A tissue microarray was constructed from the 113 patients described in Table S1 in Additional file 1. As detailed in Additional file 2, the clinical trial was conducted about 30 years ago with no consent required at that time. Cores (600 μm diameter) of histologically confirmed invasive breast carcinomas were extracted from the original paraffin blocks and re-embedded into a gridded recipient paraffin block using a tissue arrayer (Alphelys; Beecher Inc., Plaisir, France). For each case, three tumor cores and one normal breast core were taken from the original block.

Tissue microarray immunostaining was performed on a Techmate Horizon™ slide processor (Dako, Trappes, France { }). Sections were incubated with antibodies to ERα (NCL-L-ER-6F11; Novocastra, Nanterre, France), PR (PgR636 clone; Dako) and RhoB (sc-180; Santa Cruz Biotechnology, Heidelberg, Germany). The ERα and PR status was classified by immunohistochemistry expression as positive (≥ 10% immunoreactive cells) or negative (< 10% immunoreactive cells), according to the standards applied in France, as recommended by the Groupe d'Evaluation des Facteurs Pronostiques par Immunohistochimie dans les Cancers du Sein. RhoB immunostaining was analyzed by evaluation of the percentage of tumor-stained cells and staining intensity, allowing assessment of an ImmunoReactive Score:

IRS = %  score  ×  intensity score

Correlation of RhoB expression with clinical variables was assessed using Mann-Whitney and Spearman's rank tests. Univariate survival analysis was performed for disease-free survival by applying the log-rank test to RhoB expression levels stratified by median value. The Kaplan-Meier method was used to link the disease-free survival according to RhoB expression in the tumors.

The human breast adenocarcinoma cell lines MCF-7, ZR75, T47D, SK-BR-3 and MDA-MB-231 were obtained from the ATCC (Molsheim, France). Cells were grown routinely in DMEM (Lonza Levallois-Perret, France), supplemented with 5% fetal bovine serum (FBS; Pan-Biotech, Aidenbach, Germany). Hormonoresistant LCC2 cells (R Clarke, Karmanos Cancer Center, provided the parent MCF-7) were grown in F-12 (Lonza) phenol red-free medium, containing 5% dextran-coated charcoal (DCC)-treated FBS. Cryopreservation of cell cultures ranged from passage 1 to 10. Cells were used for experiments during up to 20 passages. Mouse embryonic fibroblasts (MEFs) were grown in DMEM with 10% FBS.

Where indicated, cells were deprived of E₂ by growing them in phenol red-free medium containing 5% DCC-treated FBS. Cells were treated with 50 nM E₂ or 50 ng/ml epidermal growth factor or 2 mM ICI-182, 780 (Sigma-Aldrich Chimie, Lyon, France).

Transient transfection of 40 nM siRNA was performed using Oligofectamine^® (Life Technologies - Invitrogen, Saint Aubin, France) according to the manufacturer's instructions. The siRNA sequences of the oligonucleotide duplexes (Eurogentec, Angers, France) were siB1 (5'-CCGUCUUCGAGAACUAUGU-dTdT-3') and siB2 (5'-UGAUAUCCCUUGUCUGUAA-dTdT-3'), siER (5'-GGGAGAAUGUUGAAACACA-dTdT-3') and the nonspecific sequence siControl (5'-GACGUGGGACUGAAGGGGU-dTdT-3').

Replication-defective (ΔE1, E3) adenoviral vectors expressing RhoB under the transcriptional control of the CMV promoter were constructed with the Adeasy System (Qbiogen, Illkirch, France) as described previously 31. Cells were transduced with adenoviral vectors (control empty vector or expressing RhoB) at a multiplicity of infection of 300:1.

Cells were lysed and protein analyzed by SDS-PAGE 18. Antibodies were used against RhoB (sc-180), ERα (sc-543) and, ERβ (sc-53494) from Santa Cruz Biotechnology, phospho-ERα (Ser¹¹⁸, 2511 and Ser¹⁶⁷, 2514) from Cell Signaling (Saint Quentin Yvelines, France), β-actin (MAB1501) from Chemicon (Merck Millipore; Darmstadt, Germany), and secondary antibodies were anti-mouse (MP21120) and anti-rabbit (MP23145) horseradish peroxidase from Interchim (Montluçon, France) using a chemiluminescence detection kit (ECL; Pierce, Thermo Fisher Scientific, Courtaboeuf, France). Protein abundance was quantified by Image Quant TL analysis (GE Healthcare, Velizy-Villacoublay, France).

