A meta-analysis was used to analyze the mRNA expression of the ASCL family in clinical cancer specimens by following the PRISMA guidelines (17,18) Oncomine (www.oncomine.org), a web-based microarray database, was used to analyze the mRNA expression of ASCL in clinical cancer tissue (19). According to ‘Oncomine Platform Overview Q1 2014,’ the database resource of Oncomine includes upwards of 700 independent datasets with an estimated 90,000 microarray trials. Oncomine has standardized and organized the datasets of public cancer microarray data into different cancer type and subtypes (16,20).

ASCL gene (ASCL1-5) expression in 20 cancer types was investigated. Detailed information on ASCL genes, such as tissue of origin, comparing mRNA expression with matched normal tissue types, was displayed in groups. The gene summary view in Oncomine was presented throughout the analysis with an alteration in color, reflecting the degree of expression. The expression coloration represents a gene with highest ranking in a particular type of cancer based on the threshold analysis (Fig. 1).

The cancer vs. normal filter that only displayed datasets investigating ASCL gene mRNA expression in the same tissue of origin was selected. To be included in the study, all the data had to satisfy the threshold with a P-value of <0.01, a fold change of >1.5 and a gene rank percentile of <10%. Statistical analyses were performed using the Oncomine default algorithms, such as P-values, two-tailed Student's t-test, and multiple testing corrections.

Several studies have identified potential roles for ASCL family members in cancer development; however, our overall understanding of ASCL family member function during tumor initiation and progression is incomplete. To investigate a potential alteration in ASCL family gene expression in different types of cancer, we accessed the web-based Oncomine microarray database to analyze 20 different types of cancer. Cancer tissue was compared with normal tissue (control) and thresholds were set to screen suitable datasets from the Oncomine database. To include suitable datasets for further analysis, the gene expression in cancer cells compared with that of normal tissue had to fulfill the following threshold criteria: Fold change >1.5, P<0.01 and gene rank percentile <10% (Fig. 1). Our analysis demonstrated alterations in ASCL gene family expression in multiple cancer types, which may provide useful information for future studies investigating the role of ASCL genes in tumorigenesis.

The proneural transcriptional factor ASCL1/MASH1 is essential for proper nervous system development (21). In the cerebrum, ASCL1 controls the primitive as well as the late phases of neurogenesis, with the division of radial glia progenitors and the radial migration of post-mitotic neurons (22,23). ASCL1 controls the expression of numerous target genes that are involved in cell cycle progression and cytoskeletal reorganization associated with neuronal cell migration (6,7). Recently, a potential oncogenic role for ASCL1 in lung cancer has been reported (23); however, the role of ASCL1 in cancer remains unclear. Our analysis indicated that ASCL1 is significantly overexpressed in the majority of cancer types, such as cancers of the brain, lung, head and neck, prostate, pancreas, kidney, esophagus, leukemia, lymphoma and sarcoma (Fig. 1). ASCL1 also ranked in the top 1% of overexpressed genes in leukemia, brain and lung cancer. Importantly, our analysis indicated that ASCL1 is overexpressed in the majority of brain cancers, such as glioblastoma, oligodendroglioma, anaplastic astrocytoma, anaplastic oligodendroglioma and anaplastic oligoastrocytoma. Compared with normal tissue, ASCL1 exhibited a higher expression in brain tumor tissues (P-values of 0.004–2.50E-21), and the ASCL1 gene ranked 1–7% in our meta-analysis results (Table II). We found that, in various lung cancers, such as small-cell lung carcinoma and carcinoid tumors, ASCL1 was significantly overexpressed (P-values of 0.002–3.53E-13) and the gene ranked in the top 1–8% relative to the control. In addition to brain tumors, ASCL1 was also highly expressed in acute adult T-cell leukemia/lymphoma, with the gene ranking in the top 1%, a 3.76-fold change, and a P=3.43E-5. These data indicated that ASCL1 expression varied in different types of leukemia. We also observed that ASCL1 was highly expressed in leiomyosarcoma (3.55-fold change, P=6.13E-4 and gene ranking in the top 7%), prostate carcinoma (3.21-fold change, P=0.001 and gene ranking in the top 1%), pancreatic adenocarcinoma (3.22-fold change, P=0.002 and gene ranking in the top 6%), renal oncocytoma (5.16-fold change, P=0.002 and gene ranking in the top 8%) and Barrett's esophagus (1.95-fold change, P=0.002 and gene ranking in the top 9%) (Table II).

