Prognostic Value of MUC1 and its Correlation with Tumor-Inltrating Immune Cells in Breast Cancer

Background: MUC1 is a transmembrane glycoprotein, aberrantly glycosylated and overexpressed in a variety of epithelial cancers, and plays a crucial role in cancer progression, especially in breast cancer. It is also an essential regulator for immune functionality, but the mechanisms whereby it effects immune inltration in breast cancer remain uncertain. Methods: In this research, MUC1 expression was analyzed by the Oncomine and TIMER database. The association between MUC1 and prognosis was evaluated by Kaplan-Meier plotter database. The correlations of MUC1 with immune inltration and immunological markers were assessed by TIMER database. Results: We found that MUC1 expression was signicantly correlated with outcomes in multiple cancers, with the effect being particularly pronounced in breast cancer. Pathologically, elevated MUC1 expression was related with worse prognosis depending on intrinsic subtypes, ER status, patient stage, lymph node and TP53 mutation status in breast cancer. Specically, low level of MUC1 seemed to be more favorable to luminal B patients with systemic treatment. MUC1 expression had signicant negative correlations with in ﬁ ltrating levels of CD8+ T cells, CD4+ T cells, B cells, macrophages, neutrophils, monocytes and dendritic cells (DCs) in breast cancer. Besides, MUC1 displayed strong regulations on macrophage polarization and diverse immunological gene markers. The patients with lower MUC1 and deeper immune inltration predicted better prognosis. Conclusions: MUC1 is signicantly associated with clinical prognosis and potentially plays an essential role in modulating cytotoxic T lymphocytes (CTLs), tumor associated macrophages (TAMs), natural killer cells (NK cells) and DCs in breast tumors. Collectively, MUC1 could be served as a valuable biomarker of predicting prognosis and immune in ﬁ ltration for breast cancer patients.


Background
Breast cancer is the most frequently diagnosed cancer and the second most common cause of cancer mortality in women worldwide, accounting for an estimated 28% of new cancers. It is a highly heterogeneous disease with different clinical manifestations and multiple signaling pathways can mediate its initiation and progression [1]. Based on the expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2), researchers have classi ed breast cancer into four types according to the hormone receptors: HER2 type (ER+, PR−, HER2+), luminal A (ER+, PR+, HER2−), luminal B (ER+, PR−, HER2+), and triple-negative breast cancer (ER−, PR−, HER2−,TNBC) [2].
Basically, the drugs applied in the treatment of breast cancer include chemotherapy and pharmacological agents for endocrine therapy, such as tamoxifen targeted at ER. Although there are numerous therapeutic options available for breast cancer patients, allowing for a number of approaches designed to inhibit ER or oncogene molecules, drug resistance and recurrence is the main problem for most treatments [3,4].
Previous studies showed that patients with higher level of immune in ltration showed signi cant better outcomes than lower groups in many cancers, meanwhile, deeper immune in ltration displayed intensive drug effects for chemotherapy. Therefore, cancer immunotherapy, currently as the most promising treatment, has been developed and achieved great success in immune checkpoint blockade and chimeric antigen receptor T cell (CART) therapies, especially in melanoma and non-small cell lung cancer (NSCLC). These successes also underscore the importance of understanding basic tumor immunology for successful clinical application in treating patients with cancer. Immune checkpoint inhibitors, targeting PD-1/PD-L1 or CTLA1-4, are gradually utilized in breast cancer [5]. However, unlike melanoma and NSCLC, breast cancer, as a kind of immunologically cold cancer, has lower response to this treatment, probably due to its high intra-tumoral heterogeneity, cancer stem cells, low immune in ltration and complex tumor microenvironment (TME) [6]. In some breast cancers without immunogenic features, single immune agent activity occurs in less than 10% of patients with metastatic breast cancer. This is likely due to the dominant repression of tumor immunity, particularly through down-regulating the immune in ltration like CTLs. Therefore, nding methods to activating tumor immune environment is necessary to improve prognosis in breast cancer.
MUC1 is a transmembrane glycoprotein, aberrantly glycosylated and overexpressed in a variety of epithelial cancers, and plays a crucial role in cancer progression, especially in breast cancer [7]. For the past 30 years, MUC1 has been considered as a possible therapeutic target. Recently, the mechanism of MUC1 in TME and immunotherapy has been increasingly explored [8]. In NSCLC, targeting MUC1 was associated with the downregulation of PD-L1, induction of IFN-γ and enhanced effector function of CD8 + tumor-in ltrating lymphocytes (TILs) in MUC1 transgenic mouse model [9]. In addition, MUC1 drived PD-L1 transcription by NF-κB p65 promoter occupancy and activates NF-κB to repress the toll-like receptors [9], which was associated with decreased overall survival [10]. Furthermore, targeting MUC1 in transgenic mice was correlated with suppressed PD-L1, increased TILs and better survival [7]. These studies demonstrated that MUC1 was closely related with tumor TME and possibly be an essential regulator for breast tumor immunity, so it is urgent to elucidate the correlations between MUC1 and multiple immune in ltration in breast cancer.
In this study, we conducted a comprehensive assessment of the relationship between MUC1 and patient prognosis using databases including Oncomine, PrognoScan and Kaplan-Meier plotter. We further investigated the link between MUC1 and immune cell in ltration of tumors by the TIMER database. Our results offer novel insights into the immunological role of MUC1 in breast cancer, thereby highlighting a potential relationship and an underlying mechanism between MUC1 and immune cell interactions.

