Expression of oncogenic HRAS in human Rh28 and RMS-YM rhabdomyosarcoma cells leads to oncogene-induced senescence

Rhabdomyosarcoma (RMS) is the most common pediatric soft tissue sarcoma. The two predominant histologic variants of RMS, embryonal and alveolar rhabdomyosarcoma (eRMS and aRMS, respectively), carry very different prognoses. While eRMS is associated with an intermediate prognosis, the 5-year survival rate of aRMS is less than 30%. The RMS subtypes are also different at the molecular level—eRMS frequently has multiple genetic alterations, including mutations in RAS and TP53, whereas aRMS often has chromosomal translocations resulting in PAX3-FOXO1 or PAX7-FOXO1 fusions, but otherwise has a “quiet” genome. Interestingly, mutations in RAS are rarely found in aRMS. In this study, we explored the role of oncogenic RAS in aRMS. We found that while ectopic oncogenic HRAS expression was tolerated in the human RAS-driven eRMS cell line RD, it was detrimental to cell growth and proliferation in the human aRMS cell line Rh28. Growth inhibition was mediated by oncogene-induced senescence and associated with increased RB pathway activity and expression of the cyclin-dependent kinase inhibitors p16 and p21. Unexpectedly, the human eRMS cell line RMS-YM, a RAS wild-type eRMS cell line, also exhibited growth inhibition in response to oncogenic HRAS in a manner similar to aRMS Rh28 cells. This work suggests that oncogenic RAS is expressed in a context-dependent manner in RMS and may provide insight into the differential origins and therapeutic opportunities for RMS subtypes.


Materials and methods
Generation of cell lines and constructs. Human skeletal muscle myoblasts (HSMMs; Lonza) were cultured as described 24 . Human RMS cell lines RD 25 , Rh30 26 and Rh28 27 were gifts from Tim Triche (Children's Hospital of Los Angeles, CA, USA) in 2005; Rh3 28 , Rh36 29 , and SMS-CTR 30 were gifts from Brett Hall (Columbus Children's Hospital, OH, USA) in 2006; RMS-YM RAS wild type cells were a gift from Marielle Yohe (Pediatric Oncology Branch, NCI, Bethesda, MD, USA); all human RMS cell lines were cultured as described 31 . Cell line authentication was performed in 2014 using STR analysis (Promega PowerPlex 18D) conducted by the DNA Analysis Facility at Duke University (Durham, NC, USA); Rh28 and Rh30 were reauthenticated in 2016. Of note, from the STR analysis, we retrospectively discovered that the Rh3 and Rh28 cell lines were derived from the same patient 32 and this was originally described in Ref. 33 . For serum stimulation studies, cells were serumstarved for 24-48 h, followed by induction with 10% FBS for 1 h 36 ; immunoblots were carried out to assess protein expression levels pre-and post serum stimulation.
Immunoblotting. Immunoblotting was performed as previously described, using a range of 30-100 µg of lysate per sample 36 . The following antibodies were used for immunoblotting: anti-pan-RAS (Calbiochem OP40, 1:1000), anti-pRB (Cell Signaling #8180, 1:1000), anti-RB (Cell Signaling #9313, 1:1000), anti-p16 (BD #554079 1:1000), anti-pp53 (Calbiochem PC461 1:5000), anti-p53 (Santa Cruz sc-126 1:200), anti-p21 (Santa Cruz Growth curve and BrdU assay. Cell growth was assayed using Trypan blue staining followed by manual cell counting on a hemocytometer. Cells were cultured in 6 cm dishes and counted at three or four time points between days 1 through 7 in triplicate. 1.5 × 10 5 cells were plated per replicate at each time point. Cell proliferation was measured using BrdU assays as previously described 31 . Senescence and differentiation assays. Cell senescence was assessed by expression of β-galactosidase by light microscopy, as previously described 31 . In brief, cells were plated at equal density in 6-well plates and the following day were assayed for β-gal staining using an X-Gal-based senescence detection kit (Calbiochem) according to the supplier's protocol. Myogenic differentiation was assessed by MF20 staining under light microscopy, as previously described 37 . In brief, cells were plated at equal density to about 60% confluence in 6-well plates and subject to differentiation-inducing conditions (fusion media consisting of DMEM/F12 with 2% horse serum and refreshed every other day for ∼ 5 days). Myotubes were fixed, permeabilized, and stained with primary antibody anti-sarcomere-myosin hybridoma MF20 followed by secondary biotinylated anti-mouse IgG, then HRP-streptavidin with a 3,3′-diaminobenzidine reagent. MF20 was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology, Iowa City, IA 52242.

