Retinoid X receptor suppresses a metastasis ...

Retinoid X receptor suppresses a metastasis-promoting transcriptional program in myeloid cells via a ligand-insensitive mechanism Mate Kissa, Zsolt Czimmerera,b, Gergely Nagyb, Pawel Bieniasz-Krzywiecc,d, Manuel Ehlingc,d, Attila Papa, Szilard Poliskaa,e, Pal Botof, Petros Tzerpose, Attila Horvatha, Zsuzsanna Kolostyaka, Bence Danielg, Istvan Szatmarif, Massimiliano Mazzonec,d, and Laszlo Nagya,b,g,1 a Nuclear Receptor Research Laboratory, Department of Biochemistry and Molecular Biology, University of Debrecen, 4032 Debrecen, Hungary; bMTA-DE “Lendület” Immunogenomics Research Group, University of Debrecen, 4032 Debrecen, Hungary; cLaboratory of Tumor Inflammation and Angiogenesis, Center for Cancer Biology, VIB, 3000 Leuven, Belgium; dLaboratory of Tumor Inflammation and Angiogenesis, Department of Oncology, KU Leuven, 3000 Leuven, Belgium; eUD-GenoMed Ltd., 4032 Debrecen, Hungary; fStem Cell Differentiation Laboratory, Department of Biochemistry and Molecular Biology, University of Debrecen, 4032 Debrecen, Hungary; and gGenomic Control of Metabolism Program, Sanford Burnham Prebys Medical Discovery Institute at Lake Nona, Orlando, FL 32827

Retinoid X receptor (RXR) regulates several key functions in myeloid cells, including inflammatory responses, phagocytosis, chemokine secretion, and proangiogenic activity. Its importance, however, in tumor-associated myeloid cells is unknown. In this study, we demonstrate that deletion of RXR in myeloid cells enhances lung metastasis formation while not affecting primary tumor growth. We show that RXR deficiency leads to transcriptomic changes in the lung myeloid compartment characterized by increased expression of prometastatic genes, including important determinants of premetastatic niche formation. Accordingly, RXR-deficient myeloid cells are more efficient in promoting cancer cell migration and invasion. Our results suggest that the repressive activity of RXR on prometastatic genes is mediated primarily through direct DNA binding of the receptor along with nuclear receptor corepressor (NCoR) and silencing mediator of retinoic acid and thyroid hormone receptor (SMRT) corepressors and is largely unresponsive to ligand activation. In addition, we found that expression and transcriptional activity of RXRα is down-modulated in peripheral blood mononuclear cells of patients with lung cancer, particularly in advanced and metastatic disease. Overall, our results identify RXR as a regulator in the myeloid cell-assisted metastatic process and establish lipid-sensing nuclear receptors in the microenvironmental regulation of tumor progression. retinoid X receptor

and identified a network of RXR-bound enhancers that control angiogenic genes including Vegfa, Hbegf, Litaf, and Hipk2 (9). Activation of these enhancers by RXR induced a proangiogenic transcriptional program and phenotype (9). Interestingly, we found that half of the 5,200 RXR-occupied genomic sites were transcriptionally inactive in macrophages (9). This suggested that the receptor may also possess extensive regulatory functions in myeloid cells that are distinct from ligand recognition and activation. These observations provided a rationale to investigate whether RXR influences tumor progression through regulating the phenotype of recruited or tissue-resident myeloid cells. Results RXR Deletion in Myeloid Cells Promotes Metastasis to the Lung. In

myeloid cells, RXRα shows the highest expression and RXRβ is expressed at a lower level, while RXRγ is not expressed (5). Specific deletion of RXRα in myeloid cells was achieved by crossing mice bearing lox-P-targeted RXRα (RXRαfl/fl) with mice carrying the lysozyme-M Cre (LysM-Cre) recombinase transgene. To exclude the potential compensatory effects of RXRβ in the absence of RXRα resulting from the functional Significance

| metastasis | premetastatic niche | myeloid cell | NCoR

Metastasis formation from malignant tumors is the leading cause of cancer-related deaths. There is an increasing body of evidence indicating that immune cells in distant organs actively contribute to this process by establishing a tissue environment that is hospitable for cancer cells. In this study, we show that deletion of retinoid X receptor (RXR), a cellular sensor of vitamin A metabolites, specifically in the myeloid lineage of the immune system, leads to an enhanced metastasis rate. We also demonstrate that RXR inhibits the expression of a number of genes that encode proteins involved in the promotion of metastasis formation. Surprisingly, our results suggest that this activity of RXR is independent of the presence of its activators.

R

etinoid X receptor (RXR) is a member of the nuclear receptor superfamily of ligand-dependent transcription factors that bind various lipophilic hormones and lipid metabolites. Sensing the lipid milieu by RXR, in concert with other nuclear receptors, plays critical roles in physiology, including the regulation of organogenesis, lipid metabolism, and skin homeostasis (1). RXR is able to bind vitamin A derivatives 9-cis-retinoic acid and 9-cis-13,14-dihydroretinoic acid as well as fatty acids such as docosahexanoic acid and phytanic acid (1, 2). The receptor has three isotypes, RXRα, RXRβ, and RXRγ, which appear to be functionally interchangeable in tissues in which more than one isotype is expressed (3). RXR has a unique integrative function in nuclear receptor signaling because of its ability to form homodimers as well as heterodimers with several other members of the superfamily, such as PPAR, LXR, RAR, and VDR. RXR and its heterodimers have a profound effect on the function of myeloid cells by linking cellular metabolism and immune function (4, 5). Specific activation of RXR has been shown to up-regulate chemokine expression, promote phagocytosis of apoptotic cells, and attenuate antiviral responses in myeloid cells (6–8). To further delineate the function of the receptor, we recently mapped the genomic binding sites of RXR in macrophages

www.pnas.org/cgi/doi/10.1073/pnas.1700785114

Author contributions: M.K., Z.C., I.S., M.M., and L.N. designed research; M.K., Z.C., P.B.-K., M.E., A.P., S.P., P.B., P.T., Z.K., and B.D. performed research; M.K., Z.C., G.N., P.B.-K., M.E., S.P., A.H., and Z.K. analyzed data; and M.K. and L.N. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Freely available online through the PNAS open access option. Data deposition: The data reported in this paper have been deposited in the NCBI Sequence Read Archive database, https://www.ncbi.nlm.nih.gov/sra (accession no. SRP095164). 1

To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1700785114/-/DCSupplemental.

PNAS Early Edition | 1 of 6

IMMUNOLOGY AND INFLAMMATION

Edited by Bert W. O’Malley, Baylor College of Medicine, Houston, TX, and approved August 18, 2017 (received for review January 18, 2017)

B16-F10 lung metastases (macroscopic) 15 * 10

5

0

% within total metastasis no.

D 25

B No. of metastases / coronal section

No. of surface metastases

A

RXR +/+ RXR fl/fl

B16-F10 lung metastases (coronal section) 80

C

*

60 40 20 0

RXR +/+ RXR +/+ RXR fl/fl

B16-F10 lung metastasis size distribution ( 1000 µm) RXR +/+ RXR

20 15 10

RXR fl/fl

5 0

0 0 0 0 0 0 0 0 00 00 -20 0-30 0-40 0-50 0-60 0-70 0-80 0-90 0-10 0-1 100 70 60 40 80 50 20 30 90

Diameter of metastasis (µm)

E

RXR +/+

RXR fl/fl % of Ki67+ cells in metastatic area

Ki67

100 80 60

20

% of cleaved casp-3+ cells in metastatic area

8

LLC primary tumor growth

4 2

RXR +/+ RXR fl/fl

0.6 0.4 0.2 0.0

0

5

10 15 Time (days)

20

25

RXR +/+ RXR fl/fl

LLC lung metastases (macroscopic) Number of surface metastases

Tumor volume (cm3)

0.8

B

RXR +/+ RXR fl/fl

6

0

A

ns

40

0

Cleaved casp-3

redundancy of RXR isotypes (3, 10), we crossed LysM-Cre RXRαfl/fl with RXRβ−/− mice. Given the low fertility and viability of the resulting LysM-Cre RXRαfl/fl RXRβ−/− mice (11), we transplanted their bone marrow to syngeneic C57BL/6J mice. This way, we were able to generate age-matched groups of mice bearing LysM-Cre RXRαfl/fl RXRβ−/− and LysM-Cre RXRα+/+ RXRβ−/− bone marrow (hereafter referred to as RXRαfl/fl and RXRα+/+, respectively). To investigate the effect of RXRα deletion in the myeloid infiltrate on tumor progression, we s.c. implanted RXRαfl/fl and RXRα+/+ mice with Lewis lung carcinoma (LLC), a murine model of non-small cell lung carcinoma showing abundant infiltration of myeloid cells from early stages. Primary tumor growth remained unaltered (Fig. 1A), and the immune cell infiltration in RXRαfl/fl mice was largely unaltered with a significantly lower abundance of M1-like CD11b+Ly6G−Ly6C−MHC-IIhigh macrophages (SI Appendix, Fig. S1). However, the frequency of M2-like MHC-IIlow macrophages and other immune cell populations was not affected by RXRα deficiency (SI Appendix, Fig. S1). Although primary tumor growth was not affected, we observed a higher frequency of animals with spontaneous lung metastases, as well as a higher number of metastatic nodules in the RXRα knock-outs compared with controls (Fig. 1 B and C). Thus, we further investigated the role of myeloid cell-expressed RXRα in lung colonization by using an experimental metastasis model. Consistent with the results obtained in the LLC model, we found that B16-F10 melanoma cells formed a higher number of macroscopic and microscopic metastases in the lungs of RXRαfl/fl mice after i.v. injection (Fig. 2 A–C). RXRαfl/fl lungs contained a considerably higher frequency of micrometastases with diameters less than 100 μm, demonstrating that extravasation followed by micrometastasis formation in the lung is more efficient in the presence of RXRα-deficient

*

12 10

6/15

11/15

RXR +/+

RXR fl/fl

8 6 4 2 0

C

RXR +/+

RXR fl/fl

Fig. 1. RXRα deletion in myeloid cells enhances Lewis lung carcinoma metastasis. (A) Subcutaneous LLC tumor growth in mice with myeloid cell-specific deletion of RXRα (RXRαfl/fl) and control (RXRα+/+) mice (n = 5–7/group). (B) Number of macroscopic lung metastases in RXRαfl/fl and RXRα+/+ mice (*P < 0.05, Mann-Whitney test, n = 15/group, data combine two independent experiments). (C) Representative images of hematoxylin and eosin-stained midcoronal lung sections from RXRαfl/fl and RXRα+/+ mice. Arrowheads indicate metastatic nodules. (Scale bar, 1 mm.) Graphs on A and B show mean ± SEM.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1700785114

Fig. 2. RXRα deletion in myeloid cells results in enhanced lung colonization by B16-F10 melanoma cells. (A) Number of macroscopic lung metastases in RXRαfl/fl and RXRα+/+ mice (*P < 0.05, unpaired Student’s t test, n = 10–13/ group, data combine two independent experiments). (B) Number of microscopic lung metastases on midcoronal lung sections in RXRαfl/fl and RXRα+/+ mice (*P < 0.05, unpaired Student’s t test, n = 8/group). (C) Representative images of hematoxylin and eosin-stained midcoronal lung sections from RXRαfl/fl and RXRα+/+ mice. (Scale bar, 6 mm.) (D) Size distribution of B16F10 melanoma lung metastases with a diameter ≤1,000 μm in RXRαfl/fl and RXRα+/+ mice. Metastasis diameter data are pooled from eight lungs/group. (E) Representative images and quantification of Ki67 and cleaved caspase3 stainings on midcoronal lung sections with B16-F10 melanoma metastases from RXRαfl/fl and RXRα+/+ mice (Scale bar, 100 μm.) (P > 0.05, unpaired Student’s t test, n = 5/group). Bar graphs on A, B, and E show mean ± SEM.

myeloid cells (Fig. 2D). Since survival and proliferation of extravasated cancer cells at the distant organ is a critical early step of metastasis, we also examined the effect of myeloid cell RXRα on the apoptotic and mitotic rate of metastatic cancer cells. Staining of B16-F10 lung metastases with Ki67, a marker of cell proliferation, did not show major differences between RXRαfl/fl and RXRα+/+ lungs (Fig. 2E). The apoptosis rate, as assessed by staining of cleaved caspase-3, was not significantly different in metastases of RXRαfl/fl lungs either (Fig. 2E). This suggested that myeloid-specific RXRα deficiency does not affect the proliferation or survival of metastatic cancer cells in the lung, and the enhanced metastasis rate in RXRαfl/fl mice originates principally from enhanced lung colonization. Altogether, these results indicate that deletion of RXRα in the myeloid compartment generates a lung microenvironment that is more permissive toward metastasis formation. Kiss et al.

Our results showing enhanced lung colonization of cancer cells in mice with myeloid-specific RXRα deficiency prompted us to investigate whether the absence of RXRα has any effect on the abundance or phenotype of pulmonary myeloid cells. Flow cytometry analysis of the pulmonary myeloid compartment revealed that the lack of RXRα was associated with slightly decreased numbers of CD45+CD11b+F4/80+Ly6G− lung macrophages in naive mice (Fig. 3A). In contrast, neither the resident population nor the metastasis-induced recruitment of

Macrophages

50 40 30 20 10 0

% within CD45+ cells


A

*

naive

B16-F10 met.

B

RXR +/+ RXR fl/fl

naive

B16-F10 met.

