It is widely recognized that the accumulation of various harmful genetic alterations in normal cells may induce malignant cancer cells (1). Genetic mutation is one genetic alteration, but not all genetic mutations are harmful. Genetic mutation promotes biological evolution and results in biodiversity (2,3). Genetic alterations include genetic mutation, gene copy number variation (CNV), loss of heterozygosity (LOH), allelic imbalance (AI) and microsatellite instability (MSI). Epigenetic alterations, represented by post-transcriptional control and DNA methylation, have been the focus of recent studies.

Cancer-associated fibroblasts (CAFs) are one major type of component in the tumor microenvironment (4). CAFs differ phenotypically and functionally from normal fibroblasts (NFs) (5). CAFs provide cancer cells with nutrition and promote the proliferation, invasion and metastasis of cells (6–9). CAFs maintain their phenotype for numerous passages during culture in vitro without exposure to cancer cells, while NFs cannot be infinitely proliferous like cancer cells (10). Therefore, it has been demonstrated that genetic or epigenetic alterations may be responsible for the special features of CAFs (Fig. 1) (10). Due to the critical role of CAFs during cancer progression, the genetic characterization of CAFs aids in the investigation of cancer therapy. The present review summarizes the current knowledge regarding 7 possible genetic and epigenetic alterations in CAFs.

CAFs are different from NFs, and their unique phenotypes and functions are partly determined by differences in gene expression. The differences between CAFs and NFs in gene expression have been extensively compared; in one study, 31 genes in breast CAFs, which were identified using Affymetrix Human Genome U133 Plus 2.0 and an empirical Bayesian model, were different from those in NFs (11). Of the 31 genes, the 21 upregulated genes were primarily associated with cell paracrine and intracellular signaling, transcription regulation and cell adhesion and migration, and their transcriptional products included transforming growth factor-β2 (TGF-β2), insulin-like growth factor-binding protein 2 and transcriptional factor AP-2α/γ (11). By contrast, the 10 downregulated genes were primarily associated with epithelial membrane proteins (11). Genes mainly involved in coding adhesion molecules and growth factors have also been found to be upregulated in other types of CAFs, including colon (12) and pancreatic (13). The prostaglandin-endoperoxide synthase 2 gene (PTGS2), which encodes cyclooxygenase-2, was found to upregulate the expression of TGF-β2 (14). These results are consistent with another study concerning the gene expression profiling of breast CAFs, which were detected using a complementary DNA microarray (15).

A study on the gene signature of CAFs in non-small cell lung cancer (NSCLC) revealed similar results to those observed in breast CAFs (16). In total, there were 46 genes differently expressed in CAFs compared with NFs, and the upregulated genes were enriched in cell signal transduction, cell adhesion, cell response to stress and angiogenesis (16). In total, of 6/46 genes were also associated with the TGF-β signaling pathway (16). TGF-β is a multifunctional signaling factor (17–19), which is involved in the epithelial-to-mesenchymal transition (EMT) (20,21). This may be one origin of CAFs (5). Cellular tumor antigen 53 (p53) signal transduction and genes associated with cell apoptosis and death were downregulated in colon CAFs (22). In addition, the gene expression profiling of CAFs may be prognostically significant for patients with NSCLC or colorectal cancer (CRC) (16,22).

The differences in gene expression between CAFs and NFs are in keeping with the special function of CAFs in promoting cancer progression. CAFs promote the growth of cancer cells by providing nutrition to tumor cells (23,24), and facilitate the invasion and metastasis of tumor cells by secreting matrix metalloprotease to resolve the matrix surrounding cancer cells (16,25–28). Additionally, CAFs activate tumor angiogenesis (29–31).

CAFs have a gene expression signature different from NFs, and upregulated genes are primarily associated with angiogenesis, EMT, cell adhesion and cell interactions (32,33). The gene expression profiles of CAFs are different among various tumors, despite certain general features. In the 3 subtypes of breast tumor, which are differentiated according to 3 receptors, estrogen, human epidermal growth factor receptor 2 (HER2) and progesterone, the gene expression profiles of CAFs are even subtype-specific (34), which may be used to distinguish the 3 subtypes (32). Cancer stage also affects the gene expression profiling of CAFs in breast cancer (35).