In contrast to brain cancer and lymphomas, other cancers exhibited a reduction in ASCL1 expression. Gastric cancer and melanoma were among the top 1% of tumors that exhibited ASCL downregulation. The reduction in the ASCL1 transcript level suggested a tumor suppressor role, since tumor suppressor genes tend to exhibit a low or reduced expression in tumor tissue compared with normal tissue. Our analysis indicated lower ASCL1 expression in gastric, bladder and lung cancers. Evidence of this trend is also supported by a previous study that specifically evaluated a tumor suppressor gene in breast cancer datasets from the Oncomine database, which revealed a significant downregulation and low expression of the tumor suppressor gene ADAMTS1 in breast carcinomas when compared with normal tissue (24). Another similar study on the SIRT3 tumor suppressor gene also revealed lower expression in various tumor types (25). Given the pattern of downregulation, we hypothesized that ASCL1 may also play a tumor suppressor role in a subset of tissues. ASCL1 expression was considerably downregulated in lymphoma (diffuse large B-cell lymphoma, primary effusion lymphoma and mantle cell lymphoma). The reduction in ASCL1 expression ranged from −1.61- to −2.08-fold change, with P-values of 9.25E-4-1.15E-6 and the gene ranking in the top 3–10%. Our bioinformatics analyses of gastric cancer revealed that ASCL1 exhibited a lower expression in the majority of gastric cancer subtypes, namely gastric mixed adenocarcinoma, gastrointestinal stromal tumors and gastric intestinal-type adenocarcinoma. The ASCL1 expression ranged from −1.60- to −3.64-fold downregulation, with P-values of 0.002–1.06E-6 and the gene ranking in the top 1–8%. ASCL1 exhibited a lower expression in most types of melanoma, namely cutaneous melanoma, non-neoplastic nevus, benign melanocytic skin nevus and monoclonal gammopathy of undetermined significance. The ASCL1 transcript expression ranged from −1.64- to −3.22-fold downregulation, with P-values of 0.003–2.27E-5 and the gene ranking in the top 1–10%. ASCL1 also exhibited a lower expression in bladder cancer (−1.77-fold change, P=2.36E-6 and gene ranking in the top 6%), invasive ductal breast carcinoma (−1.59-fold change, P=5.03E-6 and gene ranking in the top 3%) and colon cancer (−3.58-fold change, P=9.57E-6 and gene ranking in the top 7%) (Table II). These analyses suggest that the effect of ASCL1 downregulation on transcript expression may be an equally important alteration as increased expression in cancer biology. Interestingly, ASCL1 was found to be both up- and downregulated in brain tumors compared with normal tissue. The conflicting expression profiles of ASCL1 in the same type of cancer may be due to the wide-ranging categories for each of the cancer subtypes (Table II). This discrepancy may be a sample size issue arising from the original publications' reported data including a low number of samples from these tumor types. Collectively, our data suggest that alterations in ASCL1 expression may adversely affect tissue homeostasis, which may result in tumorigenesis.

ASCL2 is a basic helix-loop-helix transcription factor that is expressed in neuronal precursors (26). ASCL2 is a target of the Wnt signaling pathway and previous studies indicated that ASCL2 may regulate LGR5 in intestinal stem cells in response to Wnt signaling (9,27). Moreover, ASCL2 is involved in T-helper cell (TH) 1 and TH17 differentiation (28).

ASCL2 is strongly expressed in colon cancer tissues and cell lines (HT-29 and LS174T cells). Selective blockade of ASCL2 disrupts tumor cell proliferation and migration in tumor xenograft models (10,29,30), a result consistent with our bioinformatics analysis (Fig. 1). This is particularly true in colon cancer tissues compared with normal tissues; however, whether ASCL2 plays a role in initiation and progression of other tumor types remains unclear.

ASCL2 expression was altered in 8 of 21 investigated cancers and was commonly observed in colorectal, gastric, breast, ovarian, testicular, lung and head and neck cancers, as well as lymphoma (Fig. 1). However, based on our bioinformatics analysis, our results were strikingly different. Downregulation of ASCL2 was only observed in the top 5% and 9% of underexpressed genes in melanoma, and brain and gastric cancers, respectively (Fig. 1).

