HMECs were purchased from Life Technologies (Carlsbad, CA, USA). MDA-MB231 and MCF7 cells were provided by Institut Hiscia (Arlesheim, Switzerland). Human embryonic lung fibroblasts (MRC5) were purchased from RD-Biotech (Besançon, France). HMECs were cultivated in HMEC medium (Life Technologies) supplemented with HMEC supplement and bovine pituitary extract (Life Technologies). MDA-MB231, MCF7, and MRC5 were maintained in Dulbecco’s modified Eagle medium (PAN- Biotech) supplemented with fetal bovine serum (Dutscher) and penicillin- streptomycin (Life Technologies). Cultures were free of mycoplasma (VenorGem classic mycoplasma detection, Minerva biolabs).

HMECs were infected with HCMV-DB (KT959235), HCMV-BL (MW980585), HCMV-TB40/E (KF297339), HCMV-FS, HCMV-KM, and HCMV-SC. The HCMV clinical isolates were previously isolated in our laboratory (20, 21), except for HCMV-TB40/E (29). Cell-free virus stocks of HCMV were grown in HMECs as described previously (21). Virus titers were determined through the quantification of HCMV load by real-time qPCR detection using KAPA SYBR FAST Master Mix, Omega Bio-TEK) and IE1 gene primers (sense, 5’-CGACGTTCCTGCAGACTATG-3’; anti-sense, 5’-TCCTCGGTCACTTGTTCAAA-3’) according to the manufacturer’s protocol. Results were collected and analyzed using MxPro qPCR software.

Colony formation in soft agar seeded with HMECs uninfected, infected with the six strains listed above, MCF7 cells and MDA-MB231 cells were assayed using Cell Biolabs Cytosolic Cell Transformation Assay kit (Colorimetric assay, CB135; Cell Biolabs Inc., San Diego, CA) as per the manufacturer’s protocol. The detection and quantification of colony formation was assessed by MTT assay (Cayman Chemical, Ann Arbor, MI). Colonies were observed under an Olympus microscope (Olympus Corporation, Tokyo, Japan)

Confocal microscopy was performed as described previously (21). Cells were washed using 1X PBS then fixed and permeabilized using BD Cytofix/Cytoperm, and finally stained with anti-IE1 (pp72) antibody (ab53495, Abcam). To visualize the nucleus, DAPI (4′, 6′-diamidino-2- phenylindole) staining was performed according to the manufacturer’s protocol. Post-staining, confocal images were taken using 63X oil immersion objective lens with a LSM800 Carl-Zeiss confocal microscope 95 (Germany). The Ki67 immunohistochemistry (IHC) evaluation on breast biopsies was assessed using the monoclonal antibody MIB1 (Invitrogen).

Uninfected HMECs and HCMV-infected HMECs were fixed and permeabilized for 20 minutes at 4°C using 100 μl Cytofix/Cytoperm solution (BD Biosciences), washed twice with 1X PBS supplemented with 3% FBS and 0.1% Triton-X, and then re-suspended in 100ul of staining buffer (1X PBS, 3% FBS, and 0.1% Triton-X). Cell staining was done using the anti-IE1 (CMV-pp72) antibody conjugated to PE (sc-69834, Santa Cruz Biotechnology). HMECs were incubated in the dark for 60 minutes at 4°C. Cells were washed twice with 1X PBS supplemented with 3% FBS and 0.1% Triton-X and subjected to cytofluorometric analysis on a BD LSRFortessa X-20 (BD Biosciences) flow cytometer. Data obtained was analyzed and processed using FACSDiva software (BD Biosciences).

Phylogenetic analysis was performed on several HCMV strains [BL (accession number: MW980585); DB (accession number: KT959235); TB40/E (accession number: KF297339); and Davis (accession number: JX512198.1)]. Multi-sequence alignments (MSA) were implemented using CLUSTAL W. Phylogenetic tree was constructed by means of the neighbor-joining method. The conducted analysis was done using MEGA7 software (http://www.megasoftware.net/).

