• Open Access

Nuclear receptor coactivator RAC3 inhibits autophagy


  • Pablo Nicolas Fernandez Larrosa,

    1. Laboratory of Molecular Biology and Apoptosis, (IDIM-CONICET), University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Cecilia Viviana Alvarado,

    1. Laboratory of Molecular Biology and Apoptosis, (IDIM-CONICET), University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Maria Fernanda Rubio,

    1. Laboratory of Molecular Biology and Apoptosis, (IDIM-CONICET), University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
    • Member of the Argentine National Research Council (CONICET).
  • Marina Ruiz Grecco,

    1. Laboratory of Molecular Biology and Apoptosis, (IDIM-CONICET), University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
  • Sabrina Micenmacher,

    1. Laboratory of Molecular Biology and Apoptosis, (IDIM-CONICET), University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
  • Giselle Astrid Martinez-Noel,

    1. Laboratory of Molecular Biology and Apoptosis, (IDIM-CONICET), University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
    • Member of the Argentine National Research Council (CONICET).
  • Laura Panelo,

    1. Laboratory of Molecular Biology and Apoptosis, (IDIM-CONICET), University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
  • Monica Alejandra Costas

    Corresponding author
    • Laboratory of Molecular Biology and Apoptosis, (IDIM-CONICET), University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
    • These authors contributed equally to this work.
    • Member of the Argentine National Research Council (CONICET).

To whom correspondence should be addressed.

E-mail: costas.monica@lanari.fmed.uba.ar


RAC3 is an oncogene naturally overexpressed in several tumors. Besides its role as coactivator, it can exert several protumoral cytoplasmic actions. Autophagy was found to act either as a tumor suppressor during the early stages of tumor development, or as a protector of the tumor cell in later stages under hypoxic conditions. We found that RAC3 overexpression inhibits autophagy when induced by starvation or rapamycin and involves RAC3 nuclear translocation-dependent and -independent mechanisms. Moreover, hypoxia inhibits the RAC3 gene expression leading to the autophagy process, allowing tumor cells to survive until angiogenesis occurs. The interplay between RAC3, hypoxia, and autophagy could be an important mechanism for tumor progression and a good target for a future anticancer therapy.

Autophagy is a highly regulated cellular process that serves to remove damaged proteins and organelles from the cell. It contributes to an array of normal and pathological processes, and has recently emerged as a key regulator of multiple aspects of cancer biology.

Autophagy is not only triggered under nutrient deficiency, resulting in mTOR pathway inhibition, but also under cellular stress, or mitochondria or endoplasmic reticulum damage.[1] Several transcription factors can also induce autophagy, such as HIF-α and FoxoA1, which are involved in hypoxia and aging responses, respectively.[1] Autophagy and nuclear factor-kappa B (NF-κB) share common signals and regulators; both are able to control each other through positive or negative feedback loops, thus ensuring homeostatic responses.[2] Moreover, kinases were also found to have positive or negative modulation on autophagy. Class I phosphatidylinositol 3-kinase (PI3K) and protein kinase B (Akt) (both involved in the mTOR pathway), p70S6K, and p38[3] were associated with inhibitory effects, whereas ERK, death-associated protein kinase, and c-Jun N-terminal kinase were involved in positive modulation of autophagy.[4]

The molecular mechanism by which autophagy takes place is complex and implicates several steps.[5] Interference at initiation could lead to the complete inhibition of autophagy, and blockage on maturation could result in an abnormal accumulation of immature autophagosomes.[6] In normal cells, these two scenarios could be pathological. However, the role of autophagy in cancer is not at all clear and is likely dependent on tumor type, stage, and genetic context. This complexity is illustrated by the fact that autophagy has a dual effect in tumorigenesis, either inhibiting it during the early stages or promoting it at later stages or during chemotherapy[5] in order to allow tumor maintenance.