C

150 150

Lyz2 Ecm1

Hr

20 Regulation of cell migration

Marco

Regulation of intracellular signal transduction

Cytokine-mediated

Pmp22

100 100

C1qa Cpne5

Aass

Cd200r2

Epcam

50 50

Colec12 Flrt3

Npy

Slc6a8

Gpd1

Kcnk5

Baiap3

Tmem216

Ear11

-log10 (p-value)

Cd109

Mmp12

15 signaling pathway

Positive regulation of cell migration

10 Regulation of phosphate metabolic process

5

Fstl1

Greb1

0 5.0 -5.0

2.5 -2.5

0.0 0

2.5 2.5

5.0 5.0

0 0

log2FoldChange log2 (fold change) RXR fl/fl vs. RXR +/+ FDR < 0.01 significant +/+ RXR Notfl/fl

Ecm1 C1qa Gas6 Hbegf Il1a Igf1 Mmp2 Vcan Pdgfa S100a4 Tnf Ctss Pdgfb Ctsb Sema4d Il1b Lcn2 S100a8 0

LLC migration 400

***

300 200 100

0 iBMDMs: RXR +/+ RXR fl/fl

F

LLC invasion 40

*

30 20 10

0 iBMDMs: RXR +/+ RXR fl/fl

Row z-score -1

E Migrated cells per field

D Significant RXR

50 100 150 Number of associated genes

Migrated cells per field

log10(padj) - log10 (p-value)

Neutrophils

50 40 30 20 10 0

1

Fig. 3. RXRα represses a prometastatic gene set in pulmonary myeloid cells. (A) Relative numbers of myeloid cells in the lungs of RXRα+/+ and RXRαfl/fl mice in the presence or absence of B16-F10 melanoma metastases (*P < 0.05, unpaired Student’s t test, n = 3/group). (B) Volcano plot showing transcriptomic differences in CD45+CD11b+ pulmonary myeloid cells from RXRαfl/fl mice compared with RXRα+/+. Red dots represent differentially expressed genes (adjusted P < 0.01, Wald test, n = 3/group). (C) Dot plot showing significantly enriched gene ontology biological processes within the up-regulated genes in RXRαfl/fl pulmonary myeloid cells. The top five enriched GO terms by P value are highlighted. (D) Heat map showing expression changes of selected prometastatic genes in RXRα+/+ and RXRαfl/fl pulmonary myeloid cells. (E) Number of migrated LLC cells in the presence of RXRα+/+ or RXRαfl/fl iBMDMs in Transwell migration assay. The experimental setup is shown in SI Appendix, Fig. S2A (***P < 0.001, unpaired Student’s t test, n = 3/group). (F) Number of migrated LLC cells through Matrigel layer in the presence of RXRα+/+ or RXRαfl/fl iBMDMs in Transwell invasion assay. The experimental setup is shown in SI Appendix, Fig. S2B (*P < 0.05, unpaired Student’s t test, n = 3/group). Bar graphs on A, E, and F show mean ± SEM.

Kiss et al.

CD45+CD11b+F4/80−Ly6G+ neutrophils was affected by the absence of myeloid RXRα (Fig. 3A). Along with metastasisassociated macrophages differentiated from inflammatory monocytes, lung-resident macrophages have recently been reported to promote the formation of lung metastases (12). Accordingly, LysM-Cre-mediated deletion of an RXR heterodimeric partner PPARγ in myeloid cells results in a moderate decrease of alveolar macrophages (13), and these mice develop fewer pulmonary metastases (14). For these reasons, we assumed that the increase in metastasis formation cannot be explained by the decreased number of lung-resident macrophages in RXRαfl/fl mice. Thus, we decided to investigate whether the lack of RXRα can alter the phenotype of pulmonary myeloid cells, resulting in a more permissive tissue environment for metastatic cancer cells. To characterize the cellular consequences of RXRα deletion in myeloid cells, we performed transcriptomic analysis via RNA sequencing on the CD45.2+CD11b+ myeloid compartment isolated from naive lungs of RXRα+/+ and RXRαfl/fl mice. Expression of the CD45.2 pan-leukocyte antigen by CD11b+ myeloid cells confirmed their transplanted bone marrow origin. We found 1,284 differentially expressed genes (adjusted P value < 0.01) comparing RXRα+/+ and RXRαfl/fl cells (Fig. 3B and Dataset S1). Interestingly, gene ontology analysis identified “regulation of cell migration” and “positive regulation of cell migration” along with “cytokine-mediated signaling pathway” as the most significantly enriched biological processes among the genes up-regulated in RXRα-deficient pulmonary myeloid cells (Fig. 3C). Using Ingenuity Pathway Analysis as an alternative approach yielded similar results (SI Appendix, Table S1). Accordingly, within the RXRα-repressed gene set, we could identify a number of genes encoding secreted factors that have been previously described to promote metastasis, including inflammatory mediators (C1qa, Il1a, Il1b, Tnf, Lcn2), growth factors (Gas6, Hbegf, Igf1, Pdgfa, Pdgfb), extracellular matrix proteins (Ecm1, Vcan), matrix-degrading proteases (Mmp2, Ctsb, Ctss), proinflammatory S100 calcium-binding proteins (S100a4, S100a8), and the proinvasive secreted semaphorin, Sema4d (Fig. 3D and SI Appendix, Table S2). In addition, upstream regulator analysis of differentially expressed genes in RXRα-deficient myeloid cells predicted the activation of several inflammatory pathways including TNF, TLR2, and TLR4, which have a crucial role in initiating an inflammatory microenvironment in the lung, thereby promoting metastatic colonization (15, 16) (SI Appendix, Table S3). These results strongly suggested that RXRα constitutively represses a large set of genes involved in the positive regulation of cell migration and metastasis formation. This led us to hypothesize that the relief of repression (de-repression) of these genes in the absence of RXRα results in a migration-promoting phenotype in myeloid cells. To test this hypothesis, we used RXRαfl/fl and RXRα+/+ immortalized bone marrow-derived macrophages (iBMDM). As expected, based on the migration-promoting transcriptomic signature induced in the absence of RXRα, RXRαfl/fl iBMDMs were significantly more efficient in promoting the migration of LLC cells compared with RXRα+/+ cells (Fig. 3E and SI Appendix, Fig. S2A). Interestingly, RXRα deficiency increased not only the migrationpromoting activity of iBMDMs but also their proinvasive effect. LLC cells were more efficient in invading through a Matrigel extracellular matrix layer in the presence of RXRα-deficient iBMDMs (Fig. 3F and SI Appendix, Fig. S2B). While enhancing cancer cell migration and invasion, RXRα-deficient iBMDMs did not promote cancer cell proliferation (SI Appendix, Fig. S2C). Overall, these results indicate that the transcriptomic signature generated by RXRα-deficiency in myeloid cells translates to a metastasis-promoting cellular phenotype. RXR-Mediated Repression of Prometastatic Genes Is Predominantly Ligand-Insensitive. Next, we sought to determine whether the

RXR-mediated repression in myeloid cells can be enhanced or suspended by the presence of an RXR-activating ligand. To PNAS Early Edition | 3 of 6


RXR Represses a Prometastatic Gene Set in Pulmonary Myeloid Cells.

