SAMHD1 regulates human papillomavirus 16 induced cell proliferation and viral replication during differentiation of oral keratinocytes

Human papillomaviruses induce a host of anogenital cancers, and also oropharyngeal cancer (HPV+OPC); HPV16 is causative in around 90% of HPV+OPC. Using TERT immortalized “normal” oral keratinocytes (NOKs) we have identified significant host gene reprogramming by HPV16 (NOKs+HPV16), and demonstrated that NOKs+HPV16 support late stages of the viral life cycle. Expression of the cellular dNTPase and homologous recombination factor SAMHD1 is transcriptionally regulated by HPV16 in NOKs, and here we demonstrate that E6 and E7 regulate expression of SAMHD1 at the transcriptional and post-transcriptional levels. CRISPR/Cas9 removal of SAMHD1 from NOKs and NOKs+HPV16 demonstrate that SAMHD1 controls cell proliferation of NOKs only in the presence of HPV16; deletion of SAMHD1 promotes hyper-proliferation of NOKs+HPV16 cells in organotypic raft cultures but has no effect on NOKs. Viral replication is also elevated in the absence of SAMHD1. This new system has allowed us to identify a specific interaction between SAMHD1 and HPV16 that regulates host cell proliferation and viral replication; such studies are problematic in non-immortalized primary oral keratinocytes due to their limited lifespan. To confirm the relevance of our results we repeated the analysis with human tonsil keratinocytes immortalized by HPV16 (HTK16) and observe the same hyper-proliferative phenotype following CRISPR/Cas9 editing of SAMHD1. Identical results were obtained with three independent CRISPR/Cas9 guide RNAs. The isogenic pairing of NOKs with NOKs+HPV16, combined with HTK16, presents a unique system to identify host genes whose products functionally interact with HPV16 to regulate host cellular growth in oral keratinocytes. Importance Head and neck cancer is the sixth most common cancer worldwide. The incidence of HPV+OPC has been rising steadily since the 1970s and has recently reached epidemic proportions, according to the WHO. Upwards of 70% of the 600,000 new OPC cases per year are HPV positive, with high-risk type 16 present in 90% of those incidences. A better understanding of the viral life cycle will facilitate the development of novel therapeutics to combat this ongoing epidemic, as well as other HPV positive cancers. Here we present a unique oral keratinocyte model to identify host proteins that specifically interact with HPV16. Using this system, we report that a cellular gene, SAMHD1, is regulated by HPV16 at the RNA and protein level in oral keratinocytes. Elimination of SAMHD1 from these cells using CRISPR/Cas9 editing promotes enhanced cellular proliferation by HPV16 in oral keratinocytes and elevated viral replication, but not in keratinocytes that do not have HPV16. Our study demonstrates a specific intricate interplay between HPV16 and SAMHD1 during the viral life cycle and establishes a unique model system to assist exploring host factors critical for HPV pathogenesis.

Introduction between the diseases. For example, the majority of HPV+OPC retain episomal genomes 48 replicating in an E1-E2 dependent manner whereas in cervical cancer most often the viral genome 49 is integrated into that of the host (7, 8). Such differences, and the ongoing epidemic of HPV+OPC, 50 prompted us to develop and characterize an oral keratinocyte model supporting late stages of the 51 HPV16 life cycle in organotypic raft cultures (9). This was done by introducing the HPV16 genome 52 into TERT immortalized "normal" oral keratinocytes (NOKs, generated from a 28 year old male 53 (10)) and carrying out organotypic raft cultures followed by confirmation of late stages of the viral 54 life cycle including E1^E4 and E2 expression as well as amplification of the HPV16 genome in 55 the differentiated epithelial cells. We used this model to determine host reprogramming induced 56 by HPV16 in oral keratinocytes and a large number of innate immune genes were shown to be 57 downregulated, as others have demonstrated before (9). One of the innate immune genes 58 predicted to be downregulated was SAMHD1 (Sterile alpha motif and histidine-aspartic domain 59 HD containing protein 1). To our knowledge this is the first time downregulation of SAMHD1 60 expression by HPV16 has been reported. 61 SAMHD1 is a triphosphohydrolase (dNTPase) enzyme, which regulates intracellular 62 levels of dNTPs and acts as an intrinsic immune response factor (11,12). To function as a 63 dNTPase, the protein forms a homo-tetramer, which is destabilized by phosphorylation. Each 64

