The present invention relates to the use of purine nucleoside phosphorylase (PNP) inhibitors for the treatment of cancer and/or for the treatment of HIV-2 infection. In particular, the invention provides a method for stratifying cancer patients to ensure the most appropriate therapy is given. The invention also provides a method of treating these patients. More specifically, in one aspect the invention provides for the use of measuring SAMHD1 levels to stratify and/or identify cancer patients suitable for treatment with PNP inhibitors.

Cells require deoxyribonucleoside triphosphates (dNTPs) for DNA synthesis, to replicate their genome and divide. Intracellular dNTP concentrations are controlled by dNTP synthesis and degradation pathways. Generation of the dNTP pool in the cell occurs through the de novo pathway and the salvage pathway, both requiring numerous enzymes for dNTP synthesis. These pathways are critical to control the dNTP pool and to maintain genome integrity. Abnormally high or low dNTP levels at the wrong time can lead to DNA damage, increased mutation rates, defective DNA repair and possibly cell death.

In the de novo pathway, dNTPs are synthesised from intracellular precursors. The enzyme ribonucleotide reductase catalyses the rate-limiting step and converts ribonucleoside diphosphates into deoxyribonucleoside diphosphates. The salvage pathway involves uptake of deoxyribonucleosides (dNs) from the extracellular environment, followed by intracellular phosphorylation by cytosolic and mitochondrial kinases to form dNTPs.

SAMHD1 is a deoxyribonucleoside triphosphohydrolase (dNTPase) and therefore is an important enzyme involved in regulating dNTP pools. SAMHD1 cleaves all four dNTPs into the corresponding dN and inorganic triphosphate. Its roles in virus restriction and auto-immunity have been extensively studied over the last few years. Mutations in the SAMHD1 gene cause Aicardi-Goutieres syndrome, a disease characterised by the production of type I IFN. The mechanism by which SAMHD 1 prevents the production of type I IFN is not entirely clear although evidence suggests that SAMHD 1 modulates the expression of the retrotransposon LINE-1. The SAMHD1 gene is mutated in several types of cancer, including in chronic lymphocytic leukaemia (CLL), T cell prolymphocytic leukaemia (T-PLL) and colon cancer. SAMHD1 mutations lead to higher dNTP pools, which may increase mutation rates, or simply give cancer cells a growth advantage. Since SAMHD 1 has also been shown to be involved in DNA repair and/or replication, it is not clear why SAMHD1 mutations are observed in cancer cells (Daddacha et al., 2017).

Purine nucleoside phosphorylase (PNP) is an important enzyme in nucleotide metabolism, responsible for deoxyguanosine (dG) degradation into guanine, which is further catabolised into uric acid that is released by cells.

Genetic PNP deficiency causes immunodeficiency and results in loss of T cells and in some patients also affects B cell function. This observation initiated the development of PNP inhibitors to use against leukaemias. Targeting enzymes involved in the nucleotide pathway is a strategy to kill cancer cells. For example, ribonucleotide reductase (RNR) and inosine monophosphate dehydrogenase (IMPDH) are both enzymes of the nucleotide pathway that are targeted - the RNR inhibitor COH29 and the IMPDH inhibitor AVN944 have both been tested as treatments for patients with refractory solid tumours.

The small molecule forodesine (also known as Imunicillin H or BCX-1777) was developed to inhibit PNP. Upon forodesine treatment, dG accumulates intracellularly and is phosphorylated to deoxyguanosine triphosphate (dGTP). The resulting imbalance in dNTP pools is predicted to cause cell death and to eliminate leukaemic cells (Bantia et al., 2001). However, forodesine had substantial activity only in a small subset of patients with B or T cell malignancies and showed poor remission rates (Gandhi et al., 2005; Balakrishnan et al., 2006; Dummer et al., 2014; Balakrishnan et al., 2010; Maruyama et al., 2018; Alonso et al., 2009; Gandhi et al., 2007). Therefore, forodesine appeared unsuitable as an effective leukaemia treatment and was not developed further.

There is therefore a need for improved therapies to treat cancer, and in particular to treat B and T cell malignancies. The lack of good biomarkers to indicate which patients will respond to a particular therapy is a hindrance for cancer treatment. In particular, there is an unmet need for novel biomarkers that have value to facilitate the implementation of more targeted therapeutic strategies in cancer patients that could improve overall clinical outcomes.

It is therefore an aim of the present invention to provide biomarkers that may be used to select cancer patients for a particular therapy. Such biomarkers will allow therapies to be targeted to those most likely to benefit. Such biomarkers will enable stratification of patients to identify those most likely to respond to a particular course of treatment and those that probably will not. This will allow patients to be given the most appropriate treatment quickly and avoid administering treatment which will not be effective and/or has an undesirable side effect. Not only are there benefits to the patient, but it is clear that there will also be significant cost savings in identifying and selecting therapy-responsive patients.

SAMHD1 -mediated dNTP pool regulation can also restrict retroviral replication, by reducing intracellular deoxynucleoside triphosphate levels below those which are required for reverse transcription. Human immunodeficiency virus type 2 (HIV-2) and some simian immunodeficiency viruses (SIVs) encode the accessory protein viral protein X (Vpx) that counteracts nuclear SAMHD 1, by inducing its ubiquitin-mediated proteosomal degradation (Hofmann et al., 2012). HIV-2 infection is usually treated with conventional antiretroviral therapy (ART), which focusses on targeting viral elements to inhibit or reduce viral replication. HIV-2 is intrinsically resistant to non nucleoside reverse transcriptase inhibitors, and infection can still lead to AIDS. There remains a need for alternative therapies to treat HIV-2 infection. Targeting of an HIV- 2 infected cell is one such alternative therapeutic strategy. This would prevent viral replication, as the virus hijacks cellular machinery for its replication, and could be used either as an alternative or as a complement to ART. Such treatment would reduce viral burden on the infected subject, and therefore reduce the risk of progressing to AIDS, benefitting the patient who would therefore be less likely to suffer from secondary indications as a result of developing AIDS, and reducing medical costs arising from AIDS. The present inventors have surprisingly found that SAMHD1 enzyme status in cancer cells is a predictive marker for a therapeutic response to a PNP inhibitor in cancer patients.

The present inventors have also surprisingly found that the treatment of cells infected with HIV-2 with a PNP inhibitor can induce the death of those infected cells.

In a first aspect of the invention, there is provided a method for identifying cancer patients who are predicted to respond to therapy with a PNP inhibitor, the method comprising:

(i) obtaining a sample of cancer cells from a subject;

(ii) detecting the presence and/or level of SAMHD1 enzyme and/or detecting the level of activity of the SAMHD1 enzyme in the cancer cells; and

(iii) predicting that if the SAMHD1 enzyme is absent or if the SAMHD 1 enzyme is present at a reduced level, and/or predicting that if the SAMHD1 enzyme has no activity or has a reduced activity, then a PNP inhibitor will be effective in treating the cancer in the subject.

In another aspect of the invention, there is provided a method for identifying cancer patients who are predicted to respond to therapy with a PNP inhibitor, the method comprising:

(i) obtaining a sample of cancer cells from a subject;

(ii) detecting the presence or absence of a mutation in the SAMHD1 gene in the cancer cells; and

(iii) predicting that if a mutation in the SAMHD1 gene is present, then a PNP inhibitor will be effective in treating the cancer in the subject.

In another aspect of the invention, there is provided a method for identifying cancer patients who are predicted to respond to therapy with a PNP inhibitor, the method comprising:

(i) obtaining a sample of cancer cells from a subject;

(ii) detecting the presence or absence of a mutation in the SAMHD1 gene in the cancer cells; and (iii) predicting that if a mutation in the SAMHD1 gene is present which results in no SAMHD 1 enzyme or a reduced level of SAMHD1 enzyme in the cancer cells and/or predicting that if a mutation in the SAMHD1 gene is present which results in no SAMHD1 enzyme activity or a reduced level of SAMHD1 enzyme activity in the cancer cells, then a PNP inhibitor will be effective in treating cancer in the subject.

There may be no detectable SAMHD1 enzyme and/or no detectable SAMHD1 enzyme activity in the cancer cells in the sample, in which case a PNP inhibitor is predicted to be effective in treating the cancer in the subject.

