60°C for 1 min. For all genotypes, the assay success rate was >99%, and the repeated sample's results were 100% concordant.

Transfection of hsa-miR-885-5p and real-time RT-PCR analysis of CASP3 mRNA

The hsa-miR-885-5p mimic and the negative control RNA duplex (Sigma-Aldrich, Atlanta, GA, USA) were transfected into UMSCC14A cells seeded in 6-well plates using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). Cells were harvested 48 h after transfection, and RNAs were isolated for the real-time RT-PCR analysis of CASP3 mRNA. Total RNA was isolated from UMSCC14A cells and PBMCs from a subset of 118 patients with SCCHN and 103 cancer-free control subjects by using the TRIzol reagent (Invitrogen). Expression levels of CASP3 mRNA in the total RNA were measured by using TaqMan gene expression assays of quantitative RT-PCR, in which the master mix reagent was used according to the manufacturer's instructions with the ABI-Prism 7900HT Sequence Detection System (Applied Biosystems). Each sample was analyzed in duplicate, and the normalized expression levels were calculated relative to that of RNA 18S.

Reporter constructs, transfection, and luciferase assays

The 425-bp fragment of the CASP3 3′-UTR containing the rs1049253T or C allele was amplified with the forward primer 5′-GTAGACGCGTCCAAGTCTCACTGGCTGTCA-3′ and the reverse primer 5′-TCCAGTTTAAACGGAACTTCTGCGAGGACTTG-3′ from a homozygous human genomic DNA sample. The PCR products were separated in agarose gel and extracted, purified, and cloned into pMIR-Report plasmids (Applied Biosystems) with digestion by MluI and PmeI. All constructs used in the experiments were verified by direct sequencing (see Fig. 2B). Two head and neck cancer cell lines (UM-SCC-1 and UMSCC14A; ref. 36) were seeded into 24-well plates at 0.5 × 10⁵ cells/well (BD Biosciences, Bedford, MA, USA), and 24 h after plating, the cells were cotransfected with the FuGene HD reagent (Roche Applied Science, Indianapolis, IN, USA). Each cotransfection reaction contained 500 ng of the pMIR-CASP3 (rs1049253) T or C vector, 50 ng pRL-TK (Promega, Madison, WI, USA) plasmids used as a transfection internal control, and 50 pmol negative control (scrambled sequence) or miR-885-5p RNA (Sigma-Aldrich). The cells were washed and lysed with 100 μl Passive Lysis Buffer (Promega) 48 h after transfection. Both firefly and Renilla luciferase activities were quantified by a dual-luciferase reporter assay system (Promega), and the relative luciferase activity was calculated according to the manufacturer's instructions (BD Monolight 3010 Luminometer; Becton Dickinson, Mississauga, ON, Canada). Physical and biological containment procedures of recombinant DNA used in this study were practiced in accordance with the U.S. National Institutes of Health.

Figure 2.

Characterization and functional analysis of the CASP3 3′-UTR. A) The rs1049253 polymorphism and its exact location within the CASP3 3′-UTR. Sequence complementarity of miR-885-5p and CASP3 (rs1049253) is shown here. The SNP rs1049253T>C is located within the seed region of the binding site with the C allele, perfectly matching the corresponding G allele in miR-885-5p. B) Schematic drawing of the reporter gene constructs containing a 425-bp CASP3 3′-UTR region; the only difference between the two constructs was a T or C at the stop codon 1475 polymorphic site. C) Luciferase activity assays to measure T/C allele difference at rs1049253. UM-SCC-1 and UMSCC14A head and neck cancer cell lines were transiently transfected with T- or C-containing reporters. All constructs were cotransfected with pRL-TK Renilla plasmid as an internal control. Results are shown as the relative percentages of luciferase activity. Data are from 3 independent transfection experiments with assays conducted in 6 replications. D) Relative luciferase activity assay to measure the difference in expression levels of transcripts containing either rs1049253T or C allele in the presence of negative control or miR-885-5p RNA in UM-SCC-1 and UMSCC14A cells. Cells were transiently cotransfected with constructs and 50 pmol miR-885-5p mimic or negative control (NC). Results are shown as the luciferase activity relative to that of the control. Three replicates were analyzed for each group, and the experiment was repeated ≥3 times. P values were determined by 2-sided Student's t test. E) Down-regulation of CASP3 mRNA by hsa-miR-885-5p in UMSCC14A cells. Cells were transfected with hsa-miR-885-5p mimic, and RNA was harvested 48 h after transfection. cDNA was synthesized and used for the real-time reverse transcription PCR analysis of CASP3 mRNA normalized to an RNA 18S standard. Expression levels were relative to the expression of UMSCC14A cells. Data were obtained from 3 independent transfection experiments, with assays conducted in triplicate.

