ING4 Regulates a Secretory Phenotype in Primary Fibroblasts with Dual

ORIGINAL ARTICLE ING4 regulates a secretory phenotype in primary ﬁbroblasts with dual effects on cell proliferation and tumor growth

A Moreno1,3, I Soleto1,4, P Garcı´a-Sanz1,2,4, G Moreno-Bueno1,2 and I Palmero1

ING proteins have an essential role in the control of a variety of cellular functions whose deregulation is associated with tumor formation and dissemination, such as proliferation, apoptosis, senescence or invasion. Accordingly, loss of function of ING proteins is a frequent event in many types of human tumors. In this report, we have studied the function of ING4, a member of the ING family of tumor suppressors, in the context of normal, non-transformed primary fibroblasts. We show that ING4 negatively regulates cell proliferation in this cell type. The antiproliferative action of ING4 requires its ability to recognize chromatin marks, it is p53-dependent at least in part, and it is lost in an ING4 cancer-associated mutant. Gene expression analysis shows that ING4 regulates the expression and release of soluble factors of the chemokine family. The secretory phenotype regulated by ING4 in primary fibroblasts displays a selective paracrine effect on proliferation, fostering the division of tumor cells, while inhibiting division in primary fibroblasts. Consistently, ING4-expressing fibroblasts promoted tumor growth in vivo in co-injection tumorigenesis assays. Collectively, our results show that ING4 not only can regulate the proliferation of primary non-transformed human fibroblasts, but also orchestrates a secretory phenotype in these cells that promotes tumor cell proliferation in vitro and in vivo. These findings support a critical role for ING4 expression in normal cells in the non-cell-autonomous regulation of tumor growth.

Oncogene (2014) 33, 1945–1953; doi:10.1038/onc.2013.145; published online 22 April 2013 Keywords: ING4; p53; tumor suppression; secretory phenotype

INTRODUCTION tumor types.8,10,12,14 A correlation between ING4 alterations and ING proteins are putative tumor suppressors that control key high malignancy grade and increased tumor angiogenesis has 10,12 cellular functions frequently altered during tumor initiation or been reported in several studies. Besides these cancer-related progression, including cell division, apoptosis or motility.1–4 ING functions, ING4 also has a key role in control of innate immunity proteins act primarily as chromatin-mediated regulators of gene in vivo, presumably in connection with the NF-kB pathway, as expression, in association with transcription factors like p53 or indicated by defective inflammatory response in Ing4-mutant 15 NF-kB. ING proteins fulfil this function through the specific reading mice. Our current knowledge of ING4 function in cancer is based of the active histone mark H3K4me3, and their participation in mainly on studies with immortalized or tumor cells. In this complexes with histone acetylation or deacetylation activity.2,5,6 report, we wished to understand the role of ING4 in the context of The ING family includes several sequence-related proteins, normal primary cells, in order to gain insights in the physiological encoded in five loci in human and mice, many of which role of this protein. To this end, we used for our studies early- produce several alternative isoforms. Among the ING family, passage human primary embryonic fibroblasts. These cells are ING4 displays a series of distinctive features consistent with its genetically normal, and retain fully functional tumor-suppressive participation in both tumor initiation and progression. Like other pathways. Accordingly, they have been widely used as a cellular ING proteins, ING4 can control cell proliferation and apoptosis.7–9 model to study processes relevant for tumor suppression, In addition to these common functions, ING4 can also regulate cell including cellular senescence.16 Our results show that ING4 can motility, invasion and angiogenesis, supporting an important role inhibit the proliferation of human fibroblasts, and this effect for this ING protein in suppression of tumor cell invasion and requires, at least partially, the recognition of histone marks by metastatic spread.10–13 It is assumed that ING4 regulates these ING4 and the presence of functional p53. More importantly, we processes mainly via the transcriptional control of specific target show that ING4 can regulate a secretory phenotype in primary genes, although a direct role of the ING4 in the cytoplasm has also fibroblasts, which differentially regulates proliferation of primary been described.11 Consistent with a role as tumor suppressor, or transformed cells in vitro, and promotes tumor growth in vivo. ING4 function is impaired in multiple human tumors, by diverse These results unveil a novel role for ING4 in paracrine crosstalk in mechanisms. For instance, the ING4 locus is deleted in 20% tumors and suggest that ING4 expression in non-transformed of breast tumors, and reduced ING4 expression can be found cells is a key determinant in non-cell-autonomous regulation in glioblastoma, lung carcinoma or melanoma, among other of tumorigenesis.

