MTA1—a stress response protein: a master regulator of gene expression and cancer cell behavior

In humans, around 80–90 % of malignant tumors are epithelially derived carcinomas. Ever since Dr. Broaders first systemically described the in situ carcinoma lesion in 1932 [11], the lesion has been seen as the earliest form of cancer. With further morphological observations, the stepwise carcinogenesis model was gradually accepted by the scientific field. This model asserts that an epithelial cell is malignantly transformed due to gene mutation, further proliferates to form atypical hyperplasia, progresses to in situ carcinoma, and with gene mutation accumulation, it breaks down the basement membrane separating the epithelium from the connective stroma. It becomes invasive cancer in the stroma, where it can metastasize to distant sites by lymphatics or blood vessels [12]. The model was widely accepted and was taken as fact.

However, this model has never been extensively tested, and its dominance hinders researchers from thinking otherwise, i.e., normal epithelial cells displaced to the connective tissue stroma sites and developed into cancer in the wrong environment. The basic difference between these two models is that the classic model posits that epithelial cells malignantly transform first and then enter the stroma by a process called epithelial-mesenchymal transition (EMT), while the alternative model states that the epithelial cells enter the stroma first and then transform to cancer cells in the wrong environment [1]. With these two possible choices, logically, we cannot prove one model is right unless we prove the other is wrong.

Unfortunately, this exclusive study approach has never been applied to test the classic carcinogenic model. Morphological observations provide support but not evidence for the model per se. Since the alternative model has never been studied, we cannot say it is wrong. Interestingly, the classic carcinogenesis model has been studied for many decades, so we should be able to falsify it if it was wrong. In fact, paradoxes falsifying the in situ carcinoma to invasive carcinoma model are accumulating and urging us to take a different stance.

The paradoxical evidence comes from different levels. The first evidence level is of molecular pathology [1]. HER2 is a well-known oncogene often amplified and overexpressed in breast cancer. Intriguingly, ductal carcinoma in situ (DCIS), which is deemed to be the precursor lesion of invasive ductal cancer, has a much higher rate (50–60 %) of HER2 amplification and overexpression than that of invasive breast cancer, which is about 25 % positive for HER2 [13‐15]. Yet, we cannot say that HER2 inhibits DCIS progression to invasive ductal carcinoma. The second level of evidence came from histological pathology. Lobular carcinoma in situ (LCIS) and invasive lobular carcinoma (ILC) are both characterized by e-cadherin expression loss, and LCIS is thought to be the precursor lesion of ILC. Paradoxically, if simple LCIS was diagnosed, no specific treatment was needed, since it has been proven that LCIS did not necessarily further progress [16]. Where ILC comes from remains unclear. The third level of evidence includes clinical epidemiology. Evidence has shown that if DCIS was left untreated, only 20 % of patients would develop invasive breast cancer in 10 years [17]. By this speed, if all invasive breast cancer derived from DCIS, its incidence should be many times that of invasive breast cancer, but the opposite is true. Invasive breast cancer incidence is four times that of DCIS [18].

With this evidence, we concluded that not all invasive breast cancer is derived from in situ carcinoma [1]. There must be an alternative carcinogenesis path that creates epithelial-derived invasive cancer.

The above described evidence strongly suggests that the stepwise carcinogenesis model of in situ carcinoma to invasive breast cancer is logically impossible [19]. This implies that carcinoma must be grown out de novo from the stroma, i.e., developed from the displaced epithelial cells. The SCMT we proposed solved the above puzzle [1]. SCMT posits that carcinoma originates from normal/non-transformed epithelial stem cells displaced in the stroma by the damaged basement membrane (BM) [1]. All known carcinogenic factors, such as inflammation and chronic injury, can damage the BM. Historically, German pathologist Julius Cohnheim suggested carcinogenesis by displaced embryonic stem cells some 150 years ago [20].

The current question is not whether the epithelial stem cells can be displaced to the stroma, but regards the fate of the misplaced epithelial cells in the wrong environment. In most cases, we would expect that the misplaced cells die out. However, some could manage to survive. Since they are epithelial cells by nature, they will form epithelial structure types. Usually, if they can differentiate and form BM, they are benign structures like a cyst, a benign tumor, or even normal glandular tissues. However, if they failed to differentiate and form BM, they are carcinoma, i.e., cancer (Fig. 1). The MCF-DCIS cell line is an interesting example that proves the above hypothesis principle. This cell line was derived from the benign MCF10A cell line. When injected into the mammary fat pad of nude mice, it formed DCIS, meaning there is myoepithelial cell differentiation and basement membrane formation [21].