Western blot analysis of CASP3 expression

Eleven previously established head and neck cancer cell lines (36) were used in these experiments. Because the baseline CASP3 expression levels were very low in these cell lines (data not shown), we stimulated CASP3 expression by treating the cells with 250 nM camptothecin (CPT) for 24 h, as described previously (37). The cells were then lysed, and the proteins were separated by 12% SDS-PAGE, followed by transfer onto nitrocellulose membranes (Bio-Rad, Hercules, CA, USA), which were blocked with 10% nonfat dry milk (0.2% TBST) for 40 min at room temperature and incubated overnight at 4°C with the following primary antibodies: polyclonal anti-caspase3 antibody (1:500, Millipore, Temecula, CA, USA) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH; 1:5,000; Millipore). The membranes were then incubated with the secondary horseradish peroxidase-linked antibodies (1:10,000; Calbiochem, San Diego, CA, USA), and signals were visualized by using SuperSignal West Femto Maximum Sensitivity Substrate (ThermoScientific, Rockford, IL, USA). Bands were quantified using Alpha Innotech software (Alpha Innotech, Santa Clara, CA, USA).

Statistical analysis

Differences in selected demographic variables, smoking status, and alcohol use between SCCHN case and control subjects were evaluated by the χ² test. The associations between CASP3 genotypes and SCCHN risk were estimated by computing odds ratios (ORs) and their 95% confidence intervals (CIs) from both univariate and multivariate logistic regression models in case-control analysis, and the associations between CASP3 genotypes and risk of SPM were estimated in case-only analyses with/without adjustment for age, sex, smoking status, and alcohol use. We assessed hazard ratios (HRs) and their 95% confidence intervals (CIs) for a SPM development in the Cox regression models. The means ± se of fold changes in expression levels were calculated and tested with Student's t test by using Stata 10.0 (StataCorp LP, College Station, TX, USA). All tests were 2-sided, and values of P < 0.05 were considered statistically significant. All statistical analyses were performed with SAS 9.1.3 software (SAS Institute, Cary, NC, USA), unless stated otherwise.

RESULTS

Association of the selected CASP gene variants with risk of SCCHN

Selected characteristics of case and control participants are presented in Table 1. All study subjects were non-Hispanic whites. Consistent with the frequency matching by design, there were no significant differences in the distributions of age and sex between the case and control subjects, with similar mean ages of 57.1 ± 11.2 yr for case subjects and 56.7 ± 11.0 yr for control subjects. The genotype distributions of the 7 SNPs in the control subjects were in agreement with the Hardy-Weinberg equilibrium (data not shown). As shown in Table 2, compared with the CASP3 TT genotype of rs1049253, a SNP at the microRNA binding site of the gene's 3′-UTR, the variant CC and TC/CC genotypes were associated with a significantly increased risk of SCCHN (adjusted OR, 1.80; 95% CI, 1.13–2.87; P=0.014 for the CC genotype; and adjusted OR, 1.29; 95% CI, 1.07–1.56; P=0.007 for the TC/CC genotypes; P_trend=0.003); however, no cancer risk associations with the other 6 SNPs for CASP3, CASP6, and CASP7 were observed (Table 2). In the stratified analysis with the rs1049253T>C SNP, the increased risk associated with the variant CC/CT genotypes was more evident among older subjects (adjusted OR=1.51; 95% CI=1.15–2.00), males (adjusted OR=1.32; 95% CI=1.06–1.64), never and former smokers (adjusted OR=1.44 and 1.45; 95% CI=1.07–1.93, and 1.06–1.97, respectively) and never drinkers (adjusted OR=1.60; 95% CI=1.17–2.18), compared with the TT genotype (Table 3).

