Introduction
Adenoid cystic carcinoma (ACC) is an uncommon malignancy. Primary lesions usually arise from salivary glands, more frequently from minor than major ones. Other localizations include paranasal sinuses, larynx, lacrimal glands, bronchi, mammary and skin (Coca-Pelaz et al.
2015; Dillon et al.
2016). This type of cancer most frequently occurs between the fourth and the sixth decades of life. ACC is rather indolent disease with slow growth and late development of distant metastases, but it may also be fast-growing malignancy with early distant spread. When diagnosed at an early stage, it usually appears as a single and oval lesion (usually with smaller than 6 cm diameter) well defined and partially capsulated with a pronounced tendency to aggressively infiltrate surrounding tissues along nervous tissue structures. Metastases to regional lymph nodes and distant sites occur, and ACC recurrences may appear many years after initial treatment (van der Wal et al.
2002; Kokemueller et al.
2004) although the lymph node metastases are uncommon (Ferlito
2016; He et al.
2017). Treatment of these type of cancer remains limited to surgery with or without adjuvant radiotherapy (proton beam) (Gentile et al.
2017). Moreover, no systemic treatment has been found to be effective (Laurie et al.
2011; Xu et al.
2018a). Due to peculiar biology of this malignancy, various attempts have been made including the use of metronomic chemotherapy in patients with advanced or disseminated disease (Visa et al.
2010; Caballero et al.
2013; Cassidy et al.
2018).
The ACC mutational landscape was determined and revealed a low exonic somatic mutation rate and wide mutational diversity. Among others, mutations in genes involved in histone acetylation/deacetylation (
ARID4B, ARID5B, BRD1, FTSDJ1, MLL3), as well as encoding histones (
HIST1H2AL, HIST1H1E), were found suggesting that chromatin dysregulation may promote ACC development (Ho et al.
2013). Additionally, in ACC, mutations in
SMARCA2, a gene encoding the BRM ATPase of the SWI/SNF chromatin remodeling complex (CRC), were found. Moreover, 2% of ACC cases are featured by mutations in
ARID1A and
SMARCE1 genes, coding for BAF250a and BAF57 non-core SWI/SNF CRC subunits, respectively, consistently suggesting aberrant SWI/SNF CRC function in ACC (Ho et al.
2013).
In this study, we attempted to better understand the molecular features of adenoid cystic carcinoma. The re-analysis of available transcriptomic data followed by gene ontology (GO) classification indicated that ACC is featured by the upregulation of steroid hormone receptor activity, carbohydrate metabolism, cell proliferation and drug transport-related processes. By contrast, the downregulated processes were mostly related in ACC to the control of chromatin structure and function including chromatin remodeling and organization, histone deacetylase and acetyltransferase complex, telomere maintenance and methyltransferase activity. Among downregulated genes, we found genes encoding nearly all subunits of SWI/SNF chromatin remodeling complex. The subsequent immunohistochemistry (IHC) staining showed alterations in SWI/SNF CRC subunits’ abundance and ectopic expression of androgen receptor in ACC compared to margin tissue. Among BRM-target genes overexpressed in ACC, a class of genes involved in chemoresistance was identified. Taking together, our findings suggest an important role of SWI/SNF CRC impairment in the adenoid cystic carcinoma development and provided a likely explanation of the ACC resistance to various specific chemotherapy treatments.
Materials and methods
Patients’ characteristics
A group of 50 patients from, Marie Sklodowska-Curie Memorial Cancer Center and Institute of Oncology in Warsaw and Gliwice with diagnosed adenoid cystic carcinoma was subjected to this study. Primary lesions were most often localized in submandibular and parotid glands (Table
1). All 50 patients underwent primary surgical treatment. The patients represent whole area of Poland.
Table 1
List of the most important patient characteristics in a group consisting of 50 ACC patients
Gender |
Female | 25 (50%) |
Male | 25 (50%) |
Age |
Mean age | 53 (min. 22, max. 90) |
< 60 | 33 (66%) |
≥ 60 | 17 (34%) |
Localization |
Minor salivary gland | 31 (62%) |
Major salivary gland | 19 (38%) |
pT |
pT1 | 6 (12%) |
pT2 | 19 (38%) |
pT3 | 3 (6%) |
pT4 | 11 (22%) |
n/a | 11 (22%) |
Treatment |
Radiotherapy | 45 (90%) |
Chemotherapy | 6 (12%) |
Progression |
Total number | 14 (28%) |
Female | 5 (36%) |
Male | 9 (64%) |
Microarray re-analysis
Microarray datasets were obtained from Gene Expression Omnibus (GEO) database with the accession number GSE28996 for ACC samples and GSE36820 for normal salivary gland (Moskaluk et al.