Heterozygous (+/-) and homozygous (-/-) mutant strains of RhoB-deficient mice 10 were kindly provided by G Prendergast (Lankenau Institute for Medical Research, Wynnewood, PA, USA). Claudius Regaud Institute animal ethics committee approval was obtained (# ICR-2012-001-A) for the use of the animal model and the study protocols. Mice were housed in polycarbonate cages in controlled conditions. MEFs were generated 10 before lysis 18. Protein were extracted and analyzed as described above with antibody against murine ERα (sc-542) or RhoB (sc-180; Santa Cruz Biotechnology).

The level of activated RhoB and GTP-bound RhoB protein was measured using the GST fusion protein containing the Rho binding domain of Rhotekin 8. The amount of GTP-bound RhoB and the total amount of RhoB in cell lysates were determined by western blot as described above 32.

For each condition, 5 × 10⁵ cells were seeded onto glass slides dishes and were grown for 3 days in phenol red-free DMEM, containing 5% DCC-treated FBS. Cells were then treated with E₂ during 16 hours, washed in PBS and fixed in RCL2 33 for 2 minutes. Staining was performed by a Techmate Horizon™ slide processor as described for Figure 1. The primary antibody used was a monoclonal anti-ERα antibody (HC-20, sc-543, dilution 1:50; Santa Cruz Biotechnology) or an anti-PR antibody (PgR636 clone, dilution 1:50; Dako).

The level of ERα and PR staining was determined by evaluation of the percentage of tumor-stained cells and staining intensity, allowing assessment of an IRS.

Total RNA was isolated 16 hours after stimulation by E₂, extracted using the RNeasy kit following the manufacturer's instructions (Qiagen, Courtaboeuf, France) and reverse-transcribed using the iScript™ cDNA Synthesis Kit (Biorad, Marnes la Coquette, France). Quantitative PCR was performed with an iQreal-time PCR detection system (Biorad). The specific human primer pairs used were: for GAPDH, 5'-TGCACCACCAACTGCTTAGC-3' and 5'-GGCATGGACTGTGGTCATGAG-3'; for 28s, 5'-TCGCTGGGTCTTGGATGTC-3' and 5'-AGCAGATTGTGACAGACCATTCC-3'; for PR, 5'-CGCGCTCTACCCTGCACTC-3' and 5'-TGAATCCGGCCTCAGGTAGTT-3'; and for ERα, 5'-CCACCAACCAGTGCACCATT-3' and 5'-GGT CTTTTCGTATCCCACCTTTC-3' (Eurogentec).

Development of stable transfectants of MCF-7 cells (MELN cells) has been described previously 34. These cells, kindly provided by P Balaguer (INSERM 540, France), were established by transfecting MCF-7 cells with ERE-β-globin-luc-SV-Neo plasmid and thus express luciferase in an estrogen-dependent manner.

Three days after siRNA transfection, MELN cells were seeded in DMEM-DCC-treated FBS during 3 days. They were then treated with E₂ or ethanol for 16 hours. Cells were then lysed in reporter lysis buffer (Promega, Charbonnières Les Bains, France). The luciferase activity was measured with luciferase assay reagent (Promega) according to the manufacturer's instructions. Protein concentration was measured to normalize the luciferase activity data.

Chromatin immunoprecipitation assays were performed with modifications of the procedure of Metivier and colleagues 35 as described elsewhere 36. Briefly, 10⁶ cells were synchronized by 3 days of culture in DMEM 5% DCC-treated FCS and treated during 1 hour with E₂. Immunoprecipitation antibodies were ERα (sc-543), HDAC1 (sc-6299), and polII (sc-899) from Santa Cruz Biotechnology, and acetylated histone H3 (ab1791) and H4 (ab193) from Abcam (Paris, France). Quantitative PCR were performed on an iCycler (Biorad) using the primers 5'-GGCGACACAGCAGTGGGGAT-3' and 5'-TCTCCTCCCTCTGCCCCTATATTC-3' (Eurogentec) to amplify the fragment of the human PR promoter flanking the +745 AP-1 site 37.

