Regulation of early growth response 2 expression by secreted frizzled related protein 1

This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Baystate Medical Center Institutional Animal Care and Use Committee (Permit Number: 132-681). Four week old female BALB/c control mice (n = 30) and BALB/c Sfrp1 ^−/− mice (n = 22) were individually housed in plastic cages with food and water provided continuously, and maintained on a 12:12 light cycle. The animals were treated with DMBA by gavage (1 mg/week) for 4 consecutive weeks to induce mammary tumor formation. Tumor size, number, and time-to-incidence was noted and compared between control and Sfrp1 ^−/− animals. Tumors were collected from 9 control mice and 7 Sfrp1 ^−/− mice, flash frozen and stored at −80 °C until processed for RNA isolation.

The 76 N TERT cell line (obtained from Dr. Vimla Band) were stably transfected with either pSUPER.retro (TERT-pSUPER) or siSFRP1-PSUPER.retro (TERT-siSFRP1) and cultivated as previously described [20]. MCF7, T47D, and MDA-MB-231 cells were purchased from ATCC (ATTC#s HTB-22, HTB-133, and HTB-26) and TMX2–28 cells [21] were obtained from Dr. Kathleen Arcaro. Breast cancer cell lines were transfected with either pCDNA3.1 or SFRP1-pCDNA3.1 as previously described [22]. Cells were grown to 70% confluence in 6-well plates for RNA isolation. For some experiments, cells were treated with DMSO, 10 μM LY364947 (L6293; Sigma), 5 μM U0126 (U120; Sigma), or 10 μM FR108204 (sc203945; Santa Cruz Biotechnology) 24 h prior to RNA isolation. Fresh breast tissue enriched with epithelium from women undergoing elective breast surgery was grossly dissected from the surrounding adipose tissue and placed on Surgifoam gelatin sponges (Ferrosan, Sueborg, Denmark) in 60 mm tissue culture dishes containing 3 mL of medium [(phenol red free DMEM/F12 buffered with Hepes and NaHCO₃ from Gibco (Invitrogen, Carlsbad, CA)], 5 μg/mL human insulin, 1X antibiotic/antimycotic (100 U/mL penicillin/streptomycin and 0.250 μg/mL amphotericin B), and 10 μg/mL gentamycin from Sigma (Sigma, St. Louis, MO). The media was supplemented with either 0.1% BSA or 1 μg/ml rSFRP1 for 24 h and the tissue was subsequently flash frozen and stored at −80 °C until being processed for RNA isolation.

Ten week old virgin control (n = 12) or Sfrp1 ^−/− mice (n = 12) were euthanized with carbon dioxide prior to organ removal. The fourth mammary glands were harvested, minced, and finally dissociated in DMEM:F12 (Sigma) supplemented with 5% fetal bovine serum (Gibco, Waltham, MA), 2 mg/ml collagenase (Worthington Biochemical, Freehold, NJ), 100u/ml hyaluronidase (Sigma), 100u/ml pen/strep (Gibco) and 100 μg/ml gentamicin (Gibco) for 6 h. The cell pellet was collected and further dissociated with 1 ml pre-warmed 0.05% Trypsin-EDTA (Gibco) and l0 1 mg/ml DNase I (Roche, Mannheim, Germany). Cell suspensions were sequentially sieved through 100 μm and 40 μm cell strainers. Primary cells were seeded onto rat tail collagen-1 (BD Biosciences, San Jose, CA) coated tissue culture dishes in 10% serum containing mammary growth medium (EpiCult®B for Mouse Mammary Epithelial Cell Culture, Vancouver, BC) supplemented with10ng/ml EGF (Sigma), 10 ng/ml FGF (Sigma), 4 μg/ml heparin, 100u/ml pen/strep (Gibco) and 100 μg/ml gentamicin (Gibco) [23]. Cells were routinely cultivated at 37 °C in 5% CO₂. Serum containing media was removed the next day and replaced with media containing DMSO, 10 μM LY364947, 5 μM U0126, or 10 μM FR108204 24 h prior to RNA isolation. The spleens from Sfrp1 ^−/− mice were removed aseptically, placed in 100-mm² tissue culture dishes with 5 ml of phosphate buffered saline (PBS) and the cellular contents was released by macerating the spleens between frosted glass slides. The cells were collected by centrifugation, re-suspended in RPMI media (Gibco), plated in 6-well plates, and stimulated with either LPS (Sigma) or 2.5 ng/mL TGFβ1 (Sigma) and subsequently treated with either 0.1% BSA or 1 μg/ml rSFRP1. The following day, the media was removed and the adherent macrophage rich cells were harvested for RNA isolation.