answer this, we used our previously published RNA-seq data from BMDMs treated with a highly selective, synthetic RXR agonist LG268 (9). The short-term ligand activation (30/60/120 min) enabled us to exclude secondary transcriptional changes after activation of the receptor. By examining the ligand response of RXR-repressed genes identified in pulmonary myeloid cells, we found that only a small fraction of them (14.1%) showed gene expression change in response to RXR activation (8/2, 29/3, 58/ 25 up-/down-regulated from 695 after 30, 60, 120 min of LG268 stimulation, respectively) (Fig. 4A). In the previously described prometastatic gene set, only Hbegf, Pdgfa, and Pdgfb showed significant up-regulation on ligand stimulus (Fig. 4B). This suggested that the majority of RXR-repressed genes in myeloid cells (including genes with prometastatic effect) cannot be modulated by RXR agonists. This was further supported by the observation that the cancer cell migration-promoting activity of iBMDMs was not altered after pretreatment with selective RXR agonist LG268 (Fig. 4C). RXR Colocalizes with NCoR/SMRT Corepressors to Mediate Repression of Prometastatic Genes. Unliganded RXR homodimers and het-

erodimers bind the promoters and/or enhancers of their target genes in association with corepressors nuclear receptor corepressor (NCoR) or silencing mediator of retinoic acid and thyroid hormone receptor (SMRT). Using our previously published RXR ChIP-seq dataset from BMDMs (9), we determined which genes show RXR binding in the proximity of their transcription start site (±100 kb) at the genome-wide level. Intriguingly, gene ontology analysis of these RXR-bound genes revealed a significant overrepresentation of genes involved in the regulation of cell migration and activation of the immune response (SI Appendix, Fig. S3). This suggested that the major genomic targets of RXR binding (and possibly repression) in the steady state are genes that can be functionally linked to cell migration and inflammation. Next, we focused on genes that are repressed by RXR in pulmonary myeloid cells and examined whether RXR binding sites can be found in the vicinity of them. We could detect RXR binding sites within 100 kb of the transcription start sites in 66% of RXR-repressed genes. This suggested that direct DNA binding by the receptor accounts for its repressive effect by the majority of repressed genes. As

B

LG268 (min) 30 60 120

Ecm1 C1qa Gas6 Hbegf Il1a Igf1 Mmp2 Vcan Pdgfa S100a4 Tnf Ctss Pdgfb Ctsb Sema4d Il1b Lcn2 S100a8

Ligand-unresponsive (597)

Ligand-responsive (98)

Log2 fold change vs. control -1.84

0

3.43

C

LG268 (min) 30 60 120

LLC migration Migrated cells per field

A

200 150 100 50

0 iBMDMs: control LG268

Log2 fold change vs. control -1.84

0

3.43

Fig. 4. RXR-mediated repression of prometastatic genes is predominantly ligand-insensitive. (A) Heat map showing expression changes in response to synthetic RXR agonist LG268 in BMDMs among RXR-repressed genes identified in pulmonary myeloid cells (adjusted P < 0.01, Wald test, n = 2/time point). (B) Heat map showing expression changes in response to synthetic RXR agonist LG268 in BMDMs among selected RXR-repressed prometastatic genes identified in pulmonary myeloid cells. Genes with significant expression changes are shown in bold italics (adjusted P < 0.01, Wald test, n = 2/ group). (C) Number of migrated LLC cells in the presence of control or LG268-pretreated iBMDMs in Transwell migration assay (P > 0.05, unpaired Student’s t test, n = 3/group).


mentioned earlier, NCoR and SMRT corepressors are the most common interacting partners of RXR in mediating transcriptional repression. To determine whether RXR binding is associated with NCoR or SMRT binding at RXR-repressed genes, we combined our RXR ChIP-seq data with public NCoR and SMRT ChIP-seq datasets from BMDMs (17). Interestingly, 59% of RXR binding sites detected in the vicinity of RXR-repressed genes showed overlap with both NCoR and SMRT, while 12% and 7% showed overlap with only NCoR or only SMRT, respectively. In addition, NCoR and SMRT occupancy in the proximity of RXR-repressed genes was higher at genomic sites where RXR was also present, suggesting that the presence of RXR facilitates the binding of these corepressors (Fig. 5A). When we examined the genomic loci of the members of RXRrepressed prometastatic gene set identified in pulmonary myeloid cells, we detected overlapping RXR/NCoR/SMRT binding in the close proximity of 11 of the 18 genes (Fig. 5B). It has been shown previously that nuclear receptor-mediated transcriptional repression largely required either NCoR alone or both NCoR and SMRT, only a small fraction of genes showing repression specifically by SMRT (18). Accordingly, deletion of the dominant corepressor NCoR in iBMDMs resulted in the up-regulation of Il1a, Il1b, Pdgfb, Sema4d, Igf1, and Ctss, confirming the requirement of NCoR for the repression of these genes (Fig. 5C). Lack of de-repression by several RXR/NCoR/SMRT-bound genes in NCoR-deficient iBMDMs suggested the existence of compensatory mechanisms emerging in the absence of NCoR, as reported previously (19). RXR Signaling Is Down-Modulated in Peripheral Blood Mononuclear Cells of Patients with Non-Small Cell Lung Cancer. Finally, we set out

to determine whether any alterations in RXR signaling can be observed in patients with cancer. Deletion of RXRα in mice results in embryonic lethality caused by abnormal development of the heart (1). Similarly, DNA sequence data from the Exome Aggregation Consortium showed that loss-of-function mutations in genes coding any of the RXR isotypes are highly improbable, with a probability of loss-of-function intolerance of 94%, 100%, and 94% for RXRα, RXRβ, and RXRγ, respectively (20). Thus, since loss-of-function mutations of RXR are predicted to be very rare, we aimed to investigate whether any changes in RXR expression can be observed in myeloid cells of patients with cancer that could potentially influence its repressive activity. As expression data from tissue-resident or metastasis-associated myeloid cells is difficult to obtain in the case of patients with cancer, we decided to analyze gene expression changes in peripheral blood mononuclear cells (PBMCs). PBMCs have been described to exhibit distinct gene expression changes in a number of malignancies (21–23). We chose to analyze the gene expression of PBMCs from patients with non-small cell lung cancer (NSCLC) and cancer-free control patients, using the largest publicly available dataset including samples from 137 patients with NSCLC and 91 cancer-free control patients (22). Intriguingly, we found that PBMCs from patients with NSCLC showed a significantly lower expression of RXRα (Fig. 6A). In addition, the mean expression level of RXRα exhibited decreasing kinetics during disease progression, showing the lowest expression in patients with advanced and metastatic disease (stage III-IV) (Fig. 6A). RXRβ and RXRγ isotypes displayed lower average expression than RXRα in PBMCs and did not show any changes in patients with NSCLC (Fig. 6A). These data showed that RXRα, the dominant RXR isotype expressed in myeloid cells, was down-regulated in individuals with NSCLC, and its decreased expression was not compensated for by the up-regulation of other RXR isotypes. Next, we wanted to determine whether down-regulation of RXRα in patients with lung cancer is associated with its altered transcriptional activity. RXR’s activity can be assessed by examining the regulation of the target genes of its Kiss et al.

