|Home | About | Journals | Submit | Contact Us | Français|
Oncoprotein C-MYC is overexpressed in human metastatic melanomas and melanoma-derived cells where it is required for suppression of oncogene-induced senescence (OIS). The genetic events that maintain high levels of C-MYC in melanoma cells and their role in OIS are unknown. Here, we report that C-MYC in cells from several randomly chosen melanoma lines was up-regulated at the protein level, and largely due to the increased protein stability. Of all known regulators of C-MYC stability, levels of B56α subunit of the PP2A tumor suppressor complex were substantially suppressed in all human melanoma cells compared to normal melanocytes. Accordingly, immuno-histochemical analysis revealed that the lowest and the highest amounts of PP2A-B56α were predominantly detected in metastatic melanoma tissues and in primary melanomas from patients with good clinical outcome, respectively. Importantly, PP2A-B56α overexpression suppressed C-MYC in melanoma cells and induced OIS, whereas depletion of PP2A-B56α in normal human melanocytes up-regulated C-MYC protein levels and suppressed BRAFV600E- and, less efficiently, NRASQ61R-induced senescence. Our data reveal a mechanism of C-MYC overexpression in melanoma cells and identify a functional role for PP2A-B56α in OIS of melanocytic cells.
Oncogene-induced senescence (OIS) is a mechanism for the suppression of tumor growth at the pre-neoplastic stage (Bringold and Serrano, 2000; Mooi and Peeper, 2006; Bansal and Nikiforov, 2010). One of the best examples of naturally occurring OIS in humans is the nevus, a benign aggregation of non-proliferative melanocytes that often harbor activating mutations in BRAF, NRAS or HRAS genes (Michaloglou et al., 2005; Gray-Schopfer et al., 2007; Pollock et al., 2003). Intriguingly, activating mutations in the same genes have been detected in several types of malignant melanomas but at lower frequencies (Pollock et al., 2003; Curtin et al., 2005; Maldonado et al., 2003), indicating that malignant melanomas that often originate from nevi must have developed mechanisms that circumvent OIS. Ectopic expression of the activated forms of the above oncogenes (BRAFV600E, NRASQ61R, HRASG12V) in normal human melanocytes (NHM) also induced senescence (Pollock et al., 2003; Denoyelle et al., 2006) which was not prevented by individual depletion of such tumor suppressor genes as p53, p16INK4A (Denoyelle et al., 2006; Zhuang et al., 2008), p21CIP/WAF or p14ARF (Haferkamp et al., 2009). Instead, ectopic expression of the oncogenic transcription factor C-MYC has been shown to significantly suppress senescence phenotypes caused by BRAFV600E and, to much lesser extent, by NRASQ61R in NHM (Zhuang et al., 2008).
C-MYC is a member of the MYC family of oncoproteins (Grandori et al., 2003; Nikiforov et al, 2003). It regulates the expression of more than 3,000 genes involved in a large variety of cellular processes including the cell cycle regulation (Henriksson and Lüscher, 1996), differentiation (Eilers, 1999), metabolism (Dang and Gao, 2009; Mannava et al., 2008), apoptosis (Nilsson and Cleveland, 2003; Meyer et al., 2006) and senescence (Wu et al., 2007; Zhuang et al., 2008; Hydbring et al, 2010). C-MYC is required for the proliferation of all tested human cells (Wang et al., 2008) and is frequently up-regulated in human malignancies including metastatic melanoma (Nesbit et al., 1999). In cells from solid tumors, the elevated expression of C-MYC can be maintained through several mechanisms including gene amplification (Nesbit et al., 1999), enhanced cap-dependent and cap-independent translation of C-MYC mRNA (De Benedetti et al., 2004; Shi et al., 2008) and elevated protein stability (Hann, 2006). C-MYC stability is regulated in part by post-translational modifications on its two highly conserved residues: phosporylation of Ser62 by the RAS-RAF-MEK-ERK kinase cascade stabilizes C-MYC (Lutterbach and Hann, 1994; Sears et al., 1999), whereas phopshorylation of Thr58 by GSK3β kinase promotes C-MYC proteosomal degradation (Gregory and Hann, 2000; Bahram et al., 2000). C-MYC phosphorylated on Thr58 is recognized by FBW7, a receptor component of the ubiquitin-ligase complex SCF (Skp2-Cullin-F-box), and is subsequently ubiquitinated (Kim et al., 2003; Welcker et al., 2004). MYC phosphorylated at Ser62/Thr58 is also recognized by the prolylisomerase Pin1 that induces conformational changes at the peptide bond between Ser62 and Pro63 and facilitates interactions between phosphorylated Ser62 and a serine/threonine phosphatase complex PP2A (Arnold and Sears, 2006). PP2A dephosphorylates phospho-Ser62 of C-MYC thus promoting its degradation. PP2A consists of a structural subunit (A), catalytic subunit (C), and one of multiple regulatory subunits (B) that specify PP2A target binding (Arroyo and Hann, 2005). The B56α regulatory subunit of PP2A has been shown to interact with C-MYC, enhance dephosphorylation of phospho-Ser62 and promote C-MYC degradation (Arnold and Sears, 2006). Depletion of PP2A-B56α cooperated with SV40 T antigen, catalytic subunit of telomerase reverse transcriptase and HRASG12V in transformation of human kidney epithelial cells (Sablina et al., 2010). Yet, unlike other protein controlling C-MYC stability (Rajagopalan et al., 2004; Vaarala et al., 2010), deregulated expression of PP2A-B56α has never been associated with any human malignancies.
