Abstract
Background/Aim: Activin, a member of the TGF-β super family of cytokines, is involved in head and neck squamous cell carcinoma (HNSCC). This study examined the constituents of the activin axis in order to further elucidate the role of activin A in HNSCC progression. Materials and Methods: Immunohistochemistry (IHC), reverse transcription polymerase chain reaction (RT-PCR), MTT, and matrigel invasion assays, in addition to analysis of the tumor cancer genome atlas (TCGA), were employed. Results: IHC in HNSCC and oral leukoplakia (OPL) lesions demonstrated increased expression of the inhibin subunit βA (INHBA) (p<0.0001), as well as activin receptor type IB (ACVR1B) (p<0.0032) compared to normal mucosa. TCGA analysis revealed increased INHBA expression was associated with lymph node positive tumors (p=0.024), decreased overall survival (p=0.0167), and decreased promoter methylation (p<0.0001). Concomitant up-regulated expression of gene pathways strongly correlated with INHBA expression demonstrated further deleterious effects on survival (p<0.0148). Conclusion: Activin may be an important component of early carcinogenesis in OPL and HNSCC with unfavorable effects on clinical end-points such as survival.
Head and neck squamous cell carcinoma (HNSCC) accounts for an estimated 66,000 newly diagnosed cases every year, resulting in approximately 12,300 deaths annually in the United States (1). Approximately 50-55% of all HNSCC diagnoses are stage IV; individuals who are at a three to 12 times higher risk for mortality depending on the original tumor site (2). Locoregional recurrence occurs in approximately 15-50% of patients with HNSCC and is a major factor contributing to HNSCC-related deaths (3).
Our lab previously identified molecular markers associated with recurrent HNSCC. In that analysis, genes associated with tumor invasion and metastases contributed to the most significant signature (4). Of these ~70 genes, inhibin bA was highly over-expressed and chosen for further investigation (5-7).
Inhibin βA (INHBA) subunits homodimerize and form the ligand activin A. Similarly, INHBA, when combined with subunit inhibin α (INHA), forms inhibin; a ligand that competitively binds and sequesters activin receptors (8) (Figure 1). Similar to transforming growth factor-beta (TGF-β, activin signals through two types of cell surface receptors canonically (type I and type II receptors). The activin type II receptors, ACVR2 and ACVR2B, are the primary ligand-binding proteins that are able to bind to ligands without type I receptors. However, activin receptor type IB, ACVR1B (also called ALK4), is unable to bind to the ligand in the absence of type II receptors. Upon ligand binding, the type II receptor complex activates the type I receptor, resulting in phosphorylation of Smad complex proteins (Figure 1). The Smad complex translocates to the nucleus to activate downstream genes (9, 10). In short, ACVR1B is essential for the stimulation of the activin axis.
The canonical activin A pathway. Activin A is composed of inhibin βA subunits (βA) and binds to activin receptor type II and IIB (ACVR2/2B). Inhibin, composed of a βA and α subunit, competitively binds and sequesters ACVR2/2B, ultimately inhibiting the activin axis. Contrariwise, activin binding and receptor activation leads to the phosphorylation of Smad proteins, forming the Smad transcription complex (comprising Smad 2, 3, and 4). Of note, activin may have proliferative effects outside of this axis, as described in the introduction and discussion.
In oral squamous cell carcinomas (SCCs) and other aerodigestive cancers, activin A over-expression has been correlated with metastatic potential, positive lymph node status, poor patient prognosis, and recurrence (4-7, 11-15). However, the exact underlying pathophysiology resulting in these observations is not fully understood. Since activin A stimulates epithelial-to-mesenchymal transition (EMT) during embryogenesis, activin A over-expression may contribute to metastatic processes in HNSCC (7, 11, 16, 17).
Presently, we sought to investigate constituents of the activin axis in order to further elucidate the role of activin A in HNSCC progression in vivo. This was achieved by examining the expression of activin signaling molecules in normal tissue, premalignant leukoplakia lesions (OPL), malignant HNSCC, and a variety of HNSCC cell lines. The invasive and proliferative effects of activin A treatment on HNSCC cell lines in vitro were also measured. We observed an over-expression of multiple activin subunits across the spectrum of carcinogenesis (from normal tissue to invasive carcinoma). Furthermore, there was a lack of up-regulation of inhibitory components of the activin axis, which could further contribute to this increased activity. HNSCC tumors exhibited only a minimal proliferative and invasive response when treated with exogenous activin A, which was not sustained throughout treatment.
To extend our findings, we analyzed the tumor cancer genome atlas (TCGA), which demonstrated INHBA is significantly over-expressed in HSCCC and decreased promoter methylation observed in tumors vs. normal tissue samples. Further, increased INHBA expression was observed in node positive tumors and was associated with a decrease in overall survival. Multiple genes that were strongly associated with activin A expression have demonstrated carcinogenic properties in related fields of research. Tumors with concomitant up-regulated expression of components of these pathways with increased activin A expression further attenuated survival.
Materials and Methods
Informed consent was obtained from patients according to guidelines set forth by the Institutional Review Board of the Human Subjects Protection Committee at the University of Minnesota. Twelve HNSCC tumor samples and fifteen randomly selected oral leukoplakia specimens were utilized for this study. The tumor tissues were obtained from the Minnesota Cancer Center Tissue Procurement Facility and the leukoplakia samples were from the Oral Pathology Department (IRB: 0001M34501), both located at the University of Minnesota. Lab experiments were performed during the thesis work of author Ketan Patel (18).
Cell culture. The SCC cell lines, FaDu, UM-SCC-9, UM-SCC-11A, UM-SCC-11B, and UM-SCC-38, courtesy of Tom Carey (19), were maintained in Minimum Essential Media (MEM; Thermo Fisher, Waltham, MA, USA), supplemented with 10% heat inactivated fetal bovine serum (FBS; Thermo Fisher), and 2-4 mM L-Glutamine (LG; Corning, NY, USA). The HNSCC cell line UM-SCC-15 was cultured in Dulbeccos Modified Eagles Medium (DMEM; Gibco-Thermo Fisher)/F12 (ATCC), supplemented with 10% heat inactivated FBS, 400 ng/μl hydrocortisone, and 1% LG. Transformed human epidermal keratinocytes (Rhek) immortalized by Ad 12-SV40 virus were obtained from Dr. Jhong S. Rhim at the National Cancer Institute (Frederick, MD, USA) (20). Rhek cells were cultured in DMEM, high glucose, with LG and 10% FBS. The oral SCC cell lines CA 9-22 and NA and the ovarian carcinoma cell line OVCAR3 were maintained in RPMI 1640 (Invitrogen, Carlsbad, CA, USA) with 10% heat inactivated FBS, 1% PS, and 1% LG. All cells were mycoplasma free by PCR testing.
PCR. Primer sequences are shown in Table I. PCR reagents were purchased from Thermo Fisher (now owned by Invitrogen) and used according to the manufacturer’s protocol. Briefly, cDNA from cell lines, 100 mM dNTPs, 50 mM MgCl2, 5 mM Primer Mix, 10X PCR Buffer and 5 U/ml Taq were combined in a final reaction volume of 50 ml. The reaction was run in a PCR machine for 40 cycles under standard cycling conditions. After PCR was complete, 20 ml of the mixture was run on a 1% agarose gel and observed under a gelbox and an image was taken for analysis.
Primer sequences.
Immunohistochemistry. Immunohistochemical analysis of surgically resected, formalin-fixed, paraffin-embedded tissue samples was performed with the streptavidin-peroxidase technique, which allows for the localization of tissue antigens with concomitant morphologic evaluation using light microscopy. Tissues were sectioned at 5-7 μm and immunostained according to standard protocols. The expression of the various activin signaling proteins was evaluated independently by a surgical pathologist for positivity and distribution of expression (diffuse vs. focal).
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay. Cell proliferation was determined using the MTT assay. FaDu, UM-SCC-9 and Rhek cells were plated at 5×104 cells/well in 96-well tissue culture plates. Activin at 25 ng/ml (optimal concentration) or vehicle (PBS/0.2% BSA) was added on day 0 in serum-free media. On day 0, 1,2, or 3, 3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (Sigma, St. Louis, MO, USA) was added to the culture media at 0.5 mg/ml and incubated at 37°C for 4 h. Mitochondrial dehydrogenases of live cells convert MTT to a water insoluble purple formazan, which was then solubilized in isopropyl alcohol/DMSO and the absorbance was read at 560 nm.