Total RNA was extracted from uninfected HMECs and HMECs infected with HCMV-BL, HCMV-TB40/E, HCMV-FS, HCMV-KM and HCMV-SC at MOI of 1 at day 1 post-infection using an RNA extraction kit (EZNA Total kit I, Omega BIO-TEK). Then, the cDNA was synthesized using the RT² First Strand Kit (Qiagen, Germantown, MD, USA) following the manufacturer’s instructions. Two RT² Profiler™ PCR Arrays for human breast cancer (PAHS-131ZA) and human oncogenes & tumor suppressor genes (PAHS-502ZR) (both from Qiagen, Germantown, MD, USA) were performed using Mx3005P real-time PCR system as per manufacturer’s instructions. These microarrays allow the quantification of mRNA from 84 cellular genes, using qPCR with premixed primer sets and SybrGreen qPCR MasterMix. DNA contamination control, housekeeping genes, reverse transcription control, and positive control were included in each PCR array. Data were analyzed using the manufacturer web-based analysis software (http://pcrdataanalysis.sabiosciencescom/pcr/arrayanalysis.php). The analysis was done based on three independent biological experiments for each of the HCMV strains.

Mammosphere formation by uninfected HMECs and HMECs infected with HCMV-BL, HCMV-TB40/E, HCMV-FS, and HCMV-KM at MOI of 1 was assayed as described previously (19). MCF7 and MDA-MB231 were used as positive controls. Mammospheres larger than 60 microns were counted.

Cellular extracts of uninfected and infected HMECs with HCMV-DB and HCMV-TB40/E were prepared at 1 hour, 5 hours, days 1 and 3 post-infection and used to assess the expression of Rb, pRb, Akt, pAkt-Th308, pAkt-Ser473, STAT3, pSTAT3, CyclinD1, and ATM as described previously (20). Protein levels were quantified using ImageJ 1.40 software (National Institutes of Health, Bethesda, MA, USA). β-actin was used as loading control to normalize sample loading. Anti-Rb, anti-pRb, anti-Akt, anti-pAktThr308, anti-pAktSer473, anti-cyclin D1, and anti-ATM antibodies were purchased from Cell signaling (Danvers, MA, USA). Anti-pSTAT3 and anti-STAT3 antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-β-actin antibody was purchased from Sigma-Aldrich (St. Louis, MO, USA).

Nineteen frozen tumor breast biopsies (ten luminal tumor biopsies and nine basal tumor biopsies) and eight adjacent healthy breast biopsies were provided by the regional tumor bank (BB0033-00024 Tumorothèque Régionale de Franche-Comté). Healthy breast biopsies were numbered from one to eight, luminal tumor biopsies from nine to eighteen, and basal tumor biopsies from nineteen to twenty-seven. A detailed histopathological biopsy data including the histologic biopsy type, Elston-Ellis grading system, hormone percentages (estrogen and progesterone), presence or absence of the human epidermal growth factor receptor 2 (HER2) protein, vascular emboli, and TNM staging was provided as a supplementary data ( Supplementary Table 1 ). Additionally, using twenty different breast tumor biopsies, we assessed the Ki67 mRNA and Ki67 protein levels measured by RT-qPCR and IHC, respectively, on the same biopsy sample. A written informed consent for participation was obtained from all patients. The study was authorized by the local ethics committees of Besançon University Hospital (Besançon, France) and the French Research Ministry (AC-2015-2496, CNIL n°1173545, NF-S-138 96900 n°F2015). The presence of HCMV, within all the biopsies, was assessed by qPCR using IE1 and UL69 primers as mentioned above and described previously (30). For Ki67 and GAPDH mRNA quantification, RNA was extracted from the biopsies using E.Z.N.A.^® Total RNA Kit I (Omega BIO-TEK, GA, USA). Reverse transcription was performed using the SuperScript IV First-Strand Synthesis kit (Invitrogen, Carlsbad, CA, USA). The expression of Ki67 and GAPDH was assessed by real-time qPCR using KAPA SYBR FAST Master mix (KAPA BIOSYSTEMS, KK4601) and specific primer for Ki67 (sense, 5’TCCTTTGGTGGGCACCTAA GACCTG3’; anti-sense, 5’TGATGGTTGAGGTCGTTCCTTG ATG3’) and GAPDH (sense, 5’CCCCTCTTCAAGGCCTC TAC3’; anti-sense, 5’CGACCACTTTGTCAAGCTCA3’) according to the manufacturer’s protocol. The fold change expression was calculated using the delta-delta Ct method (31). Biopsies having a fold change for Ki67 gene expression <30, from 30 to 100, and >100 were considered to have a low, intermediate, and high Ki67 mRNA absolute level respectively. The Ki67 mRNA cut-off was set based on the state of healthy biopsies (32).