Table 1.

Frequency distributions of selected variables in SCCHN case subjects and cancer-free control subjects recruited from the University of Texas M. D. Anderson Cancer Center in Houston, Texas between October 1999 to October 2007

Variable	Case subjects		Control subjects		P
	n	%	n	%
Total	1066	100.0	1074	100.0
Age (yr)					0.787
≤57	568	53.3	566	52.7
>57	498	46.7	508	47.3
Sex					0.581
Female	263	24.7	254	23.7
Male	803	75.3	820	76.3
Smoking status					<0.001
Never	296	27.8	531	49.4
Former	369	34.6	389	36.2
Current	401	37.6	154	14.4
Alcohol use					<0.001
Never	290	27.2	466	43.4
Former	235	22.0	174	16.2
Current	541	50.8	434	40.4
Tumor site
Oropharynx	540	50.7
Nonoropharynxa	526	49.3

P values were calculated by 2-sided χ² test.

^aIncluded oral cavity (n=316), larynx (n=167), and hypopharyngeal (n=43).

Table 2.

Genotype frequencies of the CASP3, CASP6, and CASP7 polymorphisms among SCCHN cases and controls and their associations with SCCHN risk in a case-control study conducted at The University of Texas MD Anderson Cancer Center in Houston, Texas between October 1999 and October 2007

Polymorphism and genotype	Case subjects [n (%)]	Control subjects [n (%)]	Ptrend	Adjusted OR [OR (95% CI)]
CASP3
rs1049253 T>C	1066 (100.0)	1074 (100.0)
TT	669 (62.8)	734 (68.3)	0.003*	1.00
TC	346 (32.4)	306 (28.5)		1.24 (1.02–1.15)*
CC	51 (4.8)	34 (3.2)		1.80 (1.13–2.87)*
TC/CC	397 (37.2)	340 (31.7)		1.29 (1.07–1.56)*
rs1049216 A>G	1066 (100.0)	1074 (100.0)
AA	581 (54.5)	542 (50.5)	0.095	1.00
AG	401 (37.6)	441 (41.0)		0.83 (0.68–1.00)
GG	84 (7.9)	91 (8.5)		0.86 (0.62–1.20)
AG/GG	485 (45.5)	532 (49.5)		0.83 (0.70–0.99)
CASP6
rs1042891 C>T	1065 (100.0)	1072 (100.0)
CC	480 (44.8)	462 (43.5)	0.416	1.00
CT	453 (42.5)	473 (43.7)		0.92 (0.77–1.12)
TT	132 (12.4)	137 (12.8)		0.91 (0.69–1.21)
CT/TT	592 (54.9)	610 (56.5)		0.93 (0.77–1.11)
CASP7
rs1127687 G>A	1066 (100.0)	1074 (100.0)
GG	613 (57.5)	635 (59.1)	0.989	1.00
AG	407 (38.2)	381 (35.5)		1.09 (0.90–1.31)
AA	46 (4.3)	58 (5.4)		0.80 (0.52–1.21)
AG/AA	453 (42.5)	439 (40.9)		1.05 (0.88–1.26)
rs10787498 T>G	1062 (100.0)	1071 (100.0)
TT	466 (43.9)	470 (43.9)	0.818	1.00
GT	482 (45.4)	486 (45.4)		1.03 (0.85–1.24)
GG	114 (10.7)	115(10.7)		0.93 (0.69–1.26)
GT/GG	596 (56.1)	601 (56.1)		1.01 (0.84–1.20)
rs4353229 T>C	1064 (100.0)	1074 (100.0)
TT	611 (57.4)	614 (57.2)	0.410	1.00
CT	387 (36.4)	403 (37.5)		1.03 (0.86–1.25)
CC	66 (6.2)	57 (5.3)		1.22 (0.82–1.78)
CT/CC	453 (42.6)	460 (42.8)		1.06 (0.88–1.26)
rs12247479 A>G	1066 (100.0)	1073 (100.0)
AA	838 (78.1)	827 (77.1)	0.365	1.00
AG	219 (20.6)	228 (21.2)		0.95 (0.76–1.18)
GG	14 (1.3)	18 (1.7)		0.68 (0.32–1.41)
AG/GG	233 (21.9)	246 (22.9)		0.93 (0.75–1.15)