The SAMHD1 enzyme may be detectable, but it may be present at a level that is reduced compared to the level of the SAMHD1 enzyme in non-cancer cells, in which case a PNP inhibitor is predicted to be effective in treating the cancer in the subject. For example, for it to be predicted that a PNP inhibitor will be effective in treating cancer, the SAMHD1 enzyme in the cancer cells may be present at about 95% or less of the level of the SAMHD1 enzyme in non-cancer cells, or about 90% or less, or about 85% or less, or about 80% or less, or about 75% or less, or about 70% or less, or about 65% or less, or about 60% or less, or about 55% or less, or about 50% or less, or about 45% or less, or about 40% or less, or about 35% or less, or about 33% less, or about 30% or less, or about 25% or less, or about 20% or less, or about 15% or less, or about 10% or less, or about 5% or less, or about 2.5% or less, or about 3% or less, or about 2% or less, or about 1% or less of the level of the SAMHD1 enzyme in non cancer cells. Preferably, the SAMHD1 enzyme in the cancer cells may be present at about 5% or less of the level of SAMHD1 enzyme in non-cancer cells. The non-cancer cells may be from the same subject and optionally from the same tissue as the cancer cells of step (i). Alternatively, the non-cancer cells may be from a different subject, and optionally from the same tissue as the cancer cells of step(i) but from a different subject.

The SAMHD1 enzyme may be detectable, but its activity may be reduced compared to the activity of the SAMHD1 enzyme in non-cancer cells, in which case a PNP inhibitor is predicted to be effective in treating cancer in the subject. For example, for it to be predicted that a PNP inhibitor will be effective in treating cancer, the level of activity of the SAMHD1 enzyme in the cancer cells may be about 95% or less of the activity of the SAMHD1 enzyme in non-cancer cells, or about 90% or less, or about 85% or less, or about 80% or less, or about 75% or less, or about 70% or less, or about 65% or less, or about 60% or less, or about 55% or less, or about 50% or less, or about 45% or less, or about 40% or less, or about 35% or less, or about 33% less, or about 30% or less, or about 25% or less, or about 20% or less, or about 15% or less, or about

10% or less, or about 5% or less, or about 2.5% or less, or about 3% or less, or about 2% or less, or about 1% or less of the activity of the SAMHD1 enzyme in non-cancer cells. Preferably, the SAMHD 1 enzyme in the cancer cells may have about 5% or less of the activity of SAMHD1 enzyme in non-cancer cells. The non-cancer cells may be from the same subject and optionally from the same tissue as the cancer cells of step (i). Alternatively, the non-cancer cells may be from a different subject and optionally from the same tissue but from a different subject as the cancer cells of step (i).

Mutations in SAMHD1 as referred to herein relate to the sequence found in UniProt accession Q9Y3Z3. The mutation in the SAMHD1 gene may be any loss-of-function mutation. For example, the mutation in the SAMHD1 gene may be any nucleotide mutation that induces a change of amino acids in the catalytic domain (HD domain) of the SAMHD1 enzyme. Additionally or alternatively, the mutation in the SAMHD1 gene may result in the SAMHD1 enzyme having an amino acid substitution comprising or consisting of one or more of M1K, D 16Y, H123P, R143X/H/C, R145Q, Q149X/Q, Y155C, P158S, I201N, H206R, G209S, L244F, M254V, R290C, I300L, E355K, T365P, L369S, R371H, M385V, A389T, L431S, R442X, R451C/L, L493R, V500G, D501Y, C522X, F545L, Q548X and W572X. The amino acid substitution may be one or more of R145Q, 120 IN, H206R, A389T, and W572X. Additionally or alternatively, the mutation in the SAMHD1 gene may result in the deletion of one or more amino acids in the SAMHD1 enzyme, for example a deletion of amino acid residues 120-123 and/or amino acid residues 565-569. Additionally or alternatively, the mutation in the SAMHD1 gene may be any nucleotide deletion in the gene leading to a frameshift.

Advantageously, the methods of the invention allow subjects with cancer to be stratified into those which are expected to respond to therapy with a PNP inhibitor and those that are predicted not to respond to therapy with a PNP inhibitor or predicted to respond poorly to therapy with a PNP inhibitor. SAMHD1 enzyme levels may be detected by determining SAMHD1 protein levels in the sample, or by determining SAMHD1 mRNA levels in the sample. SAMHD 1 protein levels may be determined by directly measuring the protein and/or indirectly by measuring the activity of the protein.

SAMHD1 enzyme levels may be detected by any suitable method. For example, the level may be determined by immunoassays, spectrometry, western blot, ELISA, immunoprecipitation, slot or dot blot assay, isoelectric focusing, SDS-PAGE, antibody microarray, immunohistological staining, radioimmunoassay (RIA), fluoroimmunoassay, an immunoassay using an avidin-biotin or streptoavidin-biotin system, chromatographic techniques (i.e. immunoaffinity chromatography), flow cytometry etc. and combinations. Other methods may also be used. These methods are well known to the person skilled in the art.

SAMHD1 enzyme levels may be detected by determining the expression level of genes encoding the SAMHD1 protein. Nucleic acid-based techniques for assessing expression are well known in the art and include, for example, hybridization or amplification assays that include, but are not limited to, Southern or Northern analyses, polymerase chain reaction analyses (PCR or reverse transcription polymerase chain reaction) and probe arrays. mRNA may be assayed directly, or it may be first transcribed into cDNA, which can then serve as template for multiple rounds of transcription by the appropriate RNA polymerase.

Samples may be analysed by means of a biochip. Biochips generally comprise solid substrates and have a generally planar surface to which a capture reagent (also called an adsorbent or affinity reagent) is attached. Frequently, the surface of a biochip comprises a plurality of addressable locations, each of which has the capture reagent bound there. Protein biochips are biochips adapted for the capture of polypeptides. Many protein biochips are described in the art.

The sample used in the method of the invention may be a bodily fluid such as plasma, serum, whole blood, saliva, lymph, or cerebrospinal fluid. Alternatively, the sample may be a biopsy sample, such as a skin sample. Alternatively, the sample may an isolated cell sample which may refer to a single cell, multiple cells, more than one type of cell, cells from tissues, cells from organs and/or cells from tumours. The method of the invention may include the step of taking the sample from the subject. Alternatively, the method of the invention may not include the step of taking the sample, but instead the sample may be provided after it has been taken from the subject.

The method of the invention may be carried out in vitro.

The subject may be a mammal and is preferably a human, but may alternatively be a monkey, ape, cat, dog, cow, horse, rabbit or rodent.

In another aspect of the invention, there is provided a method of treating cancer in a subject in need thereof, comprising administering a PNP inhibitor to a subject with:

(i) a reduced level of SAMHD 1 enzyme or no SAMHD 1 enzyme in cells of the cancer to be treated;

(ii) reduced SAMHD 1 enzyme activity or no SAMHD 1 enzyme activity in cells of the cancer to be treated; and/or

(iii) a mutation in the SAMHD 1 gene in cells of the cancer to be treated.

In a further aspect the invention provides the use of a PNP inhibitor to treat a cancer in a subject, wherein the subject has:

(i) a reduced level of SAMHD 1 enzyme or no SAMHD 1 enzyme in cells of the cancer to be treated;

(ii) reduced SAMHD 1 enzyme activity or no SAMHD 1 enzyme activity in cells of the cancer to be treated; and/or

(iii) a mutation in the SAMHD 1 gene in cells of the cancer to be treated.

In a yet further aspect the invention provides a PNP inhibitor for use in treating a cancer in a subject, wherein the subject has:

(i) a reduced level of SAMHD 1 enzyme or no SAMHD 1 enzyme in cells of the cancer to be treated;

(ii) reduced SAMHD 1 enzyme activity or no SAMHD 1 enzyme activity in cells of the cancer to be treated; and/or

(iii) a mutation in the SAMHD 1 gene in cells of the cancer to be treated. In another aspect of the invention, there is provided a method of treating cancer in a subject in need thereof, the method comprising:

(i) identifying a subject predicted to respond to therapy with a PNP inhibitor according to any one of the methods described above; and

(ii) administering a PNP inhibitor to the subject.

In another aspect of the invention, there is provided a method of selecting a subject for treatment for cancer, the method comprising:

(i) obtaining a sample of cancer cells from the subject;

(ii) analysing the sample for the presence and/or level of SAMHD 1 enzyme and/or analysing the sample for the presence and/or level of SAMHD 1 enzyme activity; and

(iii) selecting the subject for treatment if the SAMHD1 enzyme is absent or if the SAMHD1 enzyme is present at a reduced level and/or if there is no SAMHD1 enzyme activity or the SAMHD 1 enzyme activity is reduced.

