Pparγ1 Facilitates ErbB2-Mammary Adenocarcinoma in Mice

Simple Summary HER2, which is associated with clinically aggressive disease, is overexpressed in 15–20% of breast cancers (BC). Peroxisome proliferator-activated receptor γ (PPARγ), is expressed in a variety of malignancies. The aim of our study was to determine the function of endogenous Pparγ1 in the onset and progression of mammary tumors induced by ErbB2 in mice. Genetic deletion of Pparγ1 slowed the rate of tumor progression and death from ErbB2-induced mammary tumors. The deletion of Pparγ1 correlated with reduced pro-tumorigenic inflammation. We conclude ErbB2 collaborates with endogenous Pparγ1 in the onset and progression of mammary tumorigenesis. Abstract HER2, which is associated with clinically aggressive disease, is overexpressed in 15–20% of breast cancers (BC). The host immune system participates in the therapeutic response of HER2+ breast cancer. Identifying genetic programs that participate in ErbB2-induced tumors may provide the rational basis for co-extinction therapeutic approaches. Peroxisome proliferator-activated receptor γ (PPARγ), which is expressed in a variety of malignancies, governs biological functions through transcriptional programs. Herein, genetic deletion of endogenous Pparγ1 restrained mammary tumor progression, lipogenesis, and induced local mammary tumor macrophage infiltration, without affecting other tissue hematopoietic stem cell pools. Endogenous Pparγ1 induced expression of both an EphA2-Amphiregulin and an inflammatory INFγ and Cxcl5 signaling module, that was recapitulated in human breast cancer. Pparγ1 bound directly to growth promoting and proinflammatory target genes in the context of chromatin. We conclude Pparγ1 promotes ErbB2-induced tumor growth and inflammation and represents a relevant target for therapeutic coextinction. Herein, endogenous Pparγ1 promoted ErbB2-mediated mammary tumor onset and progression. PPARγ1 increased expression of an EGF-EphA2 receptor tyrosine kinase module and a cytokine/chemokine 1 transcriptional module. The induction of a pro-tumorigenic inflammatory state by Pparγ1 may provide the rationale for complementary coextinction programs in ErbB2 tumors.

Recent studies have suggested an association between Peroxisome proliferator-activated receptor γ (PPARγ) signaling and lapatinib-resistant breast cancer. PPARγ is a member of the nuclear receptor (NR) superfamily that regulates diverse biological functions including lipogenesis and differentiation, inflammation, insulin sensitivity, cellular proliferation and autophagy [36][37][38][39]. Evidence for PPARγ as a tumor growth inhibitor includes the detection of heterozygous PPARγ mutations in colon cancer and the finding that PPARγ agonists reduce tumorigenesis in murine models [40][41][42]. In contrast several lines of evidence suggest PPARγ augments growth as PPARγ ligands increased gastrointestinal polyp number in the Apc mouse model of familial adenomatosis [43]. Although the preponderance of studies demonstrated PPARγ restrains inflammation [44,45], the impact on anti-tumor immune responses and the role for endogenous Pparγ in promoting local tissue and disease-specific inflammatory processes appears to be more complex as a proinflammatory effect of Pparγ [46], the induction of tissue specific pools of macrophages [47] have been described. Furthermore, studies that deployed ER-Hoxb8-immortalized bone marrow-derived macrophages from Pparγand LysM-Cre Ppar fl/f mice, showed loss of Pparγ resulted in reduced F4/80 + macrophages [46]. Although Pparγ has been implicated in the regulation of cellular growth, the role of endogenous Pparγ1 in the growth of mammary tumors and the impact on the anti-tumor immune response was not previously known. In order to determine the role of endogenous Pparγ1 in mammary tumorigenesis the Pparγ1 gene was deleted in mammary targeted ErbB2 transgenic mice.