1Instituto de Investigaciones Biome´dicas ‘Alberto Sols’ CSIC-UAM, Madrid, Spain and 2Departamento de Bioquı´mica, Universidad Autońoma de Madrid, IdiPAZ (Instituto de Investigacio´ n Sanitaria La Paz) and Fundacioń MD Anderson Internacional, Madrid, Spain. Correspondence: Dr I Palmero, Instituto de Investigaciones Biome´dicas ‘Alberto Sols’ CSIC-UAM, Arturo Duperier, 4, E-28029 Madrid, Spain. E-mail: [email protected] 3Current address: Centre for Gene Regulation and Expression, College of Life Sciences, University of Dundee, Dundee, UK. 4These authors contributed equally to this work. Received 15 May 2012; revised 20 February 2013; accepted 22 March 2013; published online 22 April 2013 The ING4 secretory phenotype promotes tumor growth A Moreno et al 1946 RESULTS ING4 inhibits proliferation of primary fibroblasts D213A To study the role of ING4 in non-transformed, primary cells, we V ING4 N214D expressed ectopically the human ING4 protein in early-passage IMR-90 human primary fibroblasts. These cells retain normal ING4 (HA) tumor-suppressive pathways and have been extensively characterized as a cellular model in studies of cellular responses actin relevant for tumorigenesis. As controls for these studies, we used two mutant versions of ING4, previously characterized: ING4N214D, henceforth N214D, a cancer-associated mutant with impaired 13 D213A *** stability and function and ING4 , henceforth D213A, a 40 ** mutation in the PHD domain that abolishes histone mark recognition.9 The expression of all the ING4 constructs was 35 * confirmed by western blot and quantitative PCR, after retroviral transduction and selection (Figure 1a, Supplementary Figures S1a 30 and b). To determine the impact on cell division of the enforced 25 expression of the different ING4 versions, we measured their rate of BrdU incorporation and the cummmulative growth over a 20 period of 8 days (Figures 1b and c). Both assays showed that wild-type ING4 provoked a significant reduction in proliferation 15 rates in IMR-90 cells, but this effect was significantly reduced when 10 either ING4 mutant was used. BrdU-positive cells (%) 5 Link to senescence and DNA damage response 0 Having shown an antiproliferative effect of ING4 in primary V ING4 N214D D213A fibroblasts, we wished to elucidate the mechanism for this phenotype. Cellular senescence is an antiproliferative response, 5 pLPC which opposes tumor formation by inhibiting the division of cells with potentially oncogenic alterations.17 It has been reported that ING4wt two ING proteins (ING1 and ING2) are regulators of cellular 4 ING4 N214D senescence, and their ectopic expression suffices to trigger ING4 D213A a senescent phenotype in primary fibroblasts.18–21 With this background, we tested if ING4 could have a similar pro-senescent 3 action that could account for the antiproliferative effect observed. To this end, we investigated if a senescent phenotype was induced in IMR-90 fibroblasts expressing wild-type or mutant

ING4. The activated version of the Ha-Ras oncogene (RasV12) was Relative cell number 2 used as a positive control for induction of senescence.22 The following senescence markers were used: senescence-associated beta galactosidase staining, senescence-associated heterochro- 1 matin foci and flattened morphology. No significant induction of 0 246 8 senescence markers was observed in cells expressing wild-type Days or mutant versions of ING4, in conditions where RasV12 caused a clear induction of the senescent phenotype (Figures 2a–c). Figure 1. Antiproliferative action of ING4 in human primary fibroblasts. (a) Western Blot showing ectopic expression of ING4 Moreover, endogenous ING4 RNA or protein did not change proteins in retrovirally transduced IMR-90 fibroblasts. (b) BrdU during Ras-induced senescence (Supplementary Figures S1a and incorporation in IMR-90 fibroblasts retrovirally infected with the b). These results indicate that, in contrast to other ING proteins indicated ING4 versions, after a BrdU pulse of 4 h. The average and (such as ING1, Supplementary Figure S1c and d), ING4 is not a s.d. of three independent experiments are shown. ***Po0.001, potent inductor of senescence in human primary fibroblast. We **Po0.01, *Po0.05. (c) Growth curve of IMR-90 fibroblasts infected also investigated if the arrest triggered by ING4 expression could with the indicated ING4 versions or empty vector. The graph be mediated by DNA damage. It has been suggested that other represents the cell number at each time point, relative to day one. ING proteins have a role in the response to genotoxic stress in Data from one representative experiment are shown. connection with the repair machinery or through changes in global chromatin structure and accessibility.23,24 To test this 25 hypothesis, we analyzed by immunofluorescence the presence of other essential responses in primary cells. For this purpose, we g-H2AX, a marker of DNA damage foci. We detected a small transduced IMR-90 cells with short-hairpin RNA (shRNA) vectors but statistically significant increase in g-H2AX-positive cells with against p53 or Rb to stably inactivate these pathways, followed by wild-type ING4, which was absent in cells with ING4 mutants ectopic expression of ING4 and measured the growth rate of these defective in cell-cycle arrest (Figure 2d). cells (Figure 3a). As a control, RasV12 was also used in these assays.26 Inactivation of p53 by itself caused a small increase in BrdU incorporation (42±11 vs 32±6%), as expected. The Connection to tumor suppressor pathways antiproliferative effect of ING4 (measured as BrdU incorporation To investigate further the mechanism of the antiproliferative effect of ING4-expressing cells relative to vector-expressing controls) was of ING4, we sought to determine the implication of the p53 and significantly reduced in the shp53 cells, compared with the effect retinoblastoma (Rb) pathways. It has been reported that ING4 is of ING4 in cells with a control shRNA vector. In contrast, the effect functionally linked to p53.7 In addition, both tumor-suppressive of ectopic ING4 was retained in cells expressing the shRNA against pathways have critical roles in control of cell proliferation and Rb (Figure 3b). Similar results were obtained when growth rate

Oncogene (2014) 1945 – 1953 & 2014 Macmillan Publishers Limited The ING4 secretory phenotype promotes tumor growth A Moreno et al 1947 100 80 90 70 80 70 60 60 50 50 40 40 30 30 20 20 10 10 0 SAHF-positive cells (%)

SA Beta-Gal-positive cells (%) 0

pLPC ING4 pLPC ING4 N214D D213ARasV12 N214D D213ARasV12

40 35 ** ** ** V ING4wt 30 25 20 15 10 ING4 N214D ING4 D213A H2AX-positive cells 5 0

