SAMHD1 promotes oncogene-induced replication stress

Oncogenes induce DNA replication stress in cancer cells. Although this was established more than a decade ago, we are still unravelling the molecular underpinnings of this phenomenon, which will be critical if we are to exploit this knowledge to improve cancer treatment. A key mediator of oncogene-induced replication stress is the availability of DNA precursors, which will limit ongoing DNA synthesis by cellular replicases. In this study, we identify a potential role for nucleotide catabolism in promoting replication stress induced by oncogenes. Specifically, we establish that the dNTPase SAMHD1 slows DNA replication fork speeds in human fibroblasts harbouring an oncogenic RAS allele, elevating levels of endogenous DNA damage, and ultimately limiting cell proliferation. We then show that oncogenic RAS-driven tumours express reduced SAMHD1 levels, suggesting they have overcome this tumour suppressor barrier, and that this correlates with worse overall survival for these patients.


Introduction
One-third of human cancers contain mutations in the RAS gene family, placing the resultant oncogenes amongst the most frequent drivers of tumour development (1). A central feature of these tumours is the presence of DNA replication stress (RS) (2), which is when DNA synthesis slows or stalls but DNA unwinding continues, resulting in exposed stretches of single-stranded DNA. If left unattended, this can lead to the formation of DNA breaks as well as aberrant mitoses, in turn promoting genome instability linked to disease aggressiveness and therapy resistance. Yet, it can also be a mechanism of tumour suppression, as it slows cell proliferation and activates anticancer barriers such as apoptosis and senescence (3). Oncogene-induced RS is thus a double-edged sword, and a complete understanding of the molecular mechanisms underpinning this process, together with the factors involved, will be central to the identification of potential biomarkers and therapeutic targets that can be utilised to improve the treatment of these cancers.
There are several mechanisms by which oncogenes such as RAS induce RS during early cancer development (2), one of which is by limiting the availability of deoxynucleoside triphosphates (dNTPs) (4). For instance, elevated origin firing, and the resulting increase in DNA synthesis, will exhaust dNTP pools at a faster rate. Oncogenic RAS signalling also downregulates key dNTP biosynthetic enzymes such as ribonucleotide reductase and thymidylate synthase (5,6), whilst the remaining purine dNTPs can become oxidised and subsequently sanitised from the dNTP pool (7). All of this is underscored by oncogeneinduced RS in cultured cells being rescued by supplementation of the culture medium with nucleosides (5,6,8,9) or folate (10), a micronutrient required for nucleotide biosynthesis. Following induction of RS by oncogenes, ATR-Chk1 signalling will be activated, which is largely responsible for stabilising and restarting stalled forks (2), and can accordingly protect cancer cells from oncogene-induced RS in a concentration-dependent manner (11).
In the present study, we hypothesised the involvement of SAM and HD-domain containing protein-1 (SAMHD1) in oncogene-induced RS for two reasons. Firstly, SAMHD1 is a dNTP triphosphohydrolase positioned as a central regulator of dNTP pools (12,13), and thus this catabolic activity could potentially limit available dNTPs in cancer cells harbouring oncogenes such as RAS. Secondly, SAMHD1 has a non-catalytic role in ATR-Chk1 activation and subsequent replication fork restart (14), and this role could potentially protect tumour cells from oncogene-induced RS.