RPKM

5

50 kb

25 kb

50 kb

Pdgfa

Prkar1b

Pdgfb 10 kb

SMRT NCoR RXR

20 kb

Mmp2

normalized expression

C

Ctss 0.008

0.008

*

0.006

0.6

0.004

0.4

0.002

0.002

0.2

0.000

0.000

0.0

Il1b

0.0015

0.0015

0.0010

0.0010

0.0005

0.0005

0.0000

0.0000

Sema4d 1.5

0.8

0.004

Il1a normalized expression

*

0.006

Ctsb

****

***

Il1b 3

*

10 8

1.0

2

0.5

1

0.0

0

*

6 4 2

Sema4d

Pdgfb 0.6

20

0

Igf1

Ctss

0.020

**** 15

0.015

0.4

NCoR +/+

10

Pdgfa

0.010 0.2

5 0

Mmp2

0.005

0.0

Ctsb

NCoR fl/fl

0.000

C1qa

Hbegf

Fig. 5. RXR colocalizes with NCOR/SMRT corepressors to mediate repression of prometastatic genes. (A) NCoR and SMRT occupancy in BMDMs at the genomic sites of RXR-repressed genes identified in pulmonary myeloid cells with/without overlap with RXR (box plots show median and 5–95 percentiles). (B) Genomic loci of selected prometastatic genes showing overlapping RXR/NCoR/SMRT ChIP-seq peaks in BMDMs. (C) Expression of selected prometastatic genes in NCoR+/+ and NCoRfl/fl iBMDMs (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, unpaired Student’s t test, n = 7/group).

heterodimeric partners, since specific target genes for homodimeric RXR signaling have not been unequivocally established. To identify transcriptional regulators that show altered activity in NSCLC, we determined the differentially expressed genes between PBMCs of patients with NSCLC and cancer-free control patients and performed an upstream regulator prediction on these genes using Ingenuity Pathway Analysis. Interestingly, among the top 10 upstream regulators whose activity was predicted to be strongly inhibited in patients with NSCLC, we found two nuclear receptors, PPARγ and LXRα, that require RXR as a heterodimeric partner for their transcriptional activity (Fig. 6B). The inhibited transcriptional activity of PPARγ/RXR and LXRα/RXR heterodimers in patients with NSCLC could not be explained by the down-regulated expression of PPARγ and LXRα, as their expression remained unaltered (SI Appendix, Fig. S4). Overall, these data suggest that the expression and transcriptional activity of RXRα is down-modulated in the immune system of patients with NSCLC, particularly in advanced and metastatic disease. Discussion Our study demonstrates that RXRα expressed by myeloid cells in the lung is essential to maintain a metastasis-resistant tissue Kiss et al.

A

RXRA

*** ***

1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0

n=91 n=94 n=43

B

RXRB 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0

NSCLC Rank 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Upstream Regulator LIPE CBFB PPARG PPP2R5C CD3 NR1H3 (LXR ) FGF7 CD28 STAP2 PSEN1

n=91 n=94 n=43

NSCLC

RXRG 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0

n=91 n=94 n=43 stage III/IV

Igf1 20 kb

stage I/II

Il1a 10 kb

control

Hbegf

10 kb

log2 median normalized expression

C1qa

stage III/IV

5 kb

RXR peak genes

stage I/II

10 kb

RXR peak + genes

SMRT NCoR RXR

5 kb

RXR+SMRT SMRT only SMRT only

RXR peak genes

control

RXR peak + genes


0

RXR+NCoR NCoR only NCoR only

stage III/IV

0

stage I/II

RPKM

10

10

microenvironment. Our data suggest that steady-state transcriptional repression by RXR, partly in concert with NCoR and SMRT corepressors in myeloid cells, controls the expression of several genes whose protein products facilitate the metastatic spread of cancer cells in distant organs. RXR appears to be unique in its ability to form homodimers, as well as heterodimers, with a diverse range of nuclear receptors. Therefore, it is possible that prometastatic genes in pulmonary myeloid cells are repressed both by RXR homodimers and heterodimers. However, the relative contribution of homo- and heterodimers is difficult to determine, since they show redundancy in their binding site preferences, and target genes regulated exclusively by RXR homodimers have not been identified yet (24). The absence of RXR-mediated repression in myeloid cells leads to elevated expression of prometastatic regulators, creating a tissue microenvironment favoring cancer cell extravasation and micrometastasis formation. This process shares similarities to the formation of the premetastatic niche, a tumor-induced receptive and supportive tissue microenvironment that supports cancer cell colonization and survival in a secondary organ site (25, 26). Interestingly, several RXR-repressed genes identified in our work have been described as important regulators involved in the formation of premetastatic niches. We found that RXR represses both S100a4 and S100a8 in pulmonary myeloid cells. The protein products of these genes act as central regulators during the course of premetastatic niche formation by generating an inflammatory microenvironment both directly and indirectly via the up-regulation of serum amyloid A3, resulting in enhanced cancer cell adherence and colonization (25, 26). In addition, we found that RXRαfl/fl pulmonary myeloid cells up-regulated TNFα, a cytokine that has been identified as a key component of the inflammatory milieu in the premetastatic niche in various tissue environments

NSCLC

Activation z-score -2.621 -2.333 -2.142 -2 -1.992 -1.987 -1.962 -1.734 -1.732 -1.687

p-value of overlap 0.00305 0.0196 0.0246 0.0161 0.0000153 0.0378 0.0411 0.00101 0.00736 0.0406

Fig. 6. RXR signaling is down-modulated in PBMCs of patients with NSCLC. (A) Expression of RXR isotypes in PBMCs of patients with NSCLC and cancerfree control patients. (Graphs show mean. ***adjusted P < 0.001, MannWhitney test) (B) Top 10 upstream regulators by z-score predicted to be inhibited in PBMCs of patients with NSCLC compared with cancer-free control patients (P < 0.05). RXR heterodimeric partners are shown in bold italics.

PNAS Early Edition | 5 of 6


15

20

B

SMRT occupancy

20

control

NCoR occupancy

30


A

Inverse agonists have the ability to stabilize the nuclear receptor–corepressor complex and enhance transcriptional repression (29). Although RXR-specific inverse agonists are not yet available, similar compounds have already been developed and used successfully in preclinical studies for several other nuclear receptors including RAR, LXR, and ROR (30–33). This class of RXR-specific ligands could enhance the repressive effect of RXR on prometastatic regulators, potentially rendering the lung microenvironment less permissive toward metastatic colonization.

(15, 25, 26). Similarly, versican, an extracellular matrix proteoglycan with multifaceted roles in premetastatic niche formation, was also up-regulated in RXRαfl/fl pulmonary myeloid cells (25, 26). In the Lewis lung carcinoma model, tumor-derived versican activated myeloid cells through TLR2 to induce an inflammatory microenvironment in the premetastatic lung (15). In contrast, in the MMTV-PyMT model of breast carcinoma, versican secreted by CD11b+Ly6Chigh myeloid cells in premetastatic lungs stimulated mesenchymal to epithelial transition of cancer cells, leading to enhanced metastatic colonization (27). Another characteristic feature of premetastatic niche is tissue remodeling and degradation of the extracellular matrix that favors extravasation of circulating cancer cells (25, 26). Analogously, RXRαfl/fl myeloid cells in the lung up-regulated several proteases capable of matrix remodeling, including MMP2, cathepsin B, and cathepsin S. As a consequence, RXR-deficient iBMDMs had an enhanced capacity to promote cancer cell invasion through an extracellular matrix layer. Administration of selective RXR agonists have a well-established beneficial effect in solid tumors because of their capacity to induce differentiation and/or apoptosis of cancer cells (28). Nevertheless, the influence of RXR modulation both on the stromal compartment of primary tumor microenvironment and on premetastatic niches remains to be determined. Our findings suggest that the majority of genes differentially expressed in RXRα-deficient pulmonary myeloid cells do not respond to RXR activation. Therefore, it appears unlikely that endogenous or synthetic RXR agonists can alter the repressive activity of the receptor on prometastatic regulators. To therapeutically exploit RXR’s ability to control the expression of several key regulators in metastasis, RXR-specific inverse agonists could be used.

