In the present study, we investigated the significance of MSI-1 for survival outcomes in breast cancer as well as the molecular rationale for targeting MSI-1. We found that MSI-1 silencing resulted in reduced stem cell marker expression and cell proliferation while increasing apoptosis. When submitting MSI-1 downregulated cells to irradiation, a radio-sensitizing effect was seen.
MSI-1 is a negative prognostic factor in breast cancer
Leveraging the availability of large-scale data analysis tools, we provide an in-depth view of the negative prognostic role of MSI-1. While MSI-1 had previously been linked to overall survival in a small cohort of 140 patients (Wang et al.
2010), our findings from a large database provide a more nuanced perspective: while we were able to establish MSI-1 as a relevant factor for DFS and DMFS, we could only confirm an association with OS in hormone receptor-positive patients. In the entire cohort, there was a trend towards worse OS in MSI-1 high-expressing patients, but no significance was found (
p = 0.2). Our findings regarding DMFS are supported by a previous study that reported that MSI-1 levels in nodal metastatic breast cancer were higher than in node-negative breast cancer (Wang et al.
2010). Besides breast cancer, MSI-1 is also prognostically relevant for other tumor entities, including colon cancer (Li et al.
2011) and glioblastoma (Vo et al.
2012).
Taken together, our analyses suggest a key role for MSI-1 for at least two of three main outcome parameters and a significant potential for MSI-1 targeting as a therapeutic approach.
MSI-1 acts as a stem cell marker and modifier
Previous studies indicate that MSI-1 is a positive regulator of the notch pathway by directly targeting the notch repressor mnumb (Lagadec et al.
2014), thus supporting BCSC maintenance (Wang et al.
2010; Troschel et al.
2020). Our investigation reconfirms these findings as notch-1 and notch-3 were downregulated after MSI-1 silencing in MCF-7 cells. Subsequently, stem cell maintenance, as quantified by mammosphere formation (Cioce et al.
2010), was strongly repressed, similar to previous results (Wang et al.
2010). Mammosphere formation has also been correlated with tumorigenity in vivo, underlining the potential anti-tumorigenic effect subsequent to MSI-1 knockdown (Cariati et al.
2008). Additionally, Tdgf1, a factor not previously described as regulated by MSI-1 that sensitizes breast cancer cells to notch signaling (Watanabe et al.
2009), was also reduced. Notch-1, notch-3 and Tdgf1 were positively correlated with MSI-1 in the database analysis, underlining the evidence for MSI-1’s role in maintaining notch stem cell signaling. Notch-1 is a poor prognostic factor in breast cancer (Zhong et al.
2016).
Nanog is another breast cancer stem cell marker (Jeter et al.
2016) that was negatively influenced by the MSI-1 knockdown. Our findings are thus in line with other breast cancer studies (Wang et al.
2010; Lagadec et al.
2014). Interestingly, Nanog is also known as a mediator of radio-resistance in breast cancer (Harati et al.
2019), offering further incentive to investigate radiation response. Besides, Nanog expression is also correlated with clinical stage (Saravi et al.
2019), well in line with MSI-1’s prognostic relevance.
Nestin is a stem cell intermediate filament (Asleh et al.
2018). While there was only a positive trend between MSI-1 and Nestin expression in MCF-7 cells, the correlation met levels of significance in the dataset analysis. Nestin is an independent predictor of worse prognosis in breast cancer (Zhang et al.
2020).
Stem cell characteristics including notch pathway elements are associated with decreased apoptosis and increased proliferation (Suman et al.
2013). Thus, stem cell inactivation is a possible molecular mechanism behind increased apoptosis and reduced colony formation we observed after MSI-1 knockdown in this study.
Role of p21 subsequent to MSI-1 knockdown for proliferation and apoptosis
p21 is a well-known, yet controversial signaling molecule in breast cancer (Kreis et al.
2019). It is known to be a direct translational MSI-1 target (Battelli et al.
2006), including in breast cancer (Wang et al.
2010), thus explaining the upregulation seen after MSI-1 knockdown. P21 has been attributed several key characteristics.
In breast cancer, stem cell maintenance is known to be negatively regulated by p21 (Han et al.
2016). Conversely, stem cell maintenance is enhanced if p21 is downregulated (Jain et al.
2015). This is in line with our findings which demonstrate an increase of p21 and a decrease of stem cell characteristics subsequent to MSI-1 knockdown. It constitutes another mechanistic explanation for the loss of BCSC characteristics besides the MSI-1-mediated targeting of the notch repressor mnumb discussed above.
P21 has an anti-proliferative effect in breast cancer. When p21 is injected into a mouse model, breast cancer tumor growth is significantly repressed (Ibnat et al.
2019). Similarly, upregulation of p21 reduced (Wang et al.
2010) and downregulation promoted cell proliferation (Li et al.
2020) in other studies. This is in line with our investigation which found proliferation to be strongly decreased after MSI silencing while p21 expression was increased.
While some investigations highlight an anti-apoptotic role for p21 (Fan et al.
2003), multiple studies have found that p21 has the potential to induce apoptosis if artificially upregulated (Jiang et al.
2014; Tor et al.
2015; Giordano et al.
2017). The specific role may hinge upon the cancer type or the exact intracellular localization of p21 with a nuclear localization linked to pro-apoptotic signaling (Crispi
2012; Shamloo and Usluer
2019). In this present case, p21 upregulation subsequent to MSI-1 downregulation had an anti-proliferative and pro-apoptotic effect.
Most studies have indicated that p21 is repressed in breast cancer cells compared to normal breast cells (Pellikainen et al.
2003). However, survival implications remain controversial (Zohny et al.