The gene RAC3 (AIB1, SRC-3, NCoA3) is a member of the nuclear receptor coactivator p160 family, together with SRC-1 and TIF-2.[7, 8] Although it was first described as a coactivator of estrogen receptor and other steroid receptors,[8] we have previously found that RAC3 is also a NF-κB coactivator.[9] In any case, it has an intrinsic histone acetyltransferase activity, allowing it to increase the transcriptional activity of nuclear receptors and transcription factors.

Although first observed overexpressed in several hormone-dependent tumors,[10] RAC3 is also overexpressed in hormone-independent ones.[11-13] It is considered to be an oncogene that promotes cell proliferation through different mechanisms that involve its association to nuclear receptors or transcription factors. Moreover, we have previously found that its overexpression has an anti-apoptotic role,[14, 15] not only through its nuclear action, but also by positively regulating the activity of p38 and Akt kinases, inhibiting caspase-8 and -9 and blocking apoptosis-inducing factor-1 (AIF-1) translocation from mitochondria to the nucleus.[14] Therefore, RAC3 seems to have both nuclear and cytoplasmic actions.[16] In addition, a splicing variant of RAC3 (RAC3-Δ4) was described by several groups[17, 18] as having a relevant role in metastasis at the cytoplasmic level.[19]

Because both the expression levels of RAC3 and the autophagy rate are altered in cancer cells, in this work, we investigated whether RAC3 overexpression has a role in autophagy and the possible molecular mechanisms by which these alterations contribute to the survival of tumor cells.

Materials and Methods

Expression vectors and reporter plasmids

The κB-Luc reporter plasmid and pCMX-RAC3 and ssIκB expression vectors were described previously.[9, 14, 20]. The RAC3ΔNLS expression vector (corresponding to SRC3-K17A/R18A) carrying nuclear localization signal (NLS)-mutated RAC3 was kindly provided by Bert O'Malley, Baylor College of Medicine.[21] The RFP-LC3 expression vector was kindly provided by Dr. Vaccaro (Biochemistry Department, School of Medicine, University of Buenos Aires, Buenos Aires, Argentina). The pSilencer-RAC3 siRNA plasmid and scrambled control were previously developed in our laboratory.[15]

A fragment of 2312 bp of the 5′ regulatory sequence of the RAC3 gene, −1970 upstream and +335 downstream from the first human exon,[22] was amplified by PCR from the RP11-456N23 human BAC clone (GenBank accession no. AL35 3777; Invitrogen, Carlsbad, CA, USA) using the following primers: forward, CGGGGTACCTTTTAGTAGAGACGGGG TTTCGC; and reverse, TTTCTCGAGCCAGCTTCGTCTCA GCTCCTAC. The band of the PCR product was digested with restriction enzymes KpnI and XhoI and cloned into pGL3-basic vector (Promega, Madison, WI, USA) in order to obtain a reporter vector for RAC3 promoter activity (pRAC3). Empty pGL3-basic vector was used as a control of basal luciferase activity.

Cell lines, transfection, and tumor samples

Non-tumor HEK293T cell line, derived from human embryonic kidney and with a low RAC3 expression, was transfected by the CaCl2 method with 90% efficiency. Tumoral T84, HCT116, Lovo, and T47D cell lines, naturally overexpressing RAC3, were also used as tumor models. T84 and HCT116 cells were transfected using Lipofectamine 2000 (Invitrogen).

The HEK293T cells were cultured in DMEM high glucose and T84, HCT116, Lovo, and T47D cells were cultured in DMEM F12, both supplemented with 1% glutamine, 1% streptomycin–penicillin and 10% FBS (complete medium [CM]).

Sections of paraffin-embedded tumors, obtained from colon adenocarcinomas, were kindly provided by the Pathological Anatomy Laboratory, Alfredo Lanari Institute (University of Buenos Aires, Buenos Aires, Argentina). The use of patients’ samples was approved by the Ethics Committee of the Institute.

Autophagy induction and assay

Autophagy was induced by starvation. Cells were incubated in EBSS medium (Sigma-Aldrich, St. Louis, MO, USA) without FBS or glutamine for 2 h (for Western blot [WB] assay) or 6 h (for microscopy assay), or with 0.5 μM rapamycin for 6 h. In parallel, cells were incubated with CM.