RAS-GTP pulldown assays.
The RAS-GTP pulldown assay in Fig. 1 was performed based on published methods 38,39 . Bacteria pellets expressing pGEX-RasBD (binding domain) were sonicated and the supernatant was incubated with prewashed glutathione sepharose-4B (GE) slurry for two hours at 4 °C. Beads were then washed with PBST buffer × 2 and resuspended in 1:1 ratio buffer to beads. 25 µL of GST-RasBD were incubated  Statistics. Data is presented as the mean ± SD. Statistical analysis was performed using GraphPad Prism software (GraphPad); one-way ANOVA test was used. A P-value of less than 0.05 was considered significant.

Activated RAS expression is lower in aRMS cells than R-eRMS cells.
We first examined overall RAS protein expression in human R-eRMS (RD, Rh36, SMS-CTR) compared to human aRMS (Rh3, Rh28, Rh30) cell lines growing in log phase using immunoblot (Fig. 1A, Supplementary Fig. S1). Pan-RAS levels were lower in all three aRMS cells compared to all three R-eRMS cells, and densitometric quantitation revealed that activated RAS (RAS-GTP) was lower in two of the three aRMS cell lines. Since RD cells contain an NRAS mutation 32 , and Rh36 and SMS-CTR cells contain HRAS mutations 40,41 , higher levels of RAS-GTP are expected. On the other hand, none of the aRMS cell lines contain mutant RAS 32,42 (Fig. 1B). RAS is crucial to cell cycle progression and the levels of activated RAS increase after serum-stimulated cell cycle entry 43 . To determine the  S3). For each cell line, the same cells growing in log phase were used as control. Once again, the baseline RAS-GTP levels were lower in aRMS cell lines compared to R-eRMS cell lines. Upon serum-stimulation, the R-eRMS cell lines were as predicted not able to generate additional RAS-GTP (since the endogenous oncogenic RAS was already increasing RAS-GTP to threshold). On the other hand, the aRMS cells were also minimally able to be stimulated, with only the Rh28 cells showing a mild increase in RAS-GTP expression. In sum, these results suggest that baseline RAS signaling activity is lower in aRMS cells compared to R-eRMS cells and that aRMS cells do not robustly activate RAS in response to serum-stimulation.

Expression of oncogenic HRAS inhibits growth and proliferation in Rh28 aRMS cells.
To assess the phenotypic impact of altered RAS signaling in aRMS, we stably expressed wild type, dominant negative, or oncogenic HRAS mutants in Rh28 aRMS cells compared to RD R-eRMS cells. We chose HRAS since this was the isoform we had used to model eRMS from primary human myoblasts 10 . Expression of oncogenic HRAS (H-RAS 12V), wild type HRAS (H-RAS WT), or dominant negative HRAS (H-RAS 17N) were confirmed via immunoblot using a pan-RAS antibody ( Fig. 2A, Supplementary Fig. S4) since isoform specific antibodies were not available, then cells expressing these constructs were examined for changes in growth. While both wild type and dominant negative HRAS moderately impaired the growth of Rh28 cells, expression of oncogenic HRAS caused growth arrest (Fig. 2B, left) and inhibition of proliferation as assessed by BrdU incorporation (Fig. 2C).
On the other hand, growth of RD cells was unaffected regardless of the RAS mutant expressed (Fig. 2B, right).

Expression of oncogenic HRAS increases differentiation in Rh28 aRMS cells. Having observed
that oncogenic RAS was detrimental to Rh28 cell growth, we continued to investigate the etiology of the growth inhibition. We first explored the possibility of increased myogenic differentiation. Under light microscopy of standard growth culturing conditions, Rh28 cells expressing vector, wild type RAS, or dominant negative RAS all displayed normal morphology of small round cells, whereas cells expressing oncogenic RAS displayed an elongated morphology (Fig. 3A). To assess for myogenic differentiation, we cultured the cells in myogenic differentiation-permissive conditions for five days and stained them for the myogenic marker MF20. Compared to the other groups, Rh28 cells expressing oncogenic RAS displayed a decreased cell density and an increase in overall staining for MF20, consistent with myogenic differentiation (Fig. 3B). However, when compared to the positive control HSMMs, cells expressing oncogenic RAS did not display the same degree of elongation and myotube formation, suggesting that while oncogenic RAS caused an increase in differentiation markers, it was not the sole mechanism of growth inhibition.