ACKNOWLEDGMENTS. We thank Ms. T. Cseh, Mr. M. Peloquin, and Mr. J. Shelley for their excellent technical assistance. We thank Dr. M. Komatsu and members of the L.N. laboratory for discussions and comments on the manuscript. We are grateful to I. Schulman for providing LysM-Cre NCoRfl/fl bone marrow. L.N. and P.T. are supported by “NR-NET” ITN PITN-GA-2013606806 from the EU-FP7 PEOPLE-2013 program. B.D. is supported by an American Heart Association postdoctoral fellowship (17POST33660450). L.N. is supported by the Hungarian Scientific Research Fund (Grants OTKA K100196, K111941, and K116855) and by the Sanford Burnham Prebys Medical Discovery Institute.

1. Szanto A, et al. (2004) Retinoid X receptors: X-ploring their (patho)physiological functions. Cell Death Differ 11:S126–S143. 2. Rühl R, et al. (2015) 9-cis-13,14-dihydroretinoic acid is an endogenous retinoid acting as RXR ligand in mice. PLoS Genet 11:e1005213. 3. Li M, et al. (2000) Skin abnormalities generated by temporally controlled RXRα mutations in mouse epidermis. Nature 407:633–636. 4. Kiss M, Czimmerer Z, Nagy L (2013) The role of lipid-activated nuclear receptors in shaping macrophage and dendritic cell function: From physiology to pathology. J Allergy Clin Immunol 132:264–286. } szer T, Menéndez-Gutiérrez MP, Cedenilla M, Ricote M (2013) Retinoid X receptors 5. Ro in macrophage biology. Trends Endocrinol Metab 24:460–468. 6. Núñez V, et al. (2010) Retinoid X receptor α controls innate inflammatory responses through the up-regulation of chemokine expression. Proc Natl Acad Sci USA 107: 10626–10631. 7. Roszer T, et al. (2011) Autoimmune kidney disease and impaired engulfment of apoptotic cells in mice with macrophage peroxisome proliferator-activated receptor gamma or retinoid X receptor alpha deficiency. J Immunol 186:621–631. 8. Ma F, et al. (2014) Retinoid X receptor α attenuates host antiviral response by suppressing type I interferon. Nat Commun 5:5494. 9. Daniel B, et al. (2014) The active enhancer network operated by liganded RXR supports angiogenic activity in macrophages. Genes Dev 28:1562–1577. 10. Krezel W, et al. (1996) RXR gamma null mice are apparently normal and compound RXR alpha +/−/RXR beta −/−/RXR gamma −/− mutant mice are viable. Proc Natl Acad Sci USA 93:9010–9014. 11. Kastner P, et al. (1996) Abnormal spermatogenesis in RXR beta mutant mice. Genes Dev 10:80–92. 12. Sharma SK, et al. (2015) Pulmonary alveolar macrophages contribute to the premetastatic niche by suppressing antitumor T cell responses in the lungs. J Immunol 194:5529–5538. 13. Schneider C, et al. (2014) Induction of the nuclear receptor PPAR-γ by the cytokine GM-CSF is critical for the differentiation of fetal monocytes into alveolar macrophages. Nat Immunol 15:1026–1037. 14. Li H, et al. (2011) Activation of PPARγ in myeloid cells promotes lung cancer progression and metastasis. PLoS One 6:e28133. 15. Kim S, et al. (2009) Carcinoma-produced factors activate myeloid cells through TLR2 to stimulate metastasis. Nature 457:102–106. 16. Hiratsuka S, et al. (2008) The S100A8-serum amyloid A3-TLR4 paracrine cascade establishes a pre-metastatic phase. Nat Cell Biol 10:1349–1355.

17. Barish GD, et al. (2012) The Bcl6-SMRT/NCoR cistrome represses inflammation to attenuate atherosclerosis. Cell Metab 15:554–562. 18. Ghisletti S, et al. (2009) Cooperative NCoR/SMRT interactions establish a corepressorbased strategy for integration of inflammatory and anti-inflammatory signaling pathways. Genes Dev 23:681–693. 19. Li P, et al. (2013) NCoR repression of LXRs restricts macrophage biosynthesis of insulinsensitizing omega 3 fatty acids. Cell 155:200–214. 20. Lek M, et al.; Exome Aggregation Consortium (2016) Analysis of protein-coding genetic variation in 60,706 humans. Nature 536:285–291. 21. Twine NC, et al. (2003) Disease-associated expression profiles in peripheral blood mononuclear cells from patients with advanced renal cell carcinoma. Cancer Res 63: 6069–6075. 22. Showe MK, et al. (2009) Gene expression profiles in peripheral blood mononuclear cells can distinguish patients with non-small cell lung cancer from patients with nonmalignant lung disease. Cancer Res 69:9202–9210. 23. Baine MJ, et al. (2011) Transcriptional profiling of peripheral blood mononuclear cells in pancreatic cancer patients identifies novel genes with potential diagnostic utility. PLoS One 6:e17014. 24. Glass CK (1994) Differential recognition of target genes by nuclear receptor monomers, dimers, and heterodimers. Endocr Rev 15:391–407. 25. Liu Y, Cao X (2016) Characteristics and significance of the pre-metastatic niche. Cancer Cell 30:668–681. 26. Peinado H, et al. (2017) Pre-metastatic niches: Organ-specific homes for metastases. Nat Rev Cancer 17:302–317. 27. Gao D, et al. (2012) Myeloid progenitor cells in the premetastatic lung promote metastases by inducing mesenchymal to epithelial transition. Cancer Res 72: 1384–1394. 28. Altucci L, Gronemeyer H (2001) The promise of retinoids to fight against cancer. Nat Rev Cancer 1:181–193. 29. de Lera AR, Bourguet W, Altucci L, Gronemeyer H (2007) Design of selective nuclear receptor modulators: RAR and RXR as a case study. Nat Rev Drug Discov 6:811–820. 30. Germain P, et al. (2009) Differential action on coregulator interaction defines inverse retinoid agonists and neutral antagonists. Chem Biol 16:479–489. 31. Khetchoumian K, et al. (2007) Loss of Trim24 (Tif1alpha) gene function confers oncogenic activity to retinoic acid receptor alpha. Nat Genet 39:1500–1506. 32. Flaveny CA, et al. (2015) Broad anti-tumor activity of a small molecule that selectively targets the Warburg effect and lipogenesis. Cancer Cell 28:42–56. 33. Solt LA, et al. (2011) Suppression of TH17 differentiation and autoimmunity by a synthetic ROR ligand. Nature 472:491–494.