Matrigel invasion assay. Invasion in response to activin stimulation was determined using BD Biocoat Matrigel 6 Well Invasion Chambers (BD Biosciences, Bedford, MA, USA). FaDu and UM-SCC-9 cells were serum-starved for 24 h. The chambers were removed from the packages and allowed to come to room temperature. Serum free media was added to the interior of the inserts and the bottom of the wells. The inserts were rehydrated for 2 h in a humidified tissue culture incubator, 37°C/5% CO2 atmosphere. The medium was carefully removed and the chambers and control inserts were transferred to wells containing 25 ng/ml activin or vehicle in 2.5 ml serum free media. Two ml of serum free media containing 2×105 cells was immediately added to each chamber. The plates were incubated for 24 h in a humidified tissue culture incubator, 37°C/5% CO2 atmosphere. Non-invading cells were then removed from the upper surface of the membrane via gentle scrubbing with a cotton tipped swap moistened in media, repeated twice. The cells on the lower surface of the membrane were fixed and stained using 3 Step Stain (Richard-Allan Scientific, Kalamazoo, MI, USA). The membranes were removed from the insert housing and placed bottom side down on a microscope slide in a drop of Fluoromount-G (SouthernBiotech, Birmingham, AL, USA) with a coverslip over the top. The slides were examined under a microscope at 200× magnification. Five fields per slide were counted. Fields were chosen from both the center and the periphery of the membrane to represent more accurately cell distribution. Data is expressed as the ratio of the percent invasion in the activin wells over the percent invasion in the control wells.
Tumor cancer genome atlas analysis. Analysis of INHBA was performed in HNSCC primary tumors (523 total samples) with respect to normal tissue samples. Analysis of lymph node positive vs. negative tumors excluded HPV (+) tumors. Survival analysis utilizing Kaplan–Meier graphs compared groups with parameters in place for age (range=45-75 years), HPV (−) status, and race (white, see results). Sex ratios were similar between groups (data not shown here). High vs. normal expression in Kaplan–Meier analysis was defined as an mRNA z-score >2 with respect to normal tissue. The TCGA reports that normal tissue samples were obtained from adjacent sites of HNSCC tumors during surgical resection. Normal tissue samples were matched to HNSCC tumor samples; thereby, decreasing the risk of potential introduction of confounding variables, such as age, sex ratio, HPV status, anatomic location, tobacco, and alcohol use. Tumor stage was determined according to Neoplasm Disease Stage American Joint Committee on Cancer Code.
Statistical analysis. Graph Pad Prism 8 (Dotmatics, Boston, MA, USA) was utilized for all statistical analyses. Unpaired, two-tailed t-tests were used to compare two treatment groups. Contingency tests of categorical variables found in immunohistochemistry sections were achieved through employment of Chi-square tests. p-Values <0.05 were considered significant. Statistical analysis for TCGA data involved Pearson and Spearman correlation analysis, Kaplan–Meier curve, and t-tests using cBioPortal (21) as well as University of Alabama at Birmingham Cancer (UALCAN) Data Analysis Portal (22).
Results
Patient characteristics. Twelve patients with HNSCC were selected from a previously conducted microarray study to validate the expression of activin signaling components in these tumors. Inhibin bA expression in the specimens of these 12 ranged from mild to intense. Expression in the 15 oral premalignant (OPL) patient specimens ranged from mild to severe dysplasia. Normal control tissues were procured from five patients who underwent elective surgical treatment for sleep apnea. A board-certified oral pathologist performed all pathologic analyses.
Immunohistochemistry for activin A and its receptors in HNSCC progression. Immunostaining was performed for inhibin a (INHA), inhibin bA (INHBA), inhibin bB (INHBB), activin receptor type IB (ACVR1B), type II (ACVR2), and type IIB (ACVR2B), to determine their expression in five normal, 15 OPL, and 12 HNSCC tumor tissues (see Table II for antibodies used to receptors and ligands in activin axis). Chi-square tests were employed for analysis. INHBB was prevalent among all groups with no statistically significant differences (Table III). Both INHA and ACVR2 were relatively unfound in all groups (Table III).
Antibodies to receptors and ligands in activin axis.
Immunohistochemical expression of receptors and ligands in activin axis in normal tissue, leukoplakia, and head and neck squamous cell carcinoma (HNSCC) tumor samples.