Case and control subject numbers were not the same for each SNP because of their different calling rates due to few uncalling samples. OR and 95% CI values were adjusted by age, sex, smoking status, and alcohol use in logistic regression models.

^*P < 0.05.

Table 3.

SCCHN risk associated with CASP3 rs1049253 genotypes by selected subject characteristics in a case-control study

Variable	Case subjects, n = 1066		Control subjects, n = 1074		Adjusted OR [OR (95% CI)]		P
	TT [n (%)]	CC/CT [n (%)]	TT [n (%)]	CC/CT [n (%)]	TT	CC/CT
Age (yr)
≤57	376 (66.2)	192 (33.8)	382 (67.5)	184 (32.5)	1.00	1.11 (0.86–1.43)	0.440
>57	293 (58.8)	205 (41.2)	352 (69.3)	156 (30.7)	1.00	1.51 (1.15–2.00)*	0.003*
Sex
Female	159 (60.5)	104 (39.5)	166 (65.4)	88 (34.6)	1.00	1.23 (0.84–1.81)	0.293
Male	510 (63.5)	293 (36.5)	568 (69.3)	252 (30.7)	1.00	1.32 (1.06–1.64)*	0.011*
Smoking status
Never	178 (60.1)	118 (39.9)	363 (68.4)	168 (31.6)	1.00	1.44 (1.07–1.93)*	0.018*
Former	232 (62.9)	137 (37.1)	277 (71.2)	112 (28.8)	1.00	1.45 (1.06–1.97)*	0.019*
Current	259 (64.6)	142 (35.4)	94 (61.0)	60 (39.0)	1.00	0.84 (0.57–1.25)	0.399
Alcohol use
Never	171 (59.0)	119 (41.0)	323 (69.3)	143 (30.7)	1.00	1.60 (1.17–2.18)*	0.003*
Former	151 (64.3)	84 (35.7)	114 (65.5)	60 (34.5)	1.00	0.94 (0.61–1.44)	0.768
Current	347 (64.1)	194 (35.9)	297 (68.4)	137 (31.6)	1.00	1.25 (0.94–1.65)	0.131

OR, 95% CI, and P values were adjusted for age, sex, smoking status, and alcohol use in a logistic regression model, where it was appropriate.

^*P < 0.05.

Association of CASP3 variant (rs1049253) with risk of SPM

Because only CASP3 genotype of rs1049253 was associated with risk of index SCCHN, we sought to provide additional support for such an association in risk of SPM and SPM-free survival among 846 patients with SCCHN treated for curative intent at our institution. We found that the rs1049253TC/CC variant genotypes were more common among patients with SPM than among patients without SPM (P=0.015), with an ∼80% increase in SPM risk (adjusted HR=1.79; 95% CI=1.02–3.16), compared with the rs1049253TT genotype (Table 4). Furthermore, the SPM-free survival duration differed by the rs1049253 genotypes; that is, the variant TC/CC genotypes were associated with worse SPM-free survival (11.7 mo), compared with the CASP3 TT genotype (17.7 mo) (overall log rank P=0.020; Fig. 1).