Statistical analysis was performed using the Wilcoxon test with the help of the Mathematics Department of the University of Franche-Comté. p-value ≤ 0.05 were considered statistically significant. In data analysis, Microsoft Excel was used to prepare the plots with standard deviation and p-values. Statistical significance was represented on figures by stars, with the following code *p-value<0.05; **p-value<0.01; ***p-value<0.001.

A phylogenetic analysis was performed to analyse the genomic sequences of four HCMV strains, namely HCMV-DB, HCMV-BL, HCMV-TB40/E, and HCMV-Davis. Since the clinical isolates FS, KM and SC genomes were not sequenced, we selected the strain Davis for its similar clinical characteristics (33). The phylogenetic analysis showed that HCMV-BL is close to HCMV-DB, and HCMV-TB40/E is close to the genomes of HCMV-Davis ( Supplementary Figure 1 ). The two observable brackets illustrate a high-risk group including DB and BL strains, and a low-risk group including TB40/E and Davis strains ( Supplementary Figure 1 ).

We assessed the replication of six HCMV clinical isolates in HMECs, namely the HCMV-BL and HCMV-DB high-risk strains, the HCMV-FS, HCMV-KM and HCMV-SC low-risk strains, and the prototypic TB40/E strain (20, 21, 29). We evaluated the expression of the immediate early protein IE1 by immunofluorescence and flow cytometry. We observed strong IE1 expression in HMECs infected with the five strains (HCMV-DB, BL, TB40/E, FS and KM) at day 1 post-infection compared to the uninfected HMECs by confocal microscopy ( Figure 1A ). Moreover, the expression of IE1 protein in HMECs infected with the different HCMV strains was confirmed by flow cytometric analysis ( Figure 1B ). Further, we noticed productive replication of all six HCMV strains in infected HMECs as assayed by IE1 detection in culture supernatants using qPCR ( Figure 1C and Supplementary Figure 2A ). An early and transient peak of viral replication was observed at day 1 post-infection for HMECs infected with HCMV-DB, HCMV-BL, and HCMV-TB40/E unlike the delayed peaks at day 3 and 5 post-infection for HMECs infected with HCMV-KM and HCMV-FS respectively. To further assess the oncogenic status of all six HCMV strains, we performed a soft agar assay. HMECs infected with the six HCMV strains were seeded in soft agar at day 1 post-infection. Cancer cell lines, MCF7 and MDA-MB231, were used as positive controls. Uninfected HMECs were used as negative control. Colony formation was detected at day 15 post-infection only with HCMV-DB and HCMV-BL compared to the uninfected HMECs (p-value _{(HCMV-DB : UI HMEC)} =0.03, p-value _{(HCMV-BL : UI HMEC)} =0.03, p-value _{(HCMV-DB : HCMV-TB40/E)} =0.03; p-value _{(HCMV-DB : HCMV-FS)} =0.03; p-value _{(HCMV-DB : HCMV-KM)} =0.03; p-value _{(HCMV-BL : HCMV-TB40/E)} =0.03; p-value _{(HCMV-BL : HCMV-FS)} =0.03; p-value _{(HCMV-BL : HCMV-KM)} =0.03). In contrast, no significant colony formation was observed with HCMV-FS, HCMV-KM, HCMV-SC, and HCMV-TB40/E infected HMECs ( Figure 1D and Supplementary Figure 2B ). Our results indicate that HMECs infected with the high-risk strains BL and DB have an anchorage-independent growth ability.

We decided to compare the transcriptome profile of the high-risk BL strain, the low-risk FS, KM and SC strains, and the prototypic TB40/E strain. Human breast cancer along with human oncogenes and tumor suppressor genes RT² profiler PCR arrays were used to screen the infected HMECs with the four strains at MOI of 1 at day 1 post-infection and the uninfected HMECs. The expression of genes that are involved in signal transduction, cell cycle, cell survival, adhesion, angiogenesis, DNA repair, were examined by the RT² Profiler PCR arrays ( Supplementary Tables 2 , 3 ).