The treatment may be the administration of a PNP inhibitor.

In another aspect of the invention, there is provided a method of selecting the most appropriate treatment for a subject with cancer, the method comprising:

(i) obtaining a sample of cancer cells from the subject;

(ii) detecting the level and/or activity of SAMHD1 enzyme; and

(iii) choosing, and optionally administering, treatment based on the level and/or activity of SAMHD 1 enzyme detected.

The treatment may be the administration of a PNP inhibitor.

In another aspect of the invention, there is provided a use of a method of the invention to detect the level of SAMHD1 enzyme in a sample of cancer cells to determine the prognosis for a subject with cancer.

The cancer may be selected from the group consisting of anal cancer, bladder cancer, bone cancer, brain cancer, breast cancer, cervical cancer, colorectal cancer, colon cancer, endometrial cancer, head and neck cancer, leukaemia, liver cancer, lung cancer, kidney cancer, mouth cancer, ovarian cancer, pancreatic cancer, prostate cancer, skin cancer, rectal cancer, stomach cancer, testicular cancer, throat cancer, thyroid cancer, urethral cancer, vulvar cancer, and any combination thereof. In one embodiment, the cancer is colon cancer.

The cancer may be leukaemia, for example acute myeloid leukaemia (AML), chronic myeloid leukaemia (CML), acute lymphocytic leukaemia (ALL), chronic lymphocytic leukaemia (CLL), T-cell prolymphocytic leukaemia (T-PLL) and/or hairy cell leukaemia. In one embodiment, the cancer is chronic lymphocytic leukaemia (CLL).

The PNP inhibitor may be selected from the group consisting of forodesine (also known as Immucilin H or BCX-1777), BCX-34 (also known as peldesine 9-(3- Pyridylmethyl)-7H-9-deazaguanine), BCX-4208 (also known as Ulodesine or DADMe-Immucilin-H), CI-1000 (also known as 2-amino 3,5-dihydro-7-(3- thienylmethyl)-4H-pyrrolo[3,2-£ ]-pyrimidin-4-one HC1), 9-deazaguanine, homo- DFPP-DG, 6C-DFPP-DG, and any combination thereof. In one embodiment, the PNP inhibitor comprises or consists of forodesine, homo-DFPP-DG and/or 6C-DFPP-DG.

In one embodiment, the subject may have already been diagnosed with cancer before a method of the invention is undertaken. The diagnosis may be based on an assessment of one or more of clinical presentation, pathology, and other biomarker levels, for example the presence of CD5 and/or CD 19 proteins on the surface of the cancerous cells in the sample.

In another aspect of the invention, there is provided a PNP inhibitor for use in treating cancer and/or a HIV-2 infection wherein the PNP inhibitor selectively kills cells in the subject, wherein the cells selectively killed have a reduced level of SAMHD 1 enzyme or no SAMHD1 or reduced SAMHD1 enzyme activity or no SAMHD1 enzyme activity.

The invention may provide a PNP inhibitor for use in treating an HIV-2 infection in a subject.

In a further aspect of the invention, there is provided a use of a PNP inhibitor to treat an HIV-2 infection in a subject. In yet a further aspect of the invention there is provided a use of a PNP inhibitor in the manufacture of a medicament for the treatment of an HIV-2 infection. In another aspect of the invention, there is provided a method of treating an HIV-2 infection is a subject comprising administering to the subject a therapeutically effective amount of a PNP inhibitor.

The skilled man will appreciate that preferred features of any one embodiment and/or aspect of the invention may be applied to all other embodiments and/or aspects of the invention.

The present invention will be further described in more detail, by way of example only, with reference to the following figures in which:

Figure 1 demonstrates that deoxyribonucleosides (dNs) are toxic in SAMHD1- deficient cells. (A) Mouse embryonic fibroblasts (MEFs), bone marrow-derived macrophages (BMDMs), and Aicardi-Goutieres syndrome (AGS) patient-derived human fibroblasts (hFibroblasts) were treated with a mix of all four dNs. MEFs were cultured with 0.8 mM of each dN for 48 hours. BMDMs and fibroblasts were treated with 0.5 mM of each dN for 24 hours. Cell viability was determined by CellTiter-Glo assay. For each genotype, values from untreated control cells were set to 100%. Data from triplicate measurements are shown with mean ± SD. P-values determined with unpaired t-tests (MEFs and BMDMs) or one-way ANOVA (hFibroblasts) are indicated. (B, C) SAMHD1 expression was reconstituted in Samhd ' BMDMs. Cells of the indicated genotype were infected with a retrovirus expressing mouse SAMHD1 or empty control retrovirus. Cells were then treated with 0.5 mM of each dN for 24 hours. (B) Cell viability was tested as in (A). Dots represent measurements using cells from different animals and are shown with mean ± SD. P-values determined with two- way ANOVA are indicated. (C) SAMHD1 expression was tested by Western blot in BMDMs from three mice per genotype. b-Actin served as a loading control. (D) BMDMs were treated with individual dNs and combinations of dNs. Cells were cultured with 0.5 mM of the indicated dN(s) for 24 hours and viability was analysed as in (a). Data from biological triplicates were averaged and are represented as a heat map. (E) BMDMs were treated with 0.5 mM dG for 24 hours. Brightfield images are shown. The scale bar represents 300 pm. (F) BMDMs were treated with increasing doses of dG for 24 hours, fixed and stained with crystal violet. The wedge denotes 0.2, 0.4, 0.8, 1.6 mM dG; NT, not treated. (G) BMDMs were treated with 0.4 mM dG. Viability was monitored with the cell-impermeable dye Yoyo3 for 24 hours using an in-incubator imaging system (Incucyte). Yoyo3 ⁺ cells were enumerated. Mean values from triplicate measurements are shown ± SD. (H, I) BMDMs (H) and MEFs (I) were treated with 0.5 mM dG for 2 hours and intracellular dNTP levels were quantified relative to NTP levels (see methods). Data from three biological replicates are shown together with mean ± SEM. P-values determined with two- way ANOVA are indicated. Panels A-C and E-G are representative of at least three independent experiments. *** p<0.001; **** p<0.0001.

Figure 2 demonstrates that deoxyguanosine (dG) treatment kills Samhdr' cells by apoptosis. (A) BMDMs were treated with 0.5 mM of each dN for 24 hours, stained with Annexin V and 7AAD and analysed by flow cytometry. Data from triplicate measurements are shown with mean ± SD. P-values determined with two-way ANOVA are indicated. (B) Caspase activity was assessed in BMDMs 6 hours after treatment with 0.5 mM dG using the Caspase Glo 3/7 assay. For each genotype, values from untreated control cells were set to 1. Data from triplicate measurements are shown with mean ± SD. The p-value determined with an unpaired t-test is indicated. (C,D) Live cell imaging of Samhdl^ BMDMs treated with 0.5 mM dG. Alexa-488-labelled Annexin V and propidium iodide (PI) were added to the culture medium to visualise apoptotic cells. (C) Representative images. Numbers show the time after dG exposure (hh:mm). (D) Enumeration of AnnexinV ⁺ PI ⁺ cells after 24 hours of treatment with 0.5 mM dG. Six images per condition were analysed and means ± SEM are shown. The p-value determined with an unpaired t-test is indicated. (E,F) BMDMs were treated with 0.5 mM dG or 1 pg/ml cycloheximide (CHX, added to WT cells in [F]) for 8 hours. (E) Levels of the indicated proteins in total cell extracts were determined by Western blot. (F) Cells were fractionated into cytosol and a pellet containing organelles. Levels of the indicated proteins were determined by Western blot. b-Actin served as a loading control. (G) WT and Samhdl^ BMDMs were co-cultured at the indicated ratios. Cell viability was determined as in Fig. 1A 24 hours after treatment with 0.5 mM dG. Data from triplicate measurements are shown with mean ± SD. Data are representative of at least three independent experiments ns p>0.05; * p<0.05; ** p<0.01; *** p<0.001.