pLPC ING4 N214DD213ARasV12

RasV12 Figure 2. Senescence markers in primary fibroblasts expressing ING4. IMR-90 fibroblasts retrovirally infected with wild-type or the indicated mutant versions of ING4 or Ha-RasV12 were analyzed to detect SA-BetaGal staining (a, c), senescence-associated heterochromatin foci (SAHF, b) or senescent morphology (c), as described in Materials and Methods. a and b show the average and s.d. from three independent experiments. c shows representative phase contrast images of SA-BetaGal-stained cells, plated at low density. (d) Percentage of cells showing g-H2AX staining in IMR-90 fibroblasts infected with the indicated versions of ING4 or Ha-RasV12. The average and s.d. from three independent experiments is shown. **Po0.01. was measured by direct cell counting over a period of 8 days study in genes involved in cytokine and chemokine signaling (Supplementary Figure S2a). As with the parental IMR-90 cells, no regulated by ING4 in primary fibroblasts. The differential expres- induction of senescence was observed in cells expressing ING4 in sion of a selection of chemokines identified in the array (and the presence of shRNAs against p53 or Rb (data not shown). interleukin (IL)-6, which has been previously associated with ING4) was validated by quantitative PCR with cells expressing wild-type and the N214D or D213A mutants (Figure 4c and Supplementary ING4 regulates the expression of chemokine-soluble factors Figure S3). Using chromatin immunoprecipitation, we showed that The best-characterized molecular function of ING4 is as a regulator ING4 can bind directly to the promoter of CXCL1, one of the of gene expression, by the recognition and establishment of soluble factors regulated in ING4-expressing cells (Figure 4d), chromatin marks in cooperation with transcription factors. To suggesting that the changes in expression we observed reflect identify genes that mediate the phenotype provoked by ectopic direct transcriptional effects of ING4, as previously described.9,27 ING4 in primary fibroblasts, we analyzed the expression profiles of To obtain further confirmation of the effect of ING4 in production these cells, using expression arrays. To this end, early-passage and release of these factors, we analyzed the presence of soluble IMR-90 fibroblasts were infected with vectors encoding wild-type factors in the conditioned medium of fibroblasts expressing ING4, ING4, the N214D mutant, or empty vector as a control. Total RNA using a commercial chemokine antibody array (see Material and was prepared at 6 days post selection, the same time as for Methods). Two independent experiments consistently showed an functional assays, and used to hybridize human whole-genome overall increase in soluble chemokines in the medium conditioned arrays. A subset of 87 genes showed differential expression in by ING4-expressing fibroblasts, relative to vector controls IMR-90 fibroblasts with wild-type ING4, relative to vector-infected (Figure 4e). Several of the chemokines more strongly upregulated fibroblasts (see Materials and Methods) of which 81 were at RNA level in the expression array and in the quantitative PCR upregulated in ING4-expressing cells and 6 genes were down- assays (such as CXCL1, CXCL2, CXCL5 or CXCL8/IL-8) also showed regulated. Next, we compared the expression of these genes in increased production and secretion, as shown by the results IMR-90 fibroblasts expressing the loss-of-function N214D mutant. from the antibody array. To confirm these results with a more We observed that 50 out of the 87 genes in this group (nearly two- quantitative assay, the levels of IL-8, one of the factors more thirds) did not show differential expression with the ING4 mutant, strongly regulated by ING4, were measured by enzyme-linked relative to vector-infected fibroblasts, supporting the specificity immunosorbent assay in the conditioned media, showing a of the ING4 signature (Figure 4a). Functional category classification modest but reproducible upregulation by wild-type ING4, of these results revealed a statistically significant enrichment in but not by ING4-mutant versions (Figure 4f). These results indicate genes involved in proliferation, cytokine activity and inflammation that ING4 positively regulates the expression and release (Figure 4b). Prompted by these results, we decided to focus our of soluble factors in primary fibroblasts.

& 2014 Macmillan Publishers Limited Oncogene (2014) 1945 – 1953 The ING4 secretory phenotype promotes tumor growth A Moreno et al 1948 RS shp53 shRb Supplementary Figure S4b). We found that the proliferative effect of ING4-conditioned medium on MDA-MB-231 cells was V ING4 V ING4 V ING4 signiﬁcantly reduced by the anti-IL-8 antibody, supporting an important role for soluble factors in the non-cell-autonomous Rb action of ING4.