SAMHD1 is downregulated in HPV16 containing oral keratinocytes. 120
Previous work from this lab developed and characterized an HPV16 life cycle model in 121 TERT immortalized oral keratinocytes (NOKs) (9). Initial comparison of two clonal cell lines 122 containing HPV16 that support late stages of the viral life cycle (NOKs+HPV16A and 123 NOKS+HPV16B) with the parental NOKs revealed a decrease in both SAMHD1 RNA and protein 124 expression ( Figures 1A and 1B, respectively). In the presence of HPV16, SAMHD1 RNA is 125 expressed at a 50% lower level than the NOKs ( Figure 1A), which validates our previous 126 observations in RNAseq analysis (9). Protein levels are decreased correspondingly ( Figure 1B). 127 This was consistent in three individual repeats, which were quantified ( Figure 1B). 128 To identify whether a single HPV16 protein is responsible for the downregulation of 129 SAMHD1, stable cell lines were generated expressing individual viral proteins E6, E7 and 130 expressing both oncoproteins together (NOKs+E6, NOKs+E7, NOKs+E6E7). Analysis of 131 SAMHD1 expression in these cells reveals a downregulation at an RNA level in NOKs+E6E7, but 132 not when the viral proteins are expressed individually (Figure 2A), suggesting a synergistic 133 mechanism of action in SAMHD1 transcriptional repression. This decrease in RNA corresponded 134 to a lower level of SAMHD1 protein in NOKs+E6E7 when compared with NOKs ( Figure 2B). A 135 downregulation in SAMHD1 protein was also observed in cells expressing single HPV16 136 oncoproteins, suggesting a post-transcriptional mechanism for downregulating SAMHD1 137 expression ( Figure 2B). Therefore, E6 and E7 use a combination of transcriptional and post-138 transcriptional mechanisms to downregulate SAMHD1 expression. 139

Downregulation of SAMHD1 by HPV16 is maintained during differentiation. 140
In order to assess the expression of SAMHD1 during the HPV16 life cycle, NOKs and 141 NOKs+HPV16 were differentiated by organotypic 'raft' culture. The rafts were then fixed and 142 subject to immunofluorescent staining to determine SAMHD1 levels and localization in 143 differentiated epithelia. SAMHD1 is expressed in NOKs throughout the organotypic section 9 ( Figure 3A), whereas fewer cells are stained when HPV16 is present (NOKs+HPV16A and 145 NOKS+HPV16B). Three independent organotypic raft cultures were stained and SAMHD1 levels 146 quantitated using a Vectra ® Polaris ™ automated imaging system; the difference in staining 147 between HPV negative and positive NOKs cells is significant ( Figure 3B). Furthermore, the 148 presence of HPV16 in NOKs leads to the loss of SAMHD1 expression in the upper layers of the 149 epithelium ( Figure 3C). This was quantified by measuring the "height" to which SAMHD1 is 150 expressed in the rafts using a Vectra ® Polaris ™ automated imaging system; in the case of HPV16  The downregulation of SAMHD1 may promote the HPV16 life cycle as this would increase 162 the nucleotide pool in the epithelium that could assist with viral genome replication. We therefore 163 generated NOKs, NOKs+HPV16 (A and B clones) and HTK16 with exogenously overexpressed 164 SAMHD1 to investigate the ability of this excess SAMHD1 to block genome amplification (Figure  165 4A). These cells were subjected to organotypic raft cultures and then fixed and stained with a V5 166 antibody ( Figure 4B). The exogenous V5 tagged SAMHD1, like endogenous SAMHD1 (Figure 3), 167 is not detectable in the upper layers of the differentiated epithelium when HPV16 is present. This 168 was repeated and quantitated using a Vectra ® Polaris ™ automated imaging system ( Figure 4C). 169 Due to the lack of SAMHD1 expression in the upper layers of the differentiated epithelium there 170 was no change in the viral life cycle in the cells over expressing V5 tagged SAMHD1 ( Figure S1). 171 to control cells (NOKs by themselves) and to cells overexpressing SAMHD1 that were described 183

Deletion of SAMHD1 results in hyper-proliferation of HPV16 positive keratinocytes in
in Figure 4 ( Figure 5B). Triplicate rafts were sectioned and stained with H&E, before measurement 184 using a Vectra ® Polaris ™ automated imaging system. Quantification revealed that this increase in 185 section thickness by HPV16 was significant ( Figure 5C). This hyperproliferation was not due to 186 the SAMHD1 depleted cell lines growing quicker in monolayer cells ( Figure 5D). Additionally, 187 there was no difference in HPV16 genome copy number in monolayer cells that under express 188

SAMHD1 (Figures 5E&5F). 189
To investigate whether this hyper-proliferation was specific to a certain layer of the 190 differentiated culture, rafts were treated with BrdU for the final 16 hours of culture, before being 191 fixed and sectioned. BrdU staining highlights the cells actively dividing in those 16 hours. In NOKs 192 cells there was no increase in BrdU positive cells in the absence of SAMHD1, Figure 6A  To further confirm the proliferative nature of these cells, cells were stained with Cyclin E, 206 a protein expressed in S phase (28). In NOKs cells there is no increase in Cyclin E staining without 207 SAMHD1, while there is increased Cyclin E staining in the absence of SAMHD1 in 208 NOKs+HPV16A ( Figure 7A). This increase in Cyclin E staining in the absence of SAMHD1 was 209 also observed in HTK16 cells lacking SAMHD1 when compared with parental HTK16. These 210 results were repeated and Cyclin E positive cells quantitated using a Vectra ® Polaris ™ automated 211 imaging system and a summary of the results is presented graphically in Figure 7B. In 212 NOKs+HPV16 and HTK16 there is a statistically significant increase in Cyclin E expression in the 213 absence of SAMHD1 while no such increase is observed in NOKs. 214 As there was a hyper-proliferation of the HPV16 positive cells in the absence of SAMHD1, 215 and an increased thickening of the differentiated epithelium, we investigated whether there was 216 an enhanced amplification of the HPV16 genome in the cells lacking SAMHD1. To do this FISH 217 analysis for the viral DNA was carried out and the intensity of the staining was measured using 218 the Vectra ® Polaris ™ automated imaging system; whole stained sections were scanned 219 computationally and the intensity and localization of staining measured relative to a negative 220 control (NOKs) and a positive control (NOKs+HPV16). Figure 8A shows representative images 12 from the staining. Figure 8B summarizes the results of these experiments and in both NOKs 222 clones that have HPV16 and SAMHD1 removed there is a statistically significant enhanced FISH 223 signal detected indicating increased HPV16 replication. There was also an increased signal in 224 HTK16 in the absence of SAMHD1. In 8Ai it is clear that there is no signal in the NOKs while in 225 8Aii and 8Aiii there is an enhanced FISH signal in the absence of SAMHD1. In HTK16 (8Aiv) there 226 is also a slight increase in signal. Please note that the measurement of the signal is quantitative 227 and non-subjective while the images are representative. There was a hint of viral genome 228 amplification in the basal layers of the HPV16 containing cells in the absence of SAMHD1. In 229 Figure 8C we have increased the exposure and it is clear that there is some HPV16 DNA signal 230 in the absence of SAMHD1 prior to the amplification stage of the viral life cycle. This was difficult 231 to quantitate but does suggest that viral genome replication is higher even in the basal layers of 232 the differentiating epithelium. 233 Finally, we reproduced our results with an additional two SAMHD1 guide RNAs that target 234 different sequences in the SAMHD1 gene. Figure S1A details the knock down of SAMHD1 235 expression by these additional CRISPR/Cas9 targeting sequences. Figure S1B   Recently we demonstrated one potential unique use of this system. We expressed the 268 HPV16 E2 protein in NOKs and determined host gene regulation by E2 in these cells that had a 269 highly significant overlap with the genes regulated by the entire HPV16 genome (44). Therefore, 270 this system has allowed us to identify and characterize a new function for the E2 protein; 271 regulation of host gene transcription during infection. This study would not have been possible in 272 primary keratinocytes as E2 would not immortalize these cells during selection. Importantly, in our 273 studies using this system, we have validated gene expression changes in human tonsil 274 keratinocytes immortalized by HPV16 (HTK16) and also in HPV16 positive head and neck 275