Expression of oncogenic HRAS induces senescence in Rh28 aRMS cells. Based on the literature
surrounding RAS and OIS, we turned our attention to the possibility of increased senescence as the source of growth arrest. In addition to elongating, Rh28 cell populations expressing oncogenic RAS also underwent other morphologic changes including enlargement and flattening, morphologic hallmarks of senescence. Therefore we  www.nature.com/scientificreports/ stained the cells for β-galactosidase, a biochemical indicator of senescence. We found that while the vector, wild type RAS, and dominant negative RAS cells remained β-gal negative, the oncogenic RAS cells stained positive as indicated by their blue color (Fig. 4A). Since both the p16INK4a-RB pathway and ARF-p53 pathway have been implicated as crucial mediators of senescence [44][45][46] , we examined the expression of RB and p53 pathway members via immunoblot in Rh28 and RD cells stably expressing RAS mutants. While the expression of oncogenic RAS in Rh28 cells led to moderate and inconsistent changes in the levels of total RB, total p53, and phospho-p53 over multiple replicates, the levels of p16 and p21 were consistently elevated, and phospho (inactive) RB levels consistently decreased with oncogenic RAS expression (Fig. 4B, Supplementary Fig. S5). This effect was not seen to the same degree in RD cells, although there was a modest increase in p16 in RD cells expressing H-RAS 12V or H-RAS WT, so that despite the lack of change in the growth curve, at least some of the RD cell population was responding at a signaling level to increased RAS activity. In sum, these data suggest that oncogenic RAS expression induces senescence in Rh28 aRMS cells and is associated with p16 and p21 upregulation and RB tumor suppressor activation.
Expression of constitutively active AKT or ERK also impairs Rh28 aRMS cell growth. Next, we moved downstream to examine the RAS effector pathways involved in OIS in Rh28 aRMS cells. We first analyzed the levels of AKT and ERK activity in Rh28 and RD cells stably expressing RAS mutants by immunoblotting for total and phospho (active) AKT and ERK. Interestingly, the levels of pAKT markedly increased with the expression of oncogenic RAS in Rh28 but less in RD, whereas the levels of pERK increased with oncogenic RAS in both Rh28 and RD (Fig. 5A, Supplementary Fig. S6). To functionally assess these RAS effector pathways, we expressed activating mutants of single RAS effector pathways in Rh28 cells and measured cell growth over seven days. Expression of myristoylated (activated) AKT led to growth arrest that was comparable to oncogenic RAS. Expression of activated MEK also led to growth inhibition, though not to the same degree as oncogenic RAS and MyrAKT (Fig. 5B). Similar to that seen with the RAS mutant studies, RD cells were agnostic in their population growth to RAS effector pathway mutants (Fig. 5C). These activating mutants (MyrAKT, MEK1DD) have www.nature.com/scientificreports/ been well documented to increase phosphorylation of their target pathways in the case of AKT and MEK [47][48][49] , which we confirmed using immunoblot ( Fig. 5D-E, Supplementary Figs. S7, S8). As assessed by densitometry, in this experiment the H-RAS V12 mutant was also effective in activating all of the downstream RAS effector pathways, as assessed by increased phospho-AKT and phospho-ERK (Fig. 5D). Interestingly, pan-RAS protein levels increased slightly in the cells expressing the activating mutants even though the pathways activated were downstream of RAS, suggesting a feedback loop. In sum, these data suggest that the AKT and ERK RAS effector pathways impair Rh28 cell growth when ectopically overexpressed and could be the trigger for OIS.