p53 Effects on tumor growth The in vitro experiments described above indicated an important role for ING4 in the regulation of soluble factors in primary ING4 fibroblasts that can selectively modulate cell proliferation. To test if this effect could be observed in vivo, we evaluated directly the impact on tumor growth. IMR-90 fibroblasts expressing ING4 or actin empty vector were mixed with MDA-MB-231 breast cancer cells carrying a luciferase–green fluorescent protein (GFP) TGL cassette and they were co-inoculated in immunodeficient nude mice, 120 either subcutaneously or orthotopically into the mammary fat pad. *** ** Taking advantage of the presence of the TGL cassette, tumor 100 growth was followed over a period of 5 weeks by non-invasive bioluminiscence. In both conditions, the co-injection with ING4-expressing fibroblasts conferred a growth advantage to 80 MDA-MB-231 breast cancer cells in vivo, resulting in increased tumor growth (Figures 6a–f). Bioluminiscence data or direct 60 measurement of tumor volume or weight gave consistent results, indicating that the differences were due to differential response of 40 the MDA-MB-231 cells (Figures 6a, b, d and e, Supplementary Figure S5a). At the end of the experiment, we analyzed the excised tumors by immunohistochemistry. We did not observe gross Relative BrdU incorporation 20 differences in the overall structure or distribution of cell types in the tumors obtained with ING4 fibroblasts or controls. Staining 0 with an anti-GFP antibody confirmed that the bulk of the tumor mass was formed by GFP-positive MDA-MB-231 cells, further ING4 ING4 ING4 pLPC pLPC pLPC supporting that the differences in tumor size are due to increased RasV12 RasV12 RasV12 tumor cell growth in the presence of ING4-expressing fibroblasts pRS shp53 shRb (Figure 6g and Supplementary Figure S5b). Figure 3. p53 dependence of the antiproliferative action of ING4. (a) Western blot showing ectopic ING4 expression, and endogenous p53 and Rb levels in IMR-90 fibroblasts serially infected with shRNA DISCUSSION vectors against p53 or Rb and then with wild-type ING4. Actin In this report, we have characterized the role of ING4 in tumor- was used as a loading control. (b) BrdU incorporation rate of suppressive mechanisms in primary non-transformed cells. There IMR-90 fibroblasts infected with the indicated expression and shRNA is extensive evidence of the role of ING4 in different cellular vectors. The BrdU incorporation relative to vector-infected cells is responses in tumor or immortalized cells, but its role in the represented. The average and s.d. of three independent experi- context of normal primary cells remains largely unexplored. Our ments are shown. ***Po0.001, **Po0.01. data now show that ING4 has a clear antiproliferative effect in primary human embryonic fibroblasts, which retain intact tumor suppression mechanisms. Control of proliferation by ING4 in this cell type is mediated by the recognition of histone marks and it Effects of ING4-conditioned medium on cell proliferation requires, at least in part, functional p53, similarly to other To characterize the non-cell-autonomous role of ING4 expression ING4 functions.7,9 The specific downstream mediators of the in primary fibroblasts, we carried out experiments to assay the antiproliferative response to ING4 in primary fibroblasts remain to paracrine effect of ING4-conditioned medium in cell proliferation be identified. We have analyzed the cell-cycle inhibitors p16, p21 of primary and transformed cells. Conditioned medium obtained or p27, but we have not observed consistent upregulation of any from IMR-90 primary fibroblasts expressing wild-type or mutant of these essential cell-cycle regulators in response to ING4 ING4 was added to different cell types to test its impact on cell (Supplementary Figure S2b). The antiproliferative effect of ING4 proliferation. We observed that the addition of ING4-conditioned in these cells may also be related to its ability to induce a weak medium to non-infected IMR-90 primary fibroblasts (Figure 5a) DNA damage response, but this needs to be investigated in more provoked a clear reduction in their growth rate, which was not detail. In addition to its cell-autonomous effect in proliferation, we present when mutant ING4 was used. On the contrary, the same show here that ING4 regulates in primary fibroblasts the synthesis assay using the aggressive breast cancer cell line MDA-MB-231 of soluble factors with a selective paracrine mitogenic effect, (Figure 5b) showed that ING4-conditioned medium enhanced the which can promote the proliferation of tumor cells but inhibits proliferation of these tumor cells, and this effect was also lost with proliferation of normal fibroblasts. More importantly, this effect the N214D mutant. This differential effect on proliferation was can be translated to tumor growth in vivo, as supported by our further confirmed with growth curves and using the D213A findings that fibroblasts expressing ING4 can promote tumorigen- mutant (Supplementary Figure S4a, and data not shown). IL-8 is esis when co-injected with breast cancer cells subcutaneoulsy or one of the chemokines more clearly regulated by ING4. To test if orthotopically in the mammary fat pad. The secretory phenotype the effect of the conditioned medium was mediated, at least in that we have characterized shares features with those associated part, by this factor, we carried out additional proliferation to diverse categories of non-transformed cells that are connected assays using a blocking antibody against IL-8 (Figure 5c and to tumorigenesis. First, cancer-associated fibroblasts, present in

Oncogene (2014) 1945 – 1953 & 2014 Macmillan Publishers Limited The ING4 secretory phenotype promotes tumor growth A Moreno et al 1949

30 11.5 12.5 ING4 anti-HA no Ab 25 Genome 20 ChIP input ChIP 15 3 no DNA input 10 12 3 CXCL1 5 0 GAPDH Percentage representation Cell activity Cell Cytokine Adhesion migration response proliferation Inflammatory

2 ** ** 25 ING4 QPCR ING4 Array *** ING4 N214D QPCR 1.5 20 ING4 N214D Array 15 1 10 0.5

5 IL-8 Fold increase Relative expression 0 0 CXCL1 CXCL2 IL8 IL6 ING4

pLPC >2 D213A N214D 2 -2 <2

Figure 4. ING4 regulates the expression of chemokine-soluble factors in primary fibroblasts. (a) Heatmap representation of the results from the expression array. The signal for the subset of genes with differential expression in ING4-infected fibroblasts is shown for each duplicate of fibroblasts expressing wild-type or mutant ING4. (b) Functional enrichment analysis of ING4-regulated genes. For each category, the percentage of genes in the genome or in the subset of genes with differential expression in ING4-infected fibroblasts is shown. The fold enrichment is indicated on top of the columns. (c) Quantitative PCR validation for a selection of genes. For each target, the expression in fibroblasts with the indicated ING4 versions relative to vector-infected fibroblasts is shown. The fold increase obtained in the array is shown for reference. The graph shows the average and s.d. of two independent experiments using two RNA samples. (d) Chromatin immunoprecipitation showing the binding of ectopic HA-tagged ING4 to the promoter region of the CXCL1 locus. Immunoprecipitation was performed with an anti-HA antibody, or a negative control without antibody, followed by PCR for the CXCL1 promoter. PCR for the GAPDH locus is shown as a negative control for the chromatin immunoprecipitation (ChIP). (e) Heatmap representation of the concentration of chemokine factors in the conditioned medium of ING4-expressing fibroblasts, using an antibody array (see Material and Methods). The results are from two independent samples (exp #1 and exp #2), each analyzed in duplicate. (f) Detection of IL-8 in conditioned media of IMR-90 cells expressing the indicated ING4 versions by enzyme-linked immunosorbent assay. The value relative to vector-infected cells is shown. The average and s.d. from three independent assays is represented. ***Po0.001, **Po0.01. the tumor microenvironment, produce cytokines, chemokines and secretory phenotype than the ING4 signature. These extracellular matrix components that can cause tumor growth, observations raise the possibility that induction of senescence is angiogenesis and metastasis.28,29 Also, cytokines and chemokines a distinct feature of the repressor INGs, not shared by all the have been associated with senescence, a tumor-suppressive members of the family. However, despite the negative data shown mechanism, as part of the senescence-associated secretory here, we cannot completely rule out a role for ING4 in senescence, phenotype.30–35 Thus, the ‘ING4-fibroblast’ signature partially and it remains possible that it might contribute to this response. overlaps with secretory phenotypes linked to senescent Many of the soluble factors upregulated in ING4-expressing cells fibroblasts or to carcinoma-associated fibroblasts. Interestingly, are known targets of NF-kB, as is the case for CXCL1, CXCL2 and IL- both secretory phenotypes have overlapping functions, such as 8. These results are consistent with a role of ING4 as a positive the selective promotion of tumor growth.28,32 In the case of the regulator of this pathway in our cellular model. Previous studies senescence-associated secretory phenotype, it has been reported with overexpression or silencing of ING4 in tumor cell lines have that these factors can have antagonistic effects similar to the ones suggested that ING4 can act as a negative regulator of some shown here for ING4, promoting senescence in non-transformed NF-kB target genes.10,27,37 In contrast, data from Ing4-deficient cells, and inducing proliferation of malignant epithelial cells macrophages support a differential role of ING4 on NF-kB- in vitro and in vivo.34–36 Our results fail to support a direct role for mediated regulation.15 Our results raise the interesting ING4 in senescence in vitro, in contrast to other ING proteins.19–21 possibility that the link between ING4 and NF-kB can vary in A role in senescence has been shown so far for ING1 and ING2, different cell types or in primary cells versus immortalized or which form a subset of ING proteins primarily associated to tumor cells. In this regard, it is worth noting that NF-kB-mediated transcriptional repression. In turn, ING4 is closely related to ING5 effects are strongly context dependent. Activation of this pathway and they mainly participate in activating complexes.5 In this has been associated to tumorigenesis in diverse tumor types.38 regard, it should be noted that ING1 and ING4 regulate similar but However, recent reports clearly indicate that the activation of the not identical sets of cytokine/chemokine-related genes in primary NF-kB pathway in primary fibroblasts is associated with an fibroblasts (Abad et al.19 and this report), with the ING1-specific antiproliferative effect mediated by regulation of soluble signature being more similar to the senescence-associated factors,31,39,40 many of which coincide with those controlled

& 2014 Macmillan Publishers Limited Oncogene (2014) 1945 – 1953 The ING4 secretory phenotype promotes tumor growth A Moreno et al 1950 IMR-90 MDA-MB-231 200 160 ** ** * 180 140 160 120 * 140 100 120 80 100 80 60 60 40

BrdU-positive cells (%) 40 BrdU-positive cells (%) 20 20 0 0 VING4 ING4 RasV12 V ING4 ING4 RasV12 N214D N214D

4 * vector ING4 3.5

2.5

1.5 Relative Growth 1

0.5

0 0 0.1 1 10 anti-IL8 antibody (µg/ml) Figure 5. Selective effects of ING4-conditioned medium on proliferation of primary or tumor cells. (a, b) BrdU incorporation rate of IMR-90 fibroblasts infected (a) or MDA-MB-231 breast cancer cells (b) grown with conditioned medium obtained from IMR-90 fibroblasts infected with the indicated vectors. The BrdU incorporation relative to cells grown with medium from vector-infected fibroblasts is represented. The average and s.d. of three independent experiments is shown. **Po0.01, *Po0.05. (c) Effect of an anti-IL-8 blocking antibody in proliferation of MDA-MB-231 cells grown with conditioned medium obtained from IMR-90 fibroblasts infected with ING4 or empty vector. The fold increase in cell number after 4 days in culture is represented. The average and s.d. of three independent experiments is shown.

by ING4 in our work. In summary, our results have unveiled an Retroviral transduction important role for ING4 in the regulation of soluble factors in Retroviral infection was performed as described previously in Abad et al.19 primary ﬁbroblasts that can inﬂuence tumor growth and The following vectors were used: pLPC, pLPC-HA-ING4, pLPC-HA-ING4D213A, progression. Our data imply that the expression level of ING4 pLPC-HA-ING4N214D, pLPC-RasV12, pRetroSuper, pRetroSuper-p53 and in non-transformed cells, and not only in tumor cells, may be pRetroSuper-Rb (a kind gift of Daniel Peeper, NKI, Amsterdam). After relevant for tumor growth and progression. infection, cells were selected with puromycin (2 mg/ml), hygromicin (75 mg/ ml) or blastycidin (10 mg/ml) for a period from 3 to 6 days.

Western blot MATERIALS AND METHODS Preparation of total lysates and western blot analysis were carried out as Mutagenesis described in Palmero et al.42 The antibodies used were: anti-HA (1:500, Mutant versions of ING4 were generated with the Quick Change Site- 12CA5, Roche, Manheim, Germany); anti-ING4: (1:500, 16188-1-AP, Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA), using Proteintech, Chicago, IL, USA); anti-cyclin A (1:1000, SC-751, Santa Cruz, the complementary DNA of the human ING4 protein cloned in the Heidelberg, Germany); anti-p21 (1:500, C-19, Santa Cruz); anti-p27 (1:500, retroviral vector pLPC. Speciﬁc mutations were incorporated 554069, BD Pharmingen, San Jose, CA, USA); anti-p16 (1:200, JC8, CRUK, by PCR using Pfu Turbo DNA Polymerase (Stratagene), and subsequent London, UK); anti-p53 (1:500, DO-1, Santa Cruz); anti-Rb (1:1000, 554136, digestion with DpnI, and they were conﬁrmed by sequencing. BD Pharmingen); anti-actin (1:10000, AC-15, Sigma, St Louis, MO, USA) and anti-tubulin (1:10000, T9026, Sigma).

Cell culture IMR-90 primary human diploid fibroblasts were obtained from the Reverse transcriptase quantitative PCR American Type Culture Collection (ATCC, Manassas, VA, USA) and used at Total RNA was prepared using Tri Reagent (Sigma) from asynchronously a population doubling level lower than 30. The breast tumor cell line MDA- growing cells. RNA was used for complementary DNA sıńtesis using M/MLV MB-231, containing the TGL cassette (luciferase/GFP), was provided by reverse transcriptase (Promega, Madison, WI, USA). One microliter of the Roger Gomis and Joan Massague´.41 Cells were grown in Dulbecco’s reaction was used to perform PCR reactions. Assays were performed using modified Eagle’s medium (Gibco, Grand Island, NY, USA) with 10% fetal calf Universal Probe Library probes in an ABI9700 machine (Invitrogen, serum and antibiotics, at 37 1C in 5% CO2. Carlsbad, CA, USA), using 18S ribosomal RNA as internal reference.