Forty-eight hours after siRNA transfection or transduction, cells were seeded in DMEM 5% FBS with ethanol or E₂ and counted daily during the next 4 days with a coulter counter (Beckman Coulter, Grenoble, France).

Using the Oncomine microarray database, a correlation between RhoB and either ERα or PR expression was documented in datasets from 19 breast cancer studies. Moreover, four studies showed an inverse correlation between RhoB expression and tumor grade. To confirm the clinical relevance of RhoB expression levels in breast cancer oncogenesis and outcome, we compared the expression levels of RhoB in breast carcinomas from a cohort of 113 patients treated or not by tamoxifen in an adjuvant setting in a randomized prospective study. Patient characteristics are described in Table S1 in Additional file 1. ERα and PR assessments were initially performed with biochemistry techniques at the time of diagnosis (between 1980 and 1983). They have all been performed again at the time of analysis, in parallel with RhoB assessment. After new pathological analysis, 65 tumors were classified as grade I and II, 74 were ERα-positive, and 59 were PR-positive. Among these tumors, 23 presented a lymphovascular invasion and 39 cases presented with positive lymph nodes. After randomization, we determined that age, ERα and PR status, histological grade, and type and lymphovascular invasion status were similar in the two groups of patients, whether treated or not. Nonetheless, patients treated with tamoxifen (n = 62) had less favorable prognostic factors regarding positive lymph nodes, pathological tumor size and number of mitoses.

Figure 1A shows the RhoB immunohistochemistry stainings of two representative tumors of patients displaying opposite hormone receptor status (ERα/PR-negative and ERα/PR-positive), illustrating the direct correlation between RhoB and both ERα and PR expression. Enlargement of the photographs (Figure 1B) illustrates in the tumor cells of the ERα/PR-negative patient that RhoB is present and specifically cytoplasmic, although the staining is weak. For the ERα/PR-positive patient, in addition to cytoplasmic staining, a strong staining was also observed at nucleus level.

Statistical analysis of the results indicated that RhoB expression in tumors is strongly correlated with the percentage of ERα (Spearman's ρ = 0.3659, P = 0.001) and PR (ρ = 0.2544, P = 0.0076) expression, but inversely correlated with histological tumor size (ρ = -0.2344, P = 0.0166) and number of mitoses (ρ = -0.2009, P = 0.0362). We divided the patients with ERα-positive tumors into two groups with either low or high RhoB expression, and in each group we further divided the patients into groups that did or did not receive tamoxifen. This analysis argued that the level of RhoB expression was not correlated with disease-free survival of patients with ERα-positive tumors, regardless of tamoxifen treatment or not (Kaplan-Meier curve shown in Figure S1 in Additional file 3).

The RhoB IRS (see Materials and methods) integrating both the percentage and intensity of the staining (Figure 1C) was significantly higher in ERα-positive tumors (median 10.5 (3 to 12)) as compared with ERα-negative tumors (median 8 (1 to 12)), in PR-positive tumors (median 9 (3 to 12)) as compared with PR-negative tumors (median 8 (1 to 12)), and in patients with tumor grade I and II (median 9 (3 to 12)) as compared with grade III (median 8 (1 to 12)). The line in the center of each box represents the median value of the distribution, and the upper and lower ends of the box are the upper and lower quartiles, respectively. The RhoB level of expression was also higher in the smaller tumor size (≤ 2 cm, median 12 (2 to 12)) as compared with larger tumors (> 2 cm, median 8 (1 to 12)). The RhoB IRS score was not correlated with the presence of lymphovascular invasion (P = 0.26), nor with the presence of lymph node invasion (P = 0.74).

We confirmed the expectation of an effect of RhoB on ERα expression in MCF-7 cells in the presence of E₂, using two different siRNA sequences to target RhoB mRNA (siB1 and siB2) (Figure 2A, left panel). We observed associated decreases in ERα expression of 60% and 62% (siB1 and siB2, respectively). This result was confirmed using two other independent siRNA sequences targeting RhoB (data not shown).

To rule out the possibility of off-target effects of the siRNA approach, we transduced MCF-7 cells with an adenoviral vector expressing RhoB (Figure 2A, right panel). The observed 4.5-fold overexpression of RhoB increased the expression of ERα (186%), further supporting the hypothesized regulatory relationship.

We extended the study of RhoB downregulation on ERα expression to additional cell lines (Figure S2 in Additional file 4), confirming a decrease of ERα expression with RhoB depletion using siB1 and siB2 in T47D and ZR75 cells (hormone-dependent cells) or in LCC2 cells (hormone-resistant cells). We confirmed the involvement of RhoB on ERα expression in vivo using mice that are genetically deficient in RhoB (Figure 2B). A major decrease of ERα expression was visualized in MEFs collected from RhoB^+/- mice with an even more dramatic decrease in MEFs collected from RhoB^-/- mice. These results extended the support for a hypothesized regulatory relationship between RhoB and ERα.

We also studied the levels of ERα phosphorylated forms whose involvement as prognostic markers is discussed below (Figure 2C). We observed that the decrease of ERα expression observed when RhoB expression is downregulated is not associated with any significant specific change of P-Ser¹¹⁸ or P-Ser¹⁶⁷ ERα level. Indeed, the ratio phosphorylated/total ERα is not modified after RhoB inhibition.

We then demonstrated that ERα expression is dramatically decreased in the presence of the pure anti-estrogen ICI-182, 780, even when RhoB is overexpressed (Figure 2D).

In examining the effect of estrogen on RhoB expression and activity and given that the activation of RhoB is important for its physiological effect 81820, we investigated the role of estrogen stimulation on the RhoB guanine nucleotide binding status (Figure 3A) to analyze rapid effects of E₂ stimulation on RhoB activation as well as expression. We observed that RhoB is quickly activated at 30 minutes without any change of total RhoB expression, as described with epidermal growth factor stimulation 820. At 1 hour and 2 hours of treatment, E₂ rapidly increased the total RhoB expression with a second peak of GTP-bound RhoB occurring at 2 hours associated with a parallel increase of total RhoB expression.

Together these results suggested that RhoB and estrogen signaling are integrated into a feed-forward loop that may positively modify the biological effects of estrogen treatment.

We analyzed further the effect of RhoB downregulation on ERα and ERβ expression in the absence and presence of E₂ (Figure 3B). As expected, E₂ treatment induced a major significant decrease of ERα expression and concomitantly of RhoB expression. In the presence of siB2, a 36 to 41% decrease of ERα expression was observed in the absence of E₂ and 48 to 52% in the presence of E₂. In contrast, ERβ expression was not clearly altered by RhoB downregulation (Figure 3B). Using immunocytochemistry, we confirmed these results suggesting that RhoB regulates ERα expression and observed no dramatic changes of RhoB subcellular localization within MCF-7 cells (Figure 3C). ERα was highly concentrated in the nucleus of the untreated control with a significant staining of the related cytoplasm. In the presence of E₂, the staining intensities of both nuclei and cytoplasms were clearly decreased with no more detectable staining in the cytoplasm, confirmed by IRS assessment. For siB2-treated cells, a similar major decrease of the labeling intensity was observed in the cytoplasm and the nucleus, both in the presence and the absence of E₂. In the presence of E₂, the IRS score was maintained at a value of 2 because, in spite of the difference from 30 to 20% of stained cells induced by siB2 treatment, the intensity is kept very low in both cases (value of 1).

We then investigated whether RhoB downregulation modulates ERα mRNA expression (Figure 3D). We first confirmed the expected decrease of ERα mRNA expression in control cells treated by E₂ alone (to 0.49). In the cells treated with siB2, a dramatic decrease of ERα mRNA was observed in the absence of E₂ (from 1 to 0.53) and much lighter in the presence of E₂ (from 0.49 to 0.37).