Total RNA was extracted from cells and tissues (n = 3/treatment) using an acid-phenol extraction procedure [24], according to the manufacturer’s instructions (Trizol, Invitrogen, Carlsbad, CA). Relative expression levels of mRNA was determined by quantitative real-time PCR using the Mx3005P® real-time PCR system (Agilent, Santa Clara, CA) and all values were normalized to the amplification of ΑctB. The PCR primer sequences for mouse Actb, mouse Tgfb1, and human ACTB have been published [12, 20]. Additionally, primer sequences were designed to cross exon junctions using GenScript Real-time PCR Primer Design (www.​genscript.​com/​tools/​real-time-pcr-tagman-primer-design-tool) and are as follows: human EGR2 forward: 5′-TCCCAGTAACTCTCAGTGGTT-3′, human EGR2 reverse: 5-TGCCATCTCCGGCCA-3′; mouse Egr2 forward: 5′- TTGACCAGATGAACGGAGTG–3′, mouse Egr2 reverse: 5′-AGCTACTCGGATACGGGAGA–3′; mouse Gpr18 forward: 5′- TGTCAACGTGCTCAACTTCA–3′, mouse Gpr18 reverse: 5′- CCTTGGGCTTCAGCTTAGA–3′; human CD163 forward: 5′-GAGTGACCTGCTCAGATGGA–3′, human CD163 reverse: 5′-CCGTCCTTGGAATTTGATCT–3′; human HLADRA forward: 5′- CATGGAGGTGATGGTGTTTC–3′, human HLADRA reverse: 5′- TGCTTTCACTGAGGTCAAGG–3′. The assays were performed using the 1-Step Brilliant® SYBRIII® Green QRT-PCR Master Mix Kit (Agilent) containing 200 nM forward primer, 200 nM reverse primer, and 100 ng total RNA. The conditions for cDNA synthesis and target mRNA amplification were performed as follows: 1 cycle of 50 °C for 30 min;1 cycle of 95 °C for 10 min; and 35 cycles each of 951C for 30 s, 55 °C for 1 min, and 72 °C for 30 s. Non-template controls were included to control for primer dimers and no reverse-transcriptase controls were included to control for genomic DNA amplification.

Treated MMECs were washed twice with cold PBS and 100 μL of cold lysis buffer [50 mM Tris-HCl, 150 mM NaCl, 100 mM NaF, 10 mM MgCl₂, 0.5% NP40, protease inhibitor cocktail, and phosphatase inhibitor I and II (Sigma)] was added directly to the plate. The cells were incubated for 30 min at 4 °C on a shaker and then harvested using a rubber policeman. The lysates were passed 4 times through a 26 gauge syringe, kept on ice for 30 min, and then centrifuged for 20 min at 12,000 rpms at 4 °C. The supernatant was transferred to a new tube and the protein was quantified utilizing the BCA™ Protein Assay Kit (Pierce, Rockford, IL). A total of 30 μg of protein was run on a 10% SDS-Page gel and transferred to a PVDF membrane. The membrane was blocked for 45 min with 5% milk in tris-buffered saline containing 0.05% Tween-20 (TBS-T). The primary antibodies used in this study were [Rabbit phospho-ERK1/2 (Thr202/Tyr204) (1:1000), #4377, Cell Signaling Technologies, Danvers, MA and Rabbit β-actin (1:1000), ab8227, Abcam, Cambridge, MA] incubated overnight at 4 °C. The secondary antibody [anti-rabbit IgG-HRP (#7074, Cell Signaling Technologies) was applied (1:1000) and incubated for 45 min at room temperature. The blot was washed and developed using a Western Blot Luminol Reagent (Denville Scientific, Holliston, MA). The integrated band densities were measured using ImageJ software (www.​imagej.​nih.​gov).