The RAS wild-type eRMS cell line RMS-YM is also growth-inhibited in response to oncogenic HRAS expression.
To further explore our original hypothesis that eRMS cell lines tolerate oncogenic RAS while aRMS cell lines do not, we examined the impact of oncogenic HRAS and activating RAS effector mutants in RAS wild type RMS cells, with the prediction that as an eRMS cell type, it would tolerate oncogenic RAS expression. We first examined RAS levels in RMS-YM cells compared with the R-eRMS RD cell line (Fig. 6A, Supplementary Fig. S9). While total RAS expression was similar between the two cell lines, RAS-GTP levels were lower in log phase RMS-YM cells, and similar to Rh28 cells, could not be significantly stimulated after 48 h serum starvation. We then, as before, ectopically overexpressed oncogenic HRAS (H-RAS 12V), wild type HRAS (H-RAS WT), or dominant negative HRAS (H-RAS 17N) in RMS-YM cells, and expression was again confirmed via immunoblot using a pan-RAS antibody (Fig. 6B, Supplementary Fig. S10). Unexpectedly, both wild type and dominant negative HRAS moderately impaired the growth of RMS-YM cells (although not statistically significant), while expression of oncogenic HRAS caused growth arrest (Fig. 6C) and inhibition of proliferation as assessed by BrdU incorporation (Fig. 6D). To mirror the prior studies in the Rh28 and RD cells, we examined the impact of the RAS mutants in the RMS-YM cells on the p53, RB, AKT and ERK pathways. H-RAS 12V and H-RAS WT effectively activated p21, while dominant negative H-RAS 17N led to increased p53 expression (Fig. 6E, Supplementary Fig. S11). H-RAS 12V and H-RAS WT both activated the AKT and ERK pathways (Fig. 6F, Supplementary Fig. S12). Overall, RMS-YM cells appear to have low expression of p53, especially com- www.nature.com/scientificreports/ pared to RD cells that harbor TP53 mutations 32 ( Supplementary Fig. S13), with Ser15 p-p53 below the limit of detection on our blots. This is likely due to the described MDM2 amplification in RMS-YM cells 50 . The growth arrest caused by oncogenic H-RAS 12V did not lead to as stark a morphology alteration in RMS-YM cells (Fig. 7A) when compared to the Rh28 cells. In myogenic assays, they were largely non-viable after multiple days in differentiation conditions (Fig. 7B). But similar to that seen in Rh28 cells, the RMS-YM cells expressing oncogenic H-RAS did exhibit increased senescence as assessed by β-galactosidase staining (Fig. 7C). Interestingly, RMS-YM H-RAS 17N cells, which grew comparably to RMS-YM cells ectopically expressing H-RAS WT under normal growth conditions, seemed sparser after the five-day differentiation period and showed increased myogenic differentiation as assessed by MF20 staining (Fig. 7B). Finally, similar to that seen in the Rh28 cells, ectopic overexpression of constitutively active AKT or MEK constructs impaired RMS-YM cell growth (Fig. 8A). A noticeable difference was that forced expression of these RAS effector mutants was associated with a decrease in overall RAS protein levels (Fig. 8B, Supplementary Fig. S14), perhaps suggesting that the RMS-YM cells are trying to restore equilibrium in RAS signaling. All together, these data indicate that RMS-YM RAS wild-type eRMS cells respond to changes in oncogenic HRAS expression in a manner more similar to aRMS than R-eRMS and predict that eRMS tumors can be further sub-classified by mutational profile.