Figure 3 demonstrates the elimination of SAMHD1 mutated leukaemic cells by forodesine and dG treatment. (A-C) Jurkat cells were reconstituted with HA- tagged wild-type or K11A mutant SAMHD1 using a lentivector . (A,B) Cells were treated for 18 hours with 10 mM dG and 1 mM forodesine. Cells were then stained with annexin V and 7AAD and analysed by flow cytometry. Representative FACS plots are shown in panel (A) and Annexin V ⁺ 7AAD cells were quantified in panel (B). (C) SAMHD1 levels in total cell extracts were determined by Western blot. b-Actin served as a loading control. (D) PBMCs from CLL patients were treated for 24 hours with dG and Forodesine as indicated. Viability was tested as in Fig. 1A. (E-K) PBMCs from healthy control subjects and CLL patients were treated or not for 24 hours with 20 pM dG and 2 pM Forodesine (Foro + dG). Cells were then analysed using CyTOF. (E) Live cells were gated (see Fig. 9C). The CD5 and CD19 staining is shown for selected samples (see Fig. 9D for all samples). (F) Percentages of untreated, live CD5 ⁺ CD19 ⁺ cells are shown. (G) SAMHD 1 expression was analysed in untreated, live CD5 ⁺ CD19 ⁺ cells and the percentage of SAMHD1 ⁺ cells is shown (see Fig. 9C for gating). (H) Live CD5 ⁺ CD19 ⁺ cells from each sample were analysed separately by viSNE using 22 lineage markers (Cytobank; settings: 1000 iterations, 30 perplexity and 0.5 theta). Representative tSNE plots are shown (see Fig. 10 for all samples) and were coloured by expression or phosphorylation of the indicated markers. (I) Left, percentages of CD5 ⁺ CD19 ⁺ cells amongst all live cells are shown in untreated and treated PBMC samples. FdG, treatment with 2 pM forodesine and 20 pM dG. Right, the frequency of live CD5 ⁺ CD19 ⁺ cells was set to 100 in untreated samples and their percentage after forodesine and dG treatment is shown. (J,K) The staining for cleaved PARP (J) and cleaved caspase3 (K) in live CD5 ⁺ CD19 ⁺ cells was analysed. Left, median values are shown in untreated and treated cells. Right, median values from untreated cells were set to 100 separately for each sample. Data in panels A-C are representative of three independent experiments. In panels D, F, G and I-K dots represent cells from different patients and the colour indicates the mutation status (grey: ATM, black: TP53, blue: SAMHD1). Horizontal bars represent means and in panels F and G box and whiskers show SD and maximum/minimum values, respectively. P-values determined with two-way ANOVA (B,D) or unpaired t-test (F,G,I-K) are indicated ns p>0.05; * p<0.05; ** p<0.01; *** p<0.001; **** pO.OOOl .

Figure 4 demonstrates that dG-induced cell death in Samhd ' cells is independent of genome replication. (A) BMDMs were treated with 0.5 mM dG for the indicated periods of time. Cells were then labelled with BrdU for 30 minutes and fixed. Cells were stained using a-BrdU antibody and PI and analysed by flow cytometry. (B) BMDMs were labelled with BrdU for 15 minutes and then treated with 0.5 mM dG for the indicated periods of time. Cells were then analysed as in (A). (C-E) After 7 days of conventional culture, BMDMs were grown in medium containing 10% FCS (R10) or in serum-free medium (R0) for 24 hours. Alternatively, BMDMs were treated with 1 mM hydroxyurea (HU) for 8 hours. (C) Cells were analysed as in (A). (D) Cells were treated with 0.5 mM dG for 24 hours and viability was analysed as described in Fig. 1A. (E) Cells were treated with 0.5 mM dG for 16 hours and viability was analysed as described in Fig. 1A. P-values determined with two-way ANOVA are indicated. **** p<0.0001. Data are representative of three independent experiments, respectively.

Figure 5 demonstrates that dG induces death of cancer cell lines. (A) HeUa and THP1 cells were infected with VUPs containing Vpx (VUP _vpx ) or not (VUP _ctri ). After 24 hours, cells were treated with 0.5 mM of each dN for an additional 24 hours. Cell viability was assessed as in Fig. 1A. (B) He Ua treated with VUPs for 24 hours were stained with a-SAMHDl antibody and analysed by flow cytometry. The mean fluorescence intensity (MFI) of the SAMHD1 signal is shown. (C,D) HeUa and MDA-MB231 cells were left un-infected (NI) or were infected with VUPs containing Vpx (VUP _vpx ) or not (VUP _ctri ). After 24 hours, cells were treated with 0.5 mM dG and brightfield images were acquired after an additional 10-12 hours. Scale bar represents 300 pm. (E) MDA-MB231 were treated as in (D) and confluency was monitored after dG addition using a live cell imaging system in the incubator (Incucyte). The mean of 9 measurements ± SD is shown. Panels A and E are representative of three independent experiments and panels B, C and D of two experiments. In panel A dots represent technical triplicates and means ± SD are shown. P-values determined with two-way ANOVA are indicated. * p<0.05; ** p<0.01; *** p<0.001; **** p O.OOOl .

Figure 6 demonstrates that SAMHD1 -deficient B 16F10 cells die upon exposure to dG. (A) Representation of the CRISPR/Cas9 strategy to generate B 16F10 Samhdr ^/ cells. The knock-out of Samhdl exon 2 was validated by PCR using the indicated primers. (B) Brightfield images of WT and Samhdl ^ B 16F10 cell 20 hours after treatment with dG. The scale bar represents 300 pm. (C-E) Wild- type and Samhdl ^ B 16F10 cells were treated with dG as indicated for 20 hours.

(C) Confluency was determined as in Fig. S2E. (D,E) Cells were then stained with annexin V and 7AAD and analysed by flow cytometry. Representative FACS plots are shown in panel (D) and Annexin V ⁺ 7AAD and Annexin V ⁺ 7AAD ⁺ cells were quantified in panel (E). Panels B and C-E are representative of two or three independent experiments, respectively. In panel C and E dots represent technical triplicates and means ± SD are shown. P-values determined with two-way ANOVA are indicated. ** p<0.01; **** p<0.0001.

Figure 7 demonstrates that SAMHD1 -deficient Jurkat cells are killed by dG. (A) Jurkat cells were treated for 20 hours with dG as indicated or with 25 pM etoposide. Cell viability was determined as in Fig. 1A. (B,C) Jurkat cells were reconstituted with HA-tagged wild-type or K11A mutant SAMHD1 using a lentivector. Un-infected cells (NI) served as control. (B) Cells were then treated with dG for 48 hours. Cell viability was determined as in Fig. 1A. (C) SAMHD 1 levels in total cell extracts were determined by Western blot. b-Actin served as a loading control. Data are representative of three independent experiments. In panels A and B dots represent technical triplicates and means ± SD are shown. P-values determined with two-way ANOVA are indicated. * p<0.05; ** p<0.01; **** pO.0001.

Figure 8 demonstrates that forodesine and dG synergistically induce cell death in cells lacking SAMHDL (A-C) BMDMs were treated with the indicated doses of dG and Forodesine. Viability was tested as in Fig. 1A after 24 or 48 hours.

(D) BMDMs treated for 24 hours with dG and Forodesine were fixed and stained with crystal violet. After washing, cell-associated dye was solubilised and quantified by absorbance at 570 nm. For each genotype, values from untreated control cells were set to 100%. (E,F) BMDMs were treated for 8 hours with dG and Forodesine. Levels of the indicated proteins in total cell extracts were determined by Western blot. b-Actin served as a loading control eld, cleaved. Data are representative of three independent experiments. In panels A-D dots represent BMDMs from individual mice. Mean ± SD is shown. P-values determined with two-way ANOVA are indicated. ** p<0.01; *** p<0.001; **** p O.0001.

Figure 9 shows CyTOF analysis of CLL patient samples and mutation status.

(A) List of CLL samples and mutation status. CLL samples carrying either TP 53 lesions (Dell7p or mutation) or ATM lesions (Dell lq) served as controls for SAMHD1 mutated samples. (B) Details of SAMHD1 mutations. cnLOH, copy neutral loss of heterozygosity; CNLoss, copy number loss. (C) CyTOF gating strategy for the experiment shown in Fig. 3D-K. (D) Expression of CD5 and CD19 is shown as in Fig. 3E for all analysed samples.

Figure 10 (A-C) demonstrates that CLL patient cells with SAMHD1 mutation are more sensitive to forodesine and dG treatment. CyTOF data from each sample were analysed separately by viSNE analysis as described in Fig. 3H. Data from all patients studied are shown; for completeness, those selected for Fig. 3H are included here.