Materials and Methods Details of mice and bone marrow transplantation, cell lines and tumor models, histology, flow cytometry and cell sorting, RNA-seq, ChIP-seq analysis, immortalization of mouse bone marrow-derived macrophages, cancer cell migration and invasion assays, real-time quantitative PCR, microarray analysis, 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) assay, and statistical analysis are described in the SI Appendix. All experiments were carried out in accordance with the institutional ethical guidelines at the University of Debrecen (license no.: 21/2011/DE MAAB), following the European regulations.

Kiss et al.

SI Appendix Retinoid X Receptor suppresses a metastasis-promoting transcriptional program in myeloid cells via a ligand-insensitive mechanism Mate Kiss, Zsolt Czimmerer, Gergely Nagy, Pawel Bieniasz-Krzywiec, Manuel Ehling, Attila Pap, Szilard Poliska, Pal Boto, Petros Tzerpos, Attila Horvath, Zsuzsanna Kolostyak, Bence Daniel, Istvan Szatmari, Massimiliano Mazzone, Laszlo Nagy Contents: Materials and Methods Figure S1 Figure S2 Figure S3 Figure S4 Table S1 Table S2 Table S3 References for Supplemental Information

Materials and Methods Mice and bone marrow transplantation Recipient 6-8-week-old female C57BL/6J CD45.1 (BoyJ) mice were lethally irradiated (11 Gy) and then intravenously injected with 5x106 bone marrow cells from female LysM-Cre RXRαfl/fl RXRβ-/- or LysM-Cre RXRα+/+ RXRβ-/mice expressing the pan-leukocyte antigen CD45.2. All animals were bred in SPF conditions. Bone marrow transplanted animals were transferred to a ventilated cabinet and kept in sterile conditions for at least 4 weeks after irradiation. Mice were fed with autoclaved rodent chow ad libitum and autoclaved drinking water completed with amoxicillin (250mg/L) and clavulanic acid (60 mg/L) antibiotics. During the tumor or metastasis induction the animals were housed under minimal disease conditions in filter top cages. All experiments were carried out in accordance with the institutional ethical guidelines

(license

no.:

21/2011/DE

MÁB),

following

the

European

regulations. Cell lines and tumor models Lewis lung carcinoma (LLC) cells were gifts from Prof. Jo A. Van Ginderachter (Vrije Universiteit Brussel). B16-F10 melanoma cells were obtained from the American Type Culture Collection. LLC cells were maintained in RPMI supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 U/ml penicillin and 100 µg/ml streptomycin. B16-F10 cells were maintained in DMEM supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 1% MEM non-essential amino acids, 1% MEM vitamins, 100 U/ml penicillin and 100 µg/ml streptomycin. 3x106 LLC cells in 200µl PBS were injected subcutaneously into the right flank. Primary tumor volumes were determined by caliper measurements and calculated using the formula: V= π x [d2 x D]/6, where d is the minor tumor axis and D is the major tumor axis. 105 B16-F10 cells in 100µl PBS were injected intravenously through the tail vein. 21 days after tumor inoculation lungs were removed and the presence of metastases was examined.

Histology For histological analysis of tumor burden, lung tissues of tumor-bearing animals were perfused with PBS and 4% formaldehyde solution. After removal, lungs were fixed with 4% formaldehyde solution for 24 hours at room temperature and embedded in paraffin following standard procedures. Midcoronal sections were prepared and stained with hematoxylin and eosin. For immunohistochemistry, mid-coronal tissue sections of 5µm thickness were prepared from paraffin-embedded samples using a Microm 350s microtome. The slides were air dried overnight and depigmented using H2O2 solution for 1h at 60°C. After washing with distilled water, the slides were placed on the Leica Bond Rx IHC staining system. For Ki67 staining, Bond ER1 antigen retrieval buffer, rabbit anti-mouse Ki67 primary antibody (clone SP6, Thermo Fisher, dilution 1:100), and Leica Bond Polymer Refine Detection Kit (DS9800) was used. For cleaved caspase-3 staining, Bond ER2 antigen retrieval buffer, rabbit anticleaved caspase-3 primary antibody (9661, Cell Signaling, dilution 1:200) and Leica Bond Polymer Refine Red Detection Kit (DS9390) was used. The Leica Aperio Scanscope XT slide scanner system with a 20x/0.75 Plan Apo Olympus objective was used for image documentation. Scanned slides were analyzed with Aperio ImageScope (Leica Biosystems) and HALO (Indica Labs). Flow cytometry and cell sorting To prepare lung single cell suspensions, lungs were removed after transcardial perfusion with PBS, cut in small pieces, treated with 10 U/ml collagenase I, 400 U/ml collagenase IV, and 30 U/ml DNase I (Worthington) for 40 min at 37°C, and filtered. For tumor single cell suspensions, tumors were minced, treated with the enzyme mix above for 30 min at 37°C, and filtered. Red blood cells were removed using erythrocyte lysis buffer. Cells were resuspended in staining medium (phenol-red free DMEM, 10 mM HEPES, 2% fetal bovine serum) and incubated with FcR blocking reagent

(Miltenyi Biotec). The cells were then stained with antibodies against CD45.2 (Clone: 104), CD11b (Clone: M1/70), Ly6G (Clone: 1A8), F4/80 (Clone: BM8), Ly6C (Clone: HK1.4), MHC-II (Clone: M5/114.15.2), CD124 (Clone: mIL4RM1), CD204 (Clone: MCA1322FA), CD206 (Clone: MCA2235FA), CD3 (Clone 17A2), CD4 (Clone: GK15), CD8 (Clone: 53-6.7), NK1.1 (Clone: PK136), CD19 (Clone: 6D5). Expression of CD124, CD204 and CD206 was compared to the binding of isotype controls: rat IgG2a (CD124, CD206), rat IgG2b (CD204). Stained cells were washed and resuspended in staining medium for flow cytometry. For cell sorting, cells were pooled from 5 lungs, and CD45+ cells were enriched using CD45 microbeads (Miltenyi Biotec) before staining with antibodies against CD45.2 and CD11b. Data acquisition and sorting was performed using a FACS Aria III instrument (BD Biosciences) and data were analyzed with FlowJo software. RNA-seq RNA from sorted pulmonary myeloid cells was isolated using Trizol reagent according to the guidelines of Immunological Genome Project (1). RNA-Seq library was prepared from three biological replicates by using TruSeq RNA Sample Preparation Kit (Illumina) according to the manufacturer’s protocol. Libraries were sequenced with Illumina HiScanSQ sequencer. RNA-seq samples were analyzed using an in-house pipeline. Briefly, the 50bp raw single-end reads were aligned using TopHat (2) to the mm10 genome assembly (GRCm38) and only the uniquely mapped reads were kept using ‘-max-multihits 1’ option, otherwise the default parameters were used. Samtools v1.0 (3) was used for indexing the alignment files. Coverage density tracks (wig files) for RNA-seq data were generated by igvtools with ‘count’ command and then converted into tdf files using ‘toTDF’ option. Differential gene expression was assessed in R with DESeq2 using a false discovery rate of 1% (4). Gene Ontology analysis was performed using the ClueGO Cytoscape app (5). Heatmaps were generated using GENE-E (Broad Institute). RNA-seq data from pulmonary myeloid cells have been deposited to the NCBI Sequence Read Archive under accession number SRP095164. RNA-seq data

from LG268-treated BMDMs have been previously published and deposited in the NCBI Sequence Read Archive under the accession number SRP019970. ChIP-seq analysis ChIP-seq data used in our study are from previously published datasets which can be downloaded from the NCBI Sequence Read Archive under accession numbers SRP073648 (RXR) and SRP005606 (SMRT and NCoR). The primary analysis of raw sequence reads has been carried out using our ChIPseq analysis command line pipeline (6) including the following steps: Alignment to the mm10 mouse genome assembly was performed by the Burrows-Wheeler Aligner (BWA) tool (7) and BAM files were created by SAMTools (3). Genome coverage (bedgraph and tdf) files were generated by makeUCSCfile.pl (HOMER) (8) and IGVtools, respectively, and used for visualization with Integrative Genomics Viewer (IGV) 2 (9). Peaks were predicted by MACS2 (10), and artifacts were removed according to the blacklist of ENCODE. Intersections were determined by intersectBed (BEDTools) (11). Annotation to the closest TSS of protein coding transcripts was performed by PeakAnnotator (EBI), and peaks within 100 kb to a TSS were considered to belong to the corresponding gene. Reads per kilobase per million mapped reads (RPKM) for corepressor density were determined by counting the unique sequence reads falling on the peak summit +/-50 bp genomic region using intersectBed and other command line tools. Immortalization of mouse bone marrow-derived macrophages Bone marrow-derived cells from wild type, LysM-Cre RXRα+/+ RXRβ-/-, LysMCre RXRαfl/fl RXRβ-/- and LysM-Cre NCoRfl/fl mice were immortalized using the J2 cell line continuously producing the J2 virus encoding v-raf and v-myc oncogenes (12). J2 cells were grown in DMEM containing 20% FBS. Bone marrow cells were seeded in immortalization media I. (90% J2 supernatant, 5% HyClone FBS, 10µg/ml Polybrene 0.1%, 5% L929 supernatant) and incubated overnight. On the second day supernatant was collected and spun

down to pellet floating cells. Adherent cells were scraped and re-plated in a new petri dish using immortalization media II. (20% J2 supernatant, 10% HyClone FBS, 10µg/ml Polybrene 0.1%, 10% L929 supernatant, 60% DMEM) and incubated for 6 days. After the immortalization cells were kept in regular macrophage differentiation media (20% FBS, 30% L929 supernatant and 50% DMEM containing 1% antibiotics). Cancer cell migration and invasion assays For the cancer cell migration assay, 100 000 LLC cells were seeded on 8 µm pore polyester membrane inserts (Corning Transwell #3464) in 100 µl DMEM (1% FBS) and placed on a 24 well plate seeded with 250 000 iBMDMs/well in 500 µl DMEM (1% FBS). After 4 h incubation, the membranes were removed, the upper sides were cleaned and the membranes were fixed with 4% formaldehyde. Transmigrated cells were quantified after DAPI staining by counting cells in 4 microscopic fields/membrane. Average transmigrated cell numbers from 3 membranes/condition were used to determine statistical significance. In some experiments, iBMDMs were pretreated with 100 nM LG268 for 6 h. For the cancer cell invasion assay, 8 µm pore polyester membrane inserts (Corning Transwell #3464) were coated with 100 µl Matrigel. 150 000 iBMDMs and 150 000 LLC cells were resuspended in 100 µl DMEM (1% FBS) and seeded on the Matrigel layer. The membrane inserts were placed on a 24 well plate with 500 µl DMEM/well containing 10% FBS as chemoattractant. After 4 h incubation, transmigrated cells were quantified as described above. Real-Time Quantitative PCR RNA isolated with Trizol reagent was reverse-transcribed with High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s

protocol.

Transcript

quantification

was

performed

by

quantitative PCR reactions using SYBR Green Master Mix (Roche). Transcript levels were normalized to Ppia. Primers are available upon request.

Microarray analysis Gene expression data from PBMCs of cancer patients and cancer-free controls were downloaded from the NCBI Gene Expression Omnibus data repository under accession number GSE13255 and analyzed using GeneSpring GX 13.1.1 software. Briefly, raw data files were transformed using

GCRMA

algorithm

and

median

normalization.

Non-parametric

Wilcoxon-Mann-Whitney test with Benjamini-Hochberg FDR was used to determine differentially expressed genes between NSCLC patients and cancer-free controls (P value < 0.05). Ingenuity Pathway Analysis was used to predict inhibited transcriptional regulators in cancer patients (P value < 0.05). MTT assay LLC cells were seeded in a 96-well plate at a density of 10.000 cells/well and treated with cell culture supernatants collected after 48 h incubation from RXRα+/+ and RXRαfl/fl iBMDMs diluted 1:5 in RPMI with 10% FBS, 1% Penicillin and 1% Streptomycin. After 24 and 48 h incubation, MTT-based cell proliferation

assay

was

performed

according

to

the

manufacturer’s

instructions (Cell Proliferation Kit I (MTT), Roche). Statistical Analysis Statistical analyses were performed using Graphpad Prism 6 and R Studio. The statistical tests used are indicated in the figure legends.

MHC-IIhigh TAM

2

fl fl/

R XR α

0.6 0.4 0.2

R XR α

fl/

fl

0.0 +/ +

fl fl/

+/ +

R XR α

+/ +

fl fl/

4

0

R XR α

fl fl/

R XR α

+/ +

50

NK cells % within CD45+ cells

0.2

60

R XR α

R XR α


0.4

0.0

R XR α

+/ +

fl/

0.6

6

+/ +

0.0

0.8

70

B cells

R XR α

0.5

fl

+/ + R XR α


1.0

R XR α

fl/

fl

+/ +

R XR α

R XR α


1.5

1.0

80

40

CD8+ T cells

CD4+ T cells 2.0


0

0

0

Gr-MDSCs

R XR α

2

10

fl

1

4

20

fl/

2

6

30

R XR α

3

**

R XR α

4

8

Mo-MDSCs % within CD45+ cells

5



MHC-IIlow TAM

Figure S1. Immune cell infiltration in LLC tumors of RXRα +/+ and RXRα fl/fl mice determined by flow cytometry. Frequency of the following populations within + + low CD45 leukocytes is shown: CD11b Ly6G Ly6C MHC-II tumor-associated low + high macrophages (MHC-II TAM), CD11b Ly6G Ly6C MHC-II tumorhigh + + low associated macrophages (MHC-II TAM), CD11b Ly6G Ly6C MHC-II + + monocytic myeloid-derived suppressor cells (Mo-MDSCs), CD11b Ly6G granulocytic myeloid-derived suppressor cells (Gr-MDSCs), CD11b + + + + + CD3 CD4 T cells, CD11b CD3 CD8 T cells, CD11b CD19 B cells and + CD11b NK1.1 natural killer cells (NK cells). Graphs show mean +/- SEM, n=8-10/group, **p