Oncomine Database Analysis
The expression level of the MUC1 gene in multiple types of cancers was identified in the Oncomine database (https://www. oncomine.org/resource/login.html) [11]. The threshold was determined according to the following values: P-value of 0.001, fold change of 1.5, and gene ranking of all.

PrognoScan Database Analysis
The correlation between MUC1 expression and survival in various types of cancers was analyzed by the PrognoScan database (http://www.abren.net/PrognoScan/) [12]. PrognoScan analyzed for relationships between gene expression and patient prognosis as overall survival (OS) and disease-free survival (DFS), across a large collection of publicly available cancer microarray datasets. The threshold was adjusted to a Cox P-value < 0.05.

Kaplan-Meier Plotter Database Analysis
The Kaplan-Meier plotter offers a means of readily exploring the impact of a wide array of genes on patient survival in 21 different types of cancer, with large sample sizes for the breast (n = 6,234), ovarian (n = 2,190), lung (n = 3,452) and gastric (n = 1,440) cancer cohorts. We therefore used this database to explore the association between MUC1 expression and outcomes in patients with breast cancer, analyzing the impact of both clinicopathological factors and MUC1 expression on patient prognosis (http://kmplot.com/analysis/) [13]. The hazard ratio (HR) with 95% confidence intervals and log-rank Pvalue were also computed. TIMER Database Analysis TIMER (https://cistrome.shinyapps.io/timer/) is a database designed for analysing immune cell in ltrates in multiple cancers [14]. This database employs pathological examination-validated statistical methodology in order to estimate tumour immune in ltration by CD4/CD8 T cells, B cells, neutrophils, macrophages and dendritic cells. Gene expression levels against tumor purity is displayed on the leftmost panel [15]. We used database to explore the association between this MUC1 expression and the degree of in ltration by particular immune cells. Then we investigated the correlations of MUC1 mutation status and sCNA with immune in ltration. We further conducted Kaplan-Meier curve analyses to explore differences in patient survival based on MUC1 expression and certain immune cell in ltration. Lastly, we assessed how MUC1 expression correlated with the expression of particular immune in ltrating cell subset markers. These gene markers are referenced in prior studies [16,17].