Figure 11 demonstrates that forodesine and dG induce apoptosis in“inactive” CLL B-cells. (A) tSNE plots using live CD5 ⁺ CD 19 ⁺ cells from SAMHD1 mutated patients, coloured by p-p38 levels, were used to manually gate“inactive” [blue] and“active” [orange] cells (left). Expression of selected markers in these sub populations is shown as a heat map (right). (B,C) The staining for cleaved PARP

(B) and cleaved caspase3 (C) was analysed in“inactive” and“active” cells. Median values are shown in untreated and treated cells. FdG, treatment with 2 mM forodesine and 20 pM dG.

Figure 12 is a graphical summary of the invention, in which the skull and cross bones indicates the cell is dead. Figure 13 demonstrates that SAMHD1 -deficient cells are sensitive to PNP inhibitors homo-DFPP-9DG and 6C-DFPP-9DG. WT and Samhd ^A BMDMs were treated as indicated for 24 hours. Cell viability was determined by CellTiter-Glo assay. For each genotype, values from untreated control cells were set to 100%. Data from triplicate measurements are shown with mean ± SD are representative of two independent experiments. P-values determined with ANOVA are indicated. ** p<0.01; *** p<0.001; **** p<0.0001.

Figure 14 demonstrates that HIV-2-infected cells die upon exposure to dG, and that this effect is mediated by the viral Vpx protein. (A) Human MDA-MB231 cells (a breast cancer cell line) were infected with VSV-G pseudotyped HIV-2 ROD9 Aenv Anef (HIV-2) or with HIV-2 ROD9 Aenv Anef Avpx (HIV-2 Avpx) [Manel et al. Nature 2010 and Lahaye et al. Immunity 2013] At the indicated time points after infection, cell lysates were prepared and analysed by Western blot using the indicated antibodies. (B) MDA-MB231 cells were infected with

HIV-2 or HIV-2 Avpx as in (A) and were subsequently treated with 0.2 mM dG. Brightfield images were acquired after 24 hours. (C) The experiment in (B) was repeated with the indicated doses of dG. Cells were fixed after 24 hours and stained with crystal violet. (D) The experiment in (C) was quantified by OD570 nm measurement. Data from untreated cells were set to 100%. Average and SD of three technical repeats are shown. Data are representative of two independent experiments.

Figure 15 demonstrates that HIV-2-infected cells die upon exposure to dG and the PNP inhibitor forodesine, and that this effect is mediated by the viral Vpx protein. MDA-MB231 cells were infected with HIV-2 or HIV-2 Avpx as in Fig 14 and were subsequently treated with dG and/or forodesine as indicated. Cells were fixed after 24 hours and stained with crystal violet. The experiment was quantified by OD570 nm measurement. Data from untreated cells were set to 100%. Average and SD of three technical repeats are shown. Data are representative of two independent experiments. ***, p<0.0001; **, p<0.001; ANOVA.

Materials and methods Plasmids - To generate pMSCVpuro-mSAMHDl, mouse Samhdl isoform 1 was PCR amplified. A kozak sequence and N-terminal 3xFLAG-tag were introduced by PCR and the PCR product was cloned into pMSCVpuro using the EcoRI site. To generate SAMHD1 -deficient mouse cells, pX458-Ruby-sgRNA-l and pX458-sgRNA-2 (Table 1) were cloned using pX458-Ruby and pX458, respectively, as described previously

(Hertzog et al., 2018).

Mice - Samhdl ¹ mice (C57BL/6 background) have been described previously (Rehwinkel et al., 2013). This work was performed in accordance with the UK Animals (Scientific Procedures) Act 1986 and institutional guidelines for animal care. This work was approved by a project license granted by the UK Home Office (PPU No. PC041D0AB) and also was approved by the Institutional Animal Ethics Committee Review Board at the University of Oxford. Cell culture - MEFs were made by standard protocols from either WT or Samhdl ¹ mice. Bone marrow cells were isolated by standard protocols and, to obtain BMDMs, were grown in petri dishes for 7 days in R10 medium [Roswell Park Memorial Institute 1640 (RPMI) medium, 10% heat-inactivated foetal calf serum (FCS), 100 U/ml penicillin and 100 pg/ml streptomycin (P/S), 2 mM L-Glutamine] supplemented with 20% L929 conditioned medium and used on day 7. Human fibroblasts from AGS patients were collected with approval by a U.K. Multicentre Research Ethics Committee (reference number 04:MRE00/19) and were provided by Y. Crow and G.I. Rice. MEFs were cultured in D10 medium [Dulbecco’s modified Eagle medium (DMEM) containing 10% heat-inactivated FCS, P/S, 2mM L-Glutamine and 20 mM HEPES buffer] . Human fibroblasts, HeLa, B 16F10 and MDA MB231 cells were cultured in D10 without P/S. HeLa cells were a gift from M. Way, MDA MB231 cells were from A. Banham and B 16F10 cells were provided by V. Cerundolo. Jurkat cells were a gift from S. Davis and originate from the American Type Tissue Collection and were cultured in R10 without P/S. All cells were cultured under 5% C02. Human fibroblasts and MEFs were cultured under low oxygen (1.2% 02).

CLL Patients’ samples - PBMCs from 19 CLL patients recruited into the ADMIRE (n=12) and ARCTIC (n=7) studies (Munir et al, 2017; Howard et al., 2017) were retrieved from the Liverpool Bio-Innovation Hub Biobank. Genetic characterisation of the tumour cells for these patients was previously published (Clifford et al., 2014). Informed consent from all patients was obtained in line with the Declaration of Helsinki. The project was covered by the Ethics approval REC 09/H1306/54. PBMCs were thawed in R10 with P/S and 50 U/ml of benzonase, washed twice before being counted and plated. For CellTiter-Glo assay, 50,000 cells were plated in U-bottom 96- well plates. Forodesine and dG treatment for CyTOF analysis was performed using 3,000,000 cells in 12-well non-coated tissue culture plates. dNTP measurements - Cells from 4 plates (90 ^c 15 mm) of BMDMs or 3 plates (150 mm x 20 mm) of MEFs were pooled for each sample. Measurements were done on cells from different mice. Cells were treated with dG for two hours and washed twice with ice-cold NaCl (9 g/F) on ice. Cells were then scraped in 550 pF of ice-cold trichloroacetic acid (15% w/v), MgC12 (30mM) solution, collected into an Eppendorf tube, frozen on dry ice and stored at -80 °C. Cells were thawed on ice and processed as described in (Kong et al., 2018). Briefly, the cell suspension was pulse-vortexed (Intellimixer) at 99 rpm for 10 min at 4°C and centrifuged at 20,000 ^c g for 1 min at 4°C. The resulting supernatant was then neutralized twice with Freon-Trioctylamine mix (78% v/v - 22%, v/v respectively) by vortexing for 30 sec and centrifugation at 20,000 x g for 1 min. 475 pF of the aqueous phase was pH-adjusted with 1 M ammonium carbonate (pH 8.9), loaded on boronate columns (Affi-Gel Boronate Gel; Bio-Rad), and eluted with a 50 mM ammonium carbonate (pH 8.9) and 15 mM MgC12 mixture to separate dNTPs from NTPs. The eluates containing dNTPs were adjusted to pH 4.5 and loaded onto an Oasis weak anion exchange (WAX) SPE cartridge. Interfering compounds were eluted off the cartridges in two steps with 1 mF ammonium acetate buffer (pH-adjusted to 4.5 with acetic acid) and 1 mF 0.5% ammonia aqueous solution in methanol (v/v), and the analytes were eluted from the cartridge with 2 mF methanol/water/ammonia solution (80/15/5, v/v/v) into a glass tube and then evaporated to dryness using a centrifugal evaporator at a temperature below 37°C. The residue was reconstituted in 1250 pF ammonium bicarbonate buffer, pH-adjusted to 3.4 and used for the HPFC analysis as described in (Jia et al., 2015). Briefly, nucleotides were isocratically eluted using 0.36 M ammonium phosphate buffer (pH 3.4, 2.5 % v/v acetonitrile) as mobile phase. dNTPs were normalized to total NTP pool of the cells.