Deletion of Endogenous PPARγ1 Reduces ErbB2-Induced Mammary Tumor Growth
Previous studies demonstrated that PPARγ may either enhance or inhibit tumor growth depending upon the tumor type and oncogenic driver [40][41][42][43]. In order to define the role of endogenous Pparγ1 in the onset and growth of ErbB2-induced mammary tumorigenesis in an immune-competent context in vivo, we generated multigenic mice in which the Pparγ1 exon 2 alleles were flanked by LoxP sites and the induction of Cre recombinase was governed through the addition of Tamoxifen (TMX) (Rosa26 ERT2Cre ) with the mT/mG transgenic reporter (The mT/mG transgenic reporter mouse cells that have not undergone recombination express membrane-targeted tdTomato (mT), but after recombination they express membrane-targeted enhanced green fluorescent protein (EGFP)). These mice were intercrossed with the transgenic mammary gland oncomouse MMTV-ErbB2 ( Figure 1A) in order to generate transgenic mice, Pparγ1 fl/fl CreERT2 + mTmG + MMTV-ErbB2 + ( Pparγ1 −/− ), and the control mice, Pparγ1 wt/wt CreERT2 + mTmG + MMTV-ErbB2 + (Pparγ1 +/+ ). The multigenic offspring were treated with a 5 days pulse of Tamoxifen ( Figure 1B and S2A), which was sufficient to induce Cre expression, deletion of the target gene and conversion of the tdTomato expressing mammary epithelium to EGFP ( Figure 1C). Tumor formation was compared between the transgenic mice (Pparγ1 −/− vs. Pparγ1 +/+ ) treated with Tamoxifen with analysis conducted over the subsequent 6 months ( Figure 1D). Statistical analysis demonstrated that the rate of MMTV-ErbB2 induced mammary tumor onset was delayed in mice in which the Pparγ1 gene had been deleted through Cre recombinase (by the log-rank test, p = 0.043; by Wilcoxon rank-sum test, p = 0.028 or by the Tarone-Ware test, p = 0.033) ( Table S1). The potential independent effect of Tamoxifen on mammary tumor rate was controlled for, as each line received identical exposure to tamoxifen. Growth rate of established ErbB2 mammary tumors upon Cre-mediated deletion of Pparγ1 was determined as well. Tumor volumes in MMTV-ErbB2, ROSA26 CreERT2/mTmG mice that were either Pparγ1 +/+ or Pparγ1 −/− were showed in Figure S2B (n = 4 of each genotype). The estimated tumor growth rate for each group was calculated based on the mixed effects models and showed in Table S2. The results showed that tumor growth in Pparγ1 −/− mice was slower than in Pparγ1 +/+ mice. In order to determine the molecular mechanisms by which Pparγ1 maintained ErbB2 mammary tumors, analysis of RNA-seq based gene expression derived from the Pparγ1 −/− vs. Pparγ1 +/+ mammary tumors was conducted, revealing deletion of endogenous Pparγ1 changed substantially (up to 24-fold) the levels of gene expression ( Figure S1A), with reduced expression of Pparγ1-responsive genes (PERM1, (Peroxisome Proliferator-Activated Receptor Gamma Coactivator 1 And Estrogen-Related Receptor-Induced Regulator In Muscle 1), PEX11G (Peroxisomal Biogenesis Factor 11 Gamma), Cyp27A1 and Pygl1 (glycogen phospho-rylase1 ( Figure 1E), and reduced lipogenic and metabolic genes ( Figure 1F). Consistent with reduced anabolism and lipogenesis Pparγ1 −/− ErbB2 tumors showed reduced expression of anabolic solute carriers ( Figure 1G) and reduced expression of growth promoting transcription factors including reduced nuclear receptors Esr1 (ERα), Tcf25, and Tcf15 ( Figure 1H). The abundance of key enzymes required for de novo lipogenesis, including sterol regulatory element-binding protein (SREBP1, SREBP2) and fatty acid synthase (FASN) was reduced in Pparγ1 −/− vs. Pparγ1 +/+ ErbB2 mammary tumors ( Figure 1I-K and Figure S1B). CD31 (cluster of differentiation 31) expression was not significantly changed in endothelial cells ( Figure 1L).
The MCF10A-NeuT-PPARγ1 cells showed reduced transwell migration compared with vector control. In order to determine the role of PPARγ1 in ErbB2 induced growth in xenografts, MCF10A-NeuT cells were transduced with PPARγ1 or vector control as indicated in Figure S2D. 5 × 10 6 cells were implanted into the mammary fat pad of immunedeficient nude mice. Tumor growth, measured at 20 days by digital caliper, showed a substantial increase in tumor volume ( Figure S2E, Data are shown as mean ± SEM for n = 13).