Autophagy induction was monitored by monodancyl cadaverine (MDC) incubation and the percentage of cells showing an aggregated stain was determined by counting a minimum of 200 cells per slide using fluorescence microscopy. In some experiments, cells were transfected with RFP-LC3, which is a specific autophagosome label, and cells with aggregated LC3 were counted from a minimum of 200 cells per slide using fluorescence microscopy. Finally, in order to determine LC3 conversion, cells were pre-incubated with 10 μg/mL E64D and pepstatin A lysosomal protease inhibitors, before incubation with CM, EBSS, or rapamycin, then analyzed by WB. LC3 conversion was estimated as the LC3-II/LC3-I ratio and relative to tubulin.

Starved cells were stained with Hoechst in order to evaluate possible apoptotic corpse formation.

Western blot analysis and immunoprecipitation

Western blot was carried out as previously described.[14] Briefly, for LC3 conversion analysis, proteins were separated in a 15% SDS-PAGE; 12% for kinase analysis and 8% for RAC3. Protein samples were blotted to a PVDF membrane. After incubation with the corresponding primary and secondary antibodies, blots were developed by enhanced chemiluminescence (New England Nuclear, Boston, MA, USA).

Primary antibodies for the following proteins were used at the recommended dilutions: RAC3, p62, and LC3 (Cell Signaling Technology, Danvers, MA, USA); p38, pp38, Akt, pAkt, and tubulin (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Anti-mouse and anti-rabbit (Vector Laboratories, Burlingame, CA, USA) IgGs, conjugated with HRP, were used.

Kinase activity

Cells were incubated with 20 μM SB202190 (p38 inhibitor), 50 μM LY294002, and the autophagy inhibitors 0.5 mM 3-MA or 50 nM wortmannin (class I and class III PI3K inhibitors, upstream Akt activation). Inhibition of Akt and p38 was confirmed by WB at 2 h after starving. As a positive control, cells were incubated with 2 mM hydrogen peroxide as described.[14]

Hypoxia induction

Hypoxia was induced in a 2% O2 incubator (Oxoid Ltd., Basingstoke, UK) for 3, 6, 12, or 24 h. Cell cultures were incubated in CO2 incubators under normal conditions of O2, as a basal condition. After incubation, cells were collected and protein and mRNA extracts were obtained.

Primers and quantitative RT-PCR

RNA was prepared from samples using TRIzol (Invitrogen) according to the manufacturer's protocol. Retro transcriptase and quantitative RT-PCR (qRT-PCR) was standardized in our laboratory to quantified RAC3 transcript and GADPH was used as the housekeeping gene. Primers were designed by Pearlprimer software and qRT-PCR data were analyzed by the iCycler IQ software (BioRad, Hercules, CA, USA).

The primers sequences were: forward, AAGTGAAGAGGGATCTGGAG and reverse, CAG ATGACTACCATTTGAGGA for RAC3; and forward, TCTCCTCTGACTTCAACAGC and reverse, GTTGTCATACCAGGAAATGAG, for GADPH.

Microscopy assays

The MDC or Hoescht stained cells and RFP-LC3 transfected cells were observed using fluorescence microscopy. Immunohistochemistry was carried out from tumor samples using anti-RAC3 (sc-25742; Santa Cruz Biotechnology) and anti-LC3 (D11, Cell Signaling Technology) primary antibodies.

Statistical analysis

At least three independent experiments were carried out in all cases. Results were expressed as the mean ± SD. The significance of differences between experimental conditions was determined using anova and the Tukey–Kramer Multiple Comparisons Test for unpaired observations.


RAC3 overexpression inhibits autophagy

In order to determine whether RAC3 overexpression may have a role in the autophagy rate, HEK293T cells were transfected with an expression vector for RAC3 or the empty vector (EV), and after 24 h, cells were cultured with EBSS medium (starvation) or CM for 6 h, or 0.5 μM rapamycin (mTOR pathway inhibitor) in CM. In order to determine autophagy levels, cells were stained with MDC, or cotransfected with an expression vector for RFP-LC3, a specific autophagosome marker.