The individual gene expression of the oncogenes Myc (MYC), Fos (FOS), Jun (JUN), KRas (KRAS), and E2F1 was assessed in HMECs infected with the five strains HCMV-BL, TB40/E, FS, KM and SC compared to uninfected HMECs ( Figures 2A , 3A and Supplementary Figure 2C ). Values for individual genes were pooled to constitute the overall expression of the oncogenes group ( Figure 2A ). In HMECs infected with HCMV-BL, the expression of the oncogenes group was 5.8-fold higher compared to HMECs infected with the TB40/E strain (p-value _{(HCMV-BL: HCMV-TB40/E)} = 0.02), 8.5-fold higher compared to HMECs infected with FS strain (p-value _{(HCMV-BL: HCMV-FS)} = 0.004) ( Figure 2A ). However, the expression of these oncogenes was downregulated in HMECs infected with the KM strain (p-value _{(HCMV-BL: HCMV-KM)} = 0.0001) ( Figure 2A ). Among the oncogenes, Myc and E2F1 were mostly increased in HMECs infected with HCMV-BL compared to HCMV-TB40/E (p-value _{(Myc- HCMV-BL: HCMV-TB40/E)} = 0.12; p-value _{(E2F1- HCMV-BL: HCMV-TB40/E)}= 0.04), HCMV-FS (p-value _{(Myc- HCMV-BL: HCMV-FS)} = 0.12; p-value _{(E2F1- HCMV-BL: HCMV-FS)} = 0.27) and HCMV-KM (p-value _{(Myc- HCMV-BL: HCMV-KM)} = 0.04; p-value _{(E2F1- HCMV-BL: HCMV-KM)} = 0.12) infected HMECs ( Figure 3A ). Furthermore, the expression of other oncogenes was tested for HMECs infected with the high-risk strain BL and the low-risk strain FS versus uninfected HMECs. We observed an upregulation of the transcripts of numerous other oncogenes such as KITLG, MYB, RARA, ROS1, and RUNX1 in HMECs infected with HCMV-BL and HCMV-FS ( Table 1 ). The upregulation of KITLG, RARA, and ROS1 genes expression was at least 2-fold higher in HMECs infected with HCMV-BL compared to HMECs infected with HCMV-FS ( Table 1 ). In HMECs infected with HCMV-SC, the oncogenes’ expression (MYC, FOS, KRAS, JUN, E2F1) was low, likewise in HMECs infected with HCMV-FS ( Supplementary Figure 2C ).

The individual gene expression of the tumor suppressor genes coding for p53 protein (TP53), retinoblastoma protein (Rb), p73 protein (TP73), SMAD4, VHL, TSC1, and MDM2 was also assessed in the previously mentioned five strains compared to uninfected HMECs ( Figures 2B , 3B and Supplementary Figure 2C ). Values for individual genes were pooled to constitute the overall expression of the tumor suppressor genes group ( Figure 2B ). The upregulation of tumor suppressor genes’ expression was higher in HMECs infected with the BL strain compared to HMECs infected with the TB40/E and FS by 2 (p-value _{(HCMV-BL: HCMV-TB40/E)} = 0.02) and 3.8 fold (p-value _{(HCMV-BL: HCMV-FS)} = 0.002) respectively ( Figures 2B ). Nevertheless, the expression of these tumor suppressor genes was downregulated in HMECs infected with HCMV-KM (p-value _{(HCMV-BL: HCMV-KM)} = 0.00001) ( Figure 2B ). Among the tumor suppressor genes, TP53, VHL, Rb, and TSC1 expression was mainly increased in HMECs infected with HCMV-BL compared to HCMV-TB40/E (p-value _{(TP53- HCMV-BL: HCMV-TB40/E)} = 0.04; p-value _{(SMAD4- HCMV-BL: HCMV-TB40/E)} = 0.82; p-value _{(VHL- HCMV-BL: HCMV-TB40/E)} = 0.27; p-value _{(Rb- HCMV-BL: HCMV-TB40/E)} = 0.04; p-value _{(TSC1- HCMV-BL: HCMV-TB40/E)} =0.51), HCMV-FS (p-value _{(TP53- HCMV-BL: HCMV-FS)} = 0.04; p-value _{(SMAD4- HCMV-BL: HCMV-FS)} = 0.12; p-value _{(VHL- HCMV-BL: HCMV-FS)} = 0.51; p-value _{(Rb- HCMV-BL: HCMV-FS)}= 0.04; p-value _{(TSC1- HCMV-BL: HCMV-FS)} =0.51), and HCMV-KM (p-value _{(TP53- HCMV-BL: HCMV-KM)} = 0.04; p-value _{(SMAD4- HCMV-BL: HCMV-KM)}= 0.04; p-value _{(VHL- HCMV-BL: HCMV-KM)} = 0.12; p-value _{(Rb- HCMV-BL: HCMV-KM)}= 0.04; p-value _{(TSC1- HCMV-BL: HCMV-KM)} =0.12) ( Figure 3B ). In HMECs infected with HCMV-SC, the tumor suppressor genes (TP53, VHL, TSC1) were slightly expressed, similarly in HMECs infected with HCMV-FS ( Supplementary Figure 2C ). The expression of TP73, a tumor suppressor gene, was markedly increased in HMECs infected with low-risk HCMV strains (KM and SC) and to a lower extent with FS and TB40/E strains compared to HMECs infected with high-risk BL strain ( Figure 3B and Supplementary Figure 2C ).