2011). The nine ACC and three healthy salivary gland corresponding sample data files were re-analyzed using GeneSpring GX (Agilent) software according to advanced guided workflow.
Immunohistochemistry
Immunohistochemical staining was performed on 4-µm FFPE tumor sections using the EnVision FLEX+, Mouse, High-pH Detection System (Dako, Glostrup, Denmark). Sections were deparaffinized with xylene and rehydrated in ethanol solutions. Heat-induced epitope retrieval was carried out in Target Retrieval Solution (Dako) for 20 min at 96 °C. After cooling, the slides were treated for 5 min with an endogenous peroxidase blocker (Dako), incubated with monoclonal antibody against androgen receptor (D6F11, Cell Signaling), BRG1, INI1, BAF155, BAF170, BAF250a and BRM (P680, D9C2, D7F8S, D8O9V, D2A8U, D9E8B, Cell Signaling) for 30 min at room temperature, and then labeled with the EnVision FLEX+, Mouse, High-pH Detection System (Dako). The color reaction product was developed with 3,3′-diaminobenzidine tetrahydrochloride (Dako) as a substrate, and nuclear contrast was achieved with hematoxylin counterstaining.
Transcript level measurements
RNA was extracted from paraffin-embedded samples using RecoverAll™ Total Nucleic Acid Isolation Kit for FFPE (Invitrogen). Contamination with genomic DNA was removed using DNAse I treatment according to the manufacturer’s protocol. Aboult 300 ng of total RNA was used for the first-strand cDNA synthesis using Transcriptor First Strand Synthesis kit (ROCHE) and oligo-dT primer and random hexamer mix. Transcript levels of studied genes were measured in triplicates using Applied Biosystems 7500 Fast device with specific primers and SybrGreen. For SMARCA2, following primers were used hBRMq2F: TCCGAGGCAAAATCAGTCAAG and hBRMq2R: TTCCTCGATTTGGCCTTTTCT. As the reference primers UBI1: ATTTGGGTCGCGGTTCTTG and UBI2: TGCCTTGACATTCTCGATGGT specific for UBI (UBIQUITIN) gene were used. For the relative expression level calculation, the 2−ΔΔCt method was used.
Statistical methods
For statistical analysis, MedCalc software and GraphPad Prism 5.0 with Shapiro–Wilk normality test, Mann–Whitney rank test for independent sample and paired t test were used. p value < 0.05 was considered to be statistically significant.
Discussion
Surgery with postoperative radiotherapy remains the most effective therapeutic modality in patients with early-stage head and neck adenoid cystic carcinoma (Coca-Pelaz et al.
2015). In the case when surgery is impossible, the radiotherapy has been used and some attempts with chemotherapy including metronomic therapy; however, the outcome of such treatments is not satisfactory (Visa et al.
2010; Caballero et al.
2013; Ko et al.
2016; Cassidy et al.
2018). As there is no efficient therapy for the advanced stage of the adenoid cystic carcinoma there was an urgent need to investigate molecular basis of this type of cancer. Therefore, we performed the transcript profiling analysis of nine ACC samples and three normal salivary glands from publicly available datasets. We found that more than 8600 genes were mis-regulated in ACC samples compared to normal salivary gland tissue. 4044 of them were upregulated and 4603 downregulated. Interestingly, the genes with downregulated expression in ACC samples were related to chromatin or chromatin involving processes such as chromatin remodeling and organization, histone modifications, telomere maintenance as well as RNA splicing. This observation is in line with previous findings demonstrating that ACC is featured by mutations in genes coding for regulators of chromatin state including
SMARCA2, a gene encoding BRM, the central ATPase of SWI/SNF ATP-dependent chromatin remodeling complex (Ho et al.
2013). The closer inspection of SWI/SNF CRCs revealed the mis-regulated expression of genes encoding its main subunits. Interestingly, almost all genes coding for core SWI/SNF subunits (
SMARCA4, SMARCB1, SMARCC1, SMARCC2) as well as
ARID1 and
PBRM1 genes coding for auxiliary SWI/SNF subunits were downregulated. Surprisingly, the expression of
SMARCA2 gene encoding BRM ATPase was more than threefold upregulated.
