Progesterone Receptor Attenuates STAT1-Mediated Interferon Signaling in Breast Cancer

Abstract

Why some tumors remain indolent and others progress to clinical relevance remains a major unanswered question in cancer biology. Interferon signaling in nascent tumors, mediated by STAT1, is a critical step through which the surveilling immune system can recognize and destroy developing tumors. Herein, we have identified an interaction between the progesterone receptor (PR) and STAT1 in breast cancer cells. This interaction inhibited efficient interferon-induced STAT1 phosphorylation, as we observed a decrease in phospho-STAT1 in response to interferon treatment in PR-positive breast cancer cell lines. This phenotype was further potentiated in the presence of PR ligand. In human breast cancer samples, PR-positive tumors exhibited lower levels of phospho-STAT1 as compared to their PR-negative counterparts, indicating that this phenotype translates to human tumors. Breast cancer cells lacking PR exhibited higher levels of interferon-stimulated gene (ISG) RNA, the transcriptional endpoint of interferon activation, indicating that unliganded PR alone could decrease transcription of ISGs. Moreover, the absence of PR led to increased recruitment of STAT1, STAT2 and IRF9 (key transcription factors necessary for ISG transcription) to ISG promoters. These data indicate that PR, both in the presence and absence of ligand, attenuates interferon-induced STAT1 signaling, culminating in significantly abrogated activation of genes transcribed in response to interferons. PR-positive tumors may use downregulation of STAT1-mediated interferon signaling to escape immune surveillance, leading to the development of clinically relevant tumors. Selective immune evasion of PR-positive tumors may be one explanation as to why over 65% of breast cancers are PR-positive at the time of diagnosis.

Introduction

Tumor escape from the surveilling immune system is a critical first step in tumor development, and essential for tumor progression (1). The cancer immunoediting hypothesis postulated by Schreiber et al highlights that the innate and adaptive immune responses work together to “flag” early neoplastic lesions for immune-mediated elimination (2). An early mediator of this elimination process is activation of type I interferon signaling (3–5). Thus, suppression of interferon signaling may help developing tumors evade the critical early steps of immune recognition and clearance.

Interferon signaling encompasses the mechanism through which interferons (IFNs, type I-III) are made and released by the host cell in response to a viral infection. Interferons (IFNs) are cytokines that bind their corresponding cell surface receptor, allowing the JAK-STAT (Janus Activated Kinases-Signal Transducer and Activator of Transcription) signaling cascade to initiate. Type I IFNs, such as IFNα/β, bind their cognate IFN-α/β receptor (IFNAR) complex, which is associated with tyrosine kinases. The activation of both JAK1 (Janus kinase 1) and TYK2 (Tyrosine Kinase 2) results in the downstream phosphorylation of both STAT1 and STAT2. The phosphorylated STAT1-STAT2 heterodimer associates with a third transcription factor, interferon regulatory factor 9 (IRF9), leading to the formation of the IFN-stimulated gene (ISG) factor 3 (ISGF3) complex. The ISGF3 complex then translocates to the nucleus and binds DNA sequences referred to as IFN-stimulated response elements (ISRE) within the ISG promoter regions, leading to ISG transcription. Collectively, the IFN-induced production of these ISGs limits the spread of virally infected cells by promoting an anti-proliferative and/or pro-apoptotic response, in addition to activation of immune clearance mechanisms (reviewed in (6)).

IFN responsiveness and signaling play a critical role during viral infections; however, emerging data suggest an important role for IFN in tumorigenesis through modulation of immune surveillance. Independent of viral infection, both IFNs and ISG expression have been found in human tumors (4, 7, 8). While the effects of IFN signaling and ISG phenotypes vary in cancer biology, it remains clinically relevant to gain a deeper understanding of IFN signaling regulation in this context (4, 9). We have previously shown that the progesterone receptor (PR), a steroid-activated nuclear receptor implicated in breast cancer, can transcriptionally repress ISGs in breast cancer via decreased recruitment of ISGF3 components to ISREs in response to progestins (10). Herein, we present a mechanism whereby PR, in the absence of its activating ligand, can attenuate interferon signaling through decreased STAT1 activation/phosphorylation. The potent loss in IFN signaling sensitivity may contribute to the escape of malignant cells from the surveilling immune system.

Materials and Methods

Cell lines and constructs

T47D-co, T47D-Y, T47D-YB have been previously described, and were a generous gift of Dr. Carol Lange (Minnesota) (11, 12). MCF7 (13) and T47D-co shRNA (10) cells have been previously described. Cells were treated with the following reagents (when applicable): R5020 (10nM; Sigma), human recombinant interferon-alpha (IFNα2A; Sigma-Aldrich, SPR4594).