We then analyzed the individual expression of genes involved in cell proliferation namely STAT3, cyclin D1 (CCND1), and Ki67 antigen (MKI67). Interestingly, the STAT3/Cyclin D1 axis is activated in several cancer types including breast cancer (34). Values for individual genes were pooled to constitute the overall expression of the genes involved in cell proliferation group ( Figure 2C ). The expression of “proliferation” genes all together was 2.81 times more upregulated in the HMECs infected with the BL strain compared to HMECs infected with the TB40/E strain (p-value _{(HCMV-BL: HCMV-TB40/E)} = 0.05) ( Figure 2C ). While the expression of “proliferation” genes was almost unchanged in HMECs infected with HCMV-FS compared to uninfected HMECs (p-value _{(HCMV-BL: HCMV-FS)} = 0.005), it was downregulated in HMECs infected with HCMV-KM (p-value _{(HCMV-BL: HCMV-KM)} = 0.0007) ( Figure 2C ). The STAT3 gene expression was up-regulated in HMECs infected with HCMV-BL in contrast to HMECs infected with HCMV-TB40/E (p-value _{(HCMV-BL: HCMV-TB40/E)} = 0.04) and HCMV-FS (p-value _{(HCMV-BL: HCMV-FS)} = 0.04) noting that it was substantially down-regulated in HMECs infected with HCMV-KM (p-value _{(HCMV-BL: HCMV-KM)} = 0.04) ( Figure 3C ). The gene expression of cyclin D1 was up-regulated in HMECs infected with HCMV-BL compared to uninfected controls. In HMECs infected with HCMV-BL, cyclin D1 gene expression was 2 times more increased when compared to HMECs infected with HCMV-TB40/E (p-value _{(HCMV-BL: HCMV-TB40/E)} = 0.12). In contrast, the expression of cyclin D1 was almost unchanged in HMECs infected with HCMV- FS compared to uninfected HMECs (p-value _{(HCMV-BL: HCMV-FS)} = 0.12), and down-regulated with in HMECs infected with HCMV-KM (p-value _{(HCMV-BL: HCMV-KM)} = 0.04) ( Figure 3C ). In agreement with the highest expression of cyclin D1 in HMECs infected with HCMV-BL and the lowest expression in HMECs infected with HCMV-KM strain, we observed the highest expression of the proliferation marker Ki67 gene (MKI67) in the former and very low expression in the latter (3-fold versus 0.6-fold increase compared to uninfected cells) ( Figure 3C ). HMECs infected with HCMV-SC showed a low expression of the genes involved in proliferation (CCDN1, MKI67) likewise in HMECs infected with the low-risk strains ( Supplementary Figure 2C ). We also assessed the expression of CDK inhibitors CDKN1A, CDKN2A, CDKN2B and CDKN3. Values for individual genes were pooled to constitute the overall expression of CDK inhibitors group ( Figure 2D ). The expression of CDK inhibitors genes was slightly decreased in HMECs infected with HCMV-BL compared to the other three strains TB40/E (p-value _{(HCMV-BL: HCMV-TB40/E)} =0.81), FS (p-value _{(HCMV-BL: HCMV-FS)} =0.02) and KM (p-value _{(HCMV-BL: HCMV-KM)} = 0.001) ( Figures 2D , 3D ). HMECs infected with HCMV-SC showed a minimal expression of the genes coding for the cyclin dependent kinase inhibitors ( Supplementary Figure 2C ).