Tissue blocks were sectioned at 4 μm on a graded slide, deparaffinized in xylene, rehydrated in graded ethanols, and rinsed in phosphate-buffered saline (PBS). Immunohistochemistry (IHC) was performed on a DakoCytomation autostainer using the Envision HRP Detection system (Dako, Carpinteria, CA). Each mammary tissue block was sectioned at 4 μm on a graded slide, deparaffinized in xylene, rehydrated in graded ethanols, and rinsed in Tris-phosphate-buffered saline (TBS). Heat induced antigen retrieval was performed in a microwave at 98 °C in 0.01 M citrate buffer. After cooling for 20 min, sections were rinsed in TBS and subjected to the primary mouse monoclonal anti-CD163 [GH1/61] antibody (1:100, Abcam, ab111250) or the primary rabbit polyclonal anti-HLA-DR antibody (1:250, Abcam, ab137832) for 30 min. Immunoreactivity was visualized by incubation with chromogen diaminobenzidine (DAB) for 5 min. Tissue sections were counterstained with hematoxylin, dehydrated through graded ethanols and xylene, and cover-slipped. Images were captured with an Olympus BX41 light microscope using SPOT Software 5.1 (SPOT™ Imaging Solutions, Detroit, MI).

EGR2 is expressed in breast cancer cell lines, particularly the more aggressive triple negative subtypes and may be regulated in part by Epidermal Growth Factor (EGF) family members [18, 25]. Considering that EGR2 and the tumor suppressor protein SFRP1 and play a role in immune function, we sought to determine whether SFRP1 regulates EGR2 expression. We found that TERT-siSFRP1 cells express significantly more EGR2 mRNA when compared with TERT-pSUPER cells and when human explant mammary tissues are treated with rSFRP1, EGR2 mRNA levels are significantly reduced (Fig. 1a). Conversely, breast cancer cells overexpressing SFRP1 (MCF7-SFRP1) express less EGR2 when compared with vector transfected cells (MCF7-pCDNA) and exogenous rSFRP1 treatment reduces EGR2 expression in MCF7 cells (Fig. 1b). When we further tested the effect of SFRP1 expression in a panel of breast cancer cell lines, we found that SFRP1 reduced the mRNA levels of EGR2 in T47D cells but not in TMX2–28 cells (a Tamoxifen resistant variant of MCF7) or MDA-MB-231 cells (Additional file 1: Figure S1).

We next sought to establish whether mouse mammary gland tissue derived from Sfrp1^−/− mice also exhibit an increase in Egr2 expression. We found that when compared with control mice, mammary tissue from Sfrp1^−/− mice have elevated levels of Egr2 (Fig. 1c, left panel) and more specifically, isolated mouse mammary epithelial cells (MMECs) from Sfrp1^−/− mice express higher levels of Egr2 (Fig. 1c, right panel).

The expression of EGR2 is regulated by TGF-β signaling in skin fibroblasts [26] and by MAPK signaling in osteoblasts and breast adipose fibroblasts [27‐29]. As we have previously demonstrated that reducing SFRP1 in immortal mammary epithelial cells exacerbates TGF-β signaling and increases migration through TGF-β mediated MAPK signaling [30], we suspected that SFRP1 mediated modulation of EGR2 may involve these pathways. In TERT-siSFRP1 cells, we confirmed that EGR2 expression is upregulated in response to TGF-β1 treatment (Additional file 2: Figure S2A). We next sought to determine whether EGR2 expression in TERT-siSFRP1 cells could be blocked by antagonizing the TGF-βR with LY364947. We found that both TERT-pSUPER and TERT-siSFRP1 cells exhibited a significant reduction in EGR2 mRNA expression when the TGF-βR is inhibited (Fig. 2a, left panel). Considering that TGF-β signaling induces ERK1/2 activation [31] and loss of SFRP1 exacerbates the MAPK pathway [30], we next tested whether EGR2 expression could be affected in response to a MEK1/2 specific inhibitor (U0126). We clearly demonstrate that while the expression of EGR2 is not affected by U0126 in TERT-pSUPER cells, MEK1/2 inhibition in TERT-siSFRP1 cells significantly decreased EGR2 expression (Fig. 2a, right panel). Moreover, TERT-siSFRP1 cells treated with a second more specific MEK1/2 inhibitor, FR108204, exhibit a decrease in EGR2 expression when compared with the effect of FR108204 on EGR2 mRNA in TERT-pSUPER cells (Additional file 3: Figure S3A).