RIME

RIME experiments were performed as previously described (14–16) in R5020-treated (10nM for 60min) T47D-YB or T47D-Y cell lines that were crosslinked with 1% formaldehyde for 10min. Immunoprecipitations were performed using a PR antibody (20ug, Santa Cruz Biotechnology, sc-7208). The peptide samples were analyzed on a Dionex Ultimate 3000 UHPLC system coupled with the LTQ Orbitrap velos mass spectrometer (Thermo Scientific). For the separation of the peptides a multistep gradient elution was used: Mobile phase (A) was composed of 2% acetonitrile, 0.1% formic acid and 5% DMSO and mobile phase (B) was composed of 80% acetonitrile, 0.1% formic acid and 5% DMSO. The gradient elution method at flow rate 300nL/min was as follows: for 65min gradient up to 45% (B), for 10min gradient up to 95% (B), for 10 min isocratic 95% (B), for 5min down to 5% (B), for 10min isocratic equilibration 5% (B) at 40°C.The full scan was performed in the Orbitrap in the range of 400–1600 m/z at 60K resolution. The MS2 scan was performed with CID collision energy 30% and exclusion duration 30sec. The raw data were processed in Proteome Discoverer 1.4 using the SequestHT search engine. The node for SequestHT included the following parameters: Precursor Mass Tolerance 20 ppm, Fragment Mass Tolerance 0.5 Da, Dynamic Modifications were Oxidation of M (+15.995 Da) and Deamidation of N, Q (+0.984 Da). Significant peptides were filtered at FDR<1% and specific interactors were considered if they were identified in both RIME replicate experiments in T47D-YB cells, but not in the T47D-Y (PR-null) cell line.

Co-Immunoprecipitations

For Co-IP experiments, cell lysates were collected in RIPA buffer (supplemented with protease/phosphatase inhibitors) and incubated on ice for 30min. Cell lysates containing equivalent protein concentrations (1000µg) were incubated overnight at 4˚C with 2µg of appropriate antibody or control IgG. Protein G agarose (Roche Diagnostics, Indianapolis, IN) was added for the final 2hr of incubation time. Immune complexes were washed three times with supplemented RIPA buffer, resuspended in Laemmli sample buffer containing β-mercaptoethanol, boiled for 5min, and subjected to Western blotting analysis.

Immunoblotting

Immunoblotting/Western blotting was performed as previously described (10–12). Membranes were probed with primary antibodies recognizing total PR (Santa Cruz Biotechnology, sc-7208 or ThermoScientific, MS-298-P), STAT1 (Cell Signaling, 9172), p-STAT1 (Cell Signaling, 7649), p-TYK2 (Cell Signaling, 68790), TYK2 (Cell Signaling, 14193), STAT2 (Cell Signaling, 4594), IRF9 (Santa Cruz, 10793), IFIT1 (Cell Signaling, 14769), IFIT2 (Santa Cruz Biotechnology, sc-390724), IFIT3 (Santa Cruz Biotechnology, sc-393512), OAS1 (Cell Signaling, 14498), topoisomerase II-alpha (Cell Signaling, 12286) and β-tubulin (Cell Signaling, 2128). All Western blotting experiments were performed in triplicate, and representative experiments are shown.

Breast Cancer Tissue Microarray

Tissue microarrays (TMAs) were constructed from archival formalin fixed, paraffin embedded samples of invasive mammary carcinoma (43 patients) as well as matched benign breast tissue (35 patients). These samples were identified from the pathology departmental archives of the University of Kansas Medical Center from 1997–2011. Based on review of the original pathology reports, the invasive mammary carcinomas were typed as invasive ductal carcinoma (39 patients), metaplastic carcinoma (1 patient), invasive mammary carcinoma with ductal and lobular features (1 patient), invasive ductal carcinoma with a minor lobular component (1 patient) and invasive lobular carcinoma (1 patient). Using the semi-automated TMArrayer (Pathology Devices, Inc., Westminster, MD) TMA paraffin blocks were assembled with 2.0 mm cores.

Immunohistochemistry

Phospho-STAT1 (Cell Signaling, #8826), STAT1 (Cell Signaling, #14994) and PR (Dako #M3568) antibodies were used for Immunohistochemical (IHC) staining according to the following procedure: Four micron paraffin sections were mounted on Fisherbrand Superfrost slides and baked for 60min at 60˚C then deparaffinized. Epitope retrieval was performed in Biocare Decloaking Chamber (pressure cooker), under pressure for 5min, using pH 6.0 Citrate buffer followed by a 10min cool down period. Endogenous peroxidase is blocked with 3% H2O2 for 10min followed by incubation with phospho-STAT1 (1:800), STAT1 (1:3200) or PR (1:2000) primary antibody for 30min (PR) or 45min (p-STAT1, STAT1), followed by Mach 2 HRP Polymer (Biocare Medical) for 30min (p-STAT1, STAT1) or Envision+ Anti-Mouse (Dako) for 30 minutes (PR) and DAB+ chromogen (Dako) for 5min. IHC staining was performed using the IntelliPATH FLX Automated Stainer at room temperature. A light hematoxylin counterstain was performed, following which the slides were dehydrated, cleared, and mounted using permanent mounting media. A pathologist then scored the slides according to the intensity of staining (0 = no staining, 1 = mild intensity, 2 = moderate intensity, 3 = strong intensity), percentage of positive cells and the subcellular localization of the staining. A second pathologist blindly scored the slides to identify any discordant results. Statistical analysis: The Wilcoxon Rank Sum test was used to assess the differences in the phospho-STAT1 and STAT1 intensities between two groups of subjects (PR-positive vs PR-negative tumors).