Our analysis revealed that ASCL2 expression is significantly upregulated in various breast cancer subtypes, such as invasive ductal, invasive lobular and medullary breast carcinoma, with P-values of 0.009–4.39E-72, gene ranking 2–10% and a fold change of 1.66–14.9 compared with normal tissues (Table III). In colorectal tumors, such as adenocarcinoma of the colon, rectum, cecum or rectosigmoid region, colonic adenoma, rectal adenoma, colon adenoma epithelia and colon carcinoma epithelia, ASCL2 also exhibited significant upregulation compared with normal tissues, with P-values of 3.60E-7-8.24E-52, gene ranking 1–9% and a fold change of 5.64–31.35 (Table III).

In gastric cancers, such as diffuse gastric adenocarcinoma, gastric intestinal-type adenocarcinoma and gastric mixed adenocarcinoma, ASCL2 exhibited significant upregulation compared with normal tissues, with P-values of 6.30E-4-1.74E-6, gene ranking 1–5% and a fold change of 2.35–4.45. Additionally, we found that ASCL2 is highly expressed in squamous cell lung carcinoma (1.84-fold change, P=7.81E-8 and gene ranking in the top 9%), nodular lymphocyte-predominant Hodgkin's lymphoma (2.65-fold change, P=7.61E-5 and gene ranking in the top 5%), nasopharyngeal carcinoma (1.56-fold change, P=5.08E-4 and gene ranking in the top 10%), ovarian endometrioid adenocarcinoma (1.76-fold change, P=0.001 and gene ranking in the top 3%); testicular seminoma (3.21-fold change, P=0.007 and gene ranking in the top 6%) (Table III).

Of note, lower ASCL2 gene expression levels were found in certain cancer subtypes, such as brain and gastric cancer, and melanoma. These subtypes included oligodendroglioma (−1.73-fold change, P=4.42E-4 and gene ranking in the top 9%), gastrointestinal stromal tumors (−2.59-fold change, P=4.59E-4 and gene ranking in the top 9%), and cutaneous melanoma (−6.74-fold change, P=6.22E-4 and gene ranking in the top 5%) (Table III). Thus, ASCL2 exhibited increased mRNA expression in some cancer tissues and decreased expression in others. Overall, our analysis indicated that ASCL2 was ranked in the top 10% of genes involved in the regulation of breast, colorectal, lung, gastric, head-neck, ovarian and testicular cancers and lymphoma, whereas in brain cancer and melanoma it exhibited significant downregulation compared with normal tissue (Table III). These findings indicate that cell context-specific alterations in ASCL2 expression may play a critical role in cancer biology.

ASCL3 (Sgn1) belongs to the MASH gene family of transcription factors that has been associated with cell fate determination and contributes to the maintenance of the adult salivary gland homeostasis (11,31). Our database analysis indicated that ASCL3 was highly expressed in invasive ductal breast carcinoma (2.26-fold change, P=0.002 and gene ranking in the top 2%) compared with normal tissues (Table IV). Of the 21 analyzed tumor types, 5 exhibited a correlation with downregulation of ASCL3 (Table IV).

Analysis of various renal tumor subtypes indicated that ASCL3 exhibited a lower expression in renal oncocytoma with a fold change of −1.60, P=5.15E-16 and gene ranking in the top 3%. ASCL3 expression is downregulated in cervical cancer with a fold change of −1.65, P=1.24E-9 and gene ranking in the top 1%. In superficial bladder cancer, we found that ASCL3 also exhibited a lower expression, with a fold change of −1.63, P=1.16E-7 and the gene ranking in the top 3%. In anaplastic large-cell lymphoma, ASCL3 exhibited lower expression, with a fold change of −1.96, P=1.58E-5 and the gene ranking in the top 8%. Melanomas and basal cell skin carcinoma (also referred to as basalioma, the most common malignant skin tumor), exhibited ASCL3 downregulation with a fold change of −1.80, P=0.007 and the gene ranking in the top 10% (Table IV). Thus, ASCL3 ranked in the top 2% of genes exhibiting upregulation in breast cancer, while in renal, cervical and bladder cancer, lymphoma and melanoma, ASCL3 displayed significant downregulation compared with normal tissues (Table IV). These findings indicated that ASCL3 may be differentially expressed in specific types of cancer and that further investigation is required to determine the mechanisms underlying the involvement of ASCL3 in tumorigenesis.

ASCL4 (HASH4, bHLHa44) expression is associated with skin development. ASCL4 exhibited a 7-fold higher expression in fetal skin compared with adult skin (12). The role of ASCL4 in cellular function remains elusive. Therefore, comparative genomic sequencing did not reveal any function for this gene (32). ASCL4 expression did not satisfy the selection criteria of the present study; therefore, it was not selected for further investigation.