Results and Discussion
To investigate the potential role of SAMHD1 in oncogene-induced RS, we began by utilising a well-described defined genetic system (15). In this system, normal human BJ fibroblasts have been transformed in a stepwise manner, first using telomerase catalytic subunit (hTERT), to generate the immortalized but non-tumorigenic BJ-hTERT cell line, and subsequently with Simian Virus 40 (SV40) Early Region (ER) (encompassing small and large  T antigen (16,17)) and an allele of oncogenic RAS (HRAS V12 ), ultimately creating BJ-ER-RAS V12 cells, which have lethal metastatic potential (15) (Fig. 1A). Expression of RAS V12 in cultured cells induces cell hyperproliferation followed by RS and RS-induced senescence (3,5,6,9), however the presence of SV40 ER prevents senescence onset and thereby phenotypically locks this cell model in a hyperproliferative state (15)(16)(17). We first confirmed an elevated proportion of replicating cells following oncogene expression by comparing cell cycle profiles between BJ-hTERT and BJ-ER-RAS V12 cells (Fig. 1B), and furthermore, confirmed the presence of replication stress by measuring the speeds of individual replication forks and endogenous DNA damage present in these cell lines (discussed in detail below). Intriguingly, when assessing SAMHD1 abundance in these cell lines, we observed that BJ-ER-RAS V12 cells exhibited elevated levels of SAMHD1 transcript (Fig. 1C) and protein (Fig. 1D) compared to BJ-hTERT cells. Furthermore, compared to >40 cancer cell lines, BJ-ER-RAS V12 cells exhibited the highest SAMHD1 transcript abundance, followed by BJ-hTERT cells transformed with SV40 ER (BJ-hTERT-ER) ( Supplementary Fig.1), supporting a potential role for SAMHD1 in the context of oncogene expression We reasoned that elevated SAMHD1 levels could either contribute to oncogeneinduced RS by limiting dNTP supply or counteract oncogene-induced RS by promoting restart of stalled replication forks. To investigate these putative roles, we utilised a pool of four SAMHD1-specific siRNAs and transfected the BJ-hTERT and BJ-ER-RAS V12 cells, achieving a partial knockdown of SAMHD1 protein (Fig. 1D). Consistent with RS induction by RAS V12 , an elevated level of DNA damage marker phosphorylated histone variant γH2AX was present in the BJ-ER-RAS V12 cells compared to BJ-hTERT cells (Fig. 1D). Interestingly, whilst siRNA-mediated reduction of SAMHD1 level in the BJ-hTERT cells resulted in elevated γH2AX signal, in BJ-ER-RAS V12 cells, a mild reduction was observed (Fig. 1D), suggesting a role for SAMHD1 in promoting RAS V12 -induced DNA damage in these cells.
Given oncogene-induced RS can limit cell proliferation, we next monitored this phenotype using a short-term metabolic cell viability assay and a long-term colony formation assay. Depletion of SAMHD1 in BJ-hTERT cells led to a small, but significant, reduction in the proportion of cells in both assays (Fig. 1E, F), concordant with the increase in γH2AX signal induction (Fig. 1D). In contrast, SAMHD1 depletion in BJ-ER-RAS V12 cells resulted in a significant increase in the proportion of cells (Fig 1E, F). This was particularly striking in the colony formation assay in which a doubling of plating efficiency following SAMHD1 depletion was observed (Fig. 1F). Given BJ-ER-RAS V12 cells express SV40 ER, as a control experiment, we also depleted SAMHD1 in BJ-hTERT cells transformed only with SV40 ER (BJ-hTERT-ER cells) and monitored cell growth. However, no difference in cell viability was observed following SAMHD1 depletion (Supplementary Fig. 2) despite a report indicating a role for SAMHD1 in telomere maintenance following SV40 large T-transformation (18).
To directly examine RS, we next measured the speeds of individual replication forks in the siRNA-treated BJ-hTERT and BJ-ER-RAS V12 cells using the DNA fibre assay ( Fig. 2A). BJ-ER-RAS V12 cells had slower fork speeds than BJ-hTERT cells (Fig. 2B, C), consistent with RAS V12 -induced RS. Notably, whilst depletion of SAMHD1 in BJ-hTERT cells led to a reduction in fork speeds, the opposite response was observed in BJ-ER-RAS V12 cells, with a significant increase in fork speeds being apparent (Fig. 2B, C). Taken together, these data demonstrate that SAMHD1 limits the proliferation of cells harbouring RAS V12 by slowing DNA replication fork speeds.