We hypothesized that the ability of RhoB to modulate ERα expression could control the major transcriptional functions of ERα. To examine this hypothesis, we used MELN cells that express luciferase in an estrogen-dependent manner. After siB1 or siB2 transfection and E₂ deprivation, cells were treated or not with E₂ and luciferase activity was quantified (Figure 4A). In the absence of E₂ (upper-right corner), the transfection of cells by both siB1 and siB2 induced a significant decrease of the luciferase activity (0.5-fold for B1 and 0.7-fold for B2). In the control cells, E₂ treatment led to a ninefold induction of the luciferase activity. In the presence of E₂, luciferase expression was significantly decreased for cells treated by the siB1 and siB2 sequences (respectively 3.12 and 4.07 compared with 9 AU). Nonetheless, E₂ induction was still observed for all cells transfected by either siB1 or siB2. We extended the analysis of RhoB inhibition on two known estrogen-regulated genes, ERα itself described above (Figure 3D) and PR, using quantitative RT-PCR (Figure 4B).

Regarding PR, a major target gene of ERα, PR mRNA expression was induced in the presence of E₂ (7.7-fold) in the siRNA control cells (Figure 4B). In cells treated with siB1, there was a significant decrease of PR mRNA levels as compared with their respective control, both without and with E₂ (rates 0.5 and 0.6, respectively). Consequently, E₂ induction was maintained for cells transfected by both siRNA control and siB1 (respectively 7.7-fold and ninefold).

Using immunocytochemistry, we confirmed that RhoB regulates ERα (Figure 3C) plus PR (Figure 4C) expression, with no dramatic changes of their subcellular localization within MCF-7 cells. PR is mainly detected in the nuclei of the untreated control. Besides, the presence of siB2 again significantly decreases IRS scores both in the absence and presence of E₂.

Cells treated with siB2 exhibit low levels of ERα protein and mRNA in both the absence and presence of E₂ (Figure 3B, C, D). Nonetheless, the effects of RhoB on PR expression may be supported in part by other mechanisms. We used chromatin immunoprecipitation analysis to study the effect of RhoB inhibition on the recruitment of ERα, ER transcriptional co-factors and RNA polymerase II onto PR gene promoter 1 hour after E₂ stimulation (Figure 4D). Following siB1 transfection, the recruitment of RNA polymerase II to the PR gene promoter dramatically decreased (rate 0.3 compared with siControl). Recruitment of acetylated H3 and H4 histones were also clearly decreased by RhoB downregulation (rates 0.36 and 0.33, respectively). Moreover, RhoB downregulation induced a significant decrease in the recruitment of ER to the PR gene promoter (rate 0.52) paralleled by an increased in recruitment of the ERα co-repressor HDAC1 (rate 1.67). Together, these results indicate that RhoB may support to some extent ERα transcriptional activation function by interfering with its cofactor recruitment, besides the direct effect on ERα expression itself.

The effect of RhoB on cell proliferation was evaluated in three cell lines exhibiting variable levels of expression of ERα, including MCF-7 (ERα/PR-positive), SK-BR-3 (ERα/PR-negative and p185^erbB2 overexpressed) and MDA-MB-231 (ERα/PR-negative and p185^erbB2-negative). As shown in Figure 5A, B, RhoB positively regulated the proliferation of MCF-7 cells both in the absence or presence of E₂. siRNA-mediated inhibition of RhoB expression produced a 30 to 35% decrease in MCF-7 cell proliferation as soon as 1 day after transfection, with a 40 to 46% decrease by day 4 (Figure 5A). Conversely, a significant increase in cell proliferation was observed in MCF-7 cells transduced with an adenoviral vector expressing RhoB (Figure 5B), with an increase of 15 to 28% in relative cell proliferation at day 1 that reached 22 to 49% by day 4. In contrast to these observations, under similar conditions for infection of SK-BR-3 or MDA-MB-231 cells, the adenoviral RhoB vector either slightly decreased or had no significant biological effect on cell proliferation (Figure 5C, D). The effect of RhoB downregulation was also analyzed in LCC2 cells, an E₂-independent, tamoxifen-resistant subline of the MCF-7 cells. As for the MCF-7 cells, a significant decrease of proliferation was observed at day 4, in parallel to ERα and RhoB downregulation (see Figures S2 and S3 in Additional files 4 and 5).

In conclusion, we documented that RhoB had stimulatory effects on proliferation via ERα signaling that paralleled its effects on hormone receptor expression.