Premalignant and malignant lesions demonstrated a statistically significant increase in the prevalence of ligand INHBA [χ2(2, N=32)=18.98, p<0.0001] as well as ACVR1B [χ2(2, N=32)=11.52, p<0.0032] (Table III). There was also a decreased prevalence of ACVR2B among pre-malignant and malignant lesions in comparison to normal mucosa [χ2(2, N=32)=0.0018, p<0.0018] (Table III).
These data suggest as oral mucosal tissues develop into premalignant lesions with dysplasia, and then into malignant squamous cell carcinoma, there is a concomitant up-regulation of INHBA and ACVR1B, which are potential agonists in the activin signaling axis (Figure 1). Furthermore, the lack of up-regulation of INHA, typically associated with sequestering intracellular responses to activin, may help facilitate this process.
Assessment of Expression in HNSCC cell lines using PCR. To determine if HNSCC cell lines express activin signaling components, six activin signaling molecules were evaluated using reverse transcription polymerase chain reaction (RT-PCR) in a variety of HNSCC cell lines. First, we found that UM-SCC-15 and CA 9-22 cells expressed detectable levels of inhibin a transcript. Inhibin bA and inhibin bB transcript expression was detectable in five of six cell lines, similar to HNSCC tissue samples. ACVR1B and ACVR2 were expressed in the FaDu cell line.
The expression of activin receptors in cell lines differed from that in primary tissues (Table III). Most notably, ACVR1B was expressed only in the FaDu cell line compared to full expression across almost all primary HNSCC tumors. However, this is consistent with other recent publications (23). Conversely, ACVR2B was present in all HNSCC cell lines, whereas in the primary HNSCC tissues only 1/15 tumors expressed this receptor.
Activin A and cell proliferation. Next, the effect of activin A on the proliferation of FaDu, UM-SCC-9, and Rhek cells was evaluated. Unpaired, two-tailed t-tests were employed. Initial evaluation involved a dose effect by incubating cells with varying concentrations of exogenous activin A and serum to determine the optimal concentrations for incubation. It was determined that a dose of 25 ng/ml in serum-free conditions was optimal (data not shown). In the FaDu cell line, which expresses transcripts for all activin signaling components, exposure to activin A resulted in a statistically significant increase in proliferation by Day 1 (t-test, p=0.02) which was again observed on day 3 (t-test, p=0.0046) (Figure 2). UM-SCC-9 cells demonstrated a statistically significant increase in proliferation in the first 48 h (t-test, p<0.05), which was lost by 72 h. Rhek cells demonstrated a statistically significant increase in proliferation by 48 h following activin exposure (Figure 2). These results demonstrated that activin A confers a statistically significant enhancement in growth proliferation. In several experiments, we did not observe that activin A inhibits oral cancer cell growth in vitro, as others have observed in a variety of other types of solid tumor malignancies (10, 24-27).
MTT assays showing the effect of activin A on the proliferation in Rhek, FaDu and SCC-9 cell lines. Activin at 25 ng/ml (optimal concentration) or vehicle (PBS/0.2% BSA) was added on day 0 in serum-free media. All the cell lines show a modest proliferation increase in the presence of activin A ligand. Experiments were conducted in triplicate and were repeated twice. *Significant differences at p<0.05 via unpaired, two-tailed t-test.
Despite the statistically significant changes in proliferation, the biological relevance of exogenous, extracellular activin A may be small. These in vitro findings possibly indicate that in vivo activin A-related proliferation may be related to differentially expressed activin A subunits (Table III). Furthermore, interactions with concomitantly up-regulated pathways found in in vivo tumor microenvironments may be necessary for activin expression to promote invasion in HNSCC. Notably, in another study, siRNA knockdown experiments showed more significant effects on proliferation than in our study with exogenous activin treatment (15). This suggests that the carcinogenic effects of activin A may be more dependent on intracellular processes as opposed to extracellular receptor activation.
Activin A and invasion. Next, the effect of activin A on invasion, as judged by a Matrigel invasion assay, was examined. Unpaired, two-tailed t-tests were employed for statistical analysis. Cells were incubated for 48 h and the number of cells that migrated through the membrane compared to control were examined. This experiment demonstrated activin A treatment elicited a minimally invasive response in the FaDu cell line when compared to vehicle control (t-test, p<0.05) (Figure 3). Activin A had no effect on the invasion of UM-SCC-9 cells (Figure 3). These results show exposure to activin A may lead to a more invasive phenotype in cell lines with ACVR1B expression. These findings are consistent with relevant research demonstrating association between the expression of activin A and metastatic potential (6, 7, 13, 28, 29). Interestingly, another related study found knock-down of INHBA with siRNA resulted in decreased proliferation and invasion as opposed to our results that relied on exogenous exposure to INHBA (15). This again suggests that the carcinogenic effects of activin A may be more dependent on intracellular processes as opposed to extracellular receptor activation.