cancers. 276
Here we report another utility of this system. We demonstrate that removal of SAMHD1 277 from NOKs+HPV16 and HTK16 results in hyper-proliferation of these cells in organotypic raft 278 cultures. This increase in proliferation was confirmed by a "thickening" of the raft culture, and an 279 increase in BrdU and Cyclin E positive cells in the absence of SAMHD1 (both S phase markers). 280 We also observed an elevation in viral genome amplification in the absence of SAMHD1. The 281 parental NOKs cells exhibit no increase in BrdU or Cyclin E positive cells, demonstrating that the 282 increased proliferation was due to the presence of HPV16. If we had carried out these studies in 283 the absence of the NOKs cells, then we would not have been able to determine whether the 284 deletion of SAMHD1 expression by itself was proliferative for the cells. Therefore, this system has 285 allowed us to identify an interaction between SAMHD1 and HPV16 that regulates the proliferation 286 of NOKs+HPV16 and HTK16. The addition of HTK16 demonstrates that the SAMHD1 deletion 287 phenotype is retained in cell lines immortalized by HPV16. Therefore, the combination of NOKs, 288 NOKs+HPV16 and HTK16 represents an excellent system for identifying interactions between 289 host proteins and HPV16 that have an effect on host cell growth and viral replication. 290 SAMHD1 is downregulated in NOKs+HPV16 and HTK16 and this is at least partially due 291 to the viral oncogenes E6 and E7. As SAMHD1 is a dNTPase this downregulation could be 292 important for promoting viral replication, particularly at the amplification stage of the viral life cycle 293 in the differentiated epithelium. To counteract this down-regulation we overexpressed exogenous 294 SAMHD1 but notably this protein was not expressed in the differentiated layers of the epithelium 295 where genome amplification occurs. As E6 and E7 can both regulate SAMHD1 post-296 transcriptionally it is possible that the viral oncogenes remove SAMHD1 expression to allow viral 297 genome amplification. 298 SAMHD1 is also a homologous recombination (HR) factor and is involved in recruiting 299 MRE11 to damaged DNA (45, 46). This role of SAMHD1 could also play an important role in its 300 interaction with HPV16 as this virus recruits a host of HR factors to its replicating DNA and it is 301 proposed that HPV16 using HR during viral replication in order to amplify its genome. However, 302 unlike downregulation of other HR factors (5, 39, 40), downregulation of SAMHD1 boosts viral 303 genome amplification and also disrupts the equilibrium between the virus and the host cell. 304 The reason for the hyper-proliferation of HPV16 containing keratinocytes in the absence 305 of SAMHD1 is not clear. This is not observed in monolayer cultures therefore only becomes 306 apparent during organotypic raft cultures indicating that there may be some involvement of the 307 collagen-fibroblast plug used to generate the differentiating cells. This collagen-fibroblast plug 308 could mimic a stroma-epithelial cell interaction and there is a known cross-talk between HPV and 309 the stroma (47). It is also apparent that there is not an increase in HPV16 genomes in the absence 310 of SAMHD1 in monolayer cells so the difference is not due to an initial increased viral genome 311 copy number in these cells upon rafting. 312 SAMHD1 is a restriction factor for HIV and other DNA viruses including HBV (26), HSV1 313 (24) and EBV (http://dx.doi.org/10.2139/ssrn.3255560) and here we demonstrate that SAMHD1 314 is also a restriction factor for HPV16. Not only does the absence of SAMHD1 promote hyper-315 proliferation of the infected cells, it also stimulates an enhanced amplification of the HPV16 316 genome during the terminal stages of differentiation. In addition, the FISH staining for the HPV16 317 genome in Figure 8C suggests that in NOKs+HPV16 in the absence of SAMHD1 there is an 318 increase in viral signal in the basal layers of the epithelium. There is also an indication of this in 319 HTK16. This is hard to quantitate but does suggest that during differentiation the increased 320 replication of the viral genome perhaps starts early in the differentiation process. 321 Our isogenic NOKs system for investigating the HPV16 life cycle has been essential at 322 revealing a specific interaction between the virus and SAMHD1 that controls host cell proliferation. 323 In addition, the enhanced genome amplification in the absence of SAMHD1 demonstrates that 324 SAMHD1 is a restriction factor for HPV16. The virus clearly downregulates SAMHD1 expression 325 but retains a level that is required for controlling both host proliferation and viral genome 326 amplification. Perhaps downregulation of SAMHD1 is required to generate an enhanced pool of 327 nucleotides that would promote viral genome amplification, but that SAMHD1 homologous 328 recombination function is also required to control the levels of viral genome replication.