Other proteins regulating C-MYC stability include: CIP2A, a C-MYC-interacting protein that specifically inhibits PP2A activity against C-MYC (Junttila et al., 2007); USP28, a de-ubiquitinating enzymes that antagonizes FBW7 and promotes C-MYC stability (Popov et al., 2007) and TRUSS, a receptor for DDB1 (damage-specific DNA-binding protein 1)-CUL4 (Cullin 4) E3 ligase complex (Choi et al., 2010).
Recently, we have demonstrated that shRNA-mediated depletion of C-MYC in metastatic melanoma cells resulted in the re-emergence of senescence phenotypes that were characteristic for NHM ectopically expressing BRAFV600E or NRASQ61R (Zhuang et al., 2009). The mechanisms maintaining high levels of C-MYC in melanoma cells and their role in suppression of OIS remain unknown. To answer this question, we report here that increased protein stability plays a major role in maintaining high amounts of C-MYC in melanoma cells and that PP2A-B56α represents a key regulator of C-MYC protein stability and oncogene-induced senescence in normal and transformed melanocytic cells.
To investigate mechanisms underlying overexpression of C-MYC in melanoma cells, we analyzed C-MYC protein and mRNA levels in cells from seven arbitrarily chosen melanoma lines derived from human melanoma metastases and two independently isolated populations of NHM. C-MYC protein levels were ~ 4 to 11 times higher in melanoma cells compared to NHM (Figure 1a), whereas C-MYC mRNA amounts in melanoma cells were at most 2.5 fold higher than that of NHM (Figure 1b). These data suggest that high amounts of C-MYC in studied melanoma cells could be due to the increased protein stability or elevated translational rates of its mRNA. To test the former possibility, C-MYC protein half-life was assessed in a population of NHM (designated as “b” on Figure 1a) and in cells from 3 arbitrary chosen melanoma cell lines, SK-Mel-19, SK-Mel-29, SK-Mel-147, in pulse-chase 35S-methionine/cysteine labeling assay. As shown in Figure 1c, C-MYC protein half-life was 2.6 to 7.3 fold higher in melanoma cells than in NHM indicating that C-MYC protein is more stable in melanoma cells compared to NHM.
Several polypeptides were implicated in the control of C-MYC protein stability: PP2A-B56α, PP2A-C, FBW7–α,–β,–γ; PIN1, SKP2, CIP2A, TRUSS and USP28. We examined the expression of these polypeptides in NHM and melanoma cells by Western blotting. Surprisingly, expression of only two of these proteins, PP2A-B56α and CIP2A, differed dramatically between NHM and all melanoma cells (Figure 2a). PP2A-B56α, was acutely suppressed in melanoma cells compared to NHM, whereas CIP2A demonstrated directly the opposite expression pattern. Interestingly, the expression of PP2A-C, a catalytic subunit of PP2A complex, did not change among the studied cells.