Matrigel invasion assay showing increased invasion of different cell lines in response to activin A exposure. 100,000 serum-starved UM-SCC-9 and FaDu cells were plated in Boyden Invasion Chambers and incubated for 24 h in the presence of ligand activin A. Cells that invaded the matrigel and migrated to the other side of the chamber were stained and counted. The experiment was conducted in triplicate and 5 random fields were used for counting cells that migrated to the other side of the chamber. The results show that only the FaDu cell line expressing all the receptors was able to show significant difference in invasion upon exposure to activin A ligand. The UM-SCC-9 cells did show any difference in invasive capability in the presence of activin A ligand. *Significant differences at p<0.05. SCC9=UM-SCC-9, Fadu=FaDu.
TCGA INHBA analysis. Kaplan–Meier analysis of individuals with high INHBA expression (N=205) indicated by mRNA expression z-scores >2 relative to normal samples, (log RNA sequence (Seq) V2 RNA-Seq by Expectation-Maximization (RSEM)) demonstrated a decrease in overall survival compared to individuals with normal INHBA expression (N=87, p=0.0167). Tumors with log RNA Seq V2 RSEM z-scores <2 compared to normal tissue samples taken from sites adjacent to tumor samples were considered to have “normal” INHBA expression. Tumor samples with log RNA Seq V2 RSEM z-score >2 compared to normal tissue samples were considered to have “high” INHBA expression. This is because, statistically, a z-score >2 indicates expression levels significantly greater than those found in normal oropharyngeal tissue samples. Median overall survival was 52.31 vs. 76.24 months, respectively (Figure 4A). It is noteworthy that this significant difference was only noted when controlling for Caucasian race. The significance disappeared when including all racial cohorts (p=0.114, data not shown), indicating a potential race-dependent finding, which warrants further investigation. Our analysis also demonstrated stage IV tumors with increased INHBA expression (N=109) had worse outcomes compared to stage IV tumors with normal levels of expression (N=48, p<0.05, Figure 4B). This conclusion is supported by another study, which noted mutations to INHBA resulted in decreased overall survival (30, 31). Our analysis further bolstered the findings of this previous study given that it included a greater number of overall tumor samples. We further demonstrated the biological relevance since high vs. low expression was defined with respect to matched normal tissue samples as opposed to expression levels relative to other tumor samples. We also matched for age, sex, HPV status, and tumor stage, providing a more in-depth analysis of activin in the head and neck TCGA.
Kaplan–Meier analysis of head and neck squamous cell carcinoma (HNSCC) tumors with high INHBA expression vs. normal. At 5 years, survival of patients with tumors with high INHBA expression was approximately 65% as compared to approximately 50% in those with low expression INHBA tumors. Data sets were stratified based on z-score of mRNA. For all HPV (−) HNSCC tumors (panel A), the high INHBA expression group (N=205) included tumors with INHBA mRNA expression z-scores >2 relative to normal samples (log RNA Seq V2 RSEM). The normal INHBA expression group (N=87) had tumors with z-scores <2. Panel B illustrates survival outcomes of INHBA expression specifically in patients with Stage IV tumors. Analysis generated in cBioPortal (21). Groups were matched based on age (45-75), HPV status (−), sex, and race (Caucasian).
The TCGA analysis also found that INHBA expression was higher in HNSCC tumor samples in patients found to be lymph node positive (N=251) in comparison to HNSCC tumor samples in patients with no positive lymph nodes identified (N=151, p=0.024, Figure 5). This included all racial backgrounds. Primary tumor samples (N=528) exhibited a significant increase in INHBA expression compared to normal tissues (N=50, p<0.0001), as demonstrated by previous analysis (22). This occurred despite INHBA being mutated only in 1.2% of tumor samples (data not shown here).
INHBA regulation and association with lymph node status in head and neck squamous cell carcinoma. A) Promoter methylation in normal samples (N=50) vs. primary tumors (N=528, p<0.0001). B) INHBA expression in lymph node positive (N+) vs. lymph node negative (N0) patients (N=251, 151, respectively, p=0.024). For lymph node status, only HPV (−) samples were included. Panel A was generated with UALCAN (22), Panel B with cBioPortal (21).