A karyotype, which is defined as the chromosomal composition in an individual cell, reflects the number and structure of chromosomes. Humans are diploid with 46 chromosomes; however, cancer cells are usually polyploidy or aneuploidy with a larger nuclei-cytoplasm ratio compared with normal cells. Polyploidy helps cancer cells to resist adverse factors, and tetraploid cell lines are more viable than diploid cell lines (36). Whether or not CAFs have various chromosome karyotypes in a way that is similar to cancer cells remains controversial. No clear differences have been identified in the chromosome karyotype of CAFs in oral carcinoma (37). By contrast, CAFs from 2 types of tumors, namely melanoma and prostate cancer (xenograft or spontaneous), are characterized by aneuploid karyotypes, which are caused by the attenuated activation of p53 in CAFs (38). Common reasons that account for the conflicting results observed include the heterogeneity of CAFs and the various anatomical locations. Furthermore, the largest difference between the two aforementioned studies is the time in cell culturing; the study that identified positive karyotype alterations had cultured CAFs for ≥20 weeks prior to testing (38). Since survival advantage in the tumor microenvironment, caused by selection pressures, is the reason for the alteration of karyotypes, a long-term culture is more likely to obtain CAFs with altered karyotypes.

Senescence that is activated by signals in CAFs, including growth-regulated oncogene-1 and the TGF-β/connective tissue growth factor pathway, may fuel and promote the growth of cancer cells (39,40). Polyploidy is one of the three steps leading to the immortality of cancer cells, and cells with normal karyotypes cannot avoid senescence (41). Consequently, whether the absence of chromosome karyotype may cause CAF senescence and whether the senescent CAFs then promote tumor growth remains unclear.

CNVs are particularly common in tumors and can contribute to tumor progression (42). The multiplex ligation-dependent probe amplification technique has previously demonstrated the presence of gene dosage alterations in ovarian cancer CAFs, and CNV was identified in 61% of genes (43). Amplifications and deletions were detected in CAFs in colorectal and breast carcinomas, and the presence of 17 CNVs was confirmed by representational oligonucleotide microarray analysis (44). Ephrin type receptor (Eph) A4, EphA2 and EphB2 were also identified in these alterations, and EphA4 may be directly associated with tumor angiogenesis and progression; thus Ephs may be attractive drug targets for cancer therapy (45,46). By contrast, CNVs were not identified in 95% of 98 patients, including 484 frozen specimens (47), and this study suggested that previous studies exhibiting a high frequency of CNVs are not entirely believable, since formalin-fixed and paraffin-embedded (FFPE) specimens have been used and, thus, the high number of CNVs may reflect an FFPE-related artifact. Finally, the authors of the aforementioned study concluded that the tumor microenvironment is genomically stable (47). The present authors hypothesize that genetic statuses are different among various types of tumor. However, CNVs in colorectal and breast carcinomas have been detected without using FFPE specimens (44). Thus, investigations using more specific and accurate methods are required.

Numerous forms of genetic alterations, including MSI, LOH, AI and CNV, have been observed in benign and malignant epithelial neoplasms. The LOH-induced tumor suppressor gene (TSG) mutation is common among cancer epithelial cells (48–50). Whether or not LOH and AI are present in CAFs has been the focus of studies in recent years.