Oncogene (2014) 1945 – 1953 & 2014 Macmillan Publishers Limited The ING4 secretory phenotype promotes tumor growth A Moreno et al 1951 800 V 18 V 700 ING4 16 ING4 600 14 500 12 10 400 8 Relative

300 Relative 6 200

Bioluminiscence 4 Bioluminiscence 100 2 0 0

0.5 V 0.07 V ING4 0.06 ING4 0.4 0.05 ) )

0.3 3 0.04 3

(cm 0.03 (cm 0.2 0.02 Tumor volume Tumor volume 0.1 0.01 0 0

* *

MDA-MB231-TGL MDA-MB231-TGL MDA-MB231-TGL + + IMR90/V IMR90/ING4 Figure 6. ING4 expression in primary fibroblasts promotes tumor growth. (a, d) Bioluminiscence emission of tumors containing fibroblasts with empty vector or ING4, after 5 weeks. The increase in luminiscence emission relative to day 1 for each individual tumor and the average are shown. (a) subcutaneous injection; (d) orthotopic injection (b, e) Size of tumors containing fibroblasts with ING4 or empty vector. The value for each individual tumor at the end of the experiment and the average are shown. (b) subcutaneous injection; (e) orthotopic injection. (c, f) Representative bioluminiscence images from mice with subcutaneous (c) or orthotopic (f) injections (c) no injection (left), MDA-MB-231-TGL and IMR-90 with empty vector (right, left flank) or MDA-MB-231-TGL and IMR-90 with ING4 (right, right flank). (f) MDA-MB- 231-TGL and IMR-90 with empty vector (left), MDA-MB-231-TGL and IMR-90 with ING4 (middle), no injection (right). (g) Representative immunohistochemistry images of subcutaneous tumors with the indicated combinations of MDA-MB-231-TGL tumor cells and IMR-90 fibroblasts. Sections were stained with anti-GFP antibody to detect GFP-expressing MDA-MB-231-TGL cells (arrowheads). Asterisks indicate GFP-negative IMR-90 fibroblasts. Scale bar: 20 mm.

Cell culture assays Santa Clara, CA, USA). Two independent retroviral infections were BrdU incorporation of IMR-90 cells was performed as previously performed in early pasaje IMR-90 fibroblasts with wild-type ING4 described19 by immunofluorescence with an anti-BrdU antibody (1:1000, and the N214D mutant. Total RNA was labeled and competitively BP40250, Megabase Research Products, Lincoln, NE, USA) except that 4-h hybridized to arrays using as a reference a reverse-labeled sample BrdU pulses were used. g-H2AX (1:500, JBW301, Upstate, Billerica, MA, USA) from vector-infected cells. One additional hybridization was performed was detected by immunofluorescence as described.19 Growth curves and labeling the RNAs with the reciprocal fluorochromes. Differentially senescence-associated beta galactosidase activity were performed as expressed genes in wild-type ING4 versus vector were identified described.19 with GEPAS (Gene Expression Pattern Analysis Suite, http://gepas3. bioinfo.cipf.es) selecting genes with a fold difference of at least two in all of the samples and s.d. o0.5. This subset was further analyzed in the Microarray gene expression profiles arrays from ING4N214D-infected cells. Functional enrichment analysis Microarray experiments were performed essentially as described,19 was performed using the FatiGO application (http://babelomics.bioinfo. using Human Whole-Genome array G4112F (Agilent Technologies, cipf.es). Microarray raw data tables have been deposited in the Gene

& 2014 Macmillan Publishers Limited Oncogene (2014) 1945 – 1953 The ING4 secretory phenotype promotes tumor growth A Moreno et al 1952 Expression Omnibus under the accession number of GSE42412 (submitter ACKNOWLEDGEMENTS GM-B). We thank R Gomis, J Massague, G Peters, D Peeper and A Cano for providing cell lines and reagents, V Miguel for help in some experiments and I Peruzza for technical Chromatin immunoprecipitation assistance. This work was supported by grants from the Spanish Ministry of Science and Innovation to IP (SAF2009-09031 and SAF2012-32117) and GM-B (SAF2010- Chromatin immunoprecipitation assays were performed essentially as 20175), and from the Madrid Regional Government (S2010/BMD-2303) to GM-B. described in Gomez-Cabello et al.43 using 293T cells transfected with HA-tagged ING4.