Evaluating potential mechanisms for increased expression, unique to our analysis, we found INHBA underwent decreased promoter methylation in primary tumor samples (N=528) compared to control normal tissue (N=50, p<0.0001, Figure 5). Furthermore, there was less methylation in Stage 4 tumors compared to stage 1 (p<0.01, data not shown here). INHBA promoter methylation was decreased in stages 1, 2, 3, and 4 tissue samples compared to normal tissue (p<0.0001 for all). With respect to specific tumor subsets, TP53 mutant cells had decreased methylation compared to normal tissue (p<0.001) and TP53 nonmutant cell lines (p<0.001). Tumors with TP53 mutations had less INHBA promoter methylation compared to TP53 wild type tumors (p<0.0001) and normal tissue samples (p<0.0001). This indicates that TP53 is involved in epigenetic suppression of the activin pathway.
In an attempt to further elucidate the relationship between INHBA and HNSCC, correlational analysis comparing expression of INHBA to 20,000+ genes in the genome of every tumor sample was performed. Given the uncertainty surrounding the exact mechanism, which results in activin A being a negative predictor of mortality, the authors sought to investigate potential pathways through a targeted literature review. Thus, the 20 genes most strongly correlated with INHBA, as determined by Pearson and Spearman correlation coefficients (Table IV), were subjected to literature review to examine whether any documented interaction existed beyond just a correlational trend. Literature review across multiple fields of carcinogenesis research revealed positive-feedback mechanisms between a select number of these genes (see discussion), which prompted a Kaplan–Meier analysis grouping these genes together. Tumors over-expressing INHBA, ITGA1, ITGA5, KLF7, MICAL2, and SNAI2 (z-scores >2 with respect to matched normal tissue in all genes, N=58) were compared to tumors with normal levels of expression (z-scores <2 with respect to matched normal tissue) of those genes (N=27). Kaplan–Meier analysis revealed a more significant difference in overall survival between groups (p<0.0148, median survival 47.01 vs. 169.38 months, respectively) compared to stratifying based on INHBA alone (Figure 6). This analysis suggests neither a causative nature of these gene interactions nor their direct effects on survival. However, it does provide targeted insight for further research.
Analysis of genes most strongly correlated with INHBA expression. Analysis of the TCGA was performed in cBioportal with available mRNA data for both genes of interest (N=496) (21). Out of 20,000+ genes, the above listed were all in the top 15 most expressed and selected based on pre-existing literature suggesting their connection to INHBA or relevant Smad pathways.
Kaplan–Meier analysis of genes found to be closely correlated to INHBA expression. At 5 years, survival of patients with tumors with high clustered expression was approximately 50% as compared to approximately 75% in those with low expression INHBA tumors. Cluster High group (N=58) included tumor samples with INHBA, ITGA5, ITGA1, KLF7, MICAL2, and SNAI2 mRNA expression z-scores >2 relative to normal samples (log RNA Seq V2 RSEM). The normal cluster group (N=27) were tumors with z-scores <2 with respect to the aforementioned genes. The overall median survival for the cluster high group (47.01 months, CI=18.97-NA) was lower compared to that of the normal clustered group (169.38 months, CI=67.86-NA). Analysis provided by cBioPortal (21). Groups were matched based on age (45-75), HPV status (−), and race (Caucasian).
Discussion
Previous gene expression analysis demonstrated over-expression of the inhibin βA gene (activin A) as part of a 72-gene-signature associated with recurrent HNSCC (4). Concordant with this data, activin A has been shown to be associated with high risk for HNSCC (5, 32). Several studies have been conducted to explore the expression of activin and its subunits in different cancers including endometrial cancer, breast cancer, and prostate cancer (33-35). To date, there is incomplete data on the role of activin A in HNSCC progression. Therefore, the aims of this investigation were to preliminarily investigate the expression of activin A and its subunits in normal, premalignant, and malignant head and neck human specimens. This analysis also included cell line and TCGA multi-omic analysis.