The expression of genes involved in cell survival, such as AKT, was upregulated in HMECs infected with BL, and TB40/E strains unlike in HMECs infected with FS and KM strains (p-value _{(HCMV-BL: HCMV-FS)} = 0.12 and p-value _{(HCMV-BL: HCMV-KM)} =0.04) ( Figure 4A ). In addition, REL was upregulated more in HCMV-BL than in HCMV-TB40/E (p-value _{(HCMV-BL: HCMV-TB40/E)} =0.04) and HCMV-FS (p-value _{(HCMV-BL: HCMV-FS)}=0.04) and was downregulated in HCMV-KM (p-value _{(HCMV-BL: HCMV-KM)} =0.04) ( Figure 4A ). The expression of genes involved in DNA reparation (ATM and MLH) was downregulated in HMECs infected with HCMV-TB40/E, FS and KM compared to HMEC infected with HCMV-BL (p-value _{(ATM-HCMV-BL: HCMV-KM)} =0.04 and p-value _{(ATM-HCMV-BL: HCMV-FS)} =0.04) ( Figure 4B ). The gene expression of the molecules involved in cell adhesion such as E-cadherin (CDH1) and beta-catenin (CTNNB1) was minimal in HMECs infected with the four strains (BL, TB40/E, FS and KM) noting that it was mostly downregulated in HCMV-KM (p-value _{(CDH1-HCMV-BL: HCMV-KM)} =0.04 and (p-value _{(CTNNB1-HCMV-BL: HCMV-KM)} =0.04) ( Figure 4C ). The genes involved in angiogenesis regulation S100A4 and EGF were highly expressed in HMECs infected with HCMV-BL compared to HMECs infected with HCMV-FS and HCMV-KM (p-value _{(S100A4-HCMV-BL: HCMV-KM)} =0.1; p-value _{(S100A4-HCMV-BL: HCMV-FS)} =0.5; p-value _{(S100A4-HCMV-BL: HCMV-TB40/E)} =0.1; p-value _{(EGF-HCMV-BL: HCMV-KM)} =0.2; p-value _{(EGF-HCMV-BL: HCMV-FS)} =0.05 p-value _{(EGF-HCMV-BL: HCMV-TB40/E)} =0.5) ( Figure 4D ). Furthermore, the expression of genes involved in proteolysis including matrix metallopeptidase 9 (MMP9) and cathepsin D (CTSD) was 7-fold and 2.4-fold higher respectively, in HMECs infected with HCMV-BL compared to HMECs infected with HCMV-FS ( Supplementary Table 4 ). The expression of the EMT markers for instance the transforming growth factor-beta 1 (TGFB1) and the proto-oncogene tyrosine-protein kinase (SRC) was enhanced in HMECs infected with HCMV-BL compared to cells infected with HCMV-TB40/E (p-value _{(TGFB1- HCMV-BL: HCMV-TB40/E)} = 0.04; p-value _{(SRC- HCMV-BL: HCMV-TB40/E)} = 0.27), FS (p-value _{(TGFB1- HCMV-BL: HCMV-FS)} = 0.12; p-value _{(SRC- HCMV-BL: HCMV-FS)}= 0.12) and KM (p-value _{(TGFB1- HCMV-BL: HCMV-KM)} = 0.04; p-value _{(SRC- HCMV-BL: HCMV-KM)}= 0.04) ( Figure 4E ). In HMECs infected with HCMV-SC, the genes involved in cell survival (AKT, REL), DNA reparation (MLH), and angiogenesis (S100A4) as well as the EMT marker (SRC) were regulated in a similar manner compared to HMECs infected with the FS strain ( Supplementary Figure 2C ).

We previously observed that the high-risk strains (HCMV-BL and DB) favor the appearance of stemness in chronically infected HMECs and CMV-transformed HMECs (CTH cells) (21), further suggesting that HCMV strains that trigger cellular transformation could also promote stemness. Therefore, HMECs were infected with the high-risk strain HCMV-BL, the low-risk strains HCMV-FS and HCMV-KM, and the prototypic HCMV-TB40/E. At day 1 post-infection, mammosphere formation assay was performed as previously described (35, 36). When we challenged these HCMV-infected cultures to form mammospheres, we detected numerous mammospheres in HCMV-BL infected HMECs as previously reported for HCMV-DB infected cells (p-value _{(HCMV-BL: HMEC UI)} =0.03) (37), whereas HCMV-TB40/E displayed a limited number of mammospheres (p-value _{(HCMV-TB40/E: HMEC UI)} = 0.07); only few mammospheres were observed in HMECs infected with HCMV-FS and HCMV-KM ( Figure 5 ). The breast cancer cell lines MCF7 and MDA-MB231 generated mammospheres as previously reported (37) ( Figure 5 ). Moreover, few mammospheres were observed in uninfected HMECs as previously described (38) ( Figure 5 ). Altogether, we noticed a preferential formation of mammospheres in HMECs infected with the high-risk HCMV-BL strain in opposition to the low-risk HCMV-TB40E (p-value _{(HCMV-BL: HCMV-TB40/E)} = 0.03), HCMV-FS (p-value _{(HCMV-BL: HCMV-FS)} = 0.03) and HCMV-KM (p-value _{(HCMV-BL: HCMV-KM)} = 0.03) strains.

Since we observed limited mammospheres in HMECs infected with HCMV-TB40/E ( Figure 5 ) and a slight increase in telomerase activity which has been previously reported (20), we decided to compare side by side the activation of oncogenic molecular pathways in HMECs infected with HCMV-TB40/E and the high-risk HCMV-DB strain. The productive HCMV-DB and TB40/E infection was confirmed in HMECs where both confocal microscopy imaging and flow cytometry data showed the IE1 staining ( Figures 1A, B ). At the protein expression level (day 3 post-infection), Rb was diminished whereas pRb was highly expressed in HMEC-HCMV-DB (p-value _{(HCMV-DB: HMEC-UI)} = 0.04) and to a lesser extent in HMEC-HCMV-TB40/E (p-value _{(HCMV-TB40/E: HMEC-UI)} = 0.04, p-value _{(HCMV-TB40/E: HCMV-DB)} = 0.8) ( Figures 6A, B ). At the same time point, pSTAT3 expression was higher in HMEC-HCMV-DB (p-value _{(HCMV-DB: HMEC-UI)} = 0.03) compared to HMEC-HCMV-TB40/E (p-value _{(HCMV-TB40/E: HMEC-UI)} = 0.06) and STAT3 expression was further increased in HMEC-HCMV-DB compared to HMEC-HCMV-TB40/E ( Figures 6A, C ). Further, Akt expression was highly increased in HMEC-HCMV-DB (p-value _{(HCMV-DB: HMEC-UI)} = 0.03), compared to HMEC-HCMV-TB40/E at day 3 post-infection (p-value _{(HCMV-TB40/E: HMEC-UI)} = 0.06), and an increase in its phosphorylation on both serine 473 and threonine 308 in HMEC-HCMV-DB was observed ( Figures 6A, D ). At day 1 post-infection, cyclin D1 expression was increased in HEMC-HCMV-DB (p-value _{(HCMV-DB: HMEC-UI)} = 0.03) compared to HMEC-HCMV-TB40/E (p-value _{(HCMV-TB40/E: HMEC-UI)} = 0.06) ( Figures 6A, E ). The protein expression of ATM was elevated in HMECs infected with HCMV-DB in comparison to HCMV-TB40/E ( Figures 6A, F ). Our results indicate that the HCMV-TB40/E strain does not induce significant activation of the molecular oncogenic pathways, especially when compared to the high-risk strain HCMV-DB, and therefore should be classified as a low-risk HCMV strain consistent with the absence of colony formation in soft agar in HMEC cultures seeded with TB40/E.