CIP2A has been recently characterized as a direct target of C-MYC gene (Khanna et al., 2009), whereas ability of C-MYC to regulate PP2A-B56α has never been investigated. To address this question, we depleted C-MYC in cells from two arbitrarily chosen melanoma lines via shRNA (Wang et al., 2008) and monitored expression of CIP2A and PP2A-B56α by Western blotting 6 days post-infection. Depletion of C-MYC substantially affected levels of CIP2A but not PP2A-B56α (Figure 2b) indicating that PP2A-B56α expression does not depend on C-MYC in studied melanoma cells.
To validate the expression pattern of PP2A-B56α observed in cultured cells, we analyzed its protein levels immunohistochemically in tissue microarrays (TMAs) containing 56 primary cutaneous melanomas from the individuals with good clinical outcome (in lieu of TMA containing adequate number of samples from benign melanocytic nevi) and 167 metastatic melanoma specimens (Table 1). Our analysis revealed a significant correlation between intensity of PP2A-B56α staining and tumor stage (Somers’ D statistic = −0.2046, p-value = 0.0036). Specifically, cores with the highest expression of PP2A-B56α (intensity score 3; more than 50% of cells stained) were 12.22 times more frequent in TMAs derived from primary tumors compared to metastases (p=0.0237), whereas cores with the lowest expression of PP2A-B56α (intensity score 0) were 3.71 times more likely to be derived from metastatic samples than primary tumors (p=0.0242). Representative cores from each group are shown in Figure 2c. These results confirm our observations made in cultured cells that PP2A-B56α is down-regulated during melanoma progression.
To determine whether PP2A-B56α affects C-MYC in melanoma cells, the cells from 4 melanoma lines were infected with a lentiviral vector encoding PP2A-B56α or empty vector. Ectopic expression of PP2A-B56α in SK-Mel-19 and SK-Mel-29 cells caused substantial down-regulation of C-MYC protein amounts and proliferative arrest between days 5 and 8 post-infection (Figure 3ac). Proliferation arrest was accompanied by increased activity of SA-β-Gal (Figure 3ad). SK-Mel-103 and -147 cells expressing ectopic PP2A-B56α underwent proliferative arrest at later time point, between days 14 and 20 post-infection (Figure 3c). Like in SK-Mel-19 and -29 cells, the arrest was accompanied by decrease in C-MYC amounts (Figure 3a) and increased activity of SA-β-Gal compared to control cells (Figure 3d). It is noteworthy that in cells from all lines the senescence-associated phenotypes induced by PP2A-B56α overexpression were very similar to those caused by direct depletion of C-MYC via shRNAs (Zhuang et al., 2008). Therefore ectopic expression of PP2A-B56α induces senescence in melanoma cells, albeit with different incubation period. Importantly, overexpression of PP2A-B56α in cells from 3 out of 4 tested lines (SK-Mel-19, -29, and -147) either did not affect C-MYC mRNA levels or depleted them roughly by ~50% (Figure 3b), demonstrating that PP2A-B56α controls C-MYC in melanoma cells predominantly at the protein level.
To identify whether C-MYC protein levels depend on endogenous PP2A-B56α, we infected NHM with lentiviral vectors containing two different shRNAs against PP2A-B56α or control shRNA. Cells were collected 5 days post-infection to assess changes in C-MYC and PP2A-B56α mRNA and protein levels. Each shRNA against PP2A-B56α efficiently depleted its mRNA and protein (Figure 3ef). Importantly, PP2A-B56α depletion resulted in 3.1-to-3.4-fold increase in the amounts of C-MYC protein, but not mRNA (Figure 3ef).
As B56α controls PP2A-dependent de-phosphorylation of C-MYC phospho-Ser62, we investigated the effect of shRNA-mediated depletion of PP2A-B56α on this modification of C-MYC. Using Western blotting with antibodies specific to total or phospho-Ser62 C-MYC, we detected an ~50% increase in C-MYC protein phosphorylated on Ser62 in PP2A-B56α-depleted cells (Figure 3e), thus providing a mechanistic explanation to the increased C-MYC protein amounts.