Pparγ1 Augments an EphA2-amphiregulin Growth Gactor Signaling Module
Ki-67 staining, a marker of cellular proliferation, demonstrated an increased percentage of Ki-67 + staining cells in the Pparγ1 +/+ ErbB2 mammary adenocarcinoma (Figure 2A,B). Pathway analysis showed Pparγ1 augmented the activity of an Eph receptor node and a protein tyrosine kinase node, which in turn activated atypical PKC, Neu3 and ADAM12, whereas FGFR2 signaling was restrained ( Figure 2C). RNA-seq identified the induction of a growth factor module regulators including EGF, Amphiregulin, IRS2, AngP2, CXCl5, and SDF2l1 ( Figure 2D). The abundance of mRNA encoding growth factors known to enhance ErbB2 signaling, including Egf, Amphiregulin, EphA2, and Adam12 was increased ( Figure 2D). ADAM12 (a disintegrin and metalloproteinase 12), is linked to the induction of EphA2-dependent cell migration [48]. EphA2, which is known to augment ErbB2-induced mammary tumorigenesis [7], was increased 4-fold in the Pparγ1 +/+ ErbB2 mammary tumors ( Figure 2E). EphB6 signaling, known to suppress breast cancer cell aggressiveness by interacting with EPHB4 and interfering with EPHB4 action [49], was repressed by Pparγ1 in ErbB2 mammary tumors ( Figure 2F). In order to examine further the relationship between Pparγ and the abundance of the growth factors maintained by endogenous Pparγ1 in the murine mammary tumors, we interrogated a public database (https://www.cbioportal.org/study/summary?id=brca_mbcproject_wagle_2017 (accessed on 2-15 December 2020)) in which mRNA expression of breast cancer samples had been EphA2 abundance, assessed by immunohistochemistry, was increased approximately 5-fold ( Figure 2N,O). Collectively these studies are consistent with a model in which endogenous Pparγ1 maintains the abundance of a growth promoting module (ADAM12, EGFR, EphA2), each component of which is known to enhance ErbB2 function.

Pparγ1 Governs the Mammary Tumor Immune Response
In order to determine the signal transduction pathways maintained by Pparγ1 in the ErbB2 mammary tumors, we conducted Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of RNA-seq from the tumors. Deletion of Pparγ1 reduced activity of the pathways "cytokines/chemokine receptors signaling" and the related "Graft versus Host disease", "NOD-like receptor signaling" and "NFκB signaling" ( Figure 3A). Analysis of significantly affected genes by Ingenuity Pathway Analysis (IPA) identified potentially altered upstream regulators related to cancer and the inflammatory response. INFγ-induced genes were decreased by deletion of endogenous Pparγ1 ( Figure 3B,C), consistent with the known induction of Interferons (IFNs) and IFN-stimulated genes by peroxisomes [50,51]. Pparγ1 +/+ ErbB2 mammary tumors showed increased expression of specific chemokines and cytokines (Cxcl5, Cxcl19, Cxcl13, IL1b and Tnfrsf13c ( Figure 3D)). As IFNG and IL6 were the most induced by number of genes and Z score (circled in Figure 3C), we examined the relationship between Pparγ and IFNG and IL6 in mRNA expression data from The Metastatic Breast Cancer Project (Provisional, February 2020) ( Figure 3E,F). These analyses revealed a significant correlation in human metastatic breast cancer between Pparγ and IFNGR1 (n = 136, Pearson 0.57, p = 17.13 × 10 −14 ), and with IL6 (n = 136, Pearson 0.55, p = 6.1 × 10 −13 ). The F4/80 + marker of murine tissue associated macrophages (TAM) confirmed a significantly increased proportion in the Pparγ1 +/+ tumor population within both the tumor and the tumor stroma ( Figure 3G,H). Collectively these findings are consistent with a model in which endogenous Pparγ1 maintains an inflammatory tumor microenvironment that includes an increase in F4/80 + macrophages.