Subcellular Fractionation

Subcellular fractionation studies were performed as described previously (17), with modifications noted in the figure legend.

RNA-seq

Approximately 500 ng Total RNA was used for stranded Total RNA-Seq library preparation by following the KAPA Stranded RNA-Seq Kit with RiboErase (HMR) sample preparation guide. The first step in the workflow involved the depletion of rRNA by hybridization of complementary DNA oligonucleotides, followed by treatment with RNase H and DNase to remove rRNA duplexed to DNA and original DNA oligonucleotides, respectively. Following rRNA removal, the RNA was fragmented into small pieces using divalent cautions under elevated temperature and magnesium. The cleaved RNA fragments were copied into first strand cDNA using reverse transcriptase and random primers. This was followed by second strand cDNA synthesis using DNA Polymerase I and RNase H. Strand specificity was achieved by replacing dTTP with dUTP in the Second Strand Marking Mix (SMM). The incorporation of dUTP in second strand synthesis effectively quenches the second strand during amplification, since the polymerase used in the assay will not incorporate past this nucleotide. These cDNA fragments then went through an end repair process, the addition of a single ‘A’ base, and then ligation of the adapters. The products were then purified and enriched with PCR to create the final RNA-Seq library. RNA-Seq libraries were subjected to quantification process, pooled for cBot amplification and subsequent sequenced with 50 bp single end sequencing run with Illumina HiSeq 3000 platform. After the sequencing run, demultiplexing with CASAVA was employed to generate the FASTQ file for each sample.

RNA-seq data processing was performed as previous described (18). Briefly, RNA-seq FASTQ files were aligned to hg19 using TopHat (version 2.0.14) and only uniquely mapped reads were used for the further downstream analysis (19). Expression levels of genes were measured with normalized counts of reads by their respective lengths using Cufflinks 2.0.2 package followed by their distribution analysis as Fragments Per Kilobase Million (FPKM) unit (20). These data have been deposited in NCBI’s Gene Expression Omnibus (21) and are accessible through GEO Series accession number GSE126517. (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE126517)

GSEA

Gene set enrichment analysis (GSEA) was performed using the javaGSEA desktop software; the c2 Molecular Signatures Database (MSigDB) version 5.2 was queried (22, 23). Dataset files were developed based on normalized Illumina expression intensities from PR-null and PR-positive T47D cells (as described in the RNA-seq methods section). Specifically, the log2 transformed expression values were compared for two phenotypes: T47D-PR null (IFNα/vehicle) and T47D-PR positive (IFNα/vehicle). GSEA was executed using the default settings, except the permutation type was set to Gene_set with 1000 permutations, and the metric for ranking genes was set to Diff_of_Classes, because normalized expression data was log2 transformed. Leading Edge analysis was performed on the 29 gene sets from MSigDB c2 analysis that achieved FDR values ≤ 0.05.

Real-Time Quantitative PCR (qPCR)

RNA isolation, cDNA creation, and qPCR were performed as previously described (10–12), with modifications noted here and in the Figure legends. qPCR was performed using the Faststart Essential DNA Green Master (Roche) on a Roche LightCycler96. Relative concentrations were quantified using the LightCycler96 (Roche, Software 1.1, Absolute Quantification Analysis), using a 6-point standard curve.

ChIP assays

ChIP was performed using the ChIP-IT Express Kit (Active Motif) according to manufacturer’s instructions using sonication for chromatin shearing. Lysates were immunoprecipitated (IP) overnight (18hr) with the following antibodies: STAT1 (Santa Cruz, sc-346), STAT2 (Santa Cruz, sc-476), IRF9 (Santa Cruz, sc-10793), or an equal amount of negative control mouse or rabbit IgG. Resulting DNA was analyzed using qPCR as described above, and data is represented as a percentage of input DNA.

Statistical Analysis

Statistical significance for all experiments was determined using an unpaired Student’s t-test, unless otherwise specified. A p value ≤ 0.05 is considered statistically significant. The Delta method was used to calculate standard deviation for the ratio of two variables using their individual standard deviations, as seen when plotting fold relative RNA expression data between two treatment groups/cell lines (24).

