Increased expression levels of ACE2 after SARS-CoV-2 infection

ACE2, which is the molecular pathway through which SARS-CoV-2 enters host cells (Fig. 1a), showed significantly higher expression (Log2FC 0.22; p value 0.01) in blood samples of COVID-19 patients. This was determined by bioinformatics analyses of transcriptomic data obtained from 24 healthy controls and 62 COVID-19 patients (COVID19db ID: COVID000010) (Fig. 1b). The increased levels of ACE2 mRNA in blood samples from COVID-19 patients were similar to those observed in A549 lung carcinoma epithelial cells following ACE2 transfection and/or SARS-CoV-2 infection (Fig. 1c; GEO: GSE147507). Furthermore, increased ACE2 expression levels were also found in response to infections with other respiratory viruses (GEO: GSE147507), such as HPIV3 infection of A549 cells (Fig. 1d) and IAVdNS1 infection of normal human bronchial epithelial (NHBE) cells (Fig. 1e).

Fig. 1

High levels of ACE2 expression in blood samples from COVID-19 patients and SARS-CoV-2 infected cell lines compared to their uninfected counterparts. a ACE2 is an entry receptor for SARS-CoV-2 and a key molecule for understanding the pathophysiology of COVID-19. b ACE2 expression in blood samples from 24 healthy controls and 62 COVID-19 patients. The data were obtained from whole blood transcriptomic data (COVID19db). c, d ACE2 expression in A549 lung carcinoma epithelial cells transfected with an ACE2 expression vector or infected with SARS-CoV-2 (C) and infected with HPIV3 (D) (GEO: GSE147507). e Expression of ACE2 in normal human bronchial epithelial (NHBE) cells infected with IAVdNS1 (GEO: GSE147507). t.test, p > 0.05; *: p < = 0.05; **: p < = 0.01; ***: p < = 0.001; ****: p < = 0.0001

Identification and function of ACE2-regulated genes in MCF-7 BC cells

RNA-seq analysis of ACE2high and ACE2low MCF-7 cells revealed a total of 2801 differentially expressed genes (DEGs) (padj < 0.05) with 1445 significantly upregulated and 1356 significantly downregulated genes (Fig. 2a). Gene ontology (GO) enrichment analysis was performed using the DEGs to assess the functional categories of biological process (BP), molecular function (MF) and cellular component (CC). The top 20 significantly enriched GO terms of the upregulated genes in ACE2high and ACE2low MCF-7 cells include the categories defense response to other organism’ (ontology: BP; gene ratio 108/1174; p-value 2.57E-37), ‘cytokine activity’ (ontology: MF; gene ratio 46/1178; p-value 1.84E-16) and ‘MHC protein complex’ (ontology: CC; gene ratio 16/1214; p-value 5.09E-16) (Fig. 2b). The top 20 enriched GO terms of downregulated genes in ACE2low MCF-7 cells contain ‘sister chromatid segregation’ (ontology: BP; gene ratio 60/1176; p-value 7.62E-16), ‘structural constituent of ribosome’ (ontology: MF; gene ratio 41/1177; p-value 1.37E-11) and ‘chromosomal region’ (ontology: CC; gene ratio 80/1236; p-value 2.95E-19) (Supplementary Fig. 1). The top 10 upregulated genes by ACE2 were IFI6, IFIT1, IFIT2, IFIT3, OAS2, OASL, HLA-B, OAS1, DDX60 and CMPK2, the top 10 downregulated genes were SCD, ABCG1, SREBF1, FGFR4, PHGDH, FBXO27, PREX1, CRAT, AIF1L and PXMP4. Disease annotation of the top upregulated genes demonstrated a link to viral infections (Supplementary Table 2), most commonly to influenza (disease id: C0021400) (Supplementary Fig. 2A), while the downregulated genes were annotated to BC (malignant tumor of breast (disease id: C0006142) and breast carcinoma (disease id: C0678222) (Supplementary Fig. 2B). The ACE2-mediated differential expression profiles were independently confirmed for selected DEGs ACEhigh and ACElow cell systems by qPCR using DEG-specific primers (data not shown).