As the SWI/SNF complex is composed of several subunits with defined stoichiometry, the protein level analysis of SWI/SNF subunits in ACC samples was performed. The IHC analysis confirmed the general overexpression of BRM protein in the most of the ACC cells compared to normal salivary gland tissues. In our study, we observed sporadically only some BRM-unstained cells but not BRM-unstained regions of ACC suggesting that the BRM protein accumulation is a general feature of ACC. The other SWI/SNF core complex subunits, namely INI1, BAF155 and BAF170 were also overexpressed in ACC at the protein level; however, the transcript levels of genes encoding these proteins were significantly downregulated in ACC cancer samples. These findings strongly suggest that core SWI/SNF complex subunits are precisely regulated at the protein level. Our findings suggest that in the case of adenoid cystic carcinoma the balance between classes of SWI/SNF complexes is shifted to the increased amount of BRM-containing rather than BRG1-containing complexes. Therefore, the proper control of gene expression may be affected in ACC due to the alterations in the balance between various classes of SWI/SNF complexes. This could lead to the aberrant expression of various genes involved in cell cycle control, proliferation and cell adhesion as these genes were found as targeted directly by BRM (Fig.
7). Additionally, the
SMARCA2 gene (coding for BRM) was found to be mutated in 5% of ACC samples. Interestingly, all mutations were located in the helicase domain of BRM protein (Ho et al.
2013) suggesting functional impairment of the mutated protein in ACC.
The SWI/SNF complex is involved in steroid hormone action and directly interacts with various steroid hormone receptors including androgen receptor (Sarnowska et al.
2016). The androgen receptor pathway was found as one of the most deregulated, with wide spectrum of AR transcript level in ACC patient samples. There was around ninefold difference between the highest and lowest AR transcript level in patient samples. It has been shown that AR overexpression is most frequently associated with salivary duct carcinomas. In several studies, the AR positivity was shown in 64–77% of all salivary duct carcinoma and 5% of ACC cases (Dalin et al.
2017). The other study showed AR abundance in about 6.6% cases of salivary gland adenoid cystic carcinoma (Ito et al.
2009). A number of patients with AR-positive salivary gland carcinoma have been treated with anti-androgen therapy. The treatment was well tolerated, but relapse was commonly observed. A 3-year progression-free survival was 12%, and a 5-year overall survival of 19% (Dalin et al.
2017).
Current clinical data indicate that the stage of ACC is the most significant factor influencing treatment outcome, whereas histological grade has no prognostic impact. Additionally, negative impact of vascular and neural infiltration on survival in male patients was shown (Spiro
1997). The disease is slightly more common in women (van der Wal et al.
2002); however, in our study, the number of women and men was equal. No factor has yet been shown to be responsible for the increased incidence of adenoid cystic carcinoma in female patients; however, its occurrence may be connected to hormones, especially that estrogen enhances the malignancy in ACC cells (Sumida et al.
2016).
In our study, we detected AR nuclear immunoreactivity in some examined cases. Noteworthy, the staining was observed in particular cancer cells only. Moreover, the statistically significant negative correlation between AR and
SMARCA2(BRM) and
SMARCB1(INI1) expression was found, suggesting transcriptional co-regulation between these genes, although further examination of the role of this interdependence needs to be performed. Thus, results presented here confirmed the hypothesis that ACC belongs to the most heterogeneous types of cancer and likely, therefore, the development of commonly used targeted therapy against this type of the disease is very difficult. Here we showed that the BRM protein is accumulating in nearly all ACC cells, therefore, representing the most homogeneous feature of ACC which has been so far discovered. Moreover, large clusters of both ACC down- and upregulated genes represent direct BRM targets. Among common ACC upregulated and BRM target genes, there were identified gene classes responsible for proliferation and chemoresistance processes, the two features strictly linked to the ACC aggressiveness. These findings are in line with the recently published association of BRM overexpression with the chemoresistance in ovarian cancer (Xu et al.
2018b). Given the observation that neither the regular chemotherapy nor metronomic one is sufficient for effective treatment of salivary gland ACC patients, the findings presented here may help to understand the molecular basis of the chemotherapy resistance in ACC cells and serve as solid basis for further investigations aiming to establish new, more effective forms of therapy.