The EV-transfected cells showed a 10% basal autophagy that increased to 70% or 55% with starvation or rapamycin treatment, respectively. However, even though RAC3 transfected cells showed higher levels of basal autophagy (35%), these levels decreased significantly to 10% when starved or treated with rapamycin (Fig. 1A,B). These observations were confirmed when cells were cotransfected with two different RAC3 vector concentrations and the specific autophagosome RFP-LC3 vector (Fig. 1C,D). Apoptotic corpses or apoptosis were not observed under any conditions in these experiments. The increase of RAC3 levels by transfection was confirmed by qRT-PCR or by Western blot analysis (Fig. S1A,B).

Figure 1.

RAC3 inhibits autophagy induced by starvation or rapamycin (RAPA) treatment. HEK293T cells were transfected with an empty (EV) or RAC3-expressing vector and incubated with complete medium (CM), or starved using EBSS (EM) or 0.5 μM RAPA. Autophagy was assessed by monodancyl cadaverine (MDC) staining (A,B) or RFP-LC3 (C,D). Autophagy in tumor cells, incubated with CM or EBSS, was assessed by MDC staining (E). (F) HCT116 cells, transfected with siRNA or a scrambled control (SC), were used to determine autophagy by RFP-LC3, p62, and RAC3 expression by Western blot analysis. Significant comparisons are shown at each panel.

In order to evaluate autophagy levels in tumor cell lines with high levels of RAC3 expression (Fig. S1C), Lovo, HCT116, and T84 cells were incubated with CM or EBSS for 6 h then stained with MDC. As shown in Figure 1(E), no significant differences were observed between basal and starvation-induced autophagy compared with the increase of autophagic cells in HEK293T. HCT116 cells were then cotransfected with the siRNA and RFP-LC3, and starvation or rapamycin-induced autophagy levels were assessed by fluorescence microscopy. As expected, HCT116 cells transfected with the vector carrying the scrambled control of the siRNA showed similar autophagy levels when cultured with CM, EBSS, or rapamycin (Fig. 1F). Similar results were also observed in T84 tumor cells (data not shown). However, when HCT116 cells transfected with siRNA were starved or incubated with rapamycin, autophagy levels showed to be significantly higher than when cultured with CM (Fig. 1F). These results demonstrate that RAC3 overexpression has an inhibitory role in autophagy.

Effect of RAC3 on autophagy involves nuclear and cytoplasmic signals

As a coactivator of steroid receptors and NF-κB, RAC3 can regulate several cellular responses in a transcriptional fashion; however, as it was previously shown, it could also have cytoplasmic actions, modulating apoptosis, kinase activation[14] and metastasis.[19] Moreover, NF-κB shares common downstream effectors with autophagy, and it can induce the expression of anti-autophagic proteins (e.g. Bcl-2) or pro-autophagic ones (e.g. Beclin-1), depending on the physiological and genetic context.[2] Considering this evidence, we decided to investigate whether RAC3 inhibits autophagy through cytoplasmic or nuclear actions, and if it involves NF-κB or kinase activation. Therefore, we determined the autophagy levels in HEK293T cells transfected with EV, or a vector expressing wild-type RAC3 or RAC3 mutated in the NLS (RAC3ΔNLS), or cotransfected with wild-type RAC3 and the IκB super-repressor (ssIκB), able to constitutively inhibit NF-κB activation.