Viability assays - CellTiter-Glo, a luminescence assay that measures ATP levels, was used according to manufacturer instructions to assess viability. To assess cell viability with crystal violet, cells were stained with 0.5% crystal violet for 20 min, washed 3 times with water and dried overnight before being resuspend in 200 mΐ methanol and absorbance was measured at 570 nm as described in (Feoktistova et al., 2016). For analysis of cell death with the Incucyte live-cell analysis system, Yoyo3 iodide viability die was used at 1/8000 final concentration and images were acquired. The Incucyte was also used to measure confluency and acquire bright-field images over time with a lOx objective.

Apoptosis assays - The Annexin V/7AAD kit was used to detect apoptotic cells by flow cytometry according to the manufacturer’s protocol. Caspase 3/7 Glo was used to measure caspase 3/7 cleavage. For live cell imaging, BMDMs were cultured in a glass chamber coated with Poly-L-lysine at 37°C and 5% C02. Culture media were supplemented with 2.5mM CaC12, 20 mM HEPES, propidium iodide (3 mΐ/well) and Annexin V AF488 (1 mΐ/well). Images were acquired with a Delta Vision microscope with lOx objective lens every 10 minutes for 24 hours.

Cell cycle analysis - BMDMs were seeded at 106 cells/well in 6-well low attachment plates and were incubated with 10 mM BrdU for 30 min. In pulse chase experiments, cells were incubated with 10 mM BrdU for 15 min, the media was then replaced, and cells were exposed to dG. At appropriate time points, cells were washed and fixed in 70% ethanol and frozen at -20°C. Cells were washed and resuspend in pepsin solution (1 mg/ml in 30 mM HC1) for 30 min at 37°C, spun down and resuspend in 2M HC1 for 15 min at room temperature (RT) and washed with PBS. Cells were then blocked with 0.5% BSA, 0.5% Tween-20 in PBS for 30 min at room temperature, washed and resuspend in FACS buffer with a-BrdU AF488 antibody at 1 : 100 dilution for 30 min at room temperature. Fix Cycle PI/RNAse A staining solution was added to the cells for 30 min at RT. Cells were acquired on a BD LSR II flow cytometer.

Western blots - Cells were lysed in NP-40 buffer (150 mM NaCl, 1% NP-40, 50 mM Tris pH 8.0) with protease and phosphatase inhibitors. After 20 min incubation on ice, lysates were centrifuged at 17,000g for 10 min at 4°C. Supernatant was collected and diluted with sample buffer before denaturation at 94°C for 5 min. Samples were loaded on pre-cast 4-12 % gradient Bis-Tris protein gels that were run with MOPS or MES buffer at 120 volts for 2 hours. Transfer to nitrocellulose membranes was performed in transfer buffer (25mM Tris, 192mM glycine, 20% methanol) at 30 volts for 3.5 hours. Membranes were blocked in 5% milk powder in Tris buffered saline with 1 % Tween-20 (TBST) for 1 hour at room temperature and were then washed 5 times for 5 min in TBST. Membranes were incubated with primary antibody in 5 % milk TBST overnight at 4°C, then washed 5 times for 5 min in TBST. ECL or ECL prime substrates were used for signal detection. In some experiments, membranes were stripped (0.2 M glycine, 0.1 % SDS at pH 2.5) for 15 minutes, washed, blocked and re-probed with a different antibody.

Retroviral vectors - VSV-G-pseudotyped retroviral vectors were produced by plasmid transfection of 293T cells (Bridgeman et al., 2015). Retroviral infections were performed in the presence of 8 pg/ml polybrene. Bone marrow cells were transduced three times by spin-infection (2500 rpm, 120 min, 32°C, no brakes) on days 1, 2 and 3 of the 7-day differentiation process with the retroviral vector expressing SAMHD1 or a control vector. Cells were seeded into new plates on day 7 and treated with dG on the next day. THP1 cells were transduced by spin infection (2500 rpm, 120 min, 32°C, no brakes), seeded and treated as indicated in the figure legends. Jurkat cells, MDA- MB231 and Hela cells were transduced by adding viral vectors to the culture medium. VLPvpx and VLPctrl were generated using the SIVmac gag-pol expression vectors SIV3+ and SIV4+ (38). Human SAMHD1 expression constructs pCSHAwtW and pCSHAkl laW were a kind gift from T. Schaller. These were used to generate lentiviruses for reconstitution of Jurkat cells (Bridgeman et al., 2015). Mouse SAMHD1 expression construct pMSCVpuro-mSAMHDl was used to generate a retroviral vector to transduce bone marrow cells as described earlier (Rehwinkel et al., 2013). HIV-2 and HIV-2 Avpx were produced using HIV-2 ROD9 Aenv Anef and with HIV-2 ROD9 Aenv Anef Avpx plasmids, respectively (Manel et al. 2010; Lahaye et al. 2013).

Stimulation, staining and mass cytometry analysis of patient samples - PBMCs were collected 24 hours after treatment with forodesine and dG, and were processed according to the Fluidigm Maxpar® protocol, using Maxpar® reagents. Antibodies are listed in Table 2. Cells were collected in 15 ml falcon tubes and were washed in PBS, using centrifugation at 300 g for 5 min. Cells were resuspend at 107/ml in R0 with Cisplatin (1 : 10,000) and incubated at 37°C for 5 minutes. Cells were washed with R10 and resuspend in Maxpar® PBS. Staining was performed on 3* 106 cells/tube. Staining with CD56, CD27, CCR4 and CCR7 was done before fixation. Cells were fixed with Maxpar Fix I Buffer at room temperature (RT) for 10 min then washed with Maxpar Cell Staining Buffer (CSB) and spun at 800 g for 5 min. Cells were barcoded (fluidigm barcoding kit) for 30 min at RT, washed twice in CSB, pooled and counted. All further steps were performed on the pooled sample. Cells were blocked in FcR block diluted in CSB (1 : 10) for 10 min at RT. Surface staining antibody mix was added to blocking solution and incubated for 30 min at RT. Cells were washed in CSB, resuspend in ice cold methanol added dropwise under the vortex and stored at - 80°C overnight. Cells were washed twice with CSB and stained with the intracellular antibody mix for 30 min at RT and stained with intercalator overnight. Next day they were washed with CSB and resuspend in water before acquisition on the Helios mass cytometer (Fluidigm). Samples were normalized, concatenated and de-barcoded using Helios software. Files were analysed with Cytobank.

Generation of Samhdl ^_/ B16F10 cells - sgRNAs were designed to excise exon 2 of mouse Samhdl (Gene ID: 56045). Exon 2 is critical to both isoforms of Samhdl (see Fig. 6A and (Rehwinkel et al., 2013)). B 16F10 cells were co-transfected with pX458- Ruby-sgRNA-1 and pX458-sgRNA-2 plasmids. GFP-Ruby double positive cells were single cell sorted and clones were expanded. A PCR screening approach was used to identify knock-out cells. PCR-1 was designed to amplify a long fragment (709 bp) from the WT allele and a short fragment (350 bp) from the KO allele using primer 1 fwd and primer 2 rvs (see Fig. 6A). PCR-2 had a primer located in exon 2 and amplified a fragment (352 bp) only from the WT allele using primer 2 rvs and primer 3 fwd (Table 1). Quantification and statistical analysis - All experiments were performed three times or more independently under similar conditions, unless specified otherwise in figure legends. Statistical significance was calculated by unpaired t-test, one-way ANOVA or two-way ANOVA as described in the figure legends. Graph pad prism 7 software was used to generate graphs and to perform statistical analysis.

Table 1. Reagents Table 2. CyTOF antibody panel.

Results

SAMHD1 protects cells against dNTP overload

To investigate the role of SAMHD1 in dNTP metabolism, equimolar concentrations of dNs were added to wild-type (WT) or SAMHD1 -deficient cells. Uptake of dNs from the extracellular environment, followed by intracellular phosphorylation, results in the formation dNTPs. Cell viability was assessed using a luminescence-based assay for intracellular ATP levels (CellTiter-Glo). Reduced viability of dN-exposed Samhdl^ mouse embryonic fibroblasts (MEFs) and bone marrow-derived macrophages (BMDMs) was observed, as well as of primary human fibroblasts from an Aicardi- Goutieres syndrome (AGS) patient homozygously carrying the Q149X nonsense mutation in SAMHD1 (Fig. 1A). Viability of WT mouse and control human cells, including fibroblasts from AGS patients carrying other mutations, was largely unaltered after addition of dNs. Reconstitution of SAMHD1 in SamhdT^ BMDMs rescued viability after treatment with dNs (Fig. 1B,C).