To identify whether PP2A-B56α depletion overcomes OIS in NHM, first, these cells were infected with lentiviral vector expressing BRAFV600E or with empty lentiviral vector followed by super-infection with control or PP2A-B56α shRNAs. The functional activity of BRAFV600E in infected cells was confirmed by monitoring phosphorylation of ERK, a downstream target of activated RAS-RAF pathway (Figure 4a). As has been shown previously (Zhuang et al., 2008), a significant decline in C-MYC amounts was observed in BRAFV600E-expressing melanocytes between days 10 and 15 post-infection (Figure 4b), concomitant with the strong manifestation of the senescence-associated phenotypes including reduction in proliferative rates (assessed by the EdU incorporation) and the increase in the percentage of SA-β-Gal-positive cells and cells containing senescence-associated heterochromatin foci (SAHF) (Figure 4cd). Depletion of PP2A-B56α in BRAFV600E-expressing melanocytes prevented downregulation of C-MYC and the emergence of senescence-associated phenotypes (Figure 4b–d). These data demonstrate that suppression of PP2A-B56α has a functional role in overcoming BRAFV600E-induced senescence most likely via sustaining endogenous C-MYC levels. Recently, we have demonstrated that C-MYC overexpression in NHM overcomes BRAFV600E-induced more effectively than NRASQ61R-induced senescence To address the role of PP2A-B56α in OIS induced by activated members of RAS protein family that commonly mutate in melanocytic cells, we suppressed PP2A-B56α in NHM followed by transduction with NRASQ61R or HRASG12V-expressing lentiviruses. The activities of infected RAS proteins were confirmed by monitoring phosphorylation of ERK (Figure 4a). Depletion of PP2A-B56α reversed downregulation of C-MYC caused by NRASQ61R or HRASG12V almost as efficiently as in case with BRAFV600E (Figure 4b). Interestingly, upregulation of C-MYC partially reversed senescence-associated phenotypes caused by NRASQ61R but did not significantly suppress these phenotypes induced by HRASG12V (Figure 4cd).
Thus, in NHM, PP2A-B56α depletion differentially affects OIS induced by members of RAS family of proteins most commonly activated in melanocytic cells.
In the current manuscript, we present data arguing that protein stability is a major factor contributing to the elevated levels of C-MYC in melanoma cells. Furthermore, we demonstrate that among all known regulators of C-MYC stability, the expression of two proteins that regulate PP2A activity against C-MYC (CIP2A and PP2A-B56α) were markedly different between NHM and melanoma cells. Therefore, it is quite remarkable that despite high genetic variability of melanoma cells (Smalley et al., 2005), C-MYC was similarly up-regulated in cells from all studied melanoma lines and that the expression of two proteins regulating a single post-translational modification of C-MYC (phosphorylation of Ser62) was also respectively uniformed among these lines. In principal, this uniformity could be due to another feature common for all studied melanoma cells: the activated state of MEK-ERK pathway. Because Ser62 is a target of ERK (Sears et al, 1999), phosphorylation of Ser62 can be a rheostat for the opposing activities of ERK and PP2A.
Activation of BRAF, NRAS or HRAS plays a controversial role in melanomagenesis. On the one hand, constitutive activity of these oncoproteins is essential for sustained proliferation of melanoma cells which requires high levels of C-MYC (Zhuang et al., 2008; Welcker et al., 2004). On the other hand, these oncogenes cause OIS in NHM accompanied by down-regulation of C-MYC (Zhuang et al., 2008). Intriguingly, BRAFV600E stimulates proliferation shortly after delivery into NHM and suppresses it at later time points (Denoyelle et al., 2006; Zhuang et al., 2008). Our previous and current data support the model that suppression of C-MYC is responsible for the transition from proliferation-promoting to proliferation-inhibitory activities of BRAFV600E and possibly activated RAS proteins.
We report that overexpression of PP2A-B56α suppressed C-MYC protein rather than mRNA levels in cells from two out of four examined melanoma lines. In the other two lines (SK-Mel-19 and SK-Mel-103), PP2A-B56α overexpression led to 50% and 77% reduction in C-MYC mRNA, respectively. This phenomenon may be due to the reported ability of PP2A-B56α to facilitate degradation of β-catenin (Arnold et al., 2009; Hart et al., 1998), a mediator of Wnt-pathway-dependent up-regulation of C-MYC transcription (He et al., 1998). Because the Wnt pathway is often activated in melanoma cells (Larue and Delmas, 2006) including SK-Mel-103 (D.Z. and M.A.N. unpublished observations), the presumed degradation of β-catenin via PP2A-B56α in these cells may affect C-MYC mRNA levels.