Conditioned media assays REFERENCES To obtain the conditioned media from infected and selected 1 Aguissa-Toure AH, Wong RP, Li G. The ING family tumor suppressors: from IMR-90 cells, cells were grown until B90% confluence, washed with structure to function. Cell Mol Life Sci 2010; 68:45–54. phosphate-buffered saline (PBS) and incubated in serum-free medium for 2 Coles AH, Jones SN. The ING gene family in the regulation of cell growth and 72 h. After that time, medium was collected, cleared by centrifugation and tumorigenesis. J Cell Physiol 2009; 218: 45–57. stored at À 80 1C until use. For growth assays, serum was added at a final 3 Jafarnejad SM, Li G. Regulation of p53 by ING family members in suppression of concentration of 1%. In some experiments, an antibody against IL-8 was tumor initiation and progression. Cancer Metastasis Rev 2012; 31: 55–73. also added at different concentrations. 4 Ythier D, Larrieu D, Brambilla C, Brambilla E, Pedeux R. The new tumor suppressor genes ING: genomic structure and status in cancer. Int J Cancer 2008; 123: 1483–1490. Enzyme-linked immunosorbent assay and chemokine arrays 5 Doyon Y, Cayrou C, Ullah M, Landry AJ, Cote V, Selleck W et al. ING tumor Conditioned media obtained from infected and selected IMR-90 cells were suppressor proteins are critical regulators of chromatin acetylation required for used for IL-8 enzyme-linked immunosorbent assay assays (D8000C, R&D genome expression and perpetuation. Mol Cell 2006; 21: 51–64. Systems, Minneapolis, MN, USA), and to hybridize a chemokine antibody 6 Pena PV, Davrazou F, Shi X, Walter KL, Verkhusha VV, Gozani O et al. Molecular array (Human Cat #AAH-CHE-1, Ray Biotech, Norcross, GA, USA) following mechanism of histone H3K4me3 recognition by plant homeodomain of ING2. manufacturer instructions in both cases. The array signal was quantified by Nature 2006; 442: 100–103. densitometry using ImageJ software (NIH, Bethesda, MD, USA) and 7 Shiseki M, Nagashima M, Pedeux RM, Kitahama-Shiseki M, Miura K, Okamura S relativized to the signal of the internal positive controls. et al. p29ING4 and p28ING5 bind to p53 and p300, and enhance p53 activity. Cancer Res 2003; 63: 2373–2378. 8 Kim S, Chin K, Gray JW, Bishop JM. A screen for genes that suppress loss of Tumor growth assays contact inhibition: identification of ING4 as a candidate tumor suppressor gene in IMR-90 infected with wild-type ING4 or empty vector, used at day 6 after human cancer. Proc Natl Acad Sci USA 2004; 101: 16251–16256. selection, and MDA-MB-231-TGL cells growing at subconfluence were 9 Hung T, Binda O, Champagne KS, Kuo AJ, Johnson K, Chang HY et al. trypsinized and mixed in a 3:1 ratio (6 Â 105 and 2 Â 105 per injection, ING4 mediates crosstalk between histone H3 K4 trimethylation and H3 acetylation respectively). Cells were resuspended in a final volume of 100 ml de PBS per to attenuate cellular transformation. Mol Cell 2009; 33: 248–256. injection, and were inoculated subcutaneously in the flanks or in the 10 Garkavtsev I, Kozin SV, Chernova O, Xu L, Winkler F, Brown E et al. The candidate mammary fat pads of female 6- to 8-week-old nu/nu mice. For detection of tumour suppressor protein ING4 regulates brain tumour growth and bioluminiscence, 200 ml of a solution of D-luciferin (15 mg/ml in PBS) was angiogenesis. Nature 2004; 428: 328–332. injected intraperitoneally in anesthetized mice. Images were obtained at 11 Shen JC, Unoki M, Ythier D, Duperray A, Varticovski L, Kumamoto K et al. Inhibitor different points, starting B5 min after the injection, with a Xenogen IVIS of growth 4 suppresses cell spreading and cell migration by interacting with a Lumina II apparatus and analyzed with the Living Image Acquisition and novel binding partner, liprin alpha1. Cancer Res 2007; 67: 2552–2558. Analysis Software (Xenogen Corporation, Alameda, CA, USA). Biolumines- 12 Li J, Martinka M, Li G. Role of ING4 in human melanoma cell migration, invasion cence intensity was measured at the peak of emission as the total photon and patient survival. Carcinogenesis 2008; 29: 1373–1379. 2 flux (photons/second/cm ) for each region of interest. Measurements were 13 Moreno A, Palacios A, Orgaz JL, Jimenez B, Blanco FJ, Palmero I. Functional impact repeated every 7 days and relativized to the data obtained in the first of cancer-associated mutations in the tumor suppressor protein ING4. Carcino- measurement for each ROI. Tumor volume was measured at the same time genesis 2010; 31: 1932–1938. 2 3 points with a caliper, using the formula V ¼ A Â B /2 (mm ), where A is the 14 Tapia C, Zlobec I, Schneider S, Kilic E, Guth U, Bubendorf L et al. Deletion of the largest diameter and B is the perpendicular diameter. Five weeks after cell inhibitor of growth 4 (ING4) tumor suppressor gene is prevalent in human injection, mice were killed and the tumors were removed, weighed and epidermal growth factor 2 (HER2)-positive breast cancer. Hum Pathol 2011; 42: processed for immunohistochemistry. 983–990. 15 Coles AH, Gannon H, Cerny A, Kurt-Jones E, Jones SN. Inhibitor of growth-4 Immunohistochemical analysis promotes IkappaB promoter activation to suppress NF-kappaB signaling and innate immunity. Proc Natl Acad Sci USA 2010; 107: 11423–11428. Tumors were fixed in formalin and processed for paraffin embedding. Five- 16 Kuilman T, Michaloglou C, Mooi WJ, Peeper DS. The essence of senescence. micrometer sections were deparaffinized and rehydrated, treated 10 min Genes Dev 2010; 24: 2463–2479. with 3% H2O2 in methanol to inhibit the endogenous peroxidase activity, 17 Collado M, Blasco MA, Serrano M. Cellular senescence in cancer and aging. followed by washes with 0.1% PBS-Triton. The slides were then blocked Cell 2007; 130: 223–233. with 1% bovine serum albumin in 0.1% PBS-Triton for 30 min, washed 18 Goeman F, Thormeyer D, Abad M, Serrano M, Schmidt O, Palmero I et al. Growth again with 0.1% PBS-Triton and incubated with anti-GFP antibody inhibition by the tumor suppressor p33ING1 in immortalized and primary cells: 1 overnight at 4 C (1:100, A6455, Invitrogen, Carlsbad, CA, USA). After involvement of two silencing domains and effect of Ras. Mol Cell Biol 2005; 25: washings with 0.1% PBS-Triton, samples were incubated 30 min with an 422–431. anti-rabbit secondary antibody HRP-labeled (K4002, DAKO, Glostrup, 19 Abad M, Moreno A, Palacios A, Narita M, Blanco F, Moreno-Bueno G et al. The Germany) at room temperature. After washing, color was developed with tumor suppressor ING1 contributes to epigenetic control of cellular senescence. diaminobenzidine (DAKO) and sections were counterstained with Aging Cell 2011; 10: 158–171. hematoxylin for 5 s. Finally, samples were dehydrated, mounted in DPX 20 Pedeux R, Sengupta S, Shen JC, Demidov ON, Saito S, Onogi H et al. ING2 (Sigma) and analyzed in a Nikon90i microscope (Nikon, Tokyo, Japan) at regulates the onset of replicative senescence by induction of p300-dependent 4 and 10 magnification. Â Â p53 acetylation. Mol Cell Biol 2005; 25: 6639–6648. 21 Menendez C, Abad M, Gomez-Cabello D, Moreno A, Palmero I. ING proteins in Statistics cellular senescence. Curr Drug Targets 2009; 10: 406–417. Statistical significance was estimated using the Student’s t-test. 22 Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW. Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a. Cell 1997; 88: 593–602. 23 Wong RP, Lin H, Khosravi S, Piche B, Jafarnejad SM, Chen DW et al. Tumour CONFLICT OF INTEREST suppressor ING1b maintains genomic stability upon replication stress. Nucleic The authors declare no conflict of interest. Acids Res 2011; 39: 3632–3642.