Pparγ1 in B Cell Differentiation
TAM can be recruited from inflammatory monocytes, tissue generated macrophages [52,53], and can be generated from Pre/proB cells during inflammation [54]. These monocytes are derived from primarily the bone marrow and also the spleen [32,55]. In order to determine whether the changes in the tumor immune environment were a function of Pparγ1 on the hematopoietic system, we conducted a detailed analysis of the hematopoietic cell system. The hematocrit (HCT), counts of white blood cells (WBC), red blood cells and platelets were not significantly changed ( Figure S3A-D). The proportion of WBC subtypes in the peripheral blood were unchanged ( Figure S3E). The F4/80 + macrophage and Mac1 + Gr1 − cellularity counts in the bone marrow, blood and spleen and were unchanged ( Figure S3F,G). The spleen showed a modest by significant increase in the percentage of Mac1 + Gr1 + granulocytes ( Figure S3H).
Subsorting for the proportion of ProB, PreB, immature and mature B cells in the bone marrow did not show significant differences between Pparγ1 genotypes ( Figure 4A,B). Analysis of the B-cell lineage in the spleen showed a decrease in the proportion of mature splenic B cells in Pparγ1 −/− mice (B220 hi CD19 + ) (54.7 vs. 45.7, p < 0.05, Figure 4C, n = 10 total mice). There was no significant change in the CD4 − CD8 + T cell compartment ( Figure S4).

Pparγ1 Does Not Influence Hematopoietic Stem Cell Population (HSC) Differentiation
Given the small but significant reduction in the mature B cell population, we assessed the impact of Pparγ1 on the hematopoietic stem cell population (HSC). In the adult mouse, all multipotent cells are contained in the Lineage −/low Sca-1 + c-Kit + (LSK) fraction of bone marrow cells, which was similar between Pparγ1 genotypes ( Figure 5A). Assays using flow cytometry to quantify HSPC (Flk2/CD34) demonstrated that HSPCs were unaltered although there was a trend towards a reduction in the Flk2 − CD34 + (ST-HSC or short term HSC) population in Pparγ1 −/− ( Figure 5B,C). The distribution of LSK CD34 − Flk2 − (representing long term-HSC) and LSK CD34 + Flk2 + multipotent progenitors (MPP) ( Figure 5B,C) was unchanged.
SLAM family markers, CD150, CD48, CD229, and CD244, can distinguish HSCs and MPPs from restricted progenitors and subdivide them into a hierarchy of functionally distinct subpopulations with stepwise changes in cell-cycle status, self-renewal, and reconstituting potential [56]. The frequency of hematopoietic stem cells and progenitors (HSPCs) assessed by SLAM markers (LSK CD150/CD48) ( Figure 5D,E), was unchanged. A quantification of colony forming progenitors and colony subtypes types (M, G, GM, GEMM; (granulocyte (G), erythrocyte, monocyte, megakaryocyte)) [57] demonstrated Pparγ1 +/+ did not affect colony forming ability ( Figure S5A,B). As Pparγ did not increase the proportion of F4/80 + macrophages in the bone marrow, blood or spleen, these studies suggest the reduction in tumor-associated F4/80 + macrophages within the ErbB2 Pparγ1 −/− mammary tumors is not due to a Pparγ1-mediated alteration in hematopoietic precursors in the bone marrow or the circulation.  Given the small but significant reduction in the mature B cell population, we assessed the impact of Pparγ1 on the hematopoietic stem cell population (HSC). In the adult mouse, all multipotent cells are contained in the Lineage −/low Sca-1 + c-Kit + (LSK) fraction of bone marrow cells, which was similar between Pparγ1 genotypes ( Figure 5A). Assays using A quantification of colony forming progenitors and colony subtypes types (M, G, GM, GEMM; (granulocyte (G), erythrocyte, monocyte, megakaryocyte)) [57] demonstrated Pparγ1 +/+ did not affect colony forming ability ( Figure S5A,B). As Pparγ did not increase the proportion of F4/80 + macrophages in the bone marrow, blood or spleen, these studies suggest the reduction in tumor-associated F4/80 + macrophages within the ErbB2 Pparγ1 −/− mammary tumors is not due to a Pparγ1-mediated alteration in hematopoietic precursors in the bone marrow or the circulation.