Discussion
Although RAS mutations are common in eRMS, it has been unclear why aRMS cells bearing PAX3-FOXO1 fusions rarely harbor RAS mutations. We reasoned that either oncogenic RAS is not required for tumorigenesis of aRMS or is not tolerated in aRMS. In this study, we show that forced expression of HRAS triggers OIS in PAX3-FOXO1-positive Rh28 aRMS cells, and this senescence is associated with increased p16, p21, and RB pathway activity. It is important to note that we only studied the effect of HRAS. NRAS and KRAS will need to be examined independently, and this will be supported by RAS isoform-specific antibodies 51 . Given the observations that H, N, and KRAS-mutant eRMS are associated with specific age ranges and anatomic locations, this will contribute to the understanding of the role of dysregulated development in R-eRMS 52 .
Though OIS has been observed in pre-malignant cells from a variety of tissues [13][14][15][16][17][18] , there are relatively few examples of activation of OIS in cancer cells. In non-small cell lung cancer cell lines harboring KRAS mutations, latent OIS can be reactivated when another oncogene, TWIST1, is silenced 53 . In a mouse model of lung cancer, OIS occurred when oncogenic KRAS and BRAF were expressed concurrently 54 . These examples reveal the mutually exclusive nature of some oncogene combinations, such as KRAS and BRAF, in human tumors, and we posit that this mutually exclusive relationship may exist between HRAS and PAX3-FOXO1. Interestingly, when oncogenic RAS has been expressed ectopically in cell lines concurrently with PAX3-FOXO1, it has been in the setting of deficient p53 activity, resulting from prior TP53 mutations such as in Rh30 cells or acquired through expression of SV40 T antigen or dominant negative p53 constructs [55][56][57] . When oncogenic NRAS was detected in PAX3-FOXO1-positive aRMS tumor tissue, it was accompanied by loss of heterozygosity of chromosome 11p 58 . It remains to be seen whether other genetic changes influence the concurrence of RAS and PAX3-FOXO1 expression in aRMS. It is important to note that while an excess of RAS signaling was not tolerated in Rh28 aRMS cells, deficient RAS signaling was also detrimental. This is shown by the effect of dominant negative HRAS in the Rh28 cells, and others have also recently found that impaired RAS signaling suppresses aRMS cell and xenograft growth 59,60 . The changes in RAS protein expression we observed in response to ectopic expression of constitutively active RAS effectors in both Rh28 and RMS-YM cells likely reflects the homeostatic feedback loops that occur in RAS signaling 61 . www.nature.com/scientificreports/ OIS is a tumor suppressor mechanism that thwarts transformation by activating the p53/RB pathways 12,46,62 . Thus, the very existence of OIS puts selective pressure on tumor cells to acquire mutations in p53/RB pathways to bypass senescence during tumorigenesis. TP53 mutations can be found in almost every type of cancer, varying in rates from 10% to close to 100% 63 , and RB mutations have been found in multiple cancers including retinoblastomas, osteosarcomas, small-cell lung, breast, and cervical carcinomas, and leukemias [64][65][66][67][68][69] . We show here that OIS in Rh28 cells is associated with increased p16, p21, and RB pathway activation. However, this raises an interesting question-since OIS can be reactivated in an aRMS cell line, does this mean that its RB/p53 pathways are intact? While aRMS tumors have low levels of p53 pathway activity 70 , only a small percentage have mutation or loss of heterozygosity of TP53 compared to eRMS tumors 58 , and some aRMS cell lines with TP53 mutations are thought to have acquired them after culture in vitro 32 . We speculate that in at least some aRMS cells, p53 and RB pathways are functionally inactivated to permit tumor formation and proliferation, but this functional inactivation is reversible. Our prior data showed that PAX3-FOXO1 cooperates with epigenetic silencing of p16 to bypass senescence 37 . Thus, senescence pathways are turned "off " during tumor formation and growth, but these pathways can be reactivated and turned "on" again by external stressors such as ectopic oncogenic RAS expression. Pro-senescence therapy via p53/RB pathway reactivation is an enticing field of study and may provide new ways to target aRMS tumors [71][72][73][74] .
Our study also revealed that in Rh28 cells, hyperactive AKT or ERK signaling inhibits cell growth. While PI3K-AKT signaling is important for cell proliferation and survival in many tumors 75,76 , it is also a known inducer of OIS 21,77,78 . A recent study detected a PIK3CA mutation in a PAX3-FOXO1 positive aRMS tumor 79 ; however, it www.nature.com/scientificreports/ is not known whether the mutation was activating or inactivating, making this data difficult to interpret. Also, in our study, the MyrAKT construct we used was not 100% specific for the AKT pathway; though the construct mainly increased pAKT levels, it also increased pERK levels. On the other hand, the MEK1DD construct only increased pERK levels, and it only partially induced growth inhibition. Similar to AKT, hyperactive ERK signaling can also induce OIS 80,81 . It is possible that cooperation between both the AKT and ERK pathways will be needed to fully induce OIS in aRMS, and further studies will be needed to definitively prove the role of these pathways in the senescence phenotype. Future complementary experiments using pharmacologic activators or inhibitors of AKT and MEK/ERK will be useful in this regard.
In summary, we find that oncogenic HRAS expression is detrimental to the growth and leads to OIS in Rh28 aRMS cells. This observation opens the door to future investigation of the mechanisms of OIS in aRMS, including evaluation of other aRMS cell lines, and possibilities for designing aRMS therapies that target reactivation of p53/RB and/or induce senescence. Oncogenic HRAS is similarly detrimental to RMS-YM RAS wild type eRMS cells, distinct from its role as an oncogenic driver in some R-eRMS cells. These data, along with oncogenic RAS isoform studies in RMS 23 , suggest that oncogenic RAS and likely other somatic mutations in eRMS are expressed in a context-dependent manner. The subgroup of eRMS tumors that are RAS pathway wild type or lacking identifiable protein-coding mutations will require further study to understand and therapeutically target their oncogenic drivers. www.nature.com/scientificreports/