Gene Correlation Analysis in GEPIA
GEPIA is an online database which facilitates the standardized analysis of RNA-seq data from 9,736 tumor samples and 8,587 normal control samples in the TCGA and GTEx data sets (http://gepia.cance rpku.cn/index.html) [18]. We used this database to assess the connection between MUC1 expression and patient prognosis in multiple tumor types, and we further evaluated the association between MUC1 expression and the expression of particular markers associated with immune cell in ltration.

Statistical Analysis
Survival curves were generated by the PrognoScan and KaplanMeier plots. The results generated in Oncomine are displayed with P-values, fold changes, and ranks. The results of KaplanMeier plots, PrognoScan, and GEPIA are displayed with HR and P or Cox P-values from a log-rank test. The correlation of gene expression was evaluated by Spearman's correlation and statistical significance, and the strength of the correlation was determined using the following guide for the absolute value: 0.00-0.19 "very weak," 0.20-0.39 "weak," 0.40-0.59 "moderate," 0.60-0.79 "strong," 0.80-1.0 "very strong." P-values <0.05 were considered statistically significant.

Results
The mRNA expression levels of MUC1 in different types of human cancers To determine differences of MUC1 expression in tumor and normal tissues, we examined RNA sequencing data in TCGA using TIMER. This analysis revealed that the MUC1 expression was higher in breast, ovarian and cervical cancers compared with the normal tissues (Fig. 1A). Lower expression of MUC1 was observed in colorectal, gastric, kidney cancers. In lung, esophageal, brain and CNS, head and neck cancer, lymphoma, pancreatic and leukemia cancers, MUC1 was either upregulated or downregulated depending on different cancer subtypes. The detailed results were summarized in Additional le 6. To further evaluate MUC1 expression in pan cancers, we examined MUC1 expression using the RNA-seq data of multiple malignancies in TCGA. Figure 1B showed that MUC1 was signi cantly higher in BRCA (breast invasive carcinoma), CHOL (cholangiocarcinoma), COAD (colon adenocarcinoma), GBM (Glioblastoma multiforme), PAAD (pancreatic adenocarcinoma), SKCM(skin cutaneous melanoma), STAD(stomach adenocarcinoma), THCA(thyroid carcinoma) and UCEC(uterine corpus endometrial carcinoma) compared with adjacent normal tissues. However, MUC1 expression was signi cantly lower in HNSC (head and neck cancer), KIRC(kidney renal clear cell carcinoma), KIRP(kidney renal clear cell carcinoma) and LUSC(lung squamous cell carcinoma) compared with corresponding adjacent normal tissues. GEPIA was also used to evaluate MUC1 levels (Additional le 1).

High MUC1 expression impacts the prognosis of breast cancer in patients with different intrinsic subtypes, clinical factors and systemic treatment
To better understand the relevance and underlying mechanisms of MUC1 expression in breast cancer, we investigated the relationship between MUC1 expression and clinical characteristics of breast cancer patients in the Kaplan-Meier plotter database (Fig. 3). The multivariate Cox regression analysis was performed to further evaluate the association between MUC1 and the prognosis of intrinsic subtypes. Generally, MUC1 had no signi cant correlations with total breast cancer prognosis. However, in ER negative breast cancer, low MUC1 expression was related with better RFS (P < 0.05). In basal like, HER2 positive and luminal B breast cancer, overexpression of MUC1 was signi cantly associated with worse OS and RFS (P < 0.05). Conversely, high MUC1 expression was correlated with better OS in luminal A breast cancer. Speci cally, high MUC1 expression was correlated with worse OS and RFS in stage 3 breast cancer patients (P < 0.01), but there was no signi cant difference observed in stage 1 or 2. On TP53 mutation status, overexpression of MUC1 was related with worse RFS in TP53 mutation patients compared with wild type (P < 0.05). We further found that elevated MUC1 was linked to better RFS in lymph node positive status instead of lymph node negative group (P < 0.05). Table 1 demonstrated the details about the clinical parameters and prognosis. Interestingly, we also noticed low level of MUC1 predicted better prognosis in breast cancer patients on the condition of systemic treatment. Although no signi cant difference was found, there was a trend among them. Patients with low MUC1 and received systemic treatment (adjavant, neoadjavant or both) had better prognosis (Additional le 4). The prognosis of luminal B patients (with low level of MUC1) who received systemic treatment was signi cantly better than those without chemotherapy (P < 0.05). On the whole, these results concluded that MUC1 impacted the prognosis in breast cancer patients depending on the intrinsic subtypes, stages, TP53 mutation status and lymph node status. Furthermore, low level of MUC1 seemed more bene cial to the luminal B patients with systemic therapy.

MUC1 expression is correlated with immune in ltration level in breast cancer
Tumor-in ltrating lymphocytes(TILs) are an independent predictor of sentinel lymph node status and survival in many cancers. As an independent predictor of response to neoadjuvant chemotherapy, the increased number of TILs in primary tumor tissue relating to good prognosis has been reported in severa cancers, including breast cancer [19]. Therefore, we investigated the correlations of MUC1 expression with immune in ltration levels in 40 cancer types from TIMER. The results showed that MUC1 expression widely had signi cant correlations with immune cells in many cancers (Additional le 8).
The association between MUC1 expression and the degree of immune in ltration in breast cancer was explored by TIMER database (Fig. 4). Tumor purity is an important factor that in uences the analysis of immune in ltration in clinical tumor samples by genomic approaches [20], and TIMER and GEPIA have most of the homologous data from TCGA [14,18]. We found that MUC1 had signi cant correlations with tumor purity in breast cancer (P = 1.69e-02) and basal breast cancer (P = 2.08e-02), however, no signi cant correlations were observed in HER2, luminal A and luminal B subtypes. Considering basal like breast cancer is more aggressive and easier to metastasis on clinical and MUC1 plays distinct roles in basal subtype from total breast cancer, implying that MUC1 could play a part in tumor metastasis or recurrence in basal type due to reshaping tumor immune environment. Overall, MUC1 was signi cantly correlated with immune in ltration in breast cancer.
We next analyzed the correlations between MUC1 copy number with immune cells in ltrating (Fig. 5).
To further investigate the role of MUC1 expression together with immune in ltration on clinical prognosis, we compared the cumulative survival in breast cancer patients carried with different levels of MUC1 and different levels of immune in ltration.

MUC1 had close relationships with several immune gene markers, particularly NK cells, CD8 + T cells and B cells (Additional le 5)
. However, the correlation between MUC1 and speci c subtype was not signi cant (Additional le 9). We further testi ed the above immunological markers in GEPIA database, which was basically the same (Table 3).  * P < 0.01; ** P < 0.001; *** P < 0.0001.

Discussion
In this study, MUC1 was validated as an independent prognostic indicator in breast cancer, mainly depending on intrinsic subtypes and clinical parameters. Furthermore, MUC1 was signi cantly correlated with immune in ltration and diverse immune marker sets and patients with low level of MUC1 and deeper immune in ltration predicted better prognosis. These ndings provide insights in understanding the potential role of MUC1 in tumor immunology and its use as a breast cancer biomarker.
We rst con rmed that MUC1 was overexpressed in breast cancer by Oncomine and TIMER databases and high level of MUC1 was correlated with poor prognosis in breast cancer. The discrepancies in levels of MUC1 expression in different cancer types from different databases may be a re ection in data collection approaches and underlying mechanisms pertinent to different biological properties. Nevertheless, we found consistent correlations between MUC1 expression and breast cancer in these databases. MUC1 had signi cant relationships with prognosis in basal, HER2+, luminal A and luminal B subtypes, implying that MUC1 was closely related with intrinsic subtypes. Although, there was no signi cance in ER positive subtype, it has been reported that MUC1 induced tamoxifen(an estrogen analogue) resistance in ER positive breast cancer through regulating growth cascades and gene transcription [21]. It has been studied that breast cancer has higher heterogeneity than other cancers, and the classi cation of breast cancer is also more complicated [2], thus analyzing MUC1 according to subtypes is more accurate and feasible. High level of MUC1 was related with worse prognosis when lymph node was positive, indicating that MUC1 is associated with cancer lymph metastasis and could be a reliable predictor for patients with lymph node positive. In addition, high level of MUC1 was correlated with poor prognosis in stage 3 patients, showing evidence that MUC1 could be used as a biomarker to predict death risk particularly in advanced breast cancer patients. In general, these ndings strongly suggested that MUC1 is a credible prognostic biomarker in breast cancer patients.
Another key nding in this study is that the expression of MUC1 correlated with the degree of immune in ltration in breast cancer. Historically, breast cancer was not thought to be immunologically active, particularly when compared with tumors such as melanoma. However, recent evidence has emerged that TILs in ltrated in breast tumor tissue prior to treatment could predict better response to therapy and improved prognosis [19]. We found that MUC1 was negatively correlated with the degree of CD8 + T cells, CD4 + T cells, B cells, DCs, NK cells, neutrophils and monocytes in ltration in breast cancer (P < 0.001,  [23]. In HER2 + and TNBC breast cancer, even incremental increases in TILs both in and surrounding the tumor have shown to predict better response to chemotherapy and improved survival in patients [24,25]. However, our results revealed that high level of MUC1 signi cantly downregulated TILs in ltration, which could partly explain the weak immunological response in breast tumor. Therefore,MUC1 possibly hinders the adaptive immune activation in breast cancer. In addition, MUC1 had signi cant negative correlations with memory T cells, such as CD8 + central memory cells (P < 0.001, Fig. 4).

Compared with naive T cells, memory T cells (central and activated), as a type of effective T cells, have
more sensitive reactions to antigens, live longer and have stronger immune responses [26]. So MUC1 possibly impairs immune attack by inhibiting memory T cells stimulation.
DCs are considered as pivot cells of the immune system due to their exclusive ability to coordinate the innate and acquired immune responses, cross-present exogenous antigens, and prime naive T cells [27].
Normally, glycosylated MUC1 could be recognized as antigen to activate MHC class II presentation, however, the adaptive immune responses against mucin antigens is in uenced by the expression of speci c proteins or enzymes that regulate the processing of cancer-associated mucins, like heat-shock protein 70 (HSP70) and cathepsins, which destroyed the function of DCs [28,29]. Tumor-associated mucins, especially MUC1, promoted an immunosuppressive microenvironment by masking Toll-like receptors (TLR) on antigen presenting cells (APCs) or inhibiting synapse formation between cytolytic NK cells and tumor cells [30]. O-linked glycosylation at either Thr3 or Ser4 sites renders MUC1 glycoforms unavailable for proteolytic processing by DCs, probably due to cleavage site masking [31]. Moreover, the formation of aberrantly glycosylated mucins conceals various receptors on the cell surface, hence obstructing ligand-receptor or cellular cascades. Membrane associated MUC1 has been reported to suppress TLR-mediated innate immune activation, inhibiting secretion of pro-in ammatory cytokines, such as IL-6, IL-8, and TNF-α [32]. Thus, high level of MUC1 is correlated with ine cient and reduced DCs in ltration in TME. NK cells usually exhibit cytolytic activity and kill cancer cells by the engagement of receptors present on their surface upon cancer cell encounter. Mucins associated antigens, expressed in various adenocarcinomas, inhibit the cytotoxic activity of NK cells, and the effect is further enhanced in the presence of ammonium ions, which are known inhibitors of NK cell function [33]. MUC16, one of mucins family, has been identi ed as an ovarian tumor marker, could downregulate the expression of CD16 and CD94/NKG2A on NK cells, thus NK cells are unable to recognize and kill tumor cells [34].
Therefore, inhibition of immune synapse formation between NK cells and cancer cells represents another effective MUC1 mediated immunosuppressive mechanism [35].
We further found MUC1 plays an essential role in regulating TAMs polarization in breast cancer. Many researchers discovered that M1 macrophage is associated with tumor elimination and pro-in ammation by producing cytokines and NO, however, M2 macrophage involves in tissue repairment, remodeling, angiogenesis and retain homeostasis to promote tumor growth [36]. TAMs are a group of macrophages, functionally similar to M2 polarized, harboring the tumor tissue and closely associated with cancer cells, and have complicated effects on tumor behaviors. In most cancers, TAMs in ltration in solid tumors is associated with poor prognosis and leading to chemotherapy resistance [37]. In breast carcer, TAMs are the prominent components and high TAMs density correlates with worse prognosis [38]. Thus, high level of MUC1 activates M2 macrophages to promote tissue repairment like aggravating blood vessel growth or inducing EMT, while inhibits M1 macrophage to stimulate T cells. Li et al found that TP53-de cient and TP53-mutant mice showed increased expression of M2 polarized macrophages to induce M1-M2 transition [39]. It has been revealed that MUC1 cytoplasmic domain integrated directly to TP53 regulatory domain [40], so maybe there is a potential mechanism that MUC1 induces TAMs polarization to affect TME meditated by directly regulating TP53. We also found high level of MUC1 was associated with worse prognosis in TP53 mutation patients, supporting the above hypothesis. These researches demonstrated that overexpression and aberrant glycosylation of MUC1 modulates both innate and adaptive immune effectors to promote tumor surviving. High ampli cation of MUC1 was negatively related with most CTLs in ltration, but neutrophil was an exception (Fig. 5C). Neutrophils typically obliterate bacterial and fungal in our body, however, the role of neutrophils in tumor immunity remains controversial [41]. Previous research showed neutrophils escorted circulating tumor cells (CTCs) within the bloodstream and facilitated the metastatic potential [42]. Fridlender et al also discovered neutrophils had opposite roles in the TME [43]. Considering the double-face of neutrophils, interactions between MUC1 and neutrophils still remain vague and more research needs to be done. Finally, MUC1 was negatively correlated with most immunological markers in breast cancer ( Table 2 and Table 3). Taken together, immune cells especially CTLs, were signi cantly downregulated by MUC1 in breast cancer.
To investigate the relationships between MUC1 expression, immune in ltration and prognosis, Kaplan Meier survival curves were made to con rm the exact role of MUC1 in immunology (Fig. 7). Overall, our results were consistent with clinical outcomes in breast cancer patients [3]. Interestingly, in low levels of MUC1 groups, the prognosis in patients with higher M1 macrophages or M2 macrophages was totally opposite, proving the different functions of TAMs in breast tumors, which was consistent with Fig. 4. However, none of such survival bene t was detected in high MUC1 groups, indicating that down regulating MUC1 could be a potential precondition for immunotherapy. Previous studies showed that traditional chemotherapy prolongs survival in breast cancer patients meditated by improving CTLs in ltration, especially in TNBC, which is more sensitive to cancer agents [44]. Cytotoxic chemotherapy has been shown to increase type 1 T cell stimulation. Several chemotherapeutic agents have been found to trigger immune recognition in tumor by releasing stress proteins from dead cells [45]. Early data has further demonstrated that combining chemotherapy and immunotherapy may increase the number of clinical responses to checkpoint inhibitors in breast cancer. In this paper, we noticed that low level of MUC1 possibly improves clinical outcome and agent response by synergistically enhancing the CTLs immune in ltration in patients especially those who have received chemotherapy. In conclusion, high level of MUC1 could be partly an explanation for the immune suppression within breast tumors, so targeting MUC1 could be necessary to arouse immune in ltration for breast cancer patients, possibly meditated by recruiting effective immune cells, especially CTLs, DCs and NK cells. In the future, precisely targeting on MUC1 combined with chemotherapy or immunotherapy to enhance immune in ltration could be a promising strategy to improve prognosis in breast cancer patients.