We were unable to obtain any data regarding ASCL5 based on the literature search through the PubMed database. Analysis of the Gene Ontology database indicated that ASCL5 may be involved in the regulation of DNA-templated transcription (33). Our bioinformatics analysis suggested that ASCL5 was upregulated in lung cancer with a fold change of 1.96–3.71, P-values of 0.002–0.003 and the gene ranking 5–9%. However, ASCL5 was downregulated in the majority of types of brain tumors, such as glioblastoma, anaplastic oligoastrocytoma, anaplastic oligodendroglioma and oligodendroglioma, with a fold change of −3.10 to −5.50, P-values of 0.002–5.25E-12 and the gene ranking 1–6% (Table V). To the best of our knowledge, our bioinformatics analysis is the first report to provide any information regarding the potential role of ASCL5 in tumorigenesis.

In this report, we presented an in silico analysis of the ASCL gene family and investigated the potential involvement of ASCL genes in various cancers. Our meta-analysis approach provided a conspectus of the clinical data related to the ASCL gene family and suggested that alterations in the ASCL genes may result in development of various tumors. Moreover, our analysis utilized the integration and validation of numerous microarray datasets, thereby allowing the use of an ASCL gene with its correlated cancers and its subtypes as future biomarkers for future cancer studies.

It was previously indicated that ASCL1 functions as an oncogene in lung cancer (23). Recent findings also demonstrated that ASCL1 is a marker for small-cell lung carcinomas (23,34). These data are consistent with our bioinformatics analyses (Fig. 1). ASCL1 exhibited significant overexpression in half of the analyzed cancer types (10 of 20 cancers), with the gene ranking in the top 10%. Moreover, a significant number of tumors exhibited ASCL1 downregulation, with the gene ranking in the top 1% (Fig. 1). ASCL1 was in the top 1% ranking of all overexpressed genes in leukemia, brain and lung cancers. Interestingly, gastric cancers and melanoma displayed downregulation of ASCL1, with the gene ranking in the top 1% of all downregulated genes.

Previous studies have suggested that ASCL2 is strongly expressed in colon cancer tissues and cell lines (HT-29 and LS174T cells) and that selective blockade of ASCL2 results in the inhibition of xenograft tumor growth, proliferation, invasion and migration (10,29,30). ASCL2 may promote colorectal (30), lung (35) and gastric cancer (36), suggesting a crucial role for ASCL2 involvement in tumor development. These data are consistent with our bioinformatics analysis (Fig. 1). Strikingly, ASCL2 expression analysis indicated increased mRNA expression in some cancer tissues and decreased expression in others. ASCL2 is in the top 10% of genes exhibiting overexpression in breast, colorectal, lung, gastric, head-neck, ovarian and testicular cancers, as well as lymphoma. However, brain tumor and melanoma subtypes exhibited significant reductions in the expression of ASCL2 when compared with normal tissues (Table III). Of note, ASCL3 expression displayed a wide range of mRNA levels in various cancers. ASCL3 was in the top 2% of overexpressed genes in breast cancer. Conversely, in lymphomas, melanomas, renal, cervical and bladder cancers, ASCL3 expression was significantly reduced compared with that in normal tissues (Table IV). Expression analysis of ASCL5 suggested a correlation between elevated ASCL5 expression and lung cancer development. ASCL5 was one of highly expressed genes, ranking 5–9% in lung cancer. Interestingly, ASCL5 was downregulated in most types of brain tumors, such as glioblastoma, anaplastic oligoastrocytoma and anaplastic oligodendroglioma. The decrease in fold change ranged from −3.10 to −5.50, the P-values ranged from 0.002 to 5.25E-12, with the gene ranking 1–6% (Table V). Intriguingly, ASCL members exhibited increased expression in some cancer tissues and decreased expression in others. This is particularly apparent for ASCL2, ASCL3 and ASCL5 that displayed mRNA expression changes (either up-or downregulated in specific cancers). According to these data, both the up- and downregulation of ASCL genes may play an important role in tumor development. The emerging view of the unique developmental niche of ASCL members in early progenitors of diverse neural lineages suggests a potentially critical role in injury response, wound healing and tumorigenesis. However, a limited number of studies to date suggest that these ASCL members may contribute significantly to cancer development. The available data collectively suggest that alterations in the expression of ASCL genes may affect cellular behavior, such as cell proliferation, thereby initiating tumor development. The present study demonstrated that ASCL members may be involved in tumor development and introduces ASCL genes as potential candidates for future prognostic and therapeutic targets.