Table 4.

SPM risk associated with CASP3 polymorphisms after index SCCHN

Genotype	Total, n = 846		SPM-free, n = 797		SPM, n = 49		P	OR (95% CI)
	n	%	n	%	n	%
rs1049253 T>C
TT (ref.)	519	61.4	497	62.4	22	44.9	0.015*
TC+CC	327	38.6	300	37.6	27	55.1		1.79 (1.02–3.16)

P values were calculated by χ² test for difference in the distribution of CASP3 genotypes between patients who developed SPM and patients who did not. OR and 95% CI values were adjusted for age, sex, smoking status, and alcohol use in a logistic regression model. Ref., reference group.

^*P < 0.05.

Figure 1.

SPM-free survival of patients with squamous cell carcinoma of the head and neck is shown for the risk genotypes of the CASP3 rs1049253 polymorphism. Overall log-rank test, P = 0.020 for rs1049253TT (top) vs. TC+CC (bottom).

Characterization and functional analysis of expression regulated by the CASP3 3′-UTR

To further support the 3′-UTR binding site SNP (rs1049253C>T) of CASP3 as a risk factor for SCCHN, we set to test the hypothesis that miR-885-5p binds to CASP3 mRNA transcripts containing the C or T allele differentially. We first replaced the 3′-UTR of a luciferase reporter gene with the 425-bp CASP3 3′-UTR containing either rs1049253T or rs1049253C (Fig. 2B). As shown in Fig. 2C, significantly lower levels of luciferase expression were observed when UM-SCC-1 and UMSCC14A cells were cotransfected with CASP3 3′-UTR luciferase reporter plasmids carrying the rs1049253C allele than with those plasmids carrying the T allele (UM-SCC-1 cell line: 0.633±0.277 for the C allele vs. 2.609±0.485 for the T allele, P<0.001; UMSCC14A cell line: 0.591±0.070 for the C allele vs. 1.074±0.056 for the T allele, P<0.001). Subsequently, in transiently transfected UM-SCC-1 and UMSCC14A cells, the luciferase activity (representing CASP3 protein expression) in the presence of miR-885-5p (50 pmol) was lower in constructs containing the C allele than in those containing the T allele and negative controls (P=0.003 and P<0.001, respectively; Fig. 2D). We also transiently transfected miR-885-5p mimic or the negative control into the UMSCC14A cell line, creating an rs1049253CC genotype carrier, and measured the endogenous CASP3 transcript levels in these cells. Compared with transfection with the negative control, transfection with miR-885-5p mimic also significantly decreased the levels of CASP3 mRNA expression in UMSCC14A cells (P<0.001; Fig. 2E).

Correlation of CASP3 (rs1049253) genotypes and the gene expression

To characterize the biological relevance of the CASP3 rs1049253 SNP at the miRNA binding sites, we conducted a correlation analysis between the rs1049253T>C genotypes and relative expression levels of CASP3 mRNA in PBMCs from a subset of 118 patients with SCCHN and 103 cancer-free control subjects. As shown in Fig. 3A, the normalized value for CASP3 expression levels by the 18S was lower for 24 carriers of the CC genotype (0.486±0.020), 30 carriers of the TC genotype (0.525±0.034), and 54 carriers of the TC/CC genotype (0.508±0.035) than for 64 carriers of the TT genotype (0.540±0.026), and these differences were statistically significant (P<0.001, P=0.025, and P<0.001, respectively). Similarly, in the cancer-free control subjects, the CASP3 mRNA expression levels were lower for 15 carriers of the CC genotype (0.477±0.039), 35 carriers of the TC genotype (0.516±0.032), and 50 carriers of the TC/CC genotype (0.505±0.039) than for 53 carriers of the TT genotype (0.537±0.024), and these differences were also statistically significant (P<0.001, P=0.001, and P<0.001, respectively; Fig. 3B).