An Sfrp1^−/− associated increase in TGF-β expression has previously been confirmed in our analysis of pubertal rodent mammary tissues [12, 32]. We next wanted to determine whether similar to human cells, treatment with TGF-β1 could induce the phosphorylation of ERK1/2 in MMECs and if the phosphorylation of ERK1/2 could be blocked by antagonizing the TGF-βR with LY364947. Our results illustrate that TGF-β1 treatment in control MMECs does not affect ERK1/2 phosphorylation, but blocking the TGF-βR abrogates phosphorylation. However, MMECs derived from Sfrp1^−/− mice exhibited a significant increase in ERK1/2 phosphorylation in response to TGF-β1 treatment, which was also blocked by LY364947 treatment (Fig. 3). To establish whether there is a connection between TGF-β1 signaling and Egr2 in our murine model, we used MMECs from our control mice to verify that Egr2 expression in MMECs is driven by TGF-β treatment (Additional file 2: Figure S2B). Consistent with our human mammary epithelial SFRP1 knockdown cells, we show that TGF-βR, MEK1/2, and ERK1/2 inhibition reduce Egr2 expression in Sfrp1^−/− MMECs (Fig. 2b; Additional file 3: Figure S3B).

We previously reported that a targeted deletion of Sfrp1 exacerbates weight gain as well as inflammation [12]. The increased macrophage infiltration and pro-inflammatory cytokine expression observed in Sfrp1^−/− mice was expected based on the link between obesity and inflammation. The assortment of cues within the microenvironment can elicit a wide range of macrophage phenotypes and functions [33]. The classical (M1) and alternative (M2) activation of macrophage subtypes are an example of the two extremes on this continuum. Interestingly, Gong et al. have demonstrated that TGF-β signaling is required for M2 activation [7] and Egr2 is also murine marker of M2-polarized macrophages [19]. Therefore, we next tested the hypothesis that SFRP1 may play a role in macrophage polarization. We isolated splenic macrophages from Sfrp1^−/− mice and treated them with either TGF-β to induce M2 polarization or LPS to induce M1 polarization. When TGF-β stimulated macrophages were treated with rSFRP1, the M2 maker Egr2 was significantly down regulated (Fig. 4a). Conversely, when LPS stimulated macrophages were treated with rSFRP1, the M1 marker Grp18 was significantly up-regulated (Fig. 4b).

Considering that rSFRP1 has been used to decrease IL-6 in macrophages and adipocytes [12, 34] and TGF-β stimulated macrophages treated with rSFRP1 exhibit a reduction in EGR2 expression, we investigated whether rSFRP1 treatment of a human mammary explant would affect the expression of the human M2 marker, CD163. Human breast explant cultures were treated with rSFRP1 for 24 h and were subsequently harvested for real-time PCR analysis as well as immunohistochemistry. We show that in response to rSFRP1, the mRNA levels and number of CD163 staining cells are significantly downregulated in the human breast tissue (Fig. 5a). However, HLA-DR expression (a human M1 marker) was not altered by rSFRP in human mammary gland explant cultures (Fig. 5b).

Previous studies have noted that SFRP1 can elicit anti-inflammatory effects [5, 6, 35]. The work described here confirms the regulatory role of SFRP1 on TGF-β and its effect on EGR2 not only in inflammatory cells, but also in epithelial cells. Our data suggests that the control of inflammation may be due in part to the concomitant effect of SFRP1 on EGR expression. Several EGR family members have been implicated in regulation of cytokine expression in allergic reactions, mesenchymal stem cells as well as prostate cancer cells [18, 36, 37]. Moreover, EGR2 plays a complex role in a variety of cell types as well as the development of cancer. In Ras transformed NIH 3 T3 cells it controls cebpb expression [38] in gastric cancer cells its expression is associated with metastasis [39] and inversely associated with the expression of miR20a [40]. Knockdown of EGR2 in leiomyoma cells increased myc and PCNA expression as well as collagen deposition [41]. However, how EGR2 contributes to breast cancer however is not clear, its expression has been suggested to drive both the expression of both Erbb2, as well as aromatase expression [25, 29, 42]. The data presented here expand upon these findings and demonstrate that SFRP1 regulates EGR2 expression in both human and murine mammary epithelium. Interestingly, SFRP1 does not affect EGR2 expression in estrogen receptor (ER) negative cells (TMX2–28 and MDA-MB-231) which hints at a role ER signaling may play in EGR2 regulation. We have previously shown that ER signaling is upregulated in response to SFRP1 loss in both human and mouse tissues [22] which further supports the hypothesis that ER signaling may be involved in EGR2 expression. Data presented by Windahl et al. demonstrate that when one of the activation functions (AF1) within the ER gene is disrupted in mice, osteoblasts exhibit a blunted Egr2 mRNA expression in response to mechanical strain [43]. Taken together, more research is required in order to elucidate the role ER plays in EGR2 regulation.