First, our group sought to further define the expression of subunits of the activin A axis across the carcinogenic process. This was accomplished using IHC staining to evaluate the expression of signaling proteins associated with activin A in HNSCC progression in five healthy controls, 15 OPL patients, and 12 HNSCC patients. In our analysis, expression of ACVR1B and INHBA were significantly increased throughout the carcinogenic process from normal mucosa to premalignant lesions, to malignant HNSCC in vivo (Table III). This is significant as ACVR1B is a transmembrane receptor with intracellular serine/threonine kinase activity, which activates Smad 2/3 (Figure 1). This analysis further unveiled the lack of a regulatory response from the inhibitory ligand inhibin α (Table III) may play an ancillary role in this process. The increase in the activating ligand (INHBA) during carcinogenic progression without a compensatory increase in the inhibitory ligand (INHA) points towards another dysregulated component, which stimulates the activin A pathway (Figure 1). This discordance in expression was most pronounced during the progression from normal to leukoplakia samples (Table III); a novel finding indicating that activin A may be more critical earlier in carcinogenesis.
With respect to the over-expression of the activin A axis, our study found decreased promoter methylation in tumor samples compared to normal matched tissue samples (Figure 5). Increased INHBA expression was also observed in the TCGA HNSCC tumors compared to expression levels found in normal oropharyngeal tissue obtained from sites adjacent to primary HNSCC tumors. This provided further confirmation to our other observations. Consequently, aberrant expression of ACVR1B, lack of INHA production, and decreased promoter methylation may contribute to the dysregulation and over-expression of activin A in head and neck cancer. A similar analysis found decreased methylation correlated with increased INHBA expression; this study noted a potential miRNA that may lead to increased INHBA expression through interference with an INHBA promoter methylator (31).
Next, we further investigated the role of activin A in invasion of HNSCC. To determine treatment with activin A enhances HNSCC tumor aggressiveness in vitro, its effects on the proliferation and invasion of HNSCC cell lines were tested. Interestingly, only small increases in proliferation were sporadically observed following treatment of cell lines with activin (Figure 2). In in vitro cell line experiments, critical influences, like those of the microenvironment, are not present and this may be one explanation of the observed effects. For example, our study revealed differential expression of the components of the activin A axis in HNSCC tumor samples (Table III) vs. in vitro cell lines.
While treatment of the cell lines with activin A did not elicit significant invasive response, our study provided findings which further support emerging research. One such example was the stepwise increase of ACVR1B expression from normal, to leukoplakia, to HNSCC tumor samples (Table III). ACVR1B was consistently expressed in all 12 HSNCC tumor samples. ACVR1B activates the Smad complex, which has been shown to induce the expression of EMT genes like SNAI1/2 and ZEB1 in breast cancer (36). Our TCGA analysis strengthens the concept of activin A being involved in the metastatic process as patients with lymph node involvement (N+) demonstrate greater INHBA expression compared to those with no lymph node involvement (N0, Figure 5).
The final objective of our study was to perform multi-omic analysis of TCGA to evaluate the effect activin A expression on metastasis and mortality. As previously stated, our study demonstrated HPV(−) N+ tumors had higher levels of INHBA expression compared to N0 tumors (Figure 5). When controlling for tumor type, age, and sex, high levels of INHBA mRNA expression resulted in decreased overall survival (median months overall 52.31 vs. 76.24, Figure 4). Our analysis revealed this disparity in survival also persisted when controlling for stage IV tumors (Figure 4B). Our study uniquely highlights the biological relevance of these findings as we defined high vs. low expression in relation to normal oropharyngeal tissue samples from regions adjacent to primary HNSCC tumors. Tumors with statistically greater INHBA expression levels than normal surrounding tissue (log RNA Seq V2 RSEM z-scores >2 compared to normal tissue) resulted in decreased overall survival.
The next part of the TCGA investigation focused on the identification of pathways potentially relevant to the role of activin A in HNSCC metastasis and mortality. Of 20,000+ genes analyzed, the top 20 genes most strongly correlated with INHBA were subjected to literature review to determine if interactions with the activin A pathway had been established in other fields of carcinogenic research. The authors chose this approach for the purposes of identifying targeted directions of future activin A pathway research; a pathway whose role specifically in HNSCC is incompletely understood at this time. Of these, ITGA5, ITGA1, KLF7, MICAL2, THBS1, and SNAI2 were identified.