When NF-κB activity was measured using a κB-Luc reporter plasmid, RAC3-transfected cells showed, as expected. increased levels of NF-κB activity with respect to EV, the mutated RAC3 did not show any difference, but the cotransfection with ssIκB completely blocked the activity (Fig. 2A). When autophagy was analyzed, either cells transfected with RAC3ΔNLS or cotransfected (RAC3/ssIκB) showed a partial but significant reversion of the starvation-induced autophagy (RAC3ΔNLS, 30%; RAC3/ssIκB, 35%; RAC3, 10%; < 0.01; Fig. 2B,C). Interestingly, RAC3/ssIκB significantly reverts autophagy under ssIκB concentrations that totally block NF-κB activity (Fig. 2A). These results suggest that, even though RAC3 can inhibit autophagy in a nuclear fashion, it does not exclusively involve NF-κB activation and the coactivator can also block autophagy by a cytoplasmic action.

Figure 2.

RAC3 inhibits autophagy by nuclear and cytoplasmic actions. HEK293T cells were transfected with an empty vector (EV), RAC3, or RAC3ΔNLS, or cotransfected with RAC3 plus ssIκB and incubated with complete medium (CM) or EBSS medium (cell starvation) for 6 h. Basal nuclear factor NF-κB activity in transfected cells was determined by a κB-Luc reporter plasmid. The diagram bars show the average ± SD of relative light units (RLU) (A). Autophagy was determined by monodancyl cadaverine (MDC) staining (B) or LC3 conversion by Western blotting (WB) (C). NF-κB activity in starved cells was determined (D). HEK293T cells, pre-incubated with SB202190 (SB), LY294002 (LY), wortmannin (W), and 3-MA, were cultured in EBSS or CM and autophagy was assessed by RFP-LC3 (E). LC3 conversion was determined by WB (F). The effect of kinase inhibitors at the selected doses was confirmed by WB. Treatment with 2 mM H2O2 was used as a positive control of kinase activation (G). Significant comparisons are shown in each panel. In all cases the diagram bar corresponds to the average of at least three independent experiments ± SD. (C,F) representative of five experiments.

In addition, we also analyzed the autophagy inhibitory role of RAC3 through the study of NF-κB activity under starvation. In starved conditions, we observed that EV-transfected cells showed a significant decrease of NF-κB activity compared to CM, probably as a consequence of autophagy induction. However, no inhibition of this transcription factor was observed by starving in RAC3 or RAC3ΔNLS-transfected cells, showing similar levels to cells cultivated with CM (Fig. 2D).

As expected, high levels of RAC3 increase NF-κB activity with or without starvation. However, the fact that high levels of both the wild-type and its mutant, which is unable to coactivate the transcription factor, prevent NF-κB inhibition by starvation, probably suggests that this is a consequence of the autophagy inhibitory role of RAC3, whereas cells that will not go to autophagy maintain their basal NF-κB activity.

In agreement with these results, although the NF-κB activation could have a protective role in the autophagy induced by starvation, this is not the key signal for the anti-autophagy action of RAC3.

As RAC3 was previously shown to increase the activity of Akt and p38 kinases,[14] and these two kinases have inhibitory effects over autophagy, we decided to investigate whether they are involved in the mechanism by which RAC3 inhibits autophagy.

Therefore, HEK293T cells were transfected with empty vector or vector expressing RAC3, then incubated for 6 h with complete medium or EBSS in the presence of the p38 inhibitor SB202190 as well as the Akt and autophagy inhibitors LY294002, wortmannin, and 3-MA. Both LY294002 and wortmannin incubation induced, as expected, a complete inhibition of autophagy in all transfected cells, in agreement with previously published data.[23] However, although a partial inhibition of autophagy occurred by p38 inhibition in EV-transfected cells (Fig. 2E), when RAC3-transfected cells were incubated with EBSS and SB, a significant increase of autophagy levels compared to the cells without p38 inhibitor was observed (EBSS + SB, 50%; EBSS, 15%; < 0.01). This increase of autophagy by the p38 inhibitor in RAC3-transfected cells was reverted when cells were also incubated with the autophagy inhibitor 3-MA (Fig. 2E).

These results were confirmed by WB analyzing the LC3-II/LC3-I ratio (Fig. 2F) and kinase inhibition was also confirmed by WB (Fig. 2G).