Pparγ1 Binds in Chromatin Immune Precipitation Assays to the Regulatory Regions of the EphA-Amphiregulin and the Chemokine Signaling Axis Genes in Breast Cancer Cells.
ChIP-Seq was used to define the genome-wide DNA sequence-specific binding characteristics regulated by Pparγ1 in breast cancer cells. High confidence Pparγ1 ChIP-Seq peaks were identified in MCF10A-NeuT cells expressing Pparγ1 Wt. Analysis of the ChIP-Seq data ( Figure 6A) demonstrated 13,488 genes were selectively bound by Pparγ1 Wt as defined by the limits of within 10 kb upstream of the transcriptional start site (TSS), 10 kb

Pparγ1 Binds in Chromatin Immune Precipitation Assays to the Regulatory Regions of the EphA-Amphiregulin and the Chemokine Signaling Axis Genes in Breast Cancer Cells
ChIP-Seq was used to define the genome-wide DNA sequence-specific binding characteristics regulated by Pparγ1 in breast cancer cells. High confidence Pparγ1 ChIP-Seq peaks were identified in MCF10A-NeuT cells expressing Pparγ1 Wt. Analysis of the ChIP-Seq data ( Figure 6A) demonstrated 13,488 genes were selectively bound by Pparγ1 Wt as defined by the limits of within 10 kb upstream of the transcriptional start site (TSS), 10 kb downstream of the TES and binding within the gene. We next compared the position of ChIP-Seq peaks relative to gene transcription start site and evaluated the number of binding peaks at decreasing intervals upstream from the transcription start site as indicated ( Figure 6B). We then determined by ChIP-Seq analysis the binding of PPARγ1 to the regulatory regions of genes involved in both the growth factor induced module and the inflammatory module. Given the induction by Pparγ1 of EphA2, Amphiregulin, Adam12 in the mammary tumors we sought to determine whether PPARγ1 directly interacted in the context of chromatin with these target genes. We first examined the known Pparγ1 target pyruvate dehydrogenase kinase 4 (PDK4) [58] the expression of which is known to regulate de novo lipogenesis. PPARγ1 was enriched in the PDK4 ChIP ( Figure 6C). Furthermore PPARγ1 was enriched in the EPHA2, AREG, ADAMTS, ADAM12 and the IL34 and IL1B ChIP ( Figure 6D-I).
ing peaks at decreasing intervals upstream from the transcription start site as indicated ( Figure 6B). We then determined by ChIP-Seq analysis the binding of PPARγ1 to the regulatory regions of genes involved in both the growth factor induced module and the inflammatory module. Given the induction by Pparγ1 of EphA2, Amphiregulin, Adam12 in the mammary tumors we sought to determine whether PPARγ1 directly interacted in the context of chromatin with these target genes. We first examined the known Pparγ1 target pyruvate dehydrogenase kinase 4 (PDK4) [58] the expression of which is known to regulate de novo lipogenesis. PPARγ1 was enriched in the PDK4 ChIP ( Figure 6C). Furthermore PPARγ1 was enriched in the EPHA2, AREG, ADAMTS, ADAM12 and the IL34 and IL1B ChIP ( Figure 6D-I).

Discussion
The current studies extend our understanding of Her2-function by identifying the critical role for Pparγ1 in mediating ErbB2-mammary tumor progression in vivo. Using tetra-transgenic inducible, gene deletion mice, we show that the Pparγ1 gene is required for the progression of ErbB2-induced mammary tumorigenesis in immune competent mice. Second, Pparγ1 gene deletion reduced the expression of genes within signaling pathways mediating growth, including growth factors and growth factor receptors (EphA2, EGF, Amphiregulin, Adam12) ( Figure 6J). Third, Pparγ1 gene deletion reduced the influx of an F4/80 + macrophage inflammatory infiltrate into the mammary tumors, without affecting hematopoietic cell differentiation. The immune infiltration of the ErbB2 tumors was associated with the induction of lipogenesis, peroxisome signaling, INFγ and Cxcl5 [59]. Collectively these studies suggest Pparγ1 may represent a useful target for coextinction strategies of ErbB2 induced breast cancer.
Pparγ1 deletion reduced expression of Pparγ1 target genes, genes governing growth factor signaling pathways (EGF, IRS2, AngP2, CXCl5, SDF2), transcription factors that promote breast tumor growth including nuclear receptors and anabolic solute carriers (Slc25a35 (oxaloacetate carrier), Slc39a13, Slc39a14, Slc15a5, Slc18a1, Slc17a7) which may participate in cellular growth [72]. The induction of EGF, EphA2 and ADAM12 are known to augment ErbB2-induced tumorigenesis ( Figure 6J). EGF is known to enhance ErbB2 signaling via the EGFR/ErbB2 heterodimer [73]. Endogenous Pparγ1 enhanced EphA2 mRNA and protein abundance in the ErbB2 adenocarcinoma. EphA2 is known to enhance ErbB2induced mammary adenocarcinoma [5,7]. Co-expression of ErbB2 and EphA2 is sufficient to induce tyrosine phosphorylation of EphA2 in the absence of ligand and EphA2 [5,7]. A disintegrin and metalloproteases (ADAMs) are transmembrane metalloproteases that process and shed the ectodomains of membrane-anchored growth factors, cytokines and receptors [74]. ADAM12 is linked to the induction of EphA2-dependent cell migration in metastatic tumors [48]. The membrane type-I matrix metalloproteinase (MT1-MMP) also promotes cancer cell migration and invasion via EphA2 receptor cleavage [75] ADAM12 modulates intracellular signaling by cleaving various membrane bound signaling receptors and their ligands. ADAM12 is highly expressed in glioblastoma multiforme, where it is linked to shedding of HB-EGF. cancer cell invasion, including ADAMs (as reviewed above), and interestingly, ADAM12 [48], and MMP2 and 9 [76] are all upregulated by TGF-β1 signaling in tumor metastasis. ADAM12, ephrin-A1, and EphA2-contribute to growth or cell migration in primary and metastatic tumors [48,77,78].
In order to determine the molecular mechanisms by which Pparγ1 enhanced ErbB2mediated mammary tumorigenesis tissue culture-based experiments were conducted to examine cell autonomous function. MCF10A-ErbB2-PPARγ1 cells displayed reduced transwell migration compared with vector control. These findings suggest that a cell autonomous effect on migration does not contribute to the tumorigenic phenotype. Prior experiments had supported a model in which Pparγ1 mediated cell non autonomous functions through the induction of secreted factors that promoted angiogenesis [79]. Herein, endogenous Pparγ1 enhanced tumorigenic inflammation. First, Infiltration of F4/80 + macrophages into the mammary tumors of transgenic mice, which are known to augment ErbB2-induced tumor progression [80,81], was reduced by Pparγ1 deletion. EGF, which was increased upon Pparγ1 gene deletion, is known to augment a feedforward loop within the TME to enhance TAM recruitment [33][34][35]. Interferon signaling (INFγ1, INFβ1), was augmented by endogenous Pparγ1. IPA analysis identified interferon signaling as a predominant pathway induced by Pparγ1. Peroxisomes induced by Pparγ1, activate Type1 interferons (IFNs) and IFN-stimulated gene expression [50,51], including Cxcl5 and Cxcl9. Peroxisome proliferation requires PEX11-type peroxisomal proliferators [82] and herein, endogenous Pparγ1 induced peroxisomal target genes in the mammary tumors as evidenced by increased expression of PEX-11, together with PPARGC1 and ESRR induced regulator, muscle 1, PGC-1 and ERR-induced regulator in muscle 1 (Perm1). Cxcl5 is a key local cue that recruits tumor associated macrophages (TAMs), which in turn promote a pro-tumorigenic environment [26,52,53]. CXCL9 which may also be pro-tumorigenic [59], is upregulated during BCa therapy [83] and participates in augmenting the response to cancer therapy with checkpoint blockade [84]. IL-1b [85,86] and IL34, which were induced by Pparγ1 are also known to induce a pro-tumorigenic tumor microenvironment (TME) with IL-34 enhancing recruitment and survival of TAM [87].
In order to determine potential mechanisms by which Pparγ1 coordinated the induction of growth factor and cytokine/chemokine signaling in ErbB2 tumors, we conducted genome wide PPARγ1 chromatin binding assays in human mammary adenocarcinoma cells. The expression of genes governing growth factor (EphA2, amphiregulin, Adam12, AdamTS2) and a chemokine module (CXCL5, IL1B,IL34, TNFRSF13C), were induced by PPARγ1 and bound in ChIP to target regions within the genes. PPARγ1 ChIP analysis in MCF10A-NeuT mammary tumors showed the number of sites bound by PPARγ1 Wt including all cis elements was comparable to the~23,000 and~21,000 sites identified in human adipocyte cell lines [88], but fewer than the~40,000 sites identified in primary in vitro differentiated human adipose derived stem cells. In adipocytes, C/EBPs co-localizes with PPARγ1 at the majority of its binding sites and cooperate in target gene transcription [88,89], whereas in macrophages, PPARγ1 ChIP is enriched with the hematopoietic transcription factor PU.1 [90]. Collectively these findings are consistent with a model in which Pparγ1 promotes the expression of gene governing lipogenesis, mammary tumor growth and a local tumor immune response. Pparγ1 should be considered further for coextinction paradigms in ErbB2-mediated adenocarcinoma.

Statistics
All means were compared using two-tailed Student's t-test. A p value < 0.05 was considered significant. RNA-seq data was aligned using bowtie2 [95] against mm10 genome and RSEM v1.2.12 software [96] was used to estimate gene-level read counts using Ensemble transcriptome information. Raw gene counts were used in DESeq2 [97] algorithm to estimate significance of differential expression between Pparγ1 +/+ and Pparγ1 −/− groups. Expression heatmaps were plotted using mean-centered log2-scaled normalized by DESeq2 expression values. Gene set enrichment analysis was done using Qiagen's Ingenuity ® Pathway Analysis software (IPA ® , Qiagen, Redwood City, CA, USA, www.qiagen.com/ingenuity) on genes that passed nominal p < 0.05 using "Upstream Regulators" and "Networks" option. Upstream regulators with significantly predicted activation state (|Z-score|>1) that in addition passed p < 0.05 target enrichment threshold with at least 5 target genes were reported. Identified gene interaction networks with IPA score of at least 25 were considered. Activation states and fold changes were reported for Pparγ1 +/+ condition relative to Pparγ1 −/− . KEGG pathway analysis was conducted as previously described [98]. Analyses of human gene expression derived from The Metastatic Breast Cancer Project (Provisional, February 2020) were conducted using the cBioPortal analytical tools including the Pearson correlation coefficients.

Colony Forming Assay
Total BM from femur was plated onto M3434 semi-solid methylcellulose media (STEM-CELL Technologies, Vancouver, BC, Canada). Colonies were enumerated 7 days after plating.

Transwell Migration and Xenografts
Transwell migration assay was performed in 24 well transwell inserts (Corning, Tewksbury, MA, USA) as previously described [103,104]. See the supplemental methods for the details. MCF10A-NeuT cells stably expressing PPARγ1, or vector control were previously established described [102] and used in mammary fat pad xenografts as described [105].

Immunohistochemistry (IHC) Staining
IHC staining was performed on the paraffin-embedded tissue blocks in the Kimmel Cancer Center Pathology Core Facility at Thomas Jefferson University. Quantification was conducted using ImageJ software.

ChIP-Seq
For the ChIP-Seq [106], ChIP and input libraries were generated from at least three distinct biological samples from MCF10A-NeuT cells stably expressing vector control or PPARγ1 Wt. Approximately 10 ng of ChIP DNA (quantified by Qubit 2.0 Fluorometer, Invitrogen) was prepared for sequencing according to the amplification protocol from Illumina using enzymes from New England Biolabs (Ipswich, MA, USA) and PCR purification (#28104) and MinElute (#28004) kits from Qiagen. Deep sequencing was performed by the Functional Genomics Core (J. Schugg and K. Kaestner) of the Penn Diabetes Research Center using Illumina HiSeq 2000 and aligned sequences were obtained using the Solexa Analysis Pipeline.

ChIP-Seq Data Analysis
Tag Alignment and Peak Calling. Binding sites for PPAR WT were inferred from ChIP-Seq analysis of chromatin occupancy. ChIP-Seq tags in FASTQ format were mapped to the human genome version hg19 using the bwa-mem aligner with default parameters. ChIP-Seq data analysis was then performed following ENCODE guidelines to generate high quality peak calls [107]. Aligned reads, or "tags," were filtered based on mapping quality to retain only those with MAPQ score of 30 or higher for further analysis. Within each IP "treatment" group (vector control, PPARγ Wt), peak calling was performed on indi-vidual replicates using pooled input as a control using MACS2 peak calling software [108]. Peak calling was also performed on pooled IP "treatment" samples versus pooled input "control" samples within each group. Peak calling was performed using an estimated 150 bp fragment size, and a 10% FDR threshold. MACS2 automatically filters out duplicate reads, and since each individual replicate had less than the recommended 15M unique tags, the MACS2 option "-to-large" was used to linearly scale up read depth to match the number of input tags; the "-to-large" option was not used during the analysis of pooled IP samples. Consistency between ChIP replicates was evaluated using the Irreproducible Discovery Rate (IDR) methodology. Per the IDR framework, individual replicate samples were used to determine the optimal number of peaks in the pooled replicates at an IDR threshold of 5%.

Annotation and Motif Finding
Annotation and motif finding within putative PPARγ binding peaks was performed using the databases and tools provided as part of the Hypergeometric Optimization of Motif EnRichment (HOMER) software suite. The HOMER find Motifs Genome command was used to search peak regions for known and novel motifs. The HOMER annotate Peaks command was used to generate region statistics and identify the closest genomic feature relative to each peak, based on the hg19 genome. The annotate Peaks command was also used in tss-centric, rna-centric, and tts-centric mode to identify the closest peak with respect to transcription start sites, transcribed mRNA regions, and transcription termination sites, respectively.

Conclusions
Peroxisome proliferator-activated receptor γ (PPARγ) has been implicated in either promoting or inhibiting tumorigenesis, primarily deduced through using agonists of the receptor. Herein, genetic deletion of Pparγ1 in a murine model of human ErbB2-mediated breast cancer, the rate of tumor progression and death was significantly reduced. Furthermore, the genetic deletion of Pparγ1 correlated with reduced pro-tumorigenic inflammation and lipogenesis in the mammary tumors of the mice. The induction of local mammary tumor macrophage infiltration occurred without affecting other tissue hematopoietic stem cell pools. Endogenous Pparγ1 induced expression of both an EphA2-Amphiregulin and an inflammatory INFγ and Cxcl5 signaling module, that was recapitulated in human breast cancer. We conclude ErbB2 collaborates with endogenous Pparγ1 in the onset and progression of mammary tumorigenesis.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/cancers13092171/s1, Supplemental materials and methods, Table S1: Test of equality of the survival distribution function (DF = 1), Table S2: Estimated tumor growth rate, Figure S1: Deletion of Pparγ1 reduces adipogenesis in MMTV-ErbB2 transgenic mammary tumors, Figure S2: Growth rate of established ErbB2 mammary tumors upon Cre-mediated deletion of Pparγ1, Figure S3: Impact of Pparγ1 gene deletion on hematopoietic cell peripheral tissue distribution in multigenic mice, Figure  S4: T cell lineage indices of Pparγ1 +/+ vs. Pparγ −/− multigenic mice, Figure S5: Colony forming hematopoietic progenitor cell assays. Institutional Review Board Statement: The appropriate Thomas Jefferson University institutional committee approved protocols were followed when working with all mice. Animals were housed and maintained at Thomas Jefferson University in a pathogen-free barrier facility under National Institutes of Health (NIH) guidelines.
Informed Consent Statement: Not applicable. As these studies did not involve humans.
Data Availability Statement: All of the data generated from this study are available upon request.
Conflicts of Interest: R.G.P. holds ownership interests in LightSeed, Inc., CytoDyn Inc., and ownership interests (value unknown) for several patents and submitted patent applications, none of which influenced the representation or interpretation of reported research results, nor had any role in the design of the study; in the collection, analyses or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. The other authors declare that they have no conflict of interest.