TGF-β and Wnt signaling regulate a variety of physiological processes including mammary gland development and tumorigenesis. SFRP1 has been associated with the control of TGF-β and Wnt signaling. Specifically, Sfrp1^−/− derived mammary tissues express increased levels of Tgfb1 and Wnt4 mRNA and TERT-siSFRP1 cells are more sensitive to TGF-β and Wnt signaling [30, 32]. Fang et al. was the first group to reveal that Egr2 is a transcriptional target of TGF-β [26]. Here we demonstrate that TGF-β stimulates and TGFBR inhibition represses EGR2 mRNA expression in both HMECs as well as MMECs (Additional file 2: Figure S2, Fig. 2). We have also observed that tumors derived from Sfrp1^−/− mice express significantly higher levels of both Tgfb1 and Egr2 mRNA (data not shown). Dillon et al. revealed that Egr2 expression is upregulated in Erbb2 driven mammary tumors [25]. However, these researchers did not evaluate the role TGF-β plays in their model of tumorigenesis and therefore future research will be directed identifying how loss of SFRP1 together with TGF-β1, EGR2 upregulation, and additional tumor initiating pathways promote mammary carcinogenesis.

We have previously shown that activated ERK1/2 levels and the migratory action of TERT-siSFRP1 cells are drastically reduced in response to TGF-βR inhibition which is consistent with work described by Imamichi et al. showing that TGF-β signaling mediates the cellular migration of breast cancer cells by ERK1/2 activation [31]. Our current data shows that inhibition MAPK signaling in Sfrp1^−/− murine mammary glands or knockdown mammary epithelial cells results in decreased expression of EGR2. These findings are consistent with findings described by Zaman et al. showing that MEK1/2 inhibition the osteoblast UMR106 cell line resulted in decreased EGR2 expression (26). Chandra et al. demonstrate that the MAPK/ERK pathway is a major downstream signaling pathway mediating the stimulatory effects of EGF on EGR2 expression and osteoprogenitor survival [28]. Finally, To et al. report that the same MEK inhibitor utilized in our experiments, U0126, was the elicited the most potent inhibition of EGR2 transcription in breast adipose fibroblasts [29].

In solid tumors, 5–40% of the tumor mass consists of tumor-associated macrophages (TAMs) and poor prognosis is associated with elevated levels of TAMs [44]. Stimuli in the tumor environment polarize TAMs towards a protumor M2 rather than an anti-tumor M1 phenotype [45]. TGF-β promotes tumor progression by recruiting TAMs to compete with dendritic cells by suppressing their antigen-presentation [34]. Zhang et al. provide evidence that TGF-β blocks M1 macrophage development while promoting the activation of M2 macrophages [46]. The expression of Egr2 has recently been utilized to describe murine the M2 macrophages [19]. Our results add to these finding by showing that the addition of extracellular recombinant SFRP1 represses TGF-β stimulated M2 polarization. Moreover, human explant mammary gland cultures treated with rSFRP1 show a marked reduction in the human M2 marker, CD163. CD163 is a monocyte/macrophage-restricted scavenge receptor [47] and the mechanism by which rSFRP1 reduces CD163 expression may be due in part to repression of Wnt signaling because Bergenfelz et al. show that breast cancer CD163⁺ TAMs correlate with Wnt5a expression, which is responsible for macrophage reprogramming to an anti-inflammatory M2 status [48].