To determine if this effect was due to specific dNs, BMDMs were treated with single dNs or dN combinations. Interestingly, the highest toxicity in Samhdl^ cells was observed when dG was used alone or in combination with dA and/or thymidine (Fig. ID). dG was therefore focussed on in subsequent experiments. Bright-field images, crystal violet staining and live-cell imaging confirmed the toxicity of dG in Samhdl^ cells (Fig. 1E-G). Intracellular concentrations of all four dNTPs were elevated in Samhdr^ cells (Fig. 1H,I). Importantly, dG treatment resulted in 46- and 6-fold increases in dGTP concentrations in SamhdT^ BMDMs and MEFs, respectively, while dGTP levels stayed largely unchanged in WT cells (Fig. 1H,I). These data show that exposure to dG led to dGTP accumulation in cells lacking SAMHD1, subsequently resulting in cell death. dG treatment induces apoptosis in Samhdl ^ cells

Annexin V and 7AAD staining showed an increased frequency of early apoptotic (AnnexinV ⁺ 7AAD ) and dead (AnnexinV ⁺ 7AAD ⁺ ) cells in dN-treated SamhdT^ BMDM cultures (Fig. 2A). Using the Caspase-Glo assay to measure activity of apoptotic caspases, it was found that addition of dG to Samhdl^ BMDMs, but not to WT cells, activated caspase 3/7 (Fig. 2B). Live cell imaging revealed that Samhdl^ BMDMs treated with dG stained positive for Annexin V around 5 hours post-treatment and subsequently for propidium iodide (PI) (Fig. 2C). These observations suggest that dG treatment induced apoptosis, followed by secondary necrosis rendering cells permeable to PI. Quantification of AnnexinV ⁺ PI ⁺ cells by microscopy 24 hours after dG exposure affirmed increased levels of dead Samhdl^ BMDMs (Fig. 2D). Treatment of Samhdl^ BMDMs with dG led to cleavage of caspase 3 and to redistribution of cytochrome C into the cytosol (Fig. 2E,F). WT and Samhdl^ BMDMs were also co-cultured in the presence of dG and found that viability decreased with increasing proportions of SamhdT^ cells in the co-culture (Fig. 2G). In sum, these results show that dG triggered intrinsic apoptosis in Samhdl^ cells. DNA replication is not required for dG-induced apoptosis

To study the role of DNA replication in dG-induced death, cell cycle progression was analysed by BrdU and PI staining. dG treatment arrested WT cells in G0/G1 (Fig. 4A). In line with dG-induced cell death, a population of Samhdl^ cells displaying sub- G0/G1 PI staining was detected. BMDMs were also labelled with BrdU first and then treated with dG. In Samhdl^ cultures, cells with sub-GO/Gl PI staining were evident and included both BrdU ⁺ and BrdU cells, suggesting that dG treatment killed both cycling and non-cycling cells (Fig. 4B). Next, BMDMs were arrested in G0/G1 by culture in serum-free medium or by using hydroxyurea (Fig. 4C). Samhdl^ cells arrested by both methods were susceptible to killing by dG (Fig. 4D, E). Together, it is concluded that dG-induced apoptosis occurred independently of nuclear DNA replication and that dGTP overload was toxic in both cycling and non-cycling SamhdT cells. dG treatment kills SAMHDl-deficient cancer cell

SAMHD1 mutations are present in several types of cancer and in many cases result in reduced mRNA and protein levels (Clifford et al., 2014; Johansson et al., 2018; Rentoft et al., 2016). It was therefore of interest to explore the finding of dN-induced cell death in the context of malignant disease. Initially, cancer cell lines were tested. Vpx is a HIV-2 accessory protein that targets SAMHD1 for proteasomal degradation. Virus-like particles (VUPs) containing Vpx were used to deplete SAMHD1 in the cervical cancer cell line HeUa, in the monocytic cell line THP1 and in the breast cancer cell line MDA-MB231. Cells treated with VUP _vpx showed reduced viability upon addition of dNs or dG (Fig. 5). A SamhdT^ B 16F10 mouse melanoma cell line was generated using CRISPR/Cas9 (Fig. 6A). SamhdT^ B 16F10 cells showed increased frequencies of early apoptotic and dead cells upon dG treatment, accompanied by reduced confluency (Fig. 6B-E). Jurkat cells, a human T-cell line that does not express SAMHD1, was also tested. Jurkat cells were exquisitely sensitive to dG treatment (Fig. 7A). Reconstitution with a lentivirus expressing human SAMHD 1 partially rescued viability upon dG treatment (Fig. 7B,C). SAMHD1 K11A, which does not localise to the cell nucleus, had similar effects. In sum, SAMHD1 protected cancer cell lines against dN-triggered toxicity. SAMHD1 protects against combined forodesine and dG treatment

Forodesine is a PNP inhibitor. It has been shown that SAMHD1 -deficient cells fed with dG died by apoptosis due to dGTP overload. Forodesine and dG may therefore work synergistically, particularly in SAMHD1 -deficient cells. Indeed, low doses of dG or forodesine alone did not compromise viability of WT or SamhdT^ BMDMs while the combination of both was toxic in SamhdT^ cells (Fig. 8). Similarly, Jurkat cells treated with low doses of forodesine and dG entered apoptosis, and this was prevented when SAMHD1 or SAMHD1 K11A were expressed (Fig. 3A-C). Thus, SAMHD1 protected cells against death that was synergistically induced by forodesine and dG.

CLL B cells with SAMHD1 mutations are highly sensitive to a combination treatment of forodesine and dG

SAMHD1 is mutated in 11% of refractory CLL patients (Clifford et al., 2014). It was hypothesised that CLL B cells with SAMHD1 mutations are susceptible to forodesine and dG. The effects of these compounds were tested on peripheral blood mononuclear cells (PBMCs) from CLL patients with or without mutations in SAMHD1 (Fig. 9A,B). When used alone, neither forodesine nor dG significantly reduced cell viability (Fig. 3D). The combination of both compounds had little effect on viability of PBMCs from patients without SAMHD1 mutations. However, significantly reduced PBMC viability was observed in the SAMHD1 -mutated group (Fig. 3D). Next, cytometry by time of flight (CyTOF) analysis was used. After dG treatment, cells were stained with antibodies recognising cell surface and intracellular markers. CLL B cells co-express CD5 and CD19 (Swerdlow, 2008). As expected, CD5 ⁺ CD19 ⁺ cells (CLL B cells) were largely absent from control PBMCs and could be detected at varying frequencies in samples from CLL patients, irrespective of SAMHD1 genotype (Fig. 3E,F and 9C,D). CLL B cells from the SAMHD1 mutated group had no detectable levels of SAMHD 1 (Fig. 3G and 9C). Next, CLL B cells were gated and viSNE plots were generated, in which each dot represents a cell. Colour can be used to show expression of a chosen parameter. This analysis revealed a marked reduction of CLL B cells upon forodesine and dG treatment in the SAMHD1 -mutated group but not in the control groups (Fig. 3H,I and 10). Increased levels of cleaved PARP and cleaved caspase3 were observed only in N4Mi7Z)7-mutated CLL B cells post-treatment, consistent with the induction of apoptosis (Fig. 3J,K). Interestingly, forodesine and dG ablated a subpopulation of SAMHD1 mutated CLL B cells, characterised by high levels of NFi B-p65, p38 and STAT1 phosphorylation (Fig. 3H and 10). Using phosphorylated (p-) p38, a MAP kinase, “active” and “inactive” cells were defined and confirmed that p-p38 positive cells also displayed higher levels of p-p65 and p-STATl (Fig. 11A). Higher levels of CD27 expression were found in“active” cells, which may indicate engagement of the B cell receptor. Staining for cleaved PARP and cleaved caspase3 was enhanced more strongly in “inactive” cells upon treatment (Fig. 11B,C). This suggests that these“inactive” CLL B cells were also affected by the treatment and induced apoptosis with delayed kinetics compared to“active” cells. Collectively, these data show that PNP inhibitors like forodesine are highly efficient at killing malignant CLL B cells with SAMHD1 mutations that cause a defect in SAMHD 1 expression, while cells with intact SAMHD1 expression are spared.