Southern Blot 434
Total cellular DNA was extracted using a phenol chloroform method and 5 micrograms 435 digested with either SphI or HindIII, to linearise the HPV16 genome or leave episomes intact, 436 respectively. All digests included DpnI to ensure that all input DNA was digested and not 437 represented as replicating viral DNA. Digested DNA was separated by electrophoresis of a 0.8% 438 agarose gel, transferred to a nitrocellulose membrane and probed with radiolabeled (32-P) HPV16 439 genome. This was then visualized by exposure to film for 24 or 72 hours. 440

Statistics 441
Standard error was calculated from three independent experiments and significance determined 442 using a student's t-test.        Results are expressed as fold change from that observed in parental NOKs and represent the 618 average of three independent experiments. B) Western blot analysis was carried out on protein 619 extracted from the NOKs (lane 1) and HPV16-containing NOKs (lanes 2 and 3). GAPDH is 620 shown as an internal control. Western blots were visualized and quantitated using a  containing NOKs (lanes 2 and 3) and NOKs expressing HPV16 oncoproteins (lanes 4, 5 and 6).

633
GAPDH is shown as an internal control. Western blots were visualized and quantitated using a 634 Licor and calculated relative to parental NOKs. Data in A and B represents the average of 3 635 independent experiments and error bars indicate standard error of the mean. * indicates p<0.05.  NOKs+HPV16A (ii), NOKs+HPV16B (iii) and human tonsil keratinocytes immortalized by HPV16 646 (HTK16, iv), were differentiated in culture, and resulting sections stained for SAMHD1. Arrows 647 indicate level at which SAMHD1 disappears. B) SAMHD1 positive cells of three sections from 648 two individual rafts were quantified computationally and the measurements averaged. 649 Immunofluorescence was quantified, using a Vectra ® Polaris ™ automated imaging system, 650 where whole stained sections were scanned computationally and the intensity calculated 651 compared to a negative background control (secondary antibody only) and a positive 652 localization control (DAPI). The same imaging parameters were used for each slide. For each 653 sample, two sections from three individual rafts were scanned, to generate average values 654 shown in this graph. Error bars indicate standard error of the mean and * indicates p<0.05.

655
C)The overall thickness and region of SAMHD1 positivity from three sections from two individual 656 rafts was quantified computationally using a Vectra ® Polaris ™ automated imaging system and 657 the measurements averaged. Error bars indicate standard error of the mean and * indicates 658 p<0.05. blasticidin. Expression of V5-tagged SAMHD1 in these cell lines was confirmed by western blot.

673
HTK16 samples (lanes 7 and 8) were run on a separate gel. All of the control cells have been 674 selected following infection with empty pLX304. B) Representative images of raft sections 675 stained for exogenous SAMHD1, using an anti-V5 antibody. C) Sections were stained for 676 exogenous SAMHD1, using an anti-V5 antibody. Immunofluorescence from three sections each 677 from two individual differentiated cultures was then quantified using a 678 Vectra ® Polaris ™ automated imaging system and the measurements averaged. Error bars 679 indicate standard error of the mean and * indicates p<0.05.