PP2A-B56α-dependent depletion of C-MYC in studied melanoma cells resulted in the emergence of senescence phenotypes similar to those caused by depletion of C-MYC via shRNA. However, the timing of senescence varied substantially: 5–6 days post-infection with PP2A-B56α in SK-Mel-19, -29 cells versus 14–20 days post-infection in SK-Mel-103, -147 cells. In the latter group, the amounts of C-MYC did not change until prior to the emergence of senescence phenotypes (data not shown), indicating that ectopic PP2A-B56α in these cells did not suppress C-MYC amounts during a long period of time. One possible explanation for this observation is that unlike BRAFV600E-expressing SK-Mel-19 and SK-Mel-29 cells that contain activated ERK (Figure 1a), SK-Mel-103 and SK-Mel-147 cells bear NRASQ61R and, therefore possess both activated ERK and suppressed GSK3β. This leads to stabilization of C-MYC by increased phosporylation of its Ser62 and hypo-phosphorylation of its Thr58 Therefore, it is conceivable that destabilization of C-MYC in SK-Mel-103 and SK-Mel-147 cells via PP2A-mediated de-phosphorylation of Ser62 is hampered or delayed compared to that in SK-Mel-19 and -29 cells.
Depletion of PP2A-B56α led to the up-regulation of endogenous C-MYC in NHM expressing BRAFV600E, NRASQ61R or HRASG12V. Yet, C-MYC increase was sufficient to effectively suppress senescence phenotypes caused only by BRAFV600E. OIS induced by NRASQ61R was inhibited only partially, whereas HRASG12V-dependent senescence phenotypes were unaffected by upregulation of C-MYC. These data are concordant with our previous observations that ectopic expression of C-MYC much more efficiently overcomes OIS caused by BRAFV600E than by NRASQ61R (Zhuang et al., 2008). Unlike BRAFV600E, HRASG12V and, to much lesser extent, NRASQ61R induce the unfolded protein response (UPR) in NHM (Denoyelle et al., 2006). In our experience, C-MYC overexpression in NHM does not suppress the UPR pathway (Zhuang et al., 2008), which could account for the inability of PP2A-B56α depletion and subsequent C-MYC up-regulation to overcome HRASG12V-induced senescence in these cells.
In summary, we demonstrated that increased protein stability contributes substantially to the elevated levels of C-MYC in melanoma cells. We showed that PP2A-B56α can account for differential C-MYC expression between NHM and melanoma cells. Lastly, we demonstrated the functional importance of PP2A-B56α downregulation for both the maintenance of high levels of C-MYC and the suppression of oncogene-induced senescence in NHM and melanoma cells.
Melanoma cell lines were originally obtained from Memorial Sloan Kettering Cancer Center. Cells were cultured in Dulbecco’s modified Eagle’s essential minimal medium as recommended by the supplier. Supplements included fetal calf serum (10%), 2 mM glutamine, and 100 units/ml penicillin G + 100 µg/ml streptomycin. All cell culture agents were purchased from Invitrogen, Inc. (Carlsbad, CA). Normal melanocytes were isolated from neonatal foreskins as previously described (8) and maintained in Medium 254 supplemented with 0.2 mM CaCl2, 16 nM TPA and melanocyte growth factors (Cascade Biologics).
For the proliferation assay, melanocytes were plated in 96 well plates at ~50% confluence 2 days prior to the assay. Cells were incubated with a nucleoside analog of thymidine, 5-ethynyl-2´-deoxyuridine (EdU), for 60 minutes followed by fixation and staining for EdU-incorporated cells using the ClickiT™ EdU Assay kit (Invitrogen). The proportion of EdU-positive cells was determined by counting cells under the fluorescent microscope in multiple view fields.
For the senescence assay, cells were plated in 12 well plates at ~30% confluence 2 days prior to the assay. Cells were fixed and incubated for different time periods (4 hours to overnight, depending on the cell line) at 37°C with the staining solution containing the X-Gal substrate (BioVision). The development of blue color was detected visually with a microscope. The proportion of positive cells was determined by counting cells under the light microscope in multiple view fields. To visualize senescence-associated heterochromatin foci (SAHF), cells were fixed with 0.5% formaldehyde and stained with DAPI as described previously (Denoyelle et al., 2006). The proportion of SAHF-positive cells was determined by counting cells with fluorescent microscopy in multiple view fields.