Results

PR interacts with STAT1

Our lab has a long-standing interest in defining how PR affects the activity of other transcription factors, particularly through protein-protein interactions. To identify novel protein binding partners of PR, we used an immunoprecipitation (IP)/mass spectrometry (MS) approach called Rapid Immunoprecipitation Mass spectrometry of Endogenous proteins, or RIME. This technique, pioneered by the Carroll lab to identify estrogen receptor (ER)-interacting proteins, has been a powerful tool to study crosstalk between steroid receptors and other nuclear proteins (14, 15). PR is an important target gene of ER and, as such, PR expression is regulated by estrogen in most tissues (25, 26). In order to differentiate between the effects of ER/estrogen and PR/progesterone, our laboratory uses PR-positive (T47D-co) and PR-null (T47D-Y) variants of the ER/PR+ breast cancer cell line, T47D (27). T47D-co cells endogenously express both isoforms of PR, PR-A and PR-B, without the need for exogenously added estrogen, allowing us to study the function of PR without the confounding effects of estrogen. T47D-Y (PR-null) cells can be used to determine the effect of PR isoform variants or mutants, such as PR-A (T47D-YA cells) and PR-B (T47D-YB cells). We have published extensively using these cell line models to define isoform- and phosphorylation-specific PR gene regulation and protein-protein interactions (10–12, 28–30), and this cell line model remains a powerful and well-established system for studying the transcriptional activity of PR (31).

We used T47D-YB (stably expressing the full-length PR-B isoform) and PR-null (T47D-Y) cells as a model system to study PR-protein interactions using RIME. Briefly, T47D-YB and T47D-Y cells were crosslinked following treatment with a synthetic PR ligand (R5020) or vehicle (EtOH). PR was immunoprecipitated from isolated nuclei, and PR-interacting proteins were identified using mass spectrometry; the complete list of PR-interacting proteins identified using RIME is provided as Supplemental Table 1. Among the top 30 identified PR-interacting proteins was Signal Transduction and Activator of Transcription 1, or STAT1 (Fig 1A); peptide sequence coverage for PR and STAT1 is shown in Fig 1B. The interaction between PR and STAT1 was validated using co-immunoprecipitation (coIP) in T47D-YB cells (cell line used for the RIME experiments): an interaction between PR and STAT1 could be detected in vehicle-treated cells, and this interaction was slightly increased in cells treated with R5020 (Fig 1C - left). The interaction between PR and STAT1 was also observed in unmodified T47D-co cells, endogenously expressing both isoforms of PR (Fig 1C - right); both PR-A (lower band) and PR-B (upper band) appear to interact with STAT1, suggesting that the interaction between PR and STAT1 is not unique to the full-length PR-B isoform. Interestingly, the interaction between STAT1 and PR-B is increased following treatment with PR ligand, as compared to the STAT1/PR-A interaction. These data suggest that the STAT1/PR-B interaction is primarily regulated by ligand, whereas the STAT1 interaction with PR-A may be a more basal interaction, unaffected by ligand. Moreover, an interaction between PR and STAT1 was also shown in MCF7 cells, an additional PR-positive breast cancer cell line (Supplemental Fig 1A). Cumulatively, these data show that PR and STAT1 interact, and this interaction is potentiated when PR is activated by ligand.

Figure 1. PR interacts with STAT1.

A. STRING network of the top 30 PR interactors identified in RIME. The size of the node increases proportionally to the number of identified peptides and thick edges denote high confidence STRING interactions (0.7-0.99). B. Sequence coverage of PR and STAT1 in both replicate RIME experiments. Green highlights high confidence peptides at FDR<1%. PR and STAT1 have been identified by 27 and 19 unique peptides, respectively. C. Left: PR was immunoprecipitated from starved T47D-YB cell lysates (+/− R5020 60min) and the resulting associated protein complexes were analyzed by Western blotting. Bottom panels represent total input cell lysates. Species-specific (rabbit or mouse) IgG was used as a control for the IP. Right: STAT1 was immunoprecipitated from T47D-co cell lysates (+/− R5020) and the resulting associated protein complexes were analyzed by Western blotting. Bottom panels represent total input cell lysates. PR-B (upper) and PR-A (lower) isoforms are both recognized by the PR antibody. Species-specific IgG was used as a control for the IP; the band visible in the IgG only lanes represents non-specific binding between PR/STAT1 and the IgG antibody. These experiments were performed in triplicate, and a representative experiment is shown here.