RAS mutations have long been established as the drivers of multiple malignancies, and our data suggests that SAMHD1 loss-of-function and/or downregulation, which occurs in a number of different malignancies (13), could confer a particular proliferative advantage to oncogenic Ras-driven tumour clones, contributing to poorer patient survival. Thus, utilising data from The Cancer Genome Atlas (TCGA), we next interrogated SAMHD1 expression in malignancies most commonly driven by oncogenic RAS mutations (Supplementary Fig. 3). Presence of an oncogenic KRAS allele in pancreatic (PAAD) (mutation rate, 59.2%), colorectal (COAD) (36.2%), and lung adenocarcinoma (LUAD) (26.7%), corresponded to significantly lower levels of SAMHD1 mRNA when compared to their KRAS wild-type (WT) counterparts (Fig. 3A). A similar trend was observed with NRASdriven tumours (Fig. 3B), but to a lesser degree for HRAS (Fig. 3C), which typically have lower mutation frequencies of below 27.9% and 10.1%, respectively. Critically, those patients with oncogenic KRAS-driven tumours and low SAMHD1 expression had a worse overall survival (OS) when compared to those with high SAMHD1 expression, in stark contrast to their WT KRAS counterparts which had similar OS regardless of SAMHD1 levels (Fig. 4A, B). This trend was particularly striking in KRAS-driven PAAD (Fig. 4B). Overall, these data suggest that SAMHD1 is a tumourigenic barrier in oncogenic RAS-driven malignancies, and accordingly, suggest overcoming this barrier may lead to a proliferation advantage for cancer cells and ultimately worse OS.
In this study we set out to test the hypothesis that SAMHD1 is involved in oncogeneinduced RS. We speculated this could be due to the catabolic role of SAMHD1 in dNTP homeostasis and/or the role in replication fork restart; our data indicates the former could play a more prominent role in this context, although further work is required to establish the precise mechanism. There is a well-established link between reduced dNTPs and oncogene-induced RS in early cancer development (2,4). Oncogenic signalling will elevate origin firing, consuming available dNTPs at a faster rate, whilst downregulating nucleotide  biosynthetic enzymes (5,6) and inducing oxidation of the dNTP pool (7), all of which will reduce available dNTPs needed for on-going DNA synthesis. Accordingly, oncogeneinduced RS in cultured cells can be reverted by nucleoside supplementation (5,6,8,9). The data presented here indicates that a reduction of dNTP catabolism via downregulation of the dNTPase SAMHD1, a well-documented consequence of which is expansion of the dNTP pool (12,13), can rescue oncogene-induced RS. The data presented here also indicates that this tumour suppressor role of SAMHD1 is clearly dictated by the oncogenic re-wiring present in cancer cells, offering an explanation for the differing proliferative phenotypes reported from loss of SAMHD1 and/or dNTP pool expansion (13).
To achieve sustainable long-term growth, oncogene-driven malignancies need to maintain sufficient dNTP pools, and this can be achieved through metabolic rewiring. This  has been demonstrated in the case of oncogenic BRAF (19) and KRAS-driven PAAD (20), which upregulate de novo dNTP synthesis via the pentose phosphate pathway. Our analysis of clinical data suggests downregulation of dNTP catabolism via SAMHD1 could be an important component of this rewiring. Critically, under these circumstances, this corresponds to worse patient survival. Altogether, this study establishes SAMHD1 as a tumourogenic barrier in these RAS-driven malignancies through promoting RS, and opens further avenues of investigation regarding the interplay of dNTP catabolism and oncogenesis.

Cell culture. Genetically modified normal human BJ fibroblast lines BJ-hTERT (BJ cells immortalised with hTERT), BJ-hTERT-ER (SV40 Early Region transformed BJ-hTERT cells),
and BJ-ER-RAS V12 (HRAS V12 -transformed BJ-hTERT-ER cells) were provided by William C. Hahn (Dana Faber Cancer Institute) and their generation has been reported previously (15). Of note, whilst the original publication referred to these cell lines being transformed with SV40 Large-T antigen (15), a later study clarified that they were instead constructed with the entire SV40 ER, encompassing both small T and large T (16). Cells were cultured in DMEM supplemented with 10% FBS and penicillin-streptomycin at 37°C with 5% CO2 in a humidified incubator. All cell culture reagents were purchased from Gibco/Thermo Fisher. Cells were typically maintained in culture for 1-2 months before thawing of a new batch and mycoplasma screening was performed routinely using the MycoAlert Mycoplasma Detection Kit (Lonza).

Western blot analysis.
Cells were scraped in lysis buffer (50 mM Tris-HCl pH 8, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% SDS) supplemented with cOmplete EDTA-free protease inhibitor cocktail (Roche) and Halt phosphatase inhibitor cocktail (Thermo Fisher). Samples were incubated on ice for 30 mins to 1 hour with occasional vortexing before centrifugation to pellet insoluble material. Protein concentration of the remaining soluble fraction was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher) and samples with equal total protein quantity were prepared with Laemmli Sample Buffer (Bio-Rad) before denaturation at 95 °C for 5 min. SDS-PAGE was performed using a Bio-Rad setup with Criterion TGX 4-20% gels (Bio-Rad) and proteins were transferred to a nitrocellulose membrane using Trans-Blot Turbo Transfer System (Bio-Rad), all according to the manufacturer's instructions. Membranes were blocked in Odyssey Blocking Buffer (Li-Cor) and probed with primary and then species-appropriate IRDye-conjugated secondary antibodies (Li-Cor) before visualization on an Odyssey Fc Imaging System (Li-Cor). Primary antibodies used in this study were anti-SAMHD1 (Abcam, ab128107; 1:1000), anti-gH2AX (Millipore, 05-636, 1:2000) and anti-β-actin (Abcam, ab6276; 1:5000). Densitometry analysis on acquired Western Blot images was performed using Image Studio Lite (Ver. 5.2.; Li-Cor), with signals subsequently normalised to β-actin signals and relative to that of siNeg-treated cells.
High content microscopy. Cells were seeded (10,000 per well) in black clear-bottomed 96well plates (BD Falcon) and allowed to grow for 24 hours. Cells were then fixed with 4% paraformaldehyde in PBS (Santa Cruz) supplemented with 0.5% Triton X-100 for 15 minutes. Wells were washed with PBS before staining cell nuclei with 1 µg/ml DAPI for 10 minutes. Wells were washed again with PBS and imaged on an ImageXpress highthroughput microscope (Molecular Devices) with a 20x objective. Images were analysed (nuclei counting and DAPI intensity measurements) with CellProfiler (21) (Broad Institute) and data handled in Excel (Microsoft) and plotted in Prism 8 (GraphPad).
Resazurin assay. Cells (48 hrs post siRNA transfection) were seeded into 96-well plates with 1500 cells per well in 100 µl medium and incubated for 4 days. Following incubation, resazurin (Sigma Aldrich, cat no. R7017) stock solution (1 mg/ml in PBS) was diluted in fresh medium and 100 µl added to wells to a final concentration of 0.1 mg/ml, and incubated for 2 hours. Resazurin reduction to resorufin, an indicator of cell viability, was subsequently measured at 530/590 nm (ex/em) using a Hidex Sense Microplate Reader. Fluorescent intensity for each well was normalised to the average of BJ-hTERT cell control wells on the same plate to give relative cell viability values.
Colony formation assay. Cells (48 hrs post siRNA transfection) were seeded onto 10 cm dishes and incubated for 10 days. For BJ-hTERT, 1000 cells were seeded per dish, whilst BJ-hTERT-ER and BJ-ER-RAS V12 , 500 cells were seeded per dish. Following incubation, colonies were fixed and stained with 4% methylene blue in methanol. Colonies were counted manually, and plating efficiency ((no. of colonies formed/no. of cells plated) x 100) calculated and normalized to the average of BJ-hTERT cell control wells.
For comparison of SAMHD1 mRNA levels in tumours containing WT RAS vs Mut RAS in Fig. 3, statistical significance was determined using one-way ANOVA Fisher's LSD test: