High expression of ACE2 in HER2 subtype of breast cancer is a marker of poor prognosis

Background ACE2 a key molecule of the Renin-Angiotensin system has been identified as the receptor for SARS-CoV-2 entry into human cells. In the context of human cancers, there is evidence that ACE2 might function as a tumor suppressor. The expression levels of ACE2 among the different subtypes of breast cancer has not been investigated. Methods We have examined the differential expression of ACE2 and its correlation with prognosis in breast cancer subtypes using the METABRIC (n = 1898) and TCGA (n = 832) cohorts. Correlations were evaluated by Pearsons's correlation co-efficient and Kaplan-Meier analysis was used to estimate differences in disease-free survival between the ACE2 high and ACE2 low groups. Results There is minimal expression of ACE2 in the luminal classes, but significantly higher levels in the Basal-like and HER2-enriched subclasses. Metastatic biopsies of these tumor types also show enhanced expression of ACE2. High levels of ACE2 correlated with decreased disease-free survival in the HER2-enriched subtype, and it was positively correlated with EGFR expression. Conclusion These observations suggest ACE2 might function as a context dependent factor driving tumor progression in breast cancer and permit new opportunities for targeted therapy.


Introduction
The COVID pandemic has catapulted angiotensin-converting enzyme 2 (ACE2) into the spotlight since it has been identified as the most likely receptor for SARS-CoV-2 entry into host cells (1). ACE2 is a key player in the Renin-Angiotensin-Aldosterone system (RAAS) that regulates vasoconstriction and dilation. The RAAS axis starts with Renin, secreted by the renal juxtaglomerular cells, that acts on angiotensinogen to release angiotensin I (Ang I). ACE is a carboxypeptidase that acts on Ang I and converts it to Ang II which is the main effector in the RAAS system with vasoconstrictive, pro-inflammatory and pro-fibrotic properties. A homologue of ACE, the ACE2 enzyme converts Ang I to Ang (1)(2)(3)(4)(5)(6)(7)(8)(9) and Ang II to Ang (1-7) that mediate vasodilation, apoptosis and antiproliferation opposing the effects of Ang II [1,2].
ACE is equally expressed in apical and basolateral surfaces whereas, ACE2 is expressed on the apical surface of polarized cells [3]. Under normal conditions, the ACE2 does not internalize into the cell. However, binding of the spike protein of SARS-CoV and Cov2 to ACE2 triggers internalization reducing its activity on the cell surface. The surface activity is also reduced by shedding caused by the sheddase ADAM17 (TACE) [4] which results in the increased presence of the enzyme in plasma and urine [5]. ACE2 is expressed in many tissues primarily vascular endothelial cells and epithelial cells and exists in both secreted and membranous forms. The expression of ACE2 has been reported in the nasal and oral mucosa, colon, skin, nasopharynx, lymph nodes, spleen, human kidney, heart, testis, lung, liver, small intestine and brain [6,7]. It is also known to be expressed in arterial and venous endothelium and arterial smooth muscles [8]. It is also known to be expressed in glucose accumulating tissues like liver, β-cells and adipose tissue [9]. Through scRNA-seq data analysis, it was reported that ACE2 expression is highest in pulmonary AT2 cells, myocardium, esophageal epithelial cells, proximal tubule cells of kidney and bladder urothelial cells [10]. In a study on pan cancer tissues it was shown that six types of cancers including breast invasive carcinoma, prostate adenocarcinoma, thyroid carcinoma, liver hepatocellular carcinoma, kidney chromophobe and stomach adenocarcinoma had differential expression of ACE2 when compared to the corresponding normal tissues [11]. The role of ACE2 has been reported to be tumor suppressive in multiple cancer subtypes. A report on breast cancer suggests that levels of ACE2 protein is negatively correlated with the migratory ability of cells and downregulation of ACE2 contributes to invasion and metastasis of breast cancer cells [12].
Breast cancer is a heterogenous disease at the molecular level comprising multiple subtypes, and based on the PAM-50 classification it includes Luminal A, Luminal B, Basal-like, HER2-enriched (Human Epidermal Growth Factor Receptor 2-enriched) and the Normal-like subtypes [13]. PAM50 is a 50-gene q-PCR assay derived by the unsupervised clustering analysis of global gene expression of human breast cancer specimens, developed to identify sub classes that are segregated by biology and this was in turn used to develop a clinically validated assay that provides a risk of relapse (ROR) score [14,15]. Since the distribution of ACE2 expression amongst the breast cancer subtypes has not been investigated we have examined its expression and explored its associations with other genes that play an important role in the biology of breast cancer.

Material and methods
We have examined the expression of ACE2 in breast cancer samples from the TCGA (n = 832) and the METABRIC cohort (n = 1898). The data was downloaded from the data portal on TCGA website (https: //tcga-data.nci.nih.gov/tcga/) and were TCGA level 3 data, the most highly processed data (https://tcga-data.nci.nih.gov/tcga/tcgaData Type.jsp) and the European Genome-Phenome Archive for METABRIC data [(https://www.ebi.ac.uk/ega/studies) accession no: EGAS00000000083] respectively. We have also analyzed the data from the Swedish multicenter and randomized trial-TEX; (http://www.clinicaltrials.gov/ct2/show/NCT01433614) which enrolled 287 breast cancer patients between December 2002 and June 2007 of whom 111 patients with a confirmed loco-regional or distant metastasis yielded at least one biopsy with adequate RNA for microarray analysis (Affymetrix array 2.0 GPL10379). All the patients had complete clinical details and follow-up information [16]. The gene expression microarray data was obtained from Gene Expression Omnibus under the accession number, GSE56493. Inter-group variability of gene expression data was analysed using the Kruskal-Wallis test. Correlations were evaluated by Pearsons's correlation co-efficient (pcc). All statistical analysis was carried out using the software XLSTAT 2015 (Windows). Additional validation of the analyses was done using the The cBio Cancer Genomics Portal. Kaplan-Meier analysis was used to estimate differences in disease-free survival between the ACE2 high and ACE2 low groups in the METABRIC cohort. To maximize the specificity, we chose to take a cut-off for ACE2 transcript at the 3rd quartile (3rd quartile at 5.59) and divided the samples into ACE2 high and low. Disease-free survival was calculated as the time from the date of first diagnosis to the time when either a local recurrence or a distance metastasis occurred. Patients without an event or those that had succumbed to non-breast cancer related causes were right-censored. Log rank test (Mantel-Cox) was used to compare the disease-free survival between groups. Both univariate and multivariate Cox-proportional hazard analysis were done to validate the prognostic importance of ACE2 in comparison to other clinico-pathological characteristics.

High levels of ACE2 is prognostic in HER2-enriched samples and is positively associated with EGFR expression
We first examined the relative abundance of ACE2 across the PAM50 subtypes. The general level of expression of ACE2 is rather low in breast cancer subtypes except for the HER2-enriched and Basal-like subtypes. ACE2 expression was highest in HER2-enriched samples (n = 236; Table 1) when compared to the other subtypes in the METABRIC cohort (p<0.0001) ( Fig. 1.a). In the TCGA dataset also we observed the same association and an average third quartile fold difference of 30 (∆∆Ct of 4.9) between the HER2 (n = 67) and the luminal subtypes (Lum A n = 416; Lum B n = 185; p<0.0001) (Fig. 1.b). Given this very significant over expression in the Basal-like and HER2 subtypes which are clinically aggressive sub classes of the disease compared to the luminal subtypes, we examined the effect of ACE2 expression on disease-free survival by performing Kaplan Meier survival analysis on the set of patients in the METABRIC cohort (n = 1898). We chose to take a cut-off for ACE2 transcript at the 3rd quartile (3rd quartile at 5.59) and based on this divided the samples into ACE2 high and low groups. Stratification of all tumor samples by ACE2 mRNA levels produced significant separation of the groups based on disease-free survival [(Hazard Ratio (HR) of 1.44 (1.20-1.73); log rank p<0.0001)]. However, given that the enhanced expression was restricted to the HER2-enriched and Basal-like sub classes we performed the Kaplan-Meyer survival analysis on these subgroups of patients. High levels of ACE2 was not found to be prognostic within the basal-like class (Fig. 1.c) but only in the HER2-enriched class, where there was a significant higher probability of disease-free survival in the ACE2 low group (n = 123) compared to the ACE2 high group (n = 113) (Fig. 1.d) Clinico-pathological characteristics of HER2-enriched patients from METABRIC dataset used for disease-free survival analysis.
and multivariate Cox-proportional hazard analysis (Table 2). Interestingly, in the HER2-enriched subtype, we also observed that

Discussion
The RAAS system is known to play an important role in regulation of blood pressure and fluid balance in the human physiology. However, clearly ACE2 has independent and distinct cellular roles. Recent reports suggest a role for RAAS in tumor progression. Ang (1-7) has been reported to reduce migration and invasion in lung adenocarcinoma cell line and xenograft mice model systems through inhibition of pathways like the MAPK and the PI3K/Akt [17,18]. It is also known to reduce metastasis and angiogenesis in prostate cancer xenografts [19]. ACE2 has also been shown to be a tumor suppressor in various cancer types and an inhibitor of EMT. ACE2 downregulates VEGFa in breast cancer cells bringing down the process of neoangiogenesis and increased ACE2/ Ang (1-7)/ Mas axis decreases the ability of breast cancer cells to invade and metastasize in vivo and in vitro [12,20,21]. We are reporting for the first time that ACE2 might have a context dependent role in driving tumor progression in breast cancer.
Most gene-expression signatures, with small gene-sets, that have demonstrated utility in clinical prognostication, either include genes directly regulating proliferation or are strongly correlated to such genes. Both Basal-like and HER-2 enriched subtypes are known to have high grade, mitotically active tumors, with high KI-67 labeling index. However, in the tumors with high ACE2 expression we did not find any positive correlation with proliferation associated genes like BCl2, CCNE1, CCND1 and BIRC5. Neither were high ACE2 levels correlated with tumor size (data not shown). These data argue against any synergistic role for ACE2 in further enhancing rates of proliferation in these tumor types. But high ACE2 expression was positively associated with EMT associated genes like Integrin β1 and Vimentin. This observation coupled with the fact that metastatic lesions of both HER2-enriched and Basal-like breast cancer examined in the Swedish TEX trial continue to have high levels of ACE2 suggests, that in conjunction with over expressed receptor tyrosine kinase (RTK) growth factors, ACE2 a carboxy peptidase, capable of promoting internalization upon binding as demonstrated by SARS-Cov-2, might help promote signaling mechanisms that enhance metastatic behavior. The fact that ACE2 had a positive association with EGFR in all the datasets further supports the conjecture that its metastasis promoting role may only be in the context of growth factor RTKs. These observations need further validation using in-vitro model systems.
The results mirror the observation in a recent report on non-small cell lung cancer (NSCLC) cells where high ACE2 levels correlated with high levels of ERBB1, ERBB2 and ERBB3 and increased sensitivity to EGFR tyrosine kinase inhibitors (TKIs) [22]. High ACE2 levels were also found to be more in an EGFR mutant lung adenocarcinoma cell line model system [23]. EGFR inhibitors for treatment of breast cancer have been evaluated in several studies but results from clinical trials have been disappointing so far unlike in colorectal and lung cancer [24]. Hence identification of signaling mechanisms and biomarkers that work hand in hand with the EGFR pathway is very crucial. Another potential therapeutic application could be to use ACE2 inhibitors as a mode of targeted therapy in this class of breast tumors that over-express ACE2. Since there are monoclonal antibodies against ACE2 like H8R64 that are known get internalized by cells [23], it may be used as a conjugate with a targeting drug against HER2 or EGFR. Moreover, since centrally active ACE2 inhibitors like captopril and fosinopril can cross the blood brain barrier [25], they may be used in conjugation with HER2 targeting drugs for the treatment of brain metastasis; the most common site of metastasis for HER2 subtype of breast cancer. However, these conjectures need further in-vitro validations.