Of these genes strongly correlated with activin A expression, ITGA-5 is associated with migration, adhesion and metastasis in monocytes through Smad signaling (Figure 1) (37-39). Knockdown of ITGA5 in pancreatic cells abrogates TGF-β/Smad2 signaling (40). Another integrin protein, ITGA1, mediates metastases in pancreas and lung cancer (41, 42). Interestingly, activin A promotes trophoblast invasion/implantation by upregulating ITGA1 through ACVR1B/Smad activation (Figure 1) (43). These findings are relevant to our in vitro molecular analysis which demonstrated an increase in ACVR1B expression in HNSCC tumors (Table III). Other analyses of KLF7, MICAL2, and THBS1 found that these genes contribute to proliferation and EMT via stabilization and upregulation of TGF-β-Smad pathways (Figure 1) (44-48). These pathways are known to be shared by INHBA (Figure 1) (49).
We then examined the potential clinical relevance of these documented biological interactions (Figure 1). HNSCC tumors demonstrating a coordinated increased expression of INHBA, ITGA1/5, MICAL2, SNAI2, KLF7 mRNA (z-score >2 compared to matched normal tissue for all genes) had shorter overall survival (47.01 months vs. 169.38 months, Figure 6). For tumors over-expressing all of the aforementioned genes, the average survival time (47.01 months) was decreased compared to tumors with increased levels of just INHBA (52.13 months).
Greater 5-year survival was found in tumor cohorts with normal levels of expression (z-score <2) of all pathways (approximately 75%) compared to tumor cohorts with normal levels of expression of just INHBA (approximately 63% survival, Figure 4). This is noteworthy, as there have been unsuccessful clinical trials with therapeutics targeting the activin receptor (50); a treatment which targets concomitantly up-regulated pathways may be beneficial. Interestingly, our treatment with exogenous activin did not yield the same invasive and proliferative effects as another study which interfered with the activin pathway through siRNA (15). These findings indicate intracellular activin as a more promising therapeutic target, potentially targeting the interactions of these pathways.
Our analyses provide a more complete overview of activin A subunit expression during the progression from normal tissue, to OPL, to HNSCC. The identified dysregulation of this pathway was found to be associated with significantly increased INHBA expression, ACVR1B expression, decreased INHA expression, and decreased promoter methylation. Numerous pathways and proteins recently discovered to be associated with INHBA, including the Smad transcription complex, providing targeted insight for future investigations.
Conclusion
This study provides further detailed insight into the role of INHBA in HNSCC progression. Notably, our subunit immunohistochemistry analysis demonstrated activating components of the pathway, like ACVR1B, are upregulated early in the carcinogenic process, seen in premalignant leukoplakia lesions. Additional findings of pathway dysregulation included a lack of inhibitory INHA ligand upregulation and decreased promoter methylation. Treatment with exogenous activin did not elicit significant sustained invasive or proliferative responses. In contrast, other studies showed abrogation of intracellular activin signaling down-regulated proliferation and invasion. The constellation of these findings demonstrates the role of INHBA in carcinogenesis may be more facilitative in nature. This suggests optimal treatment targets may reside intracellularly in pathways with cross-talk with INHBA, as opposed to blocking extracellular activin A-mediated activation, which was unsuccessful in a previous clinical trial. In response to this, our TCGA analysis identified a strong correlation of INHBA with multiple intracellular pathways and proteins in the extracellular matrix known to be involved in proliferation, invasion, and metastasis. Synergism between these pathways and INHBA have recently been demonstrated in other fields of research, providing potentially promising interactions to target and attenuate. This provides targeted insight for further research investigating the complicated, incompletely understood role of INHBA in HNSCC.
Acknowledgements
The Authors would like to thank Danielle Howard, MS, for technical assistance and Dr. Nelson Rhodus, DMD, MPH, for critical reading the manuscript and supervising the thesis of Ketan Patel.
Footnotes
Authors’ Contributions
SB contributed to study design and investigation, manuscript development, and statistical analysis. KP contributed to performing experiments. BW contributed to performing experiments, study design, and reviewing the manuscript. PG contributed to study design and reviewing the manuscript. FO contributed to study design, statistical analysis, manuscript development, and reviewing the manuscript.
Conflicts of Interest
The Authors report no conflicts of interest in relation to this study.
Funding
American Cancer Society Institutional Research Grant IRG-58-001-40IRG44. Lion’s 5M Hearing Center Grant (Minnesota) and P30 CA77598-07 (NCI/NIH).
- Received October 11, 2023.
- Revision received November 7, 2023.
- Accepted November 13, 2023.
- Copyright © 2023 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.