PR attenuates interferon-induced STAT1 phosphorylation

STAT1 is a key mediator of the innate immune response, typically in response to viral infection. However, interferon activation in cancer cells, independent of viral infection, has been widely reported (4). In response to interferon-alpha (IFNα) stimulation, STAT1 and STAT2 are phosphorylated by Janus Activated Kinase 1 (JAK1) and Tyrosine Kinase 2 (TYK2). Phosphorylated STAT1/2 form heterodimers that bind with IFN-regulatory factor 9 (IRF9) to form a transcriptional complex referred to as interferon-stimulated gene factor 3, or ISGF3. This transcriptional complex binds to interferon-stimulated response elements (ISREs) throughout the genome, activating transcription of a cohort of genes essential for mounting the cellular anti-viral response (5). To determine if the interaction between PR and STAT1 affects the function of STAT1, we looked at STAT1 phosphorylation in response to IFNα treatment. T47D cells that are positive (T47D-co) or negative (T47D-Y) for PR expression were treated for 0–2hrs with IFNα. Cells lacking PR expression (PR-null) phosphorylated STAT1 on an earlier time course and at a greater magnitude in response to IFNα, as compared to PR-positive cells (Fig 2A – left, and quantification of Fig 2A - right). There are two isoforms of STAT1, STAT1α (larger) and STAT1β (smaller); both isoforms appear to be equally affected by the presence/absence of PR. Moreover, treatment of PR-positive cells (T47D-co) with PR ligand (R5020) even further attenuated phospho-STAT1 in response to IFNα treatment (Fig 2B-left, and quantification of Fig 2B - right). Decreased STAT1 phosphorylation in the presence of activated PR was repeated in MCF7 cells (Supplementary Figs 1 B and 1 C). These data suggest that both the presence (PR-negative vs PR-positive) and activation of PR (+/− ligand) decreases the levels of phospho-STAT1 in response to interferon treatment, likely due to the interaction between PR and STAT1 (see Fig 1C). STAT2 phosphorylation, another downstream effector of IFNα signaling, was unaffected by the presence or activation of PR following IFNα-treatment (data not shown). Finally, to determine if the decrease in interferon-induced phosho-STAT1 following treatment with PR ligand was restricted to a particular cellular compartment, we used subcellular fractionation experiments to assay phospho-STAT1. We observed similar decreases in phospho-STAT1 following treatment with interferon and PR ligand in the nuclear and cytoplasmic fractions, as well as the whole cell are decreased following treatment with PR ligand.

Figure 2. PR attenuates interferon-induced STAT1 phosphorylation.

A. T47D cells that are PR-negative (T47D-Y) or PR-positive (T47D-co) were treated with interferon-alpha (IFNα) for 0-2hrs. Isolated protein lysates were analyzed by Western blotting. Densitometry of the ratio of p-STAT1/total STAT1, as determined using ImageJ analysis, is shown to the right of the immunoblot. B. T47D-co cells were starved for 18hrs in serum-free media, followed by treatment with interferon IFNα and R5020 or vehicle (EtOH) for 0-2hrs. Isolated protein lysates were analyzed by Western blotting. Densitometry of the ratio of p-STAT1/total STAT1, as determined using ImageJ analysis, is shown to the right of the immunoblot. C. T47D-co cells were treated as in B. Whole cell lysates were subjected to nuclear/cytoplasmic fractionation, and resulting subcellular lysates were analyzed by Western blotting. B-tubulin (cytoplasmic) and topoisomerase II (topo II; nuclear) are shown as fractionation markers. These experiments were performed in triplicate, and a representative experiment is shown here.

Regulation of p-STAT1/STAT1 is disrupted in PR-positive human breast tumors

To determine if there is a correlation between PR positivity and phospho-STAT1 levels in human breast tumors, we analyzed a previously described, custom designed breast tumor tissue microarray (TMA)(13). In brief, this breast cancer TMA is composed of specimens collected from 39 breast cancer patients seen at the University of Kansas Medical Center; 21 of which were PR-positive, 18 were PR-negative. We stained this breast TMA with antibodies that recognize phospho-STAT1 and total STAT1, and staining intensities were blindly scored by a clinical pathologist. Interestingly, PR-positive tumors had lower phospho-STAT1 staining intensity when compared to their PR-negative counterparts. Although statistical analysis (Wilcoxon Rank Sum test) did not show statistically significant differences in the intensities (likely due to the limited numbers of samples on the TMA), there is a clear trend towards lower staining intensity in PR-positive tumors (Fig 3A – left). The trend between lower phospho-STAT1 staining in PR-positive tumors appears to be specific for phosphorylated STAT1, as no trend or statistical significance exists for total STAT1 staining intensity between PR-positive and PR-negative tumors (Fig 3A – right). Select examples from the TMA are shown in Fig 3B. As STAT1 itself is an ISG regulated by interferon signaling (32), higher activation of STAT1 (measured via phosphorylation) is normally positively correlated with total STAT1 levels, representing a positive-feedback regulatory loop. A scatter plot (Fig 3C) of p-STAT1 and STAT1 staining intensities for each tumor highlights the positive correlation between p-STAT1 and total STAT1 in PR-negative tumors (blue line; r=0.7188, p=0.0291).; this correlation is not present in PR-positive tumors (red line; r=0.0291, p=0.9364). These data suggest that PR disrupts the activation of STAT1, as measured via p-STAT1 levels (Fig 3A) and the correlation between p-STAT1 and total STAT1 (Fig 3C). Cumulatively, these data suggest that PR-positivity affects phospho-STAT1 levels in human breast tumors, in addition to human breast cancer cell lines.

Figure 3. Regulation of p-STAT1/STAT1 is disrupted in PR-positive tumors.

