Regulation of deoxynucleotide metabolism in cancer: novel mechanisms and therapeutic implications

Coordinated synthesis and degradation of dNTPs resulting in a balanced intracellular dNTP pool is critical for numerous cellular processes, such as fidelity of DNA synthesis and DNA damage repair [22]. Two distinct pathways that synthesize dNTPs are the de novo synthesis in the cytoplasm, and the salvage pathway that takes place both in cytoplasm and mitochondria. The rate-limiting step of de novo dNTP synthesis is catalyzed by ribonucleotide reductase (RNR) that converts ribonucleotide diphosphates to deoxyribonucleotides [3]. Degradation of dNTPs as part of the salvage pathway is accomplished by deaminases and phosphorylases, as well as the mammalian triphosphohydrolase, SAMHD1 [7, 8]. Optimization of dNTP pools is achieved by cell cycle-dependent activity and allosteric regulation of RNR and SAMHD1 [23]. Actively proliferating cells have an approximately 10-fold higher dNTP pool than quiescent cells that are in G₀/G₁ phase [24]. The dNTP pool is greatly expanded during G₁ to S-phase transition, and remains abundant until DNA synthesis is complete [24]. This biphasic regulation is critical to supply dNTPs for DNA synthesis, and to prevent excess intracellular dNTPs in the absence of DNA replication, which can contribute to innate immune activation [25] and cancer development [22].

The complement of intracellular dNTPs has numerous implications for DNA replication, mutagenesis, DNA repair, and therefore in cancer development. Recent progress in literature suggests that RNR-mediated increase in dNTP pools is accompanied by higher mutation rates due to reduced fidelity of DNA replication or activation of translesion synthesis [26]. These studies suggest that increased dNTP pools upon altered RNR activity may cause increased mutation rates. However, it is important to note that altered RNR activity can also affect its function in DNA repair (25). Therefore, further studies are required to rule out the possibility of altered DNA repair functions of RNR leading to increased mutation rate. Indeed, dNTP pools are generally greater in transformed cell lines compared to normal cells [27]. Mutator phenotypes are characterized by increased somatic mutation frequency in pre-cancerous cells that accounts for high number of mutations in cancer cells, consistent with what is observed with dNTP pool imbalances [24, 28, 29]. These pre-cancerous cells are characterized by enhanced mutagenesis, stimulation of genetic recombination, increased frequency of chromosomal abnormalities, DNA strand breaks and cell death [26]. Imbalance in cellular dNTP pool causes a hypermutator phenotype, associated with DNA replication stress and altered replication fork velocity [3, 11]. Mechanisms of mutagenesis conferred by imbalanced dNTP pool are not fully understood, but likely include nucleotide misinsertion during DNA replication, indirect inhibition of proof-reading, or forced frameshift mutations [26].

In response to DNA damage, dNTP levels in S-phase increase approximately by 4-fold [30]. Both RNR and SAMHD1 can be recruited to sites of DNA damage to tightly regulate the dNTPs supplied for DNA repair machinery [11, 30]. Replication stress from increased mutation rate contributes to genomic instability and activates the DNA damage response, which then may further exacerbate mutagenesis by altering the balance of the dNTP pool [31, 32]. It is interesting to note that pre-cancerous cells show evidence of increased DNA damage response activation [31, 32]. Together, these studies suggest that in cells with elevated activation of DNA damage responses, increased dNTP levels may lead to malignancy.

Nucleotide metabolism plays an important role in senescence and autophagy in cancer cells. Senescence may occur as a tumor suppressive mechanism early on in tumor initiation, while autophagy is a common mechanism that tumor cells use to survive under metabolic stress [3, 33]. Nucleotide metabolism therefore has implications in genomic instability and mutation as part of tumor initiation, and resistance to apoptosis during tumor promotion, two of the important hallmarks of cancer development. Thus, regulators of dNTP pool are important targets of cancer therapy.

Given the important role of nucleotide metabolism in cell proliferation, transformation and tumor progression, inhