Chromatin remodeling protein HELLS is upregulated by inactivation of the RB-E2F pathway and is nonessential for osteosarcoma tumorigenesis

Osteosarcoma is the most common primary bone malignancy in children and adolescents. Among the various molecular mechanisms implicated in osteosarcomagenesis, the RB-E2F pathway is of particular importance as virtually all cases of osteosarcoma display alterations in the RB-E2F pathway. In this study, we examined the transcription factor E2F family members that are associated with increased malignancy in Rb1-null osteosarcoma tumors. Using genetically engineered mouse models of osteosarcoma, we found that loss of activator E2Fs, E2F1 and E2F3, significantly delays tumor progression and increases the overall survival of the p53/Rb1-deficient osteosarcoma mouse model. We also studied the role of helicase, lymphoid specific (HELLS), a chromatin remodeling protein identified as a critical downstream effector of the RB-E2F signaling pathway in various cancers. In this study, we confirmed that the RB-E2F pathway directly regulates HELLS gene expression. We also found that HELLS mRNA is upregulated and its protein overexpressed in osteosarcoma. Using loss-of-function assays to study the role of HELLS in human osteosarcoma, we observed that HELLS has no effect on tumor proliferation and migration. Further, we pioneered the study of Hells in developmental tumor models by generating Hells conditional knockout osteosarcoma mouse models to examine the role of HELLS in osteosarcoma tumor development. We found that loss of Hells in osteosarcoma has no effect in tumor initiation and overall survival of mice. This suggests that while HELLS may serve as a biomarker for tumorigenesis and for RB-E2F pathway status, it is unlikely to serve as a relevant target for therapeutics in osteosarcoma.


INTRODUCTION
Osteosarcoma is a mesenchymal tumor that is histologically characterized by the presence of malignant mesenchymal cells and the production of a bone stroma. Osteosarcomas are commonly characterized by an appendicular primary tumor with a high rate of metastasis to the lungs [1]. The most common genetic findings in osteosarcoma are the dysregulation of the TP53 and RB1 tumor suppressor genes [2,3]. Loss-of-function mutations of the RB1 gene in osteosarcoma are associated with poor therapeutic outcome, as defined by increased mortality, metastasis, and poor response to chemotherapy [4][5][6][7][8]. Unfortunately, clinical outcomes for osteosarcoma patients have not substantially improved in over 30 years. As a result, the 5-year overall survival rate has remained stable at ~65% in case of local disease and ~20% for patients with metastatic disease [9][10][11][12].
The RB1 gene encodes the tumor suppressor protein RB. RB forms a transcriptional repression complex Research Paper www.oncotarget.com with the E2F family of transcription factors and thereby negatively regulates G 1 /S transition during the cell cycle through the suppression of E2F target genes. There are six E2F family members that bind to the RB family and are classified as activator E2Fs (aE2Fs: E2F1, E2F2 and E2F3a) and repressor E2Fs (rE2Fs: E2F3b, E2F4, and E2F5) [13]. Of these, aE2Fs show preferential binding to RB protein. In response to growth stimuli, the cyclin D-CDK4/6 complex phosphorylates RB, relieving aE2Fs to facilitate the transcription of genes required for G 1 /S transition. In addition to E2Fs, RB associates with a large number of nuclear proteins, including a variety of chromatin-associated proteins that have diverse activities [14][15][16]. RB1 mutations are present in up to 70% of osteosarcoma cases [17]. Furthermore, multiple studies have shown that RB1 loss is correlated with poor prognosis for patients with osteosarcoma [4][5][6][7][8]. Specifically, loss of RB1 function is associated with increased risk of osteosarcoma metastasis and poor response to chemotherapy, compared with osteosarcoma patients with intact RB1 [5,18,19]. A previous study performed in retinoblastoma, a childhood cancer caused by bi-allelic loss of RB1, indicated that retinal tumorigenesis was driven by E2F1-and E2F3-transcribed genes following the lack of transcriptional repression due to RB loss [20].
HELLS (helicase, lymphoid specific; also known as LSH) is a protein that belongs to the SNF2 family of chromatin-remodeling ATPases [21]. HELLS is critical for normal development of mammals by establishing DNA methylation patterns across the genome [22]. In retinoblastoma, HELLS was identified as a critical contributor of Rb-mediated tumorigenesis [20]. In addition to retinoblastoma, several reports have shown that HELLS overexpression contributes to malignant progression including renal cell carcinoma, gliomas, prostate cancer, melanoma, and nasopharyngeal carcinoma [23][24][25][26][27]. In astrocytomas and glioblastomas, upregulation of E2F1 correlated with increased HELLS expression and increased along with tumor grades [24]. However, the role of HELLS in tumor initiation, particularly in osteosarcoma, has never been examined.
In this study, we generated a series of genetically engineered mouse models of osteosarcoma to study the role of E2Fs in tumor development. We focused on E2F1 and E2F3, two aE2F previously described as critical for Rb-mediated tumorigenesis, and E2F5 as a rE2F control [20]. Similar to what was observed in retinoblastoma, we found that E2F1 and E2F3 contribute to increased malignancy associated with loss of Rb1. Since HELLS was identified as aE2F target gene and critical contributor of tumorigenesis in retinoblastoma, we evaluated the role of HELLS in osteosarcoma tumorigenesis. We demonstrated that activator E2Fs, E2F1 and E2F3, directly regulate HELLS expression. We also showed that RB-E2F pathway inactivation, present in most osteosarcomas, results in increased HELLS expression. Further, we evaluated the role of HELLS in osteosarcoma cell survival, proliferation, migration, and tumorigenesis both in vitro and in vivo. We showed that, unlike what has been observed in other malignancies, targeting HELLS in human osteosarcoma has a modest effect in osteosarcoma survival and no effect on migration. Further, using a Hells conditional knockout mouse model, we found that loss of Hells has no effect on osteosarcoma tumor incidence and overall survival. This suggests that HELLS is not critical for tumor initiation and progression in osteosarcoma.

HELLS is overexpressed in both human and mouse osteosarcoma
A previous study in retinoblastoma identified HELLS as a key target gene downstream of the RB-E2F signaling pathway that is overexpressed following the loss of RB, and contributes to tumorigenesis [20]. Given that RB1 loss in osteosarcoma patients is associated with poor prognosis, we hypothesized that loss of RB potentiates osteosarcoma through transcriptional deregulation of chromatin remodeling genes including HELLS. Initial analysis of The Cancer Genome Atlas (TCGA) data set of 263 sarcoma tumor samples indicated that HELLS is upregulated by 4.27-fold compared to normal controls, making HELLS an attractive target to study in osteosarcoma ( Figure 2A). Further analyses of HELLS gene expression using real-time qPCR in human osteosarcoma cell lines (143B, SJSA-1, SaOS-2, and U-2 OS) and patient derived xenografts (PDX1-5) showed a significant upregulation of HELLS mRNA level in most osteosarcoma samples when compared to human mesenchymal stem cells (MSCs) ( Figure 2B-2C). At the protein level, HELLS was also highly overexpressed across all analyzed osteosarcoma subjects compared to MSCs and human osteoblasts (hOB) controls ( Figure 2D). Increased HELLS protein expression was observed even in U-2 OS and PDX2, which showed no upregulation at the mRNA level. Further, HELLS protein overexpression was also observed in the p53/ Rb1 DKO osteosarcoma mouse model ( Figure 2E). We also noted that some primary tumors, both in human and mouse, express a longer HELLS protein. Whether this is representative of post-translational modifications (HELLS can be phosphorylated, acetylated, ubiquitinated, methylated and sumoylated) or a HELLS isoform was not investigated.

HELLS is a direct target of the RB-E2F signaling pathway in the osteogenic lineage
Evidence of direct regulation of HELLS by the RB-E2F signaling pathway is limited to a single in vitro study in gliomas [24]. To confirm that the RB-E2F pathway transcriptionally represses HELLS, we performed chromatin immunoprecipitation analysis (ChIP) in the osteosarcoma lineage of origin using MSCs. Motif mapping using MotifMap revealed a putative E2F1 binding motif within the promoter region of HELLS, 97 bp upstream of the start codon [30]. PCR primers were designed to flank this area of putative E2F1 binding motif. As seen in Figure  3A, we observed enrichment of E2F1 at the consensus binding site at the HELLS promoter ( Figure 3A).
To confirm direct transcriptional regulation of E2F1 on HELLS in human osteosarcoma, we acquired lentiviral vectors encoding shRNAs targeting E2F1 (shE2F1) and transduced the four human osteosarcoma cell lines in which we had previously observed HELLS overexpression ( Figure 2D). Interestingly, we found that E2F1 knockdown was not effective at reducing HELLS protein levels in any of the osteosarcoma cell lines used ( Figure 3B). To determine if other aE2F family members could compensate for the loss of E2F1, we acquired lentiviral vectors encoding shRNAs targeting E2F2 (shE2F2) and E2F3 (shE2F3). Knockdown of E2F2 or E2F3 using shE2F2 or shE2F3 alone did not affect HELLS protein expression (Supplementary Figure 2B); neither did the combination of shE2F1 and shE2F2 (Supplementary Figure 2A). However, combinatorial knockdown of E2F1 and E2F3 using shE2F1 and shE2F3 achieved a significant reduction in HELLS protein levels ( Figure 3C).
Of the four osteosarcoma cell lines used in this study, SaOS-2 is the only one bearing an RB1 mutation. To assess the mechanism through which the RB-E2F pathway is being deregulated in the osteosarcoma cell lines without RB1 mutations, we analyzed CDK4 and CDK6 gene expression. CDK4 and CDK6 amplification and overexpression has been reported in some osteosarcomas, resulting in RB hyperphosphorylation [31]. Real time qPCR analysis of CDK4 showed a significant increase in its transcription in all osteosarcoma cell lines ( Figure 3D). A particular high expression of CDK4 was observed in SJSA-1, a cell line with known CDK4 amplification [32]. CDK6 was also upregulated in the 3 osteosarcoma cell lines without RB1 mutations: 143B, SJSA-1 and U-2 OS ( Figure 3E). In order to confirm that increased CDK4 and CDK6 activity results in RB hyperphosphorylation and leads to overexpression of HELLS, we treated all four human osteosarcoma cell lines with CDK4/6 inhibitor, palbociclib [33]. Western blot analysis confirmed reduction in RB phosphorylation following palbociclib treatment (Supplementary Figure 2C). Real-time qPCR analysis of HELLS mRNA levels revealed significant decreases in HELLS expression upon palbociclib treatment, with the exception of the RB1-null cell line, SaOS-2, which served as a negative control ( Figure 3F).

HELLS has limited effect on human osteosarcoma cell proliferation and no effect on migration
Emerging reports have linked HELLS overexpression in cancers with its ability to promote proliferation [20,25]. To study the role of HELLS overexpression in osteosarcoma, we acquired a lentivirus encoding an shRNA sequence complimentary to HELLS (shHELLS) to facilitate HELLS gene knockdown in osteosarcoma cell lines, as well as vector-only control. Western blot analysis was used to validate successful HELLS knockdown in each osteosarcoma cell line ( Figure 4A). We performed colony-forming assay on transduced osteosarcoma cell lines to assess their ability to give rise to colonies at a single-cell level. HELLS knockdown in osteosarcoma cell lines results in modest reduction (31.6-40.5%) in the ability to form colonies when compared to controls, but reached statistical significance in 143B and U-2 OS cell lines ( Figure 4B-4C). Further evaluation of cellular proliferation and survival upon HELLS knockdown using CellTiter-Glo cell viability assay revealed a significant increase in cell survival and proliferation in 143B cells, but no significant change in the relative number of cells in SJSA-1, SaOS-2 and U-2 OS cells ( Figure 4D-4G). This confirms the modest decreases or absence of changes observed in the colony-forming assay.
HELLS expression is positively correlated with metastatic potential, shown by increased migratory capacity of cells [27,34]. Using scratch-wound healing assay, we assessed changes in migration potential in HELLS knockdown and control osteosarcoma cell lines. We detected no significant changes in the ability to heal the wounds within 8 h in any of the four cell lines analyzed ( Figure 4H-4I).