A. Tissue microarray analysis was performed using immunohistochemical staining with phospho-STAT1 (p-STAT1), STAT1 and PR antibodies. Boxplots and Stripcharts show the distribution of p-STAT1 (left) and STAT1 (right) staining intensities in PR-negative and PR-positive breast cancer samples. B. Select PR-negative (left) and PR-positive (right) breast cancer (BrCa) cases stained for PR and p-STAT1 are shown here at 20x magnification. Negative control isotype-only control staining is shown. C. Scatter plot of pSTAT1 and STAT1 with different colors for PR-positive (red) and PR-negative (blue) with trend lines for each. The Spearman’s correlation coefficient and p-values for each group are shown on top. Only tumors with greater than zero total STAT1 staining were included in the analysis.

Interferon-induced TYK2 activity is attenuated by activated PR

Interestingly, phospho-STAT1 following treatment with IFN-gamma (IFNγ−type II interferon signaling ligand) was unaffected by PR activation (data not shown), suggesting the effect of PR on STAT1 phosphorylation is exclusive to pathways/proteins involved in type I interferon signaling. TYK2 is the primary kinase responsible for phosphorylating STAT1 in response to IFNα (reviewed in (6)), whereas JAK1 and JAK2 are responsible for STAT1 phosphorylation following IFNγ. Because TYK2 is specific to type I interferon signaling, we sought to determine how TYK2 activation was impacted by PR. We measured phospho-TYK2 (indicative of TYK2 activation) following treatment with IFNα in the presence/absence of PR activation. In T47D-co cells treated with PR ligand, IFNα-induced TYK2 phosphorylation was attenuated (Fig 4A-left and quantification of Fig 4A-right). Importantly, JAK1 and JAK2 phosphorylation was unaffected by PR activation (data not shown), thus reinforcing the specificity of this effect to type I interferon signaling. As such, like PR and STAT1, we identified an interaction between PR and TYK2 that was increased in response to PR ligand (Fig 4B). Again, similar to PR-STAT1, the PR-TYK2 interaction in response to PR ligand appears to be primarily driven by PR-B (larger isoform, upper band). Together, these data suggest that an interaction between PR and TYK2 leads to decreased TYK2 activation, which subsequently translates to decreased STAT1 phosphorylation when PR is present/activated.

Figure 4. TYK2 phosphorylation is attenuated by activated PR.

A. T47D-co cells were starved for 18hrs in serum-free media, followed by treatment with interferon-alpha (IFNα) and R5020 or vehicle (EtOH) for 0-30min. Isolated protein lysates were immunoprecipitated with the total TYK2 antibody, and the resulting protein complex was blotted with phospho-TYK2 antibody. Species-specific IgG was used as a control for the IP. Densitometry of the ratio of p-TYK2/total TYK2, as determined using ImageJ analysis, is shown to the right of the immunoblot. B. TYK2 was immunoprecipitated from T47D-co cell lysates (+/− R5020; 60min) and the resulting associated protein complexes were alanyzed by Western blotting. Bottom panels represent total input cell lysates. PR-B (upper) and PR-A (lower) isoforms are both recognized by the PR antibody. Species-specific IgG was used as a control for the IP. These experiments were performed in triplicate, and a representative experiment is shown here.

PR activation disrupts the ISGF3 complex

Following STAT1 phosphorylation, a transcriptional complex is formed containing STAT1, STAT2 and IRF9. Formation of this complex is key to transcriptional activation of genes turned on in response to interferon treatment, such as ISGs. To determine if PR, due to the interaction between PR and STAT1, affects the formation or integrity of the ISGF3 complex, we used co-IP assays to interrogate the interactions within the ISGF3 complex. In T47D-co cells treated with IFNα, the interaction between STAT2 and STAT1 decreased with the addition of PR ligand (R5020; Fig 5). A decreased interaction was also observed between IRF9 and STAT2 in the presence of R5020 (Fig 5). These data suggest that PR activation decreases the integrity of the ISGF3 complex, either through promoting disassembly of the complex or preventing efficient assembly. Cumulatively, these data suggest that the interaction between PR and STAT1 disrupts the functionality of STAT1, through decreased phosphorylation (Fig 2–3) and disruption of STAT1-containing protein complexes (Fig 5).

Figure 5. PR activation disrupts the ISGF3 complex.

STAT2 and IRF9 (or a species-specific IgG control) was immunoprecipitated from interferon-alpha (IFNα)-treated (2 hrs) starved T47D-co cell lysates (+/− R5020 60min) and the resulting associated protein complexes were analyzed by Western blotting. This experiment was performed in triplicate, and a representative experiment is shown here.

PR decreases ISG transcriptional response and protein levels

The endpoint of type I interferon signaling is transcriptional activation of ISGs, genes that canonically orchestrate the cellular response to viral pathogens. As PR presence attenuated STAT1 phosphorylation in response to interferon treatment and promoted the dissociation of the ISGF3 complex, we performed RNA-seq on interferon-treated PR-positive and PR-negative breast cancer cells to determine how the presence of unliganded PR affects global interferon-activated transcriptional programs. T47D cells that are positive (T47D-co) or negative (T47D-Y) for PR expression were treated with IFNα (or vehicle) for 18hr. Gene Set Enrichment Analysis (GSEA) analysis revealed that multiple interferon-associated gene sets were enriched in the PR-null RNA-seq dataset; this enrichment was lost in cells expressing PR (22, 23). The top significantly regulated gene sets in PR-null cells (as compared to PR-positive cells) are shown in Fig 6A; select examples of enrichment are shown in Fig 6B. Leading Edge (LE) analysis, a component of GSEA that allows the identification of core genes that drive the enrichment of a particular gene set, identified multiple genes that are transcriptional targets (ISGs) of interferon signaling pathways (i.e. IFITs, MX1, OASs) whose regulation is enriched in PR-null cells, but lost in PR-positive cells (Fig 6C). Using qPCR, we validated a core set of ISGs and confirmed on the individual gene level that interferon-activation of IGSs is transcriptionally attenuated (ranging from 50–85% reduction) in T47D cells expressing PR, as compared to PR-null cells (Fig 6D). Protein expression for select ISGs mimics the same phenotype; interferon-activation of ISG protein levels is higher in PR-negative cells as compared to PR-positive cells (Fig 6E). Cumulatively, these data suggest that unliganded PR attenuates the transcriptional response to IFNα, through downregulation of ISG RNA and protein levels.

Figure 6. PR decreases interferon-induced gene expression.

A. RNA-seq was performed on RNA isolated from interferon-treated (20IU/ml IFNα for 18hrs) T47D PR-null or PR-positive cell lines. GSEA analysis was performed using the c2 MSigDB collection comparing RNA-seq gene expression datasets obtained from interferon-treated PR-null and PR-positive cells. Shown here are the top most significantly-enriched gene sets (FDR < 0.05); select enrichment examples are shown in (B). C. Top 20 ranking genes as identified using Leading Edge (LE) analysis on 29 gene sets referred to in (A). D. T47D PR-null or PR-positive cell lines were treated as in (A). Isolated RNA was analyzed for multiple ISGs using qPCR. Gene values were normalized to an internal control (β-actin). Error bars represent standard deviation between biological triplicates. Asterisks represent statistical significance between groups; p < 0.01, as determined using an unpaired Student’s t-test. E. Cells were treated as in (D), and isolated protein was analyzed via Western blotting with respective antibodies. The experiments presented in (D) and (E) were performed in triplicate, and a representative experiment is shown here.

ISGF3 recruitment to ISREs is attenuated in PR-positive cells

ISGs are transcriptionally activated following ISGF3 (composed of STAT1, STAT2 and IRF9) binding to ISRE sequences in the proximal promoter regions of these genes. To determine if the interaction between PR and STAT1 may affect recruitment of the ISGF3 complex to ISREs, we used chromatin immunoprecipitation (ChIP) assays to measure STAT1, STAT2 and IRF9 recruitment in T47D-co cells expressing NS or PR shRNA (previously described in (10) and Supplemental Fig 2). In response to IFNα treatment, ISGF3 components (STAT1, STAT2 and IRF9) were potently recruited to ISRE sequences of the IFIT genes (Fig 7). Significantly, in cells lacking PR expression (PR shRNA), ISGF3 components exhibited more robust recruitment to ISREs as compared to cells expressing PR (NS shRNA). These data indicate that the interaction between PR and STAT1 decreases DNA-recruitment of the transcriptional complex required for ISG transcription, thereby leading to a decrease in ISG RNA levels.

Figure 7. PR decreases ISGF3 recruitment to ISG promoters.

T47D-co NS and PR shRNA cells were serum-starved for 18hr, and then treated with 1000IU/ml IFNα (or vehicle) for 4 hrs. Fixed lysates were subjected to ChIP with antibodies against STAT1, STAT2, IRF9 or a species-specific IgG (control; not shown), and qPCR was performed on the isolated DNA using primers designed to amplify select ISG promoters. A percentage of ChIP’d DNA over input DNA is shown. All ChIP experiments were performed in triplicate; a representative experiment is shown here. Fold-recruitment in IFNα–treated conditions, as compared to vehicle treatment, is displayed above each bar. Error bars represent standard deviation of technical replicates.

Discussion

The data presented herein support a model whereby PR positivity can influence the potency of type I interferon signaling through an interaction with STAT1. We show that PR-positivity leads to a decrease in interferon-stimulated STAT1 phosphorylation, a collapse of the ISGF3 transcriptional complex, decreased recruitment of STAT1 to interferon-stimulated gene enhancers, and, subsequently, lower levels of ISG RNA and protein. The inhibitory effects of PR on interferon signaling are seen in the absence of PR ligand (ligand-independent), and are further potentiated when PR ligand is added. These observations are clinically relevant as we see decreased phospho-STAT1 in PR-positive breast tumors as compared to their PR-negative counterparts. These data regarding the actions of unliganded (ligand-independent) PR, together with our recently published data showing PR ligand-dependent transcriptional repression of ISGs (10), suggest a novel, multifactorial role for PR in attenuation of interferon signaling.