Oncogene (2014) 1945 – 1953 & 2014 Macmillan Publishers Limited The ING4 secretory phenotype promotes tumor growth A Moreno et al 1953 24 Larrieu D, Ythier D, Binet R, Brambilla C, Brambilla E, Sengupta S et al. ING2 34 Kuilman T, Peeper DS. Senescence-messaging secretome: SMS-ing cellular stress. controls the progression of DNA replication forks to maintain genome stability. Nat Rev Cancer 2009; 9: 81–94. EMBO Rep 2009; 10: 1168–1174. 35 Coppe JP, Desprez PY, Krtolica A, Campisi J. The senescence-associated secretory 25 Campisi J, d’Adda di Fagagna F. Cellular senescence: when bad things happen to phenotype: the dark side of tumor suppression. Annu Rev Pathol 2010; 5: good cells. Nat Rev Mol Cell Biol 2007; 8: 729–740. 99–118. 26 Chicas A, Wang X, Zhang C, McCurrach M, Zhao Z, Mert O et al. Dissecting the 36 Krtolica A, Parrinello S, Lockett S, Desprez PY, Campisi J. Senescent fibroblasts unique role of the retinoblastoma tumor suppressor during cellular senescence. promote epithelial cell growth and tumorigenesis: a link between cancer and Cancer Cell 2010; 17: 376–387. aging. Proc Natl Acad Sci USA 2001; 98: 12072–12077. 27 Nozell S, Laver T, Moseley D, Nowoslawski L, De Vos M, Atkinson GP et al. 37 Colla S, Tagliaferri S, Morandi F, Lunghi P, Donofrio G, Martorana D et al. The ING4 tumor suppressor attenuates NF-kappaB activity at the promoters of The new tumor-suppressor gene inhibitor of growth family member 4 target genes. Mol Cell Biol 2008; 28: 6632–6645. (ING4) regulates the production of proangiogenic molecules by myeloma 28 Erez N, Truitt M, Olson P, Arron ST, Hanahan D. Cancer-associated fibroblasts are cells and suppresses hypoxia-inducible factor-1 alpha (HIF-1alpha) activity: activated in incipient neoplasia to orchestrate tumor-promoting inflammation in involvement in myeloma-induced angiogenesis. Blood 2007; 110: an NF-kappaB-dependent manner. Cancer Cell 2010; 17: 135–147. 4464–4475. 29 Lazennec G, Richmond A. Chemokines and chemokine receptors: new insights 38 Perkins ND. The diverse and complex roles of NF-kappaB subunits in cancer. into cancer-related inflammation. Trends Mol Med 2010; 16: 133–144. Nat Rev Cancer 2012; 12: 121–132. 30 Acosta JC, O’Loghlen A, Banito A, Guijarro MV, Augert A, Raguz S et al. Chemokine 39 Acosta JC, O’Loghlen A, Banito A, Raguz S, Gil J. Control of senescence by CXCR2 signaling via the CXCR2 receptor reinforces senescence. Cell 2008; 133: and its ligands. Cell Cycle 2008; 7: 2956–2959. 1006–1018. 40 Liu F, Wu S, Ren H, Gu J. Klotho suppresses RIG-I-mediated senescence-associated 31 Chien Y, Scuoppo C, Wang X, Fang X, Balgley B, Bolden JE et al. Control of the inflammation. Nat Cell Biol 2011; 13: 254–262. senescence-associated secretory phenotype by NF-kappaB promotes senescence 41 Minn AJ, Gupta GP, Siegel PM, Bos PD, Shu W, Giri DD et al. Genes that mediate and enhances chemosensitivity. Genes Dev 2011; 25: 2125–2136. breast cancer metastasis to lung. Nature 2005; 436: 518–524. 32 Coppe JP, Patil CK, Rodier F, Sun Y, Munoz DP, Goldstein J et al. Senescence- 42 Palmero I, Murga M, Zubiaga A, Serrano M. Activation of ARF by oncogenic associated secretory phenotypes reveal cell-nonautonomous functions stress in mouse fibroblasts is independent of E2F1 and E2F2. Oncogene 2002; 21: of oncogenic RAS and the p53 tumor suppressor. PLoS Biol 2008; 6: 2853–2868. 2939–2947. 33 Kuilman T, Michaloglou C, Vredeveld LC, Douma S, van Doorn R, Desmet CJ et al. 43 Gomez-Cabello D, Callejas S, Benguria A, Moreno A, Alonso J, Palmero I. Oncogene-induced senescence relayed by an interleukin-dependent Regulation of the microRNA processor DGCR8 by the tumor suppressor ING1. inflammatory network. Cell 2008; 133: 1019–1031. Cancer Res 2010; 70: 1866–1874.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)

& 2014 Macmillan Publishers Limited Oncogene (2014) 1945 – 1953