HIV-2 infected cells can be selectively killed using the PNP inhibitor forodesine.

HIV-2 encodes the Vpx protein which induces the ubiquitination and degradation of cellular SAMDH1. Figure 14 shows that HIV-2 infected cells begin to produce the Vpx protein after 2 hours, which correlates with a decrease in cellular SAMHD1 levels (Fig 14A). Cells infected with HIV-2 which encodes Vpx are susceptible to cellular killing by treatment with dG, as opposed to cells infected with HIV-2 lacking Vpx (Fig 14B-D). The effect of inducing cellular death can also be achieved by treatment with forodesine and significantly reduced levels of dG (Fig 15). The data presented demonstrates the ability of PNP inhibitors, such as forodesine, to kill HIV-2 infected cells.

Discussion

The present data reveals an unexpected role of SAMHD1 in safeguarding cells against death resulting from imbalances in dNTP pools (Fig. 12, top). Given that dN-triggered cell death did not require ongoing genomic DNA synthesis, it is possible that dNTP imbalances disrupt replication or repair of mitochondrial DNA, resulting in mitochondrial stress and subsequent apoptosis (Arpaia et al., 2000; Franzolin et al., 2015). Synergy is further shown between dG exposure and forodesine, which blocks dG degradation by PNP, in cells lacking SAMHD1. Importantly, the combination of dG and forodesine selectively killed SAMHD1 -deficient CLL B cells, while other normal cells or SAMHD1 -sufficient CLL B cells remained unaffected (Fig. 12, bottom). Clinical trials showed that forodesine has beneficial effects in some but not all patients with B or T cell malignancies (Alonso et al., 2009; Balakrishnan et al., 2006; Balakrishnan et al., 2013; Balakrishnan et al., 2010; Dummer et al., 2014; Gandhi and Balakrishnan, 2007; Gandhi et al., 2005; Maruyama et al., 2018; Ogura et al., 2012), an observation that thus far has lacked an explanation. The present data shows that forodesine-sensitive leukaemias harbour mutations that ablate SAMHD 1 expression or inactivate its enzymatic activity. It is therefore possible to stratify cancer patients by SAMHD1 genotype or expression level into those that will and those that will not respond to PNP inhibitors such as forodesine. Further, as the HIV-2 virus encodes the Vpx protein which induces the cellular degradation of SAMHD1, the invention also provides an opportunity to target and selectively kill cells infected with HIV-2 using PNP inhibitors such as forodesine.

References

1. W. Daddacha et al., SAMHD1 Promotes DNA End Resection to Facilitate DNA Repair by Homologous Recombination, Cell Rep., 20(8): 1921-1935, 2017

2. M. Rentoft et al., Heterozygous colon cancer-associated mutations of SAMHD1 have functional significance. Proceedings of the National Academy of Sciences of the United States of America 113, 4723-4728 (2016).

3. R. Clifford et al., SAMHD1 is mutated recurrently in chronic lymphocytic leukemia and is involved in response to DNA damage. Blood 123, 1021-1031 (2014).

4. P. Johansson et al., SAMHD1 is recurrently mutated in T-cell prolymphocytic leukemia. Blood Cancer J 8, 11 (2018).

5. S. H. Swerdlow, WHO classification of tumours of haematopoietic and lymphoid tissues. (International Agency for Research on Cancer, Lyon, 2008).

6. S. Bantia et al., Purine nucleoside phosphorylase inhibitor BCX-1777 (Immucillin-H)— a novel potent and orally active immunosuppressive agent. International immunopharmacology 1, 1199-1210 (2001).

7. V. Gandhi et al., A proof-of-principle pharmacokinetic, pharmacodynamic, and clinical study with purine nucleoside phosphorylase inhibitor immucillin-H (BCX- 1777, forodesine). Blood 106, 4253-4260 (2005).

8. K. Balakrishnan, R. Nimmanapalli, F. Ravandi, M. J. Keating, V. Gandhi, Forodesine, an inhibitor of purine nucleoside phosphorylase, induces apoptosis in chronic lymphocytic leukemia cells. Blood 108, 2392-2398 (2006). 9. R. Dummer et al., Final results of a multicenter phase II study of the purine nucleoside phosphorylase (PNP) inhibitor forodesine in patients with advanced cutaneous T-cell lymphomas (CTCL) (Mycosis fungoides and Sezary syndrome). Ann Oncol 25, 1807- 1812 (2014).

10. K. Balakrishnan et al., Phase 2 and pharmacodynamic study of oral forodesine in patients with advanced, fludarabine-treated chronic lymphocytic leukemia. Blood 1 16, 886-892 (2010).

1 1. D . Maruyama et al., Multicenter phase 1/2 study of forodesine in patients with relapsed peripheral T cell lymphoma. Ann Hematol, (2018).

12. V. Gandhi, K. Balakrishnan, Pharmacology and mechanism of action of forodesine, a T-cell targeted agent. Semin Oncol 34, S8- 12 (2007).

13. R. Alonso et al., Forodesine has high antitumor activity in chronic lymphocytic leukemia and activates p53 -independent mitochondrial apoptosis by induction of p73 and BIM. Blood 1 14, 1563- 1575 (2009).

14. K. Balakrishnan, F. Ravandi, S . Bantia, A. Franklin, V. Gandhi, Preclinical and clinical evaluation of forodesine in pediatric and adult B-cell acute lymphoblastic leukemia. Clin Lymphoma Myeloma Leuk 13, 458-466 (2013).

15. M. Ogura et al., Phase I study of BCX 1777 (forodesine) in patients with relapsed or refractory peripheral T/natural killer-cell malignancies. Cancer Sci 103, 1290- 1295 (2012).

16. E. Arpaia et al., Mitochondrial basis for immune deficiency. Evidence from purine nucleoside phosphorylase-deficient mice . The Journal of experimental medicine 191 , 2197-2208 (2000).

17. E. Franzolin, C. Salata, V. Bianchi, C. Rampazzo, The Deoxynucleoside Triphosphate Triphosphohydrolase Activity of SAMHD 1 Protein Contributes to the Mitochondrial DNA Depletion Associated with Genetic Deficiency of Deoxyguanosine Kinase . J Biol Chem 290, 25986-25996 (2015).

18. J. Hertzog et al., Infection with a Brazilian isolate of Zika virus generates RIG-I stimulatory RNA and the viral NS5 protein blocks type I IFN induction and signaling. Eur J Immunol, (2018).

19. J. Rehwinkel et al., SAMHD 1 -dependent retroviral control and escape in mice. The EMBO journal 32, 2454-2462 (2013).

20. T. Munir et al., Results of the randomized phase IIB ADMIRE trial of FCR with or without mitoxantrone in previously untreated CLL. Leukemia 3 1 , 2085-2093 (2017). 21. D. R. Howard et al., Results of the randomized phase IIB ARCTIC trial of low- dose rituximab in previously untreated CLL. Leukemia 3 1 , 2416-2425 (2017).

22. Z. Kong et al., Simultaneous determination of ribonucleoside and deoxyribonucleoside triphosphates in biological samples by hydrophilic interaction liquid chromatography coupled with tandem mass spectrometry. Nucleic acids research 46, e66 (2018) .

23. S . Jia, L. Marjavaara, R. Buckland, S . Sharma, A. Chabes, Determination of deoxyribonucleoside triphosphate concentrations in yeast cells by strong anion- exchange high-performance liquid chromatography coupled with ultraviolet detection. Methods Mol Biol 1300, 1 13- 121 (2015).

24. M. Feoktistova, P. Geserick, M. Leverkus, Crystal Violet Assay for Determining Viability of Cultured Cells. Cold Spring Harb Protoc 2016, pdb prot087379 (2016).

25. A. Bridgeman et al., Viruses transfer the antiviral second messenger cGAMP between cells. Science 349, 1228- 1232 (2015).

26. N. Manel et al., A cryptic sensor for HIV- 1 activates antiviral innate immunity in dendritic cells, Nature, 467(73 12) :214-7 (2010).

27. Lahaye et al., Immunity, The capsids of HIV- 1 and HIV-2 determine immune detection of the viral cDNA by the innate sensor cGAS in dendritic cells, Immunity, 39(6): 1 132-42 (2013).

28. Hofmann et al., The Vpx lentiviral accessory protein targets SAMHD 1 for degradation in the nucleus, J Virol., 86(23): 12552-60, 2012.