The primary cause for epithelial cells to transform to cancer cells in the stroma is survival pressure (Fig. 2). The misplaced epithelial cells in the stroma are in a stressful state. Compared to the epithelium microenvironment, the stroma is a place of hyperoxia and rich in immune cells, antibodies, and other immune-related cytokines. Moreover, the situations introducing epithelial cell displacement to the stroma are often associated with inflammatory reactions. The misplaced epithelial cells must fight for their existence. Interestingly, increasing population is an effective way of maintaining existence. Therefore, the higher the environmental pressure and greater the cell death, the faster the misplaced epithelial cells must grow [19, 22]. Also, pathologists saw that increased apoptosis was associated with higher cancer malignancy and poor clinical outcome [22‐28].

Although it is still widely believed that resistance to apoptosis is a hallmark of cancer [29, 30], the evidence favors the opposite view [19, 22, 23]. So far, there are no documented carcinogenic agents that can promote cell survival. Instead, they are mostly cytotoxic and induce cell death. For example, aflatoxin and various viruses whose infection induces liver cancer all induce liver cell death. The HBV virus X protein is the most potent factor of the HBV virus’s carcinogenic effects and is an apoptosis-inducing protein [31‐38]. Of the other known apoptosis-inducing genes, such as the cell death receptor CD95 and death executor protein Caspase-3, all are known to promote tumor growth [39, 40]. Conversely, anti-apoptosis factors inhibit carcinogenesis and cancer growth. Autophagy inhibits apoptosis and carcinogenesis [41]. Bcl-2, the anti-apoptotic protein prototype, inhibits carcinogenesis and cancer cell growth both in vitro and in vivo [42‐44]. In various cancers, including breast cancer, colon cancer, and non-small cell lung cancer, Bcl-2 overexpression is a predicting factor of favorable clinical outcomes [45‐51].

Inducing apoptosis as a therapeutic strategy has been touted for the past two decades in both academics and industrial labs without much success. A recent study showed that the IAP inhibitor, which was developed to treat cancer, promotes breast cancer metastasis to bone [52]. Similarly, anti-angiogenic agents and radiotherapy were all found to stimulate cancer metastasis [53‐55]. Therefore, metastasis is a basic response of cancer cells to stress [56].

Epithelial mesenchymal transition has been extensively studied in cancer metastasis research over the past decade. Most studies focused on the mechanisms and signaling pathways involved in EMT with the aim of targeting therapy. However, people rarely asked why cancer cells would ever start EMT. Obviously, the notion of in situ carcinoma progressing to invasive cancer by EMT does not hold up, as the invasive cancer does not derive from the in situ carcinoma. LCIS is characterized by e-cadherin expression loss [16], an EMT hallmark. Ironically, it has been clinically proven that LCIS lesions do not further develop and do not need special treatment [16].

I propose that EMT is a way of immune escape. We know that by nature, carcinoma consists of epithelial cells trapped in mesenchymal tissue. The mesenchymal tissue is not the home of epithelial cells. These epithelial cancer cells thus become the easy target of the immune system. Interestingly, although immune response against cancer has been found for almost six decades [57], there have been no cancer-specific antigens identified for most cancer types. Though the issue has not been explored before by immunologists, I believe that epithelial cell invasion to mesenchyme would provoke an immune response, and the antigen might be the epithelial marker that discriminates cancer cells from the surrounding mesenchymal cells. Thus, to lower the risk of being targeted, the cancer cells would reduce epithelial marker expressions, and, as an adaptation response, express some mesenchymal cell-type proteins. This is in accordance with the biosphere law. We see that jungle animals exhibit colors and patterns similar to their environment to lower their chances of being targeted. Therefore, EMT is a way of immune escape by the strategy of camouflage (Fig. 2).

Adaptation is an important pathology concept and a general biosphere phenomenon. The esophageal epithelium is stratified squamous epithelium, which is resistance to wear and tear but not resistant to acid. Therefore, when gastric acid reflux happens often, the epithelium of the lower part of the esophagus would turn to columnar epithelium, which is more resistant to gastric acid. This is termed metaplasia, a form of pathological adaptation. Similarly, we proposed the concept of molecular adaptation [58]. The molecular adaptations include adaptive mutations and adaptive epigenetic modifications. The former includes point mutations, amplifications, and deletions.

The concept of adaptive mutation was proposed by Cairns three decades ago [59] and has been a controversial issue since then. Most contemporary molecular geneticists are New Darwinists and hold that gene mutations are stochastic in nature. They do not believe in adaptive mutation. It is true that we do not know the adaptive mutation mechanism, but that does not mean it does not exist. We can use the Braf V600E mutation in nevus cells as an example. Around 80 % of nevus cells have this point mutation [60]. Obviously, we cannot explain this phenomenon as a random mutation. Cell cycle regulators are also good examples of molecular adaptions. It is well known that cyclins are often overexpressed and cyclin-dependent kinases (CDKs) are over-activated in cancer, yet the cancer cell proliferation cycle duration is not shorter than the corresponding normal cells but prolonged or showing no change [61]. This paradox is explained by molecular adaptation. The prolonged cell cycle means there is increased resistance and thus requires more cyclins and more active CDKs. Otherwise, cells cannot divide.