Several types of cancer, including head and neck squamous cell carcinoma (HNSCC), breast cancer, ovarian cancer, bladder cancer and CRC, and their associated stromata have been investigated since 2000. LOH or AI was revealed in all these cancer stromata. The LOH of the p53 and BRCA1 genes in chromosome 17 was identified in ulcerative colitis (UC)-associated carcinogenesis (51) and CRC (52). The LOH in stromal cells is associated with the colitis-associated carcinogenesis, and the frequency is increased compared with cancer cells (51). In CAFs of HNSCC, LOH is identified in the tumor protein 53 (TP53) and phosphatase and tensin homolog (PTEN) genes (two important tumor-suppressing genes) and the frequency is associated with tumor size and lymph node metastasis (53). LOH in TP53 is also observed in bladder cancer CAFs (54). The frequency of LOH in chromosome 3p21.3, a region that harbors 3 hyaluronidase genes potentially encoding the hyaluronidase tumor suppressor, in the stromal cells of ovarian cancer is similar to the frequency in cancer cells (55). However, LOH in these genes does not affect the accumulation of hyaluronidase or tumor progress (55). Breast cancer is most commonly used in the detection of LOH and AI, and numerous studies focus on TP53 and PTEN (56). The LOHs of TP53 and PTEN have been identified in hereditary and sporadic breast cancers: Hereditary patients are more prone to TP53 mutation; however, it is only in sporadic breast cancer that LOHs of TP53 and PTEN are associated with cancer lymph node metastasis, tumor grade and type of estrogen receptor (57,58). The frequency of LOH is high in sporadic breast cancer, and LOH at certain loci is noted only in stromal cells; these loci primarily code TSGs (51,53). LOH and AI of certain genes occur in tumor and stromal cells, while some occur only in tumor or stromal cells (53,56). The occurrence of LOH in the cancer epithelium or stromal cells is not completely random; certain unique loci may contain an increased frequency of LOH compared with other loci, whilst several loci may act as trigger points, at which a LOH may result in a LOH at another locus (59,60). p53 mutations in CAFs also affect the CAF in question. LOH or complete loss of p53 in CAFs leads to the accumulation of stroma in cancer tissues, an increased proliferation of mesenchymal cells and a mesenchymal response (61). p53 status is involved in clinical treatment, since a LOH of p53 in CAFs weakens the response of cancer cells to radiotherapy and chemotherapy (38,62). Recently, the acquired resistance to targeted chemotherapy of pulmonary adenocarcinoma, including epidermal growth factor receptor tyrosine kinase inhibitor (EGFR-TKI), has attracted more attention, and EMT may be one major reason (63,64). Since EMT may be one origin of CAFs (5), the present study hypothesizes that the p53 status or LOH of p53 in CAFs of pulmonary adenocarcinoma may play a role in the resistance against EGFR-TKI. p53 status in cancer cells has been shown to be associated with tumor angiogenesis (65–68). Regarding the reason for the multifunctionality of CAFs, altered p53 signal transduction in CAFs may also be associated with angiogenesis.

Overall, the present review concludes that genetic alterations in cancer and stromal cells may occur independently, at least in breast cancer. In addition, interactions and cross-talk exist between tumor epithelia and supportive stromal cells. Stromal cells should be recognized as an equal and independent interactive component, rather than a response or an accessory to the carcinoma.

MSI represents a defective mismatch repair and is one feature of tumors (69). MSI occurs not only in cancer epithelial cells, but also in CAFs. MSI occurs in 41% of CRC CAFs vs. 34% of cancer cells (70). MSI in cancer cells is positively correlated with differentiation and Duke's stage; however, in CAFs, MSI is more frequent in samples with Duke's stage A (55%) and well-differentiated CRCs (54%) (70). In addition, MSI occurs in UC-associated CRC CAFs, since MSI in CAFs is associated with TSGs, including p53 (71). Furthermore, TSG instability is more important at the early stage of UC-associated oncogenesis (72). MSI also occurs at the early stage of esophageal cancer CAFs, but the frequency is increased in squamous cell carcinoma compared with in Barrett's adenocarcinoma (73).

With the exception of MSI in the nuclei of CAFs, MSI in DNA occurs in the cytoplasmic mitochondria (Mt) of CRC CAFs. The Mt genome contains 16,569 nucleotides that encode 2 ribosomal RNAs and 22 transfer RNAs, the production of which is critical for oxidative phosphorylation (74). MtMSI has already been reported in several types of cancer cells (75–77). MtMSI in CAFs is primarily detected in the D-loop of MtDNA, which contains transcription and replication regulatory elements (78). Different from nuclear MSI, MtMSI in CRC CAFs is not related to Duke's stage, or to nuclear MSI of CAFs or cancer cells. MtMSI may work only in the development, but not progression, of tumorigenesis (79).

Perspective on the clinical significance. Clinical trials have already used miRNAs as drug targets or biomarkers for the stratification of patients, prognosis and drug efficacy (128). miR-21 is upregulated in the CAFs of several cancer types and is selected as a diagnostic and prognostic marker. mir-21 is associated with the Duke's stage of CRC. Highly expressed stromal miR-21 predicts a poor recurrence-free survival (129). Combined with other miRNAs, plasma miR-21 could be a novel method for early cancer diagnosis. However, miR-143 usually shows a decreased expression and is considered as a cancer suppressor (104,130,131). Erb-b2 receptor tyrosine kinase 3 (ERBB3), one of the four members of the ERBB family of receptor tyrosine kinases, is targeted by miR-143, which in turn prevents breast cancer cells from proliferation and invasion. EGFR-TKI drugs have a special effect in targeting cancer cells, but patients have to endure drug-resistance for several months after administration of the drugs. miR-143 could potentially be a novel pathway for targeted drugs as it can inhibit cancer cell invasion through an EGFR signalling pathway (132,133). With regard to upregulated miR-221, a drug called 2′-O-methylphosphorothioate-modified anti-miRNA-221 has already demonstrated an antitumor effect in mouse models (134). The in vivo administration of tumor suppressor microRNA or inhibitors of tumor-promoting microRNA that target CAFs may be an emerging tumor treatment; this may be more effective compared with existing techniques, considering that CAFs have a relatively steady phenotype compared with cancer cells (134).

DNA methyltransferase inhibitors and DNA methyltransferase maybe be useful for combating CAFs possessing the corresponding hypermethylation and hypomethylation. The drugs, 5-azacytidine and 5-aza-2′-deoxycytidine, which have already been used for treatment of acute and chronic myeloid leukemia and myelodysplastic syndrome, have been investigated in the treatment other tumors; however, they are less effective in the treatment of solid tumors and have associated side effects (135).

Whether or not CAFs have somatic or genetic alterations remains controversial, despite certain alterations that have already been found. Studies that found a positive somatic alteration in CAFs, without exception, have extracted DNA from archival tissues. In other studies (10,47,136,137), CAFs were isolated from fresh frozen tissues or flow cytometry; no somatic genetic alterations were found, while CAFs from FFPE show a high frequency of LOH and AI. This pattern indicates that FFPE tissues could result in highly fragmented DNA and RNA molecules, which are not suitable for large-scale genetic analysis. PCR amplification then exacerbates the artificial false positive result (138–143). Thus, certain studies refute genetic alterations and instead attribute the results to epigenetic alterations More reliable and accurate methods should be found to solve these controversies.

CAFs retain diploid rather than polyploid cancer cells, which may account for the senescence of CAFs. Since telomerase is closely linked with cell senescence, particularly in cancer cells, the activity and status of telomerase in CAFs may be worth additional studies, though certain studies have already been performed (144–146).

CAFs, which are phenotypically and functionally different from NFs due to numerous potential genetic alterations and epigenetic changes, are critical for tumor progression and have attracted increasing numbers of studies in recent years (Fig. 1). With the potential genetic and epigenetic alterations changes found in CAFs, the underlying mechanisms regarding the features of CAFs are gradually revealed. With the exception of genetic and epigenetic alterations, other relevant changes may also explain the differential expressions of genes in CAFs, and certain special genetic or epigenetic alterations may be confirmed in the future. Further investigation into the detailed genetic and epigenetic alterations of CAFs in the tumor microenvironment increases the understanding of CAFs and provides novel approaches for clinical application.