Figure 3.

Genotype-phenotype correlation for rs1049253T>C and the relative expression levels of CASP3 mRNA in PBMCs from a subset of 118 SCCHN case subjects and 104 cancer-free control subjects. Quantitative real-time RT-PCR was used to measure the relative expression of CASP3 mRNA. Housekeeping gene 18S was used as the reference. P values were determined by 2-sided Student's t test. A) In PBMCs of 118 SCCHN case subjects, the relative CASP3 mRNA expression level was lower for the CC genotype (0.486±0.020), the TC genotype (0.525±0.034), and the TC/CC genotypes (0.508±0.035) than for the TT genotype (0.540±0.026), and these differences were statistically significant (P<0.001, P=0.025, and P<0.001, respectively). B) Similarly, in the 103 cancer-free control subjects, the relative CASP3 mRNA expression level was also lower for the CC genotype (0.477±0.039), the TC genotype (0.516±0.032), and the TC/CC genotypes (0.505±0.039) than for the TT genotype (0.537±0.024), and these differences were also statistically significant (P<0.001, P=0.001, and P<0.001, respectively).

Because our previous findings have shown that SNPs at the miRNA binding sites affect the expression levels of the target mRNA but not that of the miRNA (38, 39), we did not test for miRNA expression levels in the present study. Our mRNA expression data demonstrated that the rs1049253CC genotype was associated with lower levels of CASP3 mRNA, compared with the TT genotype. Subsequently, we evaluated the effects of these two target SNPs on endogenous CASP3 protein levels with and without CPT (a cytotoxic quinoline alkaloid that induces apoptosis by inhibiting the DNA enzyme topoisomerase I) stimulation in the cultures of 11 head and neck cancer cell lines carrying different rs1049253T>C genotypes. Consistent with the luciferase and mRNA results, CASP3 protein levels were preferentially down-regulated in cell lines carrying the CC genotype, compared with those carrying the TT or TC genotypes (Fig. 4 and Table 5). Taken together, these results suggested that individuals carrying the rs1049253CC genotype, which has the perfect complementary seed sequence for miR-885-5p, had lower expression levels of both CASP3 mRNA and protein than did individuals carrying the TT genotype.

Figure 4.

Western blot analysis for CASP3 levels, with or without stimulation by CPT, in 11 head and neck cancer cell lines carrying different rs1049253T>C variants. Genotypes are reported at the top; at bottom are reported CASP3 expression levels normalized for that of the GAPDH protein and compared with and without stimulation by CPT for each cell line. Bands were quantified using Alpha Innotech software (Alpha Innotech, Santa Clara, CA, USA). Mean protein expression levels are reported in Table 5.

Table 5.

CASP3 protein expression levels in 11 head and neck cell lines

Genotype	n	Mean ± sd	P
TT	4	1.303 ± 0.165	1.00
CC	4	1.027 ± 0.116	0.034*
CT	3	1.314 ± 0.240	0.947
CC/CT	7	1.150 ± 0.222	0.263

See Fig. 4. P values were calculated by t test.

^*P < 0.05.

Finally, we calculated the false-positive reporting probability (FPRP) values at different prior probability levels for all significant association findings. For a prior probability of 0.25, assuming the OR for a specific genotype is 1.50, with a statistical power of 0.940, the FPRP value was 0.027 for an association between CASP3 rs1049253C>T SNP and risk of primary SCCHN, but the corresponding value for an association with risk of SPM was 0.331 (Supplemental Table S2), suggesting the inadequacy of the events in the patient follow-up.

DISCUSSION

To our knowledge, this is the first study to evaluate the associations between polymorphisms at the miRNA binding sites of 3 CASP genes (i.e., CASP3, CASP6, and CASP7) and risk of primary SCCHN and SPM in the same patients. We found that only the CASP3 rs1049253 variant C genotypes conferred a significantly increased risk of primary SCCHN, compared with the TT genotype, in the initial case-control analysis. Subsequently, we found that this association was also true for SPM risk, which increased by ∼80% in the case-only analysis and that the variant C carriers also had shorter SPM-free survival.