HELLS is dispensable for tumor initiation and progression in mouse OS
In order to study the role of HELLS in osteosarcomagenesis, we generated genetically engineered mice to facilitate preosteoblast-specific knockout of Hells in two distinct genetic engineered mouse models of osteosarcoma: p53 cKO and p53/Rb1 DKO [28]. The resulting p53/Hells DKO (Osx-cre p53 lox/lox ; Hells lox/lox ) and p53/Rb1/Hells triple knockout (TKO; Osx-cre p53 lox/lox ; Rb1 lox/lox ; Hells lox/lox ) were compared to their corresponding littermate controls (p53 cKO and p53/Rb1 DKO, respectively) for the assessment of the role of HELLS in tumor initiation and promotion during osteosarcoma development. All genotypes lead to the development of osteosarcoma in mice with 100% penetrance. As seen in Figure 5, no differences were detected between the mice bearing Hells conditional knockout alleles compared to their littermate controls. The median survival in p53/Rb1/ Hells TKO mice was 27.7 weeks (n = 23) compared to 26.9 weeks (n = 29; p = 0.8805) in p53/Rb1 DKO mice ( Figure 5A). The median survival in p53/Hells DKO mice was 52.7 weeks (n = 22), compared to 58.45 weeks in p53 cKO (n = 35; p = 0.9739) ( Figure 5B). Intriguingly, analysis of gender differences between the Hells-null mice indicated a significantly lower survival in p53/Hells DKO females compared to males (p = 0.0057; Supplementary  Figure 3). Despite this difference, both males and females show no significant difference when compared to p53 cKO mice (p = 0.2076 and p = 0.0851, respectively). Western blot analysis of Hells wild-type (p53/Rb1 DKO and p53 cKO) and Hells-null (p53/Rb1/Hells TKO and p53/Hells DKO) tumors confirmed that HELLS was efficiently knocked-out in these osteosarcomas ( Figure 5C).

DISCUSSION
Osteosarcoma is the most common primary cancer of bone and typically occurs in children and young adults. As a highly metastatic cancer, 15-20% of osteosarcoma patients are diagnosed after the cancer has already metastasized (typically to the lungs), which translates to 5-year survival rates of <40% [35]. Thus, there is a pressing clinical need to determine the factors responsible for metastasis in osteosarcoma to facilitate development of new therapeutic strategies. Loss of function mutations at the RB1 gene are associated with increased mortality, metastasis and poor therapeutic outcome in osteosarcoma [4][5][6][7][8]. However, the mechanism(s) through which RB1 loss leads to poor prognosis remain to be elucidated. In this study, we demonstrate that poor clinical outcome associated with loss of RB1 is mediated, at least in part, by de-repression of aE2F family members. We also examined the role of HELLS, a chromatin remodeling protein transcriptionally regulated by E2F1 that promotes tumorigenesis in several cancers [20,24,27]. Here, we show that HELLS is overexpressed in osteosarcoma; however, our results indicate that while HELLS is a good biomarker for RB pathway deregulation, HELLS does not contribute to osteosarcoma tumorigenesis. www.oncotarget.com

Activator E2Fs roles in osteosarcoma
Preosteoblast-specific loss of p53 is sufficient in driving osteosarcoma tumorigenesis in mouse [28]. Disease presentation in this model is potentiated by the additional loss of Rb1, which we were able to recapitulate. Here, we tried to elucidate the mechanism through which RB inactivation contributes to poor clinical outcome in osteosarcoma. We found that loss of either E2f1 or E2f3 significantly increased the median survival age of p53/ Rb1 DKO mice; whereas loss of E2f5 did not improve or worsen the outcome. These results are consistent with other studies reporting that loss of transcriptional control by aE2Fs, particularly E2F1 and E2F3, is responsible for tumorigenesis following Rb1 inactivation [20,29,[36][37][38]. However, in osteosarcoma, loss of neither aE2F was sufficient to completely reverse the potentiation of the disease caused by Rb1 loss. This could be due, at least in part, to compensation by other E2F family members (discussed in the next section). The observation that inactivation of either E2F1 or E2F3 could significantly delay the onset of tumors and increase overall survival of osteosarcoma-bearing animals suggests that the increase malignancy upon Rb1 loss might be reversible, and that

HELLS is transcriptionally regulated by the RB-E2F pathway
Others and we have identified HELLS as a transcriptional downstream target gene of aE2Fs that is overexpressed in cancer and contributes to tumor progression [20,24,27]. To determine if HELLS is overexpressed in osteosarcoma, we blasted The Cancer Genome Atlas (TCGA) data set and found that HELLS is upregulated in sarcoma tumor samples compared to normal controls. We confirmed HELLS gene upregulation in the human osteosarcoma cell lines 143B, SJSA-1, and SaOS-2, as well as in four of the five independent  Despite the lack of mRNA upregulation in some cell lines and tumors, HELLS protein overexpression was observed in all the human osteosarcoma cell lines and xenografts analyzed in this study. The discordance between mRNA and protein levels in some of the samples suggests that while HELLS protein levels are predominantly transcriptionally regulated there are also translational and post-transcriptionally mechanisms of protein expression regulation. Further, the HELLS protein overexpression observed in human osteosarcoma is recapitulated in the p53/Rb1 DKO osteosarcoma mouse model, strengthening its validity as a study model for this disease.
Evidence of direct regulation of HELLS by the RB-E2F signaling pathway was limited to a single in vitro study in gliomas [24]. In this study, we confirmed that the RB-E2F signaling pathway directly regulates transcriptional activation of HELLS in the osteogenic lineage. ChIP analysis revealed enrichment of E2F1 on the promoter region of HELLS where putative E2F1 binding motifs reside. Interestingly, significant reduction of HELLS protein level was only achieved through combined knockdown of E2F1 and E2F3, and not by E2F1 or E2F3 knockdown alone. This suggests that E2F1 and E2F3 compensate for the loss of each other to regulate HELLS expression. This observation could begin to explain why losses of neither E2f1 (p53/Rb1/E2f1 TKO) nor E2F3 (p53/Rb1/E3f3 TKO) were sufficient to reverse the increased malignancy observed upon Rb1 loss (p53/ Rb1 DKO) in osteosarcoma but rather provided only partial improvements. The observation that rescue of altered expression of RB-E2F transcriptional gene targets is orchestrated by more than one E2F family member is an important consideration for therapeutic approaches being developed that aim to inhibit E2Fs directly.
RB-E2F signaling pathway is often inactivated in cancer due to various reasons including RB1 loss, RB hyperphosphorylation through increased activity of CDK4/CDK6, or E2F hyperphosphorylation, all of which prevent repression of E2Fs from binding to downstream targets. Therefore, it is not surprising that HELLS upregulation and overexpression in osteosarcoma is not restricted to RB1-null cases, but rather a generalized event observed across samples that bear RB-E2F pathway inactivation. Of the osteosarcoma cell lines examined in this study, SaOS-2 harbors a deletion in the RB1 gene that leads to a non-functional truncated protein [39], SJSA-1 has CDK4 amplification [32], and U-2 OS is p16 null [40]. Here, we detected upregulation of CDK4 and/or CDK6 mRNA levels in all osteosarcomas, which suggests a predominant mechanism of RB-E2F pathway inactivation through RB hyperphosphorylation in this malignancy. This was further confirmed by our observation of HELLS down-regulation upon treatment with CDK4/6 inhibitor, palbociclib.

HELLS overexpression in osteosarcoma
The expression of HELLS has been positively associated with proliferation in both normal and malignant cells [20,21,27,41]. Multiple studies have shown that HELLS knockdown in vitro results in decreased proliferation of various cancer cell types including retinoblastoma [20], gliomas [24], nasopharyngeal [27] and breast cancer [34] cells, and that engraftment of cells carrying ectopically expressed HELLS gives rise to heavier tumor burden. In some cases, HELLS expression is positively correlated with metastatic potential, shown by increased migratory capacity of cells and promotion of epithelial to mesenchymal transition [27,34]. In this study, we demonstrated that HELLS knockdown in osteosarcoma cell lines modestly decreased colonizing capacity and had no effect on proliferation, except in 143B cells where we found increased proliferative capacity, opposite of expected. Further, we found no effect of HELLS on migratory potential in osteosarcoma.
Beyond in vitro loss-of-function studies in human osteosarcoma models, we generated novel genetic engineered mouse models to study the role of Hells in osteosarcoma tumor formation. We observed that p53/ Hells DKO and p53/Rb1/Hells TKO mice showed no improvement in tumor incidence or overall survival compared to p53 cKO and p53/Rb1 DKO controls, respectively. This result suggests that HELLS is not a critical driver of Rb1-mediated malignancy in murine osteosarcoma tumors.
Taken together, our studies and others indicate that HELLS may be a reliable biomarker for RB-E2F pathway inactivation. However, HELLS upregulation may not always be synonymous to a critical role in tumorigenesis or tumor progression, or a warranted target for therapeutics in all malignancies where HELLS overexpression is observed.
The p53 lox/lox and Rb1 lox/lox conditional-knockout mice were obtained from the Mouse Models of Human Cancer Consortium at the National Cancer Institute; E2f1-knockout and E2f3 lox/lox conditional knockout mice were obtained from Dr. Gustavo Leone (The Ohio State University); E2f5 lox/lox mice were obtained from Dr. Joseph Nevins (Duke University); the Osx-cre mice were obtained www.oncotarget.com from The Jackson Laboratory. The Hells-conditional knockout mouse was generated from mice obtained from the European Mouse Mutant Archive, backcrossed to Flp mice for removal of the neo-cassette (tm1c conversion), and then backcrossed to C57BL/6N mice for Flp removal. Mice were monitored weekly for signs of osteosarcoma. Moribund status was defined as the point when tumors had reached 10% body weight or induced paralysis in the mouse. The University of California Irvine Institutional Animal Care and Use Committee approved all animal procedures. Survival curves were generated using GraphPad Prism. Mantel-Cox test was used for statistical analyzes of the curves.

Palbociclib treatment
Osteosarcoma cells were seeded at a density of 500000 cells/well of 6-well plate and allowed to adhere overnight. Two concentrations of Palbociclib, 0.5 μM and 1 μM, were used to treat cells for 24 h prior to collection. DMSO was used as drug vehicle control.

Cell viability assay
All cells were seeded into 96-well black assay plates (Costar) at a density of 3,000 cells per well. Cell viability was determined at 24, 72, and 96 hours after seeding using CellTiter-Glo (Promega) according to manufacturer's instructions. Luminescence readings were normalized to the 24 h post-seeding reading. Survival curves were generated using GraphPad Prism.

Colony formation assay
Lentiviral transduced cells were seeded at a lowdensity (20 cells/cm 2 ) as single cells onto 6-well cell culture plates in the appropriate cell culture medium. Fresh medium was supplemented every 4 days. Cells were incubated at 37° C with 5% CO 2 until cells in control dishes have formed sufficiently large colonies (~50 cells). Cells were washed once in PBS before fixing and staining in a mixture consisting of 6% glutaraldehyde and 0.5% crystal violet for 30 min. Stains were washed out with tap water and the plates left to dry at room temperature. Colonies were counted from triplicate experiments.

Scratch-wound healing assay
Scratch-wound assay was performed as previously described [44]. Cells (1 × 10 6 /well) were plated on 6-well plates. The day after, once cells were grown to a confluence of about 80%, the monolayer was scratched using a 10 µl sterile pipette tip. Images were captured immediately and 8 h after the wound. The exact location of the image was marked to identify the same gap. The distances between the boundaries of the wound at 0 and 8 h at 10 different locations were measured in pixels using Zen software (Zeiss).

Author contributions
S.W. and C.A.B. conceptualized the study, performed research, designed experiments, analyzed and interpreted data, and wrote the manuscript. All authors read and approved the final manuscript.