2019) and p21 has been described as both oncogenic and tumor-suppressive (Kreis et al.
2019). In this study, functional analyses indicate that MSI-1-dependent p21 upregulation is tumor-suppressive, similar to findings in endometrial carcinoma (Götte et al.
2011).
MSI-1 downregulation radiosensitizes MCF-7 breast cancer cells and may lead to chemosensitization
We found that MSI-1 downregulation results in a loss of clonogenic ability of MCF-7 cancer cells after irradiation. We irradiated cells with 6 Gy to reflect the 5 Gy radiation dose from the GSE59732 database we had previously used. The small discrepancy is due to technical reasons but is highly unlikely to have changed results in a significant way.
Presently, radiosensitization is an important aim for breast cancer research and therapy (Yahyanejad et al.
2016). This study suggests MSI-1 may be an interesting potential target to that end. Two possible explanations come to mind.
First, stem cells are known to be radio-resistant, specifically if notch and nanog signaling are high (Harati et al.
2019). The decrease in notch molecules and nanog expression seen after MSI-1 downregulation may thus explain the radio-sensitizing effect.
Second, the MSI-1-mediated p21 increase may also play a role. In our study, we show a similar effect of MSI-1 knockdown and irradiation on p21 expression: both independently increase p21 levels. Subsequently, we demonstrate that both treatments combined lead to higher levels of p21 than irradiation alone, suggesting synergistic potential. In turn, p21 overexpression may reduce proliferation while increasing apoptosis, resulting in a radiosensitizing effect with reduced clonogenic cell survival. In fact, previous investigations have linked p21 expression to radiosensitization in breast cancer (Yang et al.
2012; Kim et al.
2015; Xie et al.
2016).
MSI-1-mediated modulation of radiation resistance observed in the present study is in line with previous findings in glioblastoma (Lin et al.
2018), colon cancer (Sureban et al.
2008), and our previous findings in MDA-MB-231 TNBC cells (Troschel et al.
2020). They help expand the relevance of the radiosensitizing effect and establish MSI-1 as a potential target for improved radiotherapy-based cancer cell eradication.
Additionally, our database research demonstrates compelling evidence that MSI-1 knockdown may be associated with chemosensitization as low MSI-1-expressing tumors were more likely to have a less malignant NPI grade and to show a complete response after chemotherapy. However, AUC statistics indicated prognostic value was somewhat limited. We hypothesize that this may be due to breast cancer heterogeneity and the lack of granularity in the outcome variable (differentiating complete response vs. no complete response only). Additionally, previously published links between p21 and chemoresistance showing inverse correlations demonstrate a plausible mechanistic explanation for the effects (Hou et al.
2017; Mu et al.
2019). Unfortunately, our MCF-7 cell line data did not support the findings above, so the picture is not entirely clear here. We believe that the primary patient data offer arguably more compelling evidence. In vitro, the absence of the tumor microenvironment in our MCF-7 cell culture may have acted as a confounder (and even the MCF-7 data showed a strong decrease in cell viability through MSI-1 knockdown alone). Nonetheless, additional data are needed to substantiate MSI-1-based chemosensitization effects. In any case, MSI-1 knockdown demonstrated overwhelmingly favorable outcomes in the present research and continues to be an exciting scientific research opportunity.
Finally, it is worth discussing the effect of MSI-1 knockdown on cancer stem cells versus the general cancer cell population. As mentioned above, multiple studies, including the present work, suggest that MSI-1 may be a marker of cancer stem cells and/or linked to cancer stem cell maintenance. Limiting the effects mediated by MSI-1 to cancer stem cells only, however, would be insufficient to explain the findings of this study. While cancer stem cells undoubtedly play key roles in MCF-7, their percentage relative to the general population is very limited (Engelmann et al.
2008; Cioce et al.
2010; Li et al.
2017; Xu et al.
2018). As the changes in cell characteristics quantified in our experiments oftentimes reach significant strength, this points to changes in non-stem cells also. Our flow cytometric experiments further underline the effect on non-stem cells as cell histograms demonstrate uniform changes of the entire population towards a higher likelihood of apoptotic features. Importantly, effects do not seem to be limited to a subset of cells here. We thus assume that both cancer stem and normal cancer cells are changed by the MSI-1 knockdown.
However, normal cancer cells may not be directly affected by MSI-1 loss, but rather indirectly by loss of cancer stem cells in the tumor environment. A cancer stem cell-rich microenvironment has been shown to substantially affect non-cancer stem cells (Bhat et al.
2019; López de Andrés et al.
2020). In any case, whether directly or indirectly, our findings support that MSI-1 loss affects a broad segment of cancer cells and is likely not limited to cancer stem cells.
There are several limitations to this study. First, while we rely on primary patient data for nearly all analyses, experimental proof-of-concept is performed in MCF-7 breast cancer cells only. Given the well-known heterogeneity of breast cancer, this limits applicability of the findings. Second, the database gene expression correlations only demonstrate weak strength of associations. However, this is likely tumor heterogeneity in the database, as underlined by the fact that even neighboring notch-1 and notch-2 only demonstrate a Pearson’s
r of 0.13 in the dataset. Thus, we hypothesize that low Pearson’s
r values should not prematurely be discounted. Third, in western blot analyses, MSI-1 expression decrease after knockdown was less pronounced when compared to qPCR analyses. This is likely due to the binding of the other MSI protein family member, MSI-2, that is nearly identical with MSI-1 (Sakakibara et al.
2001), but was not targeted for knockdown. Finally, in accordance with this study’s aim, not all associations seen via qPCR were confirmed via Western Blot, especially when not relevant to the key findings.