The observed associations were supported by additional experiments for their possible underlying molecular mechanisms. For example, in the luciferase reporter assay, the rs1049253 variant C allele was found to be associated with significantly lower luciferase activity, compared with the T allele. Furthermore, the transfected cells carrying the rs1049253C allele with miR-885-5p mimic also showed substantial inhibition of the constitutive expression of CASP3 mRNA. These experimental findings suggest that miR-885-5p may be a post-transcriptional regulator of CASP3 expression. Also, in a subset of 118 SCCHN case subjects and 103 control subjects, the PBMCs from carriers of rs1049253 variant C allele exhibited significantly lower levels CASP3 mRNA than the TT carriers. Western blot analysis further demonstrated that rs1049253 variant C carriers had lower CASP3 protein expression than the TT genotype carriers in 11 head and neck cancer cell lines. Taken together, these data strongly suggest that the CASP3 miRNA-binding-site rs1049253 SNP may modulate expression levels of CASP3 and thus play a role in the etiology of SCCHN.

These findings are biologically plausible, especially in light of the crucial roles of CASP3 in the apoptosis pathways and carcinogenesis, because CASP3 is the effector in the process of apoptosis, and its altered expression has been found in many human cancers, including SCCHN (40–42). Down-regulation of CASP3 could be a possible mechanism of altered chemoresistance as well (40), and inhibition of CASP3 activation may contribute to the escape of cancer cells from apoptosis (41). Several studies have shown associations between caspase SNPs and cancer risk (43–45), but only one study indicated that a genetic variation—in the first intron of CASP3 (rs4647601), which is not in linkage disequilibrium with rs1049253, which was investigated in the present study—may contribute to SCCHN risk (21).

Our findings also agree with those of a recent report (46), in which the researchers observed some indirect evidence of CASP3 targeting by miR-885-5p in neuroblastoma cells with elevated p53 protein levels. However, the FPRP value for the significant finding of an association between CASP3 rs1049253C>T SNP and risk of primary SCCHN was 0.05, only when the prior probability was set at 0.25 with a sufficient statistical power (0.940), whereas the finding of an association with the subsequent SPM had an insufficient power, likely due to the reduction in the number of patients who might not have enough follow-up time to develop SPM, suggesting that these findings may not be noteworthy. Therefore, additional larger replication studies are needed to confirm these results.

The present study also has several limitations. First, this hospital-based case-control study included control subjects who were not randomly selected from the same target population as the case subjects. Second, clinical outcomes of the patients, including SPM, were collected retrospectively, without a strictly defined screening or follow-up regimen. Furthermore, the follow-up time was somewhat limited, so that the case cohort may not have had sufficient time to develop SPM. Finally, HPV16, one of the major risk factors for SCCHN, was not evaluated for patients included in this study. The joint effects of HPV16 and CASP3 genotypes should be explored by additional prospective studies with well-designed clinical investigations with ethnic stratification.

In summary, the present study investigated 7 selected miRNA binding site SNPs in CASP3, CASP6, and CASP7 and provided, for the first time, the evidence that the CASP3 rs1049253 variant C genotypes at the miR-885-5p-binding site in the 3′-UTR of CASP3 were significantly associated with an increased risk of both SCCHN and SPM, as well as a shorter SPM-free survival, compared with the TT genotype, in a non-Hispanic white population. Additional laboratory experiments provided further support for the biological plausibility of the association findings. However, further larger studies and mechanistic investigation of the molecular mechanisms underlying the observed associations and possible interaction with HPV infection would shed light on CASP3-associated pathogenesis and facilitate the targeted prevention, earlier diagnosis, and more effective treatment of head and neck cancer.