Although p38 activity could be considered a cytoplasmic signal that RAC3 requires for starving-induced autophagy, the effect of SB when the cells were transfected with the expression vector for RAC3ΔNLS were not similar to that obtained in cells overexpressing RAC3 (Fig. 2E), suggesting that this kinase is not a key pathway for the cytoplasmic action of RAC3. Taking all of these results into account, we may conclude that RAC3 requires p38 activity in order to inhibit autophagy, but this is not the key cytoplasmic signal that the coactivator uses for protection.

Hypoxia induces a decrease of RAC3 expression and increase of autophagy levels in tumor cell lines

Hypoxia is well known to positively regulate autophagy in normal cells but also in cancer cells at the internal mass of poorly vascularised tumors.[24] Because we found that RAC3 inhibits autophagy induced by starvation or mTOR inhibition, we analyzed what happens with RAC3 expression and autophagy levels in hypoxic conditions, in order to recreate the internal center of the tumor. For these experiments, tumor cell lines HCT116, T84, Lovo, and T47D, with high levels of endogenous RAC3 expression, were used.

Hypoxia induced an increase of autophagy levels in HCT116 cells transfected with RFP-LC3 (Fig. 3A). This was also observed for different tumor cell lines that express high levels of RAC3, by analyzing the LC3-II/LC3-I ratio (Fig. 3B). Interestingly, similar levels of hypoxia-induced autophagy were obtained in HCT116 cells transfected with a siRNA for RAC3 (Fig. 3A) in a concentration that clearly inhibits the levels of RAC3 mRNA (Fig. 3D). In view of this evidence, we decided to investigate a possible change of RAC3 expression under hypoxic conditions.

Figure 3.

Hypoxia induces the inhibition of RAC3 expression and autophagy increase. Autophagy was determined by RFP-LC3 under normoxia or hypoxia in HCT116 (A) or HEK293 cells (G) and the diagram bars correspond to the average of three independent experiments ± SD. LC3 conversion and p62 expression were determined in several tumor cell lines under normoxia and hypoxia (B) and HEK293 cells transfected with the expression vector for RAC3 or RAC3ΔNLS (H). RAC3 transcript expression was determined by quantitative RT-PCR (using GADPH as the housekeeping gene) in several cell lines (C), HCT116 cells transfected with the siRNA or the scrambled control (D), and HEK293 cells transfected with the expression vector for RAC3 or RAC3ΔNLS (F). The diagram bars correspond to the average of three independent experiments ± SD. RAC3 promoter activity was analyzed by a pRAC3-Luc reporter vector in HEK293T cells under normoxia or hypoxia (E). RAC3 and LC3 expression was analyzed in tumor samples obtained from patients, at the vascular zone (dashed lined) (I). Significant comparisons are shown in each panel.

Interestingly, RAC3 transcript, determined by qRT-PCR, was expressed at 30% under hypoxic conditions compared to 100% under normal levels (normoxia) in all the cell lines that were analyzed (Fig. 3C). The protein levels correlated with the mRNA expression (data not shown). Similar results are shown in Figure 3(D) for HCT116 cells in the presence or absence of siRNA, where similar very low levels of RAC3 mRNA were detected under hypoxia.

In order to determine whether the hypoxic inhibition of RAC3 levels could be a consequence of a minor mRNA stability and degradation or an effect at the transcriptional level, HEK293T cells were used to evaluate RAC3 promoter activity under these conditions, using a reporter vector carrying the luciferase gene under the control of the RAC3 promoter. We found that hypoxia significantly decreases RAC3 promoter activity (Fig. 3E) indicating that hypoxia inhibits RAC3 gene expression. In order to determine if this action is specifically dependent on the promoter, or whether any additional mechanism could be affecting the mRNA stability, we analyzed mRNA levels under normoxia or hypoxia in cells transfected with expression vectors for RAC3 or RAC3ΔNLS, where the expression of both genes is under the control of a different constitutive promoter. Figure 3(F) shows that under normoxic conditions, as expected, the levels of RAC3 mRNA are increased in transfected cells; however, hypoxia induces a significant inhibition in all the cases, suggesting that additional, not promoter-dependent mechanisms are involved in the reduction of RAC3 mRNA levels.

In view of these results, it could be suggested that constitutive RAC3 overexpression could probably not revert hypoxia-induced autophagy. Therefore, we determined autophagy under these experimental conditions. As shown in Figure 3(G,H), RAC3 or RAC3ΔNLS overexpression was unable to inhibit hypoxia-induced autophagy.

These results suggest that hypoxia induce the inhibition of RAC3 expression, allowing autophagy to occur. In order to evaluate whether this happens in tumors, we analyzed RAC3 and LC3 expression by immunohistochemistry in different tumor samples obtained from patients. In all cases, RAC3 was more likely found in zones close to the vasculature, whereas LC3 was observed mainly in zones surrounding necrotic tissues (Fig. 3I). In support of the original hypothesis, in tumor samples, autophagy and RAC3 were shown to be mutually exclusive.


During tumor development, many fundamental changes in basic cellular processes are required for the initiation and maintenance of the transformed state. These include the regulation of key processes of cancer cell physiology, such as autophagy, apoptosis, and cell cycle checkpoint control, as well as interactions with the surrounding microenvironment.[25] Most of these processes are regulated at the transcriptional level by nuclear receptors, transcription factors, and coactivators.

As a nuclear receptor coactivator overexpressed in many tumor types, RAC3 contributes to oncogenesis through several mechanisms. The primary mechanism to exert these actions is acting as a coactivator of steroid receptors[8] and NF-κB transactivation,[9] thus increasing the expression of the target genes. However, RAC3 seems to be more than a simple coactivator, as cytoplasmic actions, such as modulation of apoptosis and kinase activation[14] and metastasis,[19] have been previously reported to play a central role in tumorigenesis. This evidence has led to RAC3 being considered as an oncogene.[16]

In spite of the clear RAC3 relevance in cancer, its role in autophagy has not been previously investigated. At early stages of tumor development, autophagy plays a suppressive role, leading the cancer cell to a senescent phenotype.[26] In fact, partial or total absence of autophagy is associated with genomic instability enhancing the tumor phenotype where p62 complex degradation fails.[27] Characterization of several tumors have indicated autophagy failure by Atg′ mutations.[6] However, a complete deficiency of autophagy could be detrimental for tumor development. In support of this, defective autophagy and tumor necrosis were found in the center of the tumor (later stages) under hypoxic conditions, prior to angiogenesis, when Atg or Beclin-1 genes were knocked out or down in vivo.[24] Therefore, autophagy seems to be reversibly inhibited, having a dual role depending on the stage of tumor development.

Our results indicate that RAC3 overexpression is a signal strong enough to block autophagy induced by nutrient deprivation or mTOR pathway inhibition. However, most of the tumor cell lines expressing naturally high levels of RAC3, as well as the RAC3-transfected HEK293T cells, show higher levels of basal LC3 positive vesicle accumulation, mainly at the cytoplasm periphery or pseudopodia extensions. This is in agreement with previously reported works concerning the role of autophagy as a basal and natural mechanism of tumor cell survival.[5]

Initially, we though that RAC3 could be affecting autophagy through NF-κB transactivation, because this transcription factor is known to regulate the expression of proteins that inhibit autophagy.[2, 4] However, we found that although RAC3 overexpression enhances the basal NF-κB activity, its blockage by the expression of the IκB super repressor only partially reverts the autophagy protective effect of the coactivator, suggesting that additional anti-autophagy signals not related to the transcription factor could be involved.

Moreover, the overexpression of the ΔNLS-mutated RAC3, which is unable to translocate to the nucleus or to act as a coactivator, and of course does not affect NF-κB activity, also has a protective role, indicating that RAC3 could be using another non-nuclear signal that leads to autophagy inhibition.