Fig. 2

Comparison of transcriptional profiles in ACE2high vs. ACE2low MCF-7 cells and COVID-19 PBMNCs using RNA Sequencing. a Volcano plot of the DEGs between ACE2high and ACE2low MCF-7 groups. Significantly down-regulated genes are marked in dark blue, significantly upregulated genes in red and non-significantly regulated genes in grey. b The top 20 enriched GO terms from upregulated genes of ACE2high vs. ACE2low MCF-7 cells. c Volcano plot of the DEGs between COVID-19 and healthy blood samples (COVID19db). d The top 10 enriched GO terms from upregulated genes between COVID-19 and healthy blood samples. The enriched GO terms commonly found in both ACE2high and COVID-19 were represented by green circles. e mRNA expression profiles of commonly upregulated genes in the top five GOs enrichments of ACE2high blood samples (COVID19db) from COVID-19 patients. The gene expressions of OASL, STAT1, and IRF1 are shown here as representative

Correlation of the GO terms and DEGs between ACE2high MCF-7 cells and SARS-CoV-2-infected PBMNCs

Comparison of the GO terms of the significantly upregulated genes in ACE2high vs. ACE2low MCF-7 cells with those in whole blood obtained from 62 COVID-19 patients and 24 healthy volunteers (COVID19db ID: COVID000010) demonstrated that 9/10 selected upregulated genes in ACE2high MCF-7 cells were expressed at higher levels in blood samples of COVID-19 patients as visualized by a volcano plot (Fig. 2c). Additionally, 8/10 GO terms were commonly enriched in both ACE2high MCF-7 cells and blood samples of COVID-19 patients, as indicated by green circles. These enriched GO terms included ‘response to virus’, ‘nucleosome’, and ‘type I interferon signaling pathway’ (Fig. 2d). As expected, the GO terms ‘neutrophil degranulation and neutrophil activation’ were only found in blood samples of COVID-19 patients, but not in ACE2high MCF-7 cells. The analysis further focused on the significantly upregulated genes within the top five GO terms, namely ‘defense response to other organism’, ‘response to type I IFN’, ‘defense response to virus’, ‘response to virus’, and ‘type I IFN signaling pathway’. Notably, 14 common genes upregulated in ACE2high MCF-7 cells within these top five GO enrichments (Supplementary Fig. 3) were also enhanced in blood samples from COVID-19 patients (COVID19db). These genes include OAS1 (Log2FC 1.78; p value 5.47E-09), OAS2 (Log2FC 1.34; p value 0), OAS3 (Log2FC 1.67; p value 0), OASL (Log2FC 1.87; p value 2.91E-10), STAT1 (Log2FC 0.93; p value 9.34E-08), IFITM3 (Log2FC 1.61; p value 5.94E-09), IRF1 (Log2FC 0.48; p value 3.38E-06), IRF2 (Log2FC 0.19; p value 0.01), IRF7 (Log2FC 1.32; p value 6.86E-07), IRF9 (Log2FC 0.54; p value 0.001), BST2 (Log2FC 0.68; p value 0), IFITM1 (Log2FC 1.19; p value 2.26E-08), IFITM2 (Log2FC 0.67; p value 4.91E-06) and NLRC5 (Log2FC 0.36; p value 0.002). All 14 genes upregulated in genes of ACE2high MCF-7 cells were statistically higher (p < 0.05) in COVID-19 blood samples compared to that of healthy controls (Fig. 2e). Additionally, the top 10 up- and downregulated genes of ACEhigh MCF-7 cells were compared to those of SARS-CoV-2-infected Calu3, A549 and NHBE cells (Supplementary Fig. 4A) as well as to infection with other respiratory viruses, such as IAV, IAVdNS1, HPIV3, and RSV (Supplementary Fig. 4B). Interestingly, except for IFIT2 in SARS-CoV-2-infected NHBE cells, the expression of the top 10 genes in ACE2high MCF-7 cells exhibited a similar increased trend upon viral infections. Among the top 10 downregulated genes, SREBF1, FGFR4, CRAT and PXMP4 showed a similar decrease following different viral infections. Hence, the global transcriptomic profile and functional annotations of ACE2high MCF-7 cells were mainly comparable to those of SARS-CoV-2-infected cells and COVID-19 patients.