As described above, carcinoma cells live in a stressful environment quite different from the epithelium. Initially, when epithelial cells just land to the stroma, it is hyperoxic, with ample reactive oxygen species and reactive nitrogen intermediates, immune cells, and cytokines. When they proliferate and grow, it is hypoxic due to lack of blood supply. Therefore, most, if not all, stress response proteins are highly expressed in cancer cells. For example, heat shock proteins, hypoxia inducible factors, and MAPK kinases such as p38MAPK, MAPK13, p53, and MTA1 are all stress-related proteins and proposed therapeutic cancer targets.

The first factor known to be able to stimulate MTA1 expression in breast cancer cells is the growth factor heregulin [74]. We know that growth factors are often released during trauma, so heregulin can be regarded as a stress-related factor. Korean scientists later found that hypoxia induced MTA1 expression, which helped stabilize HIF1α by recruiting histone deacetylase complex 1(HDAC1) [69]. Since HIF1a plays an important role in angiogenesis and promotion of cancer metastasis, MTA1 also has a part to play in both the normal wound healing and cancer. Li et al. further found that ionic radiation induced marked elevation of MTA1 protein expression in U2OS osteosarcoma cells, mammary glands, thymus, and skin of mice [70]. The increased amount of MTA1 helps stabilize p53 and thus plays a role in repairing damaged DNA [71]. Moreover, MTA1 protein levels were elevated in a germ cell tumor cell line after heat shock and protected the cell from heat shock-induced apoptosis [68]. The X protein of hepatitis B virus is generally believed to be responsible for the virus’s carcinogenic effect [75]. It promotes both liver cell apoptosis and proliferation. Interestingly, X protein strongly induced MTA1 protein expression [66, 76]. Dimethynitrosamine (DEN) has a strong toxicity to the liver. Long-term treatment of rodents with DEN can induce liver cancer. We found that along with the increased liver cell damage, the MTA1 expression level was also increased not only in the nuclei but in the cytoplasm (Fig. 3) [77].

After MTA proteins were found to be a component of nucleosome remodeling and the deacetylation (NuRD) complex, many downstream targets were discovered. Mazumdar et al. first found that MTA1 inhibits ER transactivation activity by recruiting HDAC2 to the promoters of ER targeting genes [74]. It was later found that MTA1 binds transcription factor Six3 [100] and in a negative feedback fashion inhibits Six3 expression and its downstream targets [101]. Paradoxically, MTA1 was also found to be a coactivator protein [102]. By binding and recruiting Pol II and C-Jun to the FosB promoter, MTA1 stimulates FosB expression [102]. Further, MTA1 binds FosB in the E-cadherin promoter region and recruits HDAC2 to downregulate E-cadherin expression, a hallmark of the epithelial mesenchymal transition [102]. Since ER, Six3, FosB, and possibly many other transcription factors regulate the expression of a broad spectrum of genes, MTA1 may thus exert a wide range of regulatory functions.

Except for functioning as a transcriptional coregulator, MTA1 was also found to stabilize proteins by directly binding to them and inhibiting their break down through ubiquitination inhibition. For example, the expressions of both MTA1 and p53 were upregulated in cells exposed to ionizing radiation [71]. MTA1 was found to bind and stabilize p53, which is notoriously known to have a very short half-life [71]. Similarly, in cells cultured under hypoxic conditions, MTA1 and HIF1α expression were both upregulated, and MTA1 bonded to and stabilized HIF1α. They collaborate to promote angiogenesis and may improve the condition of nourishment and oxygen supply [68].

MTA1 was found to block p53-induced apoptosis [103], induce cell proliferation [101], and promote epithelial mesenchymal transition in different studies [102]. However, promotion of cell survival is not positively linked with cell proliferation. In fact, in most cases, or by principle, apoptosis reduction and cell proliferation are negatively correlated [19]. The less apoptosis, the slower the cell grows [19]. Bcl-2, the typical anti-apoptotic protein, inhibits cell growth [43] as well as p202, an interferon-induced antiapoptotic protein [104, 105]. Conversely, CD95, caspase 3, and HBVx, all induce apoptosis and promote tumor cell growth [39, 40, 75]. The case of MTA1 is more complex. Environmental stress stimulates its expression, which may protect cells from apoptosis in a certain context, but is still not enough to make cells live long in an adverse environment. Therefore, cells may still show increased proliferation. As for EMT and metastasis, it is quite natural that MTA1 mediates these processes, but it cannot be the only molecule. Without MTA1, cells would still be able to migrate and invade, though perhaps be compromised.

Studies from MTA1 gene-modified mice have revealed a large range of functions MTA1 may play. MTA1 was found to play roles in circadian rhythm maintenance [106], embryonic development regulation, and visual performance by regulating rhodopsin expression [102]. A reduced rate of breast cancer metastasis to lung was observed in the MTA1 null genetic background [107]. More functions of MTA1 at body level are expected to be revealed by gene-modified animal studies.