Interferon signaling has been classically defined as a critical part of the innate immune system’s response to viral infection. However, there is an emerging role for interferons in aiding the immune system in detection and clearance of early malignancies. The cancer immunoediting hypothesis, first put forth by Robert Schreiber, highlights that immunoediting can both prevent and promote cancer progression: clearing early tumors, but helping shape the immunogenicity of more developed tumors (2, 33). The three stages of cancer immunoediting are elimination, equilibrium and escape. The elimination phase involves recognition and destruction of nascent tumors, before they can advance to tumors of biological relevance. There is a clear role for type I interferons in this process, reviewed in (34). Both tumor cells and host immune cells can produce and are responsive to interferons, and this plays a key role in tumor cell elimination. As such, alterations in interferon production, as well as the cellular response to interferons, can have dramatic effects on recognition and clearance of developing tumors. Multiple studies suggest this is the case, showing increased tumor formation in mice lacking key components of the interferon response pathway, such as interferon receptors (IFNΑR and IFGAR) and key interferon-signaling molecules (JAK2, STAT1); these data are thoroughly reviewed in (34). Situations in which the cellular response to interferon signaling is decreased or abrogated, such as that seen in the PR-positive breast cell line models shown herein, may have a significant impact on the efficacy of immune-mediated tumor elimination. Decreased interferon/STAT1-signaling, as mediated by PR, may aid early PR-positive tumors in evading immune surveillance, allowing for the development of clinically relevant tumors.

There is a growing body of evidence to suggest that STAT1 is involved in breast tumorigenesis, and has been extensively reviewed in (35–37). Significantly, STAT1 knockout mice develop mammary gland adenocarcinomas, 90% of which are ER/PR-positive (38). The molecular signatures of these tumors overlap with human luminal A/B tumors, which are predominantly ER/PR-positive tumors. Further, dysregulation of the JAK2/STAT1 signaling axis leads to the survival and proliferation of luminal progenitor cells, the precursors to ER/PR-positive tumors, in the murine mammary gland (39). These data implicate STAT1 loss as an early event in the development of mammary gland tumors. Our data suggest that one putative mechanism for STAT1 loss (i.e. loss of function) is through the PR-STAT1 interaction, and subsequent decreased STAT1 signaling, presented herein. Cumulatively our data, together with previously published data, suggest PR-STAT1 crosstalk is critical to the development of ER/PR-positive tumors.

These data discussed thus far present a role for STAT1 as a tumor suppressor in multiple tumor tissues. Conversely, work from Andy Minn, Ralph Weichselbaum, and others has shown that expression of a core set of genes (mostly ISGs) canonically regulated by STAT1 are correlated with therapy resistance in multiple tumors types, including breast. This gene set, referred to as the interferon-related DNA damage resistance signature (IRDS), is upregulated in tumors that develop resistance to chemotherapy, endocrine therapy, radiation, and immunotherapy, and can be used to predict poor prognosis (35, 40–44). Moreover, recent work from the Peter lab suggests that CD95-mediated activation of interferon signaling via STAT1 promotes a cancer-stemness phenotype in breast cancer cells (45). Together, these data highlight putative differences for the role of interferon/STAT1-signaling in early (tumor initiating/immune evasion) vs late (resistance to therapy) tumor events, and underscore the complexity of events that culminate in tumor formation and ultimately, progression.

One major unanswered question in breast cancer biology is why the overwhelming majority of breast cancers, over 65%, are ER/PR-positive at the time of diagnosis (46, 47). Breast cancer is a hormonally driven disease, largely driven by exposure to estrogens (48). Estrogen, when binding to the estrogen receptor, drives proliferative and survival gene programs that promote oncogenesis (49). However, extensive animal model and clinical data suggest that both estrogen and progesterone are needed for breast tumor development (50–52). Despite these data, the role for progesterone and PR in breast cancer remains largely unanswered (31). While ER and PR likely provide a proliferative advantage, affording one explanation for this preponderance of ER/PR-positive tumors, another non-mutually exclusive explanation is that these tumors have an immune-privileged phenotype, allowing for their escape from the surveilling immune system. PR-positive tumors, due to their ability to downregulate interferon signaling under liganded (10) and unliganded conditions (data presented herein and (10)), may evade the critical first step of elimination via immune clearance. Animal experiments are currently underway in our laboratory to test this model in the context of an intact immune system.

Finally, although ligand-dependent gene regulation remains the predominant pathway through which PR regulates gene expression, a role for ligand-independent (in the absence of ligand/progestin) PR gene regulation is emerging as a potent, but not well understood, mechanism through which PR can affect target gene regulation (53–56). In addition to our recently published work highlighting PR ligand-dependent ISG transcriptional repression (10), these data presented herein suggest a novel role for unliganded PR in blocking STAT1-dependent interferon signaling. Ligand-independent effects of PR may be a significant contributor to PR-dependent gene regulation in post-menopausal women, and warrant further investigation.

Key points.

• Progesterone receptor interacts with STAT1 in breast cancer cells

• Progesterone receptor attenuates interferon-induced STAT1 phosphorylation in breast cancer

• Interferon signaling via STAT1 is more robust in breast cancer cells lacking PR