C-MYC protein stability was determined using pulse-chase labeling with 35S-methionine/35S-cysteine. NHM and melanoma cells were pulse-labeled with the mixture of 35S-methionine/35S-cysteine (300µCi/ml, Perkin-Elmer) in the cysteine- and methionine-free media for 3 hours and 1 hour, respectively. After that, cells were washed and incubated in media containing non-radioactive cysteine and methionine for 0, 20, 40 or 60 minutes. After completion of incubation, cells were harvested, lysed in immunoprecipitation buffer and total cellular extracts were immunoprecipitated with mouse anti-C-MYC monoclonal antibodies (C33, SantaCruz). Immunoprecipitated materials were separated on 10% SDS-PAGE, and labeled C-MYC was quantified using STORM phospho-imager.
Lentiviral infection protocols and vectors containing control shRNA, C-MYC shRNA, cDNAs of BRAFV600E, NRASQ61R and HRASG12V were described previously (Denoyelle et al., 2006; Arnold and Sears, 2006). The pLKO-1 lentiviral vectors containing shRNAs specific for the human gene encoding PP2A-B56α were purchased from Sigma.
Total RNA was isolated from cells using the RNeasy Mini Kit (Qiagen). cDNA was prepared using the cDNA reverse transcription kit (Invitrogen). Q-RT-PCR was performed on 7900HT Fast Real-Time PCR System (Applied Biosystems) using TaqMan® Universal Master Mix II (Applied Biosystems) and probes and PCR primers specific for human β-actin, human C-MYC or human PP2A-B56α genes. PCR data were analyzed using sequence detection software (SDS) 2.4 (Applied Biosystems).
Antibodies were used against the following human proteins: p53, C-MYC, SKP2, CIP2A, ERK1/2 and ERK1/2-γ from Santa Cruz Biotechnology (sc-99, sc-42, sc-7164, sc-80659, sc-93, sc-7383, respectively); C-MYC-Ser-62, PP2A-B56α and Fbw7 from Abcam (ab78318, ab1084 and ab12292, respectively); p16INK4A PIN1, USP28, from Cell Signaling Technology (4824, 3722, 4217s); from SPP2A-C from BD Bioscience (bd-610555); and TRUSS from Proteintech (12606-1-AP). Membranes were developed using alkaline phosphatase-conjugated secondary antibodies and the Alpha-Innotech FluorChem HD2® imaging system. Quantification of the signal was performed using the ImageQuant 2.0 program package. The background was calculated from an equivalent area in each lane and subtracted from the value for the protein signal in that lane.
Three 1-millimeter tissue cores of formalin-fixed and paraffin-embedded primary cutaneous tissue from patients with good prognosis (56 cores) or metastatic melanoma (167 cores) were precisely arrayed into paraffin blocks. Immunohistochemical (IHC) staining was performed using the PP2A-B56α mouse monoclonal antibody (clone 23/B56α; BD Biosciences; dilution 1:50). Briefly, this procedure consisted of deparaffinization and rehydration followed by epitope retrieval in which the tissue microarrays were immersed in Target Retrieval Solution (Dako) and incubated with steam for 40 minutes. Incubation with PP2A-B56α antibodies was performed for 45 minutes and was followed by the serial application of biotinylated goat anti-mouse IgG (Jackson ImmunoResearch), alkaline phosphatase-conjugated streptavidin (Invitrogen), and the Fast-Red Substrate System (Dako). Results were recorded as semi-quantitative/ordered categorical, according to the American Society of Clinical Oncology Guideline Recommendations. The neoplastic cells for any given core were scored for i) intensity of staining using the following scale: no staining (0), weak staining (1), moderate staining (2), strong staining (3). The Somers’ D statistic (Newson, 2006) was used to assess the correlation between the intensity of immunohistochemical staining and tumor stage (primary or metastatic).
We are grateful to Dr. Catherine Burkhart (Cleveland Biolabs) and Dr. Shoshanna Zucker (Roswell Park Cancer Institute) for critical reading of the manuscript, Ms. Carina Fung (University of Sydney) for the experimental help. This work was supported by NIH R01 CA120244 and ACS RSG-10-121-01 grants to M.A.N. The work of J.C.M. was supported in part by the NIH 5P30CA016056-34 grant.
Conflict of interest
The authors declare no conflict of interest.