Upregulation of HLA class I surface expression after ACE2 overexpression and SARS-CoV-2 infection

Since the impact of SARS-CoV-2 infection relevant molecules on the expression of immunemodulatory molecules has thoroughly not been analyzed, we first determined the effect of ACE2 overexpression in MCF-7, RKO, A549 and EA.Hy926 cells on the expression of HLA class I surface antigens using flow cytometry. As shown in Fig. 3a, ACE2 overexpression resulted in an increased HLA class I surface expression in all cell lines, but the levels highly varied. Notably, MCF-7 and EA.Hy926 cells exhibited higher HLA class I expression compared to A549 and RKO cells, which might be due to differences in the cancer types or other factors involved in the regulation of HLA class I antigens. This finding is consistent with high levels of HLA-B (Log2FC 0.2; p value 0.02) and -C (Log2FC 0.59; p value 1.17E-06) expression in blood samples of COVID-19 patients compared to healthy controls (Fig. 3b).

Fig. 3

Impact of ACE2 overexpression and respiratory viral infections on the expression of HLA class I. a The ACE2 transfectants and mock controls of EA.Hy926, A549, RKO and MCF-7 cells were analyzed for HLA-I surface expression by flow cytometry described in Material and Methods. The results are presented as a histogram and MFI of HLA-ABC (n = 3). b HLA-A, -B, and -C mRNA expression in blood samples of 62 COVID-19 patients vs. 24 healthy controls (COVID19db). c, d Increased HLA-A, -B and – C expression levels in SARS-CoV-2-infected cells (GEO accession: GSE147507), ACE2high A549 cells (c) and Calu3 (d)

These data were confirmed by RNA-seq results obtained from lung biopsies of COVID-19 patients (GEO: GSE1488290), which displayed a similar correlation with higher mRNA levels of HLA class I antigens (Supplementary Fig. 5). SARS-CoV-2 infection of ACE2high A549 cells (Fig. 3c) and infection with other respiratory viruses, such as IAV, IAVdNS1, HPIV3 and RSV, upregulated HLA class I antigens (Supplementary Fig. 6) when compared to the uninfected controls. HLA-B, but not HLA-A and HLA-C antigens were enhanced in SARS-CoV-2-infected Calu3 cells (Fig. 3d).

Association of the ACE2-mediated upregulation of HLA class I surface antigens with increased APM and IFN signaling component expression

In order to determine whether the ACE2-mediated increase of HLA class I surface expression was due to an enhanced expression of HLA class I APM components, the human ACE2high and ACE2low model systems were analyzed for the mRNA and protein expression of the major HLA class I APM molecules, such as the transporter associated with antigen processing (TAP)1, TAP2, TAPBP, β2-microglobulin (B2M), the IFN-γ inducible proteasome subunits, the low molecular weight proteins PSMB8, PSMB9 and PSMB10 as well as the chaperones calreticulin (CALR) and calnexin (CANX). With the exception of calnexin, calreticulin and tapasin, an ACE2-mediated upregulation of the mRNA expression of all other HLA class I APM components analyzed was detected (Fig. 4a).

Fig. 4

Increased APM and IFN signaling components in ACE2 transfectants and their impact on NK cell activity. a The ACE2 transfectants were analyzed for the expression of major HLA-I APM components using qPCR as described in Material and Methods. The results are presented as x-fold upregulation of APM components in ACE2 transfectants vs. mock controls (set = 1). b The ACE2 transfectants were analyzed by qPCR for the expression of type I and type II IFN signaling components. The results are represented as an x-fold induction of the expression of IFN signaling components in ACE2 transfectants compared to mock controls (set = 1). c A representative Western blot analysis of ACEhigh, mock transfected and potential cells using anti-TAP1 as loading control anti-IRF1 and ACE2 antibodies is shown, staining with an anti-GAPDH antibody served as loading control. d In silico analysis of TCGA data compared ACE2 expression to the expression of the APM component, TAP1 and IFN component, IRF1 in a pan-cancer dataset (11003 samples). e Reduced NK cell activity in ACE2high vs. ACE2low MCF-7 cells. CD107a degranulation assay was performed by co-culture with NK cells from three different donors with ACE2low vs. ACE2high MCF-7 cells as described in Material and Methods. The mean ± SE of the CD107a degranulation of ACE2low vs. ACE2high MCF-7 cells using NK cells representing total NK cell activity are shown

Despite SARS-CoV-2 infection has been reported to influence cytokine signaling, including the IFN signaling pathway [48], and IFN-γ has been shown to increase ACE2 surface expression [49], a possible link between ACE2 overexpression and IFN signaling in tumors has not yet been determined. Expression analyses of various IFN type I and II signaling components revealed a strong upregulation of the mRNA expression of IRF1, IRF9, JAK2, STAT1, STING and TYK2 in ACE2high transfectants compared to ACE2low mock controls (Fig. 4b). These data were confirmed by Western blot analyses as representatively shown for TAP1 and IRF1 with an increased TAP1 and IRF1 protein expression in ACE2high vs. ACE2low cells (Fig. 4c).

The possible link between ACE2 expression and immune response relevant profiles was also examined by in silico analyses of cancer genome databases. As shown in Fig. 4d and Supplementary Table 3, a positive correlation of ACE2 with the expression of components of the HLA class I APM as well as the IFN type I and II pathways in pan-cancer and breast cancer samples was found.

The functional relevance of the upregulation of HLA class I by ACE2 was determined by co-culture of ACE2low and ACE2high MCF-7 cells by a CD107a degranulation assay and the NK cell-mediated recognition was determined. As expected, a decreased cytotoxicity of NK cells of ACE2high MCF-7 cells compared to ACE2low MCF-7 cells was detected (Fig. 4e).

Correlation of ACE2 expression with the expression of the immune checkpoint molecule PD-L1 (CD274)

It was postulated that the ACE2-mediated upregulation of immune modulatory molecules might be associated with an increased response to immunotherapy [50], since the treatment of SARS-CoV-2-infected patients with ICPi might enhance anti-viral T cell responses by affecting PD-L1 expression [51]. Indeed, ACE2high MCF-7 cells expressed higher PD-L1 levels than the ACE2low control cells (Fig. 5a). This was accompanied by an increased expression of different IFN-γ signaling pathway components in ACE2high EA.Hy926, A549, RKO and MCF-7 cells and is in line with the IFN-γ-mediated upregulation of PD-L1. Comparable results were retrieved from in silico data of blood samples (Log2FC 1.19; p value 4.43E-06) (Fig. 5b) and lung biopsies from COVID-19 patients compared to their healthy counterparts (Supplementary Fig. 5) as well as from cells infected by SARS-CoV-2 (Fig. 5c) or other respiratory viruses (Fig. 5d).

Fig. 5

Increased PD-L1 expression of ACE2high cells and respiratory viral infections. a The ACE2 transfectants of EA.Hy926, A549, RKO and MCF-7 cells were analyzed for PD-L1 surface expression using flow cytometry (n = 3). The results are shown as MFI of x-fold regulation to mock-transfected cells. b PD-L1 mRNA levels in blood samples of COVID-19 patients (62 COVID-19 patients vs. 24 healthy controls) c, d The levels of PD-L1 mRNA were analyzed from RNA-seq data of SARS-CoV-2-infected Calu3 (c) and ACE2high A549 cells (d) and other respiratory viral infections on A549 (d) (GEO: GSE147507)

ACE2high, not ACE2low MCF-7 cells, increased immune cell migration and apoptosis upon nivolumab treatment

In the next step, immune cell migration, cancer cell proliferation and apoptosis was investigated over a period of three days upon co-culturing ACE2low/high MCF-7 cells with immune cells in the presence and absence of the anti-PD1 monoclonal antibody nivolumab. The PD-1 inhibitor induced significant immune cell infiltration towards ACE2high MCF-7 cells (Fig. 6a), which significantly increased over time. In contrast, the number of immune cells migrating towards cancer cells was minimal in ACE2mock/low MCF-7 and was influenced by nivolumab treatment (Fig. 6c). In addition, nivolumab treatment showed a tendency of increased apoptosis in ACE2high (Fig. 6b), but not of ACE2low MCF-7 cells (Fig. 6d). At the same time, nivolumab had neither an effect on the proliferation of ACE2high nor of ACE2low MCF7 cells (data not shown). The higher expression of HLA class I on ACE2high MCF-7 cells might be responsible for the increased migration and apoptosis, potentially leading to the activation of T cells, while the inhibitory effect of the increased PD-L1 expression in ACE2high MCF7 cells might be blocked by nivolumab (Fig. 7a).