Tumoral cells (with natural RAC3 overexpression) and non-tumor cells transfected with RAC3 were shown to have similar autophagy levels and Rel-A activity when incubated with or without nutrients, suggesting that these cells have a higher basal NF-κB activity. In fact, starvation did not induce NF-κB activation per se until 6 h, but a decrease was observed in cells expressing normal levels of RAC3, probably because autophagy was taking place.[2] Therefore, RAC3 seems to block autophagy partially by an NF-κB-independent mechanism, and when autophagy is active (because self-limited levels of RAC3 are insufficient to inhibit it), the transcription factor activity is decreased.

In view of our results suggesting some cytoplasmic anti-autophagy signalling of RAC3 and the knowledge that this coactivator can modulate the activity of kinases[14] that have been previously described to inhibit autophagy,[4] we analyzed whether some of these pathways could be involved in the RAC3 effect. Unsurprisingly, class III PI3K inhibition by wortmannin and LY294002 resulted in autophagy inhibition, in spite of the fact that LY294002 also inhibits class I PI3K, which is downstream Akt and mTOR activation. However, p38 inhibition induces an increase of autophagy even in the presence of high levels of RAC3, suggesting that this kinase is involved in the autophagy protective action of RAC3. Although this kinase activation could be a cytoplasmic signal involved in RAC3 protection, it could not be confirmed by the results obtained in cells transfected with RAC3ΔNLS (Fig. 2E). Perhaps the mutated version of RAC3 is unable to associate to some molecules, or it may be processed or phosphorylated differently to wild-type RAC3, and does not require p38 activity. In addition, it can not be excluded that this kinase activity could be required for the nuclear action of RAC3. Although p38 activity could be involved in the cytoplasmic action of RAC3, our results suggest there are additional mechanisms by which RAC3 can inhibit autophagy without a nuclear translocation.

In conclusion, RAC3 can certainly inhibit autophagy at least by nuclear-dependent and -independent (cytoplasmic) mechanisms. The nuclear process involves NF-κB activity and the cytoplasmic process probably involves p38 activity, but additional signals are required and the mechanism remains to be determined.

We also analyzed what happens with RAC3 and autophagy in a model that reproduces the lalter stages of tumor development, when the tumor mass is growing and its center is hypoxic. In such conditions, autophagy increased as expected,[1] and interestingly, RAC3 expression was inhibited at the transcriptional level. Therefore, although a basal autophagy is usually observed in tumor cells, RAC3 downregulation could be a required step in order to allow autophagy to increase in tumors under hypoxia.

Moreover, the immunohistochemistry observations of tumor samples confirmed that RAC3 expression and autophagy detection were mutually excluded. These results are actually relevant as they allow us to explain why autophagy is active in cells at the tumor center where hypoxia induces autophagy and decreases RAC3 expression.

Clearly, autophagy is a key process in tumor development. Research concerning the molecular mechanisms of this process has increased in recent years. However, the fact that it can play a suppressive role in the early stages of tumor development but is necessary in later stages for tumor maintenance, was a question that remained unresolved.

In this work, we propose for the first time a model where a coactivator molecule plays an essential role regulating autophagy (Fig. 4), and both RAC3 and autophagy are regulated by hypoxia. Thus, hypoxia contributes to tumor cell survival, allowing autophagy under oxygen and nutrient deficiency, as well as helping angiogenesis. Then the triangle of RAC3–autophagy–hypoxia seems to be very important at the first stages of tumor development and maintenance. In this sense, the RAC3 and autophagy machinery, as well as this interplay, could be useful cellular targets to be considered for future therapies against cancer.

Figure 4.

RAC3, autophagy, and hypoxia modulation model. RAC3 inhibits autophagy by nuclear and cytoplasmic actions. Under hypoxic conditions, RAC3 expression is decreased and autophagy occurs, allowing tumor maintenance until angiogenesis occurs. NF-κB, nuclear factor-κB.


This work was supported by grants from Agencia Nacional de Promoción Científica y Tecnológica, the Argentine National Research council (CONICET), and the University of Buenos Aires. We thank to Dr Gabriel Rabinovich, Norberto Zwirner, and Monica Kotler for their help and support.

Disclosure Statement

The authors have no conflict of interest.