To assess the impact of HCMV strains on cellular gene expression in vivo, Ki67 expression was quantified by RT-qPCR in breast tumor biopsies, including luminal and basal-like tumors compared to healthy breast tissues. First, we measured the percentage of healthy and tumor biopsies with low (<30), intermediate (30–100) or high (>100) Ki67 mRNA expression ( Figures 7A, B ). Basal-like biopsies accounted for the highest percentage of biopsies with high Ki67 mRNA expression level (% of basal-like: 44%) compared to healthy (2%) and luminal biopsies (20%) (p-value _{(High Ki67 - Basal : Healthy)} = 0.02) ( Figure 7A ). To further study the role of HCMV associated with Ki67 transcript expression, we assessed the presence of HCMV in the biopsies by quantifying IE1 and UL69 genes using qPCR (21, 30). In HCMV-positive biopsies, basal-like biopsies showed the highest percentage of biopsies with high levels of Ki67 mRNA (40%) in contrast to the healthy (0%) and luminal (0%) biopsies ( Figure 7B ). Second, we assessed the absolute level of Ki67 mRNA per biopsy of healthy tissues, luminal and basal-like tumors taking into account the presence or absence of HCMV ( Figure 7C ). Healthy biopsies showed low absolute levels of Ki67 mRNA compared to tumor biopsies independent of HCMV status (presence or absence) (p-value = 0.02). In tumor biopsies, the absolute level of Ki67 mRNA was enhanced by 2.5-fold in HCMV-positive biopsies compared to HCMV-negative biopsies (p-value = 0.2). In HCMV-positive biopsies, the absolute level of Ki67 mRNA in basal tumors was enhanced by 31.45-fold compared to luminal tumors (p-value = 0.17) ( Figure 7C ). The absolute level of Ki67 mRNA was shown separately for each biopsy where cutoff lines were set to differentiate between low and high Ki67 expression and allow a closer look on the variability of Ki67 mRNA expression per sample ( Supplementary Figure 3 ). Thus, when we independently measured the absolute level of Ki67 mRNA in each HCMV-positive biopsy, we detected two basal biopsies (biopsies #20 and #21) showing the highest absolute levels of Ki67 mRNA with 315 and 4015-fold increase versus the mean value of healthy tissue, respectively ( Figure 7D ). Meanwhile, much lower absolute levels of Ki67 mRNA were detected in the other HCMV-positive tumor and healthy biopsies. Low absolute levels of Ki67 mRNA were detected in HCMV-positive luminal biopsies (less than 30-fold increase versus mean value of healthy tissue) except for one biopsy (biopsy #15 with 108-fold increase) ( Figure 7D ).

Within our data, we were able to identify two biopsies, namely biopsies #20 and #21 ( Figure 7D , arrows), that could harbor high-risk HCMV strains. Given the total number of biopsies, we might not be able to properly identify all the high-risk strains, which effect might be partially masked in the overall HCMV-positive biopsies. To confirm the high-risk status of biopsies #20 and #21, we isolated the strains for each biopsy and used it to infect HMECs ( Supplementary Figure 4A ). After few weeks in culture, we observed dense structures ( Supplementary Figure 4B ), similar to the morphology of the previously reported CTH cells (20, 21). These emerging structures were not present in HMECs infected with low-risk strains that were isolated from other biopsies in which we observed cell lysis. Overall, we identified two high-risk strains (n=2; 15%) out of the HCMV-positive biopsies (n=13); upon assessing the proportion of high-risk strain within tumor biopsies only, we identified two high-risk strains (n=2; 20%) out of the total HCMV-positive tumors (n=10).

Finally, using twenty different breast tumor biopsies, we aimed at evaluating the value of Ki67 expression as an index for increased proliferation in breast cancer. Ki67 mRNA and protein levels were measured by RT-qPCR and IHC, respectively, on the same biopsy sample. A correlation was observed between Ki67 protein and mRNA levels ( Supplementary Figure 5 ). This data indicates that Ki67 mRNA expression could be considered as a reliable marker for the measurement of increased proliferation in breast cancer.