Fig. 6

Nivolumab treatment induces migration of immune cells towards ACE2high cells and apoptosis of ACE2 transfected cancer cells. a, c Over a period of 3 days, immune cells migrated notably towards ACE2high MCF-7 cells treated with nivolumab (a), whereas the nivolumab treatment had no significant migration of immune cells towards ACE2low cells (c). The immunofluorescent images were taken on day 3, displaying a representative image. Cancer cells are depicted in red, lymphocytes in blue, and apoptotic cells in green in the fluorescence images. b, d Percentage of apoptotic cells of ACE2high/low MCF-7 cells. Nivolumab treatment of immune cells exhibited an increased apoptotic rate in ACE2high cells (b) than in the untreated control but not in the ACE2low cells (d). C - Cancer cells, CI - Cancer cells co-cultured with Immune cells, CIN – CI treated with nivolumab

Fig. 7

Analysis of cytokine expression in PBMNCs of COVID-19 patients and the impact of nivolumab treatment on cytokine release in ACE2high/low cells. a Nivolumab as programmed death-1 (PD-1) inhibitor for targeted immunotherapy of tumor cells to activate T cells. b Nivolumab treatment significantly increased IL2 and decreased CCL2 and IL-10 release on ACE2high MCF-7 cells compared to mock cells. c Data from scRNA-seq of PBMNCs from a cohort of 425,398 single cells (CZ CELLxGENE Discover). UMAP of various cell types, including plasmablasts, B cells, CD4+, CD8+, and γδ T cells, NK cells, conventional and plasmacytoid dendritic cells, classical and non-classical monocytes, and hematopoietic progenitor cells. d, e, f, g Higher mRNA expression levels of IFNG (d), IL2 (e), CCL2 (f) and IL10 (g) in different cell types of PBMNCs

Association of altered cytokine release with ACE2 and PD-L1 expression upon nivolumab treatment

Despite a cytokine storm has been widely reported to be caused by viral respiratory infections of influenza viruses and SARS-CoV-2 [52, 53], it has not been directly linked to immune cells. To investigate the impact of nivolumab on cytokine release in cell supernatants of ACE2high and ACE2low MCF-7 cells during co-culture with PBMNCs were analyzed with a human FirePlex®-96 key cytokine immunoassay panel, which consists of the 17 cytokines CSF2, IL1B, IL2, IL4, IL5, IL6, CXCL8, IL9, IL10, IL12A, IL13, IL17A, IFNG, CCL2, CCL3, CCL4 and TNF. As shown in Fig. 7b, treatment of ACE2high and ACE2low MCF-7 cells with nivolumab decreased the release of the innate immunity-related cytokine CCL2 and increased the secretion of the adaptive immunity-related cytokine IL-2. Furthermore, the secretion of the anti-inflammatory cytokine IL-10 known to inhibit MHC class I expression was significantly reduced following nivolumab treatment (Fig. 7b), while the other cytokines analyzed did not show statistically significant changes (Supplementary Table 4A).

Altered cytokine expression profile in peripheral blood cells and lung epithelium in COVID-19 patients

To assess the expression status of the differentially secreted cytokines upon ACE2 overexpression and treatment with nivolumab, single cell (sc) RNA-seq data from PBMNCs obtained from a cohort of 425,398 single cells from COVID-19 patients [54] were examined using CZ CELLxGENE Discover. Using the uniform manifold approximation and projection (UMAP), the expression of IFNG, IL6, IL2, IL10, and CCL2 was determined across various immune cell subtypes, including plasma blasts, B cells, CD4+, CD8+, γδ T cells, NK cells, conventional and plasmacytoid dendritic cells, classical and non-classical monocytes and hematopoietic progenitor cells (Fig. 7c). The UMAP revealed high IFNG mRNA levels in CD4+, CD8+ and γδ T cells as well as NK cells (Fig. 7d), while IL2 expression was found to be high in CD4+ T cells (Fig. 7e). In addition, CCL2 expression exhibited higher levels in classical monocytes and dendritic cells (Fig. 7f), while an upregulation of IL10 mRNA was detected in monocytes and CD4+ T cells of COVID-19 patients (Fig. 7g). Since the lung epithelium is a major target of the cytokine storm [44], a separate in silico analysis was performed on lung biopsies from severe COVID-19 patients (GEO: GSE147507) to assess the expression of these cytokines. The analysis revealed low IL2, but high CCL2 and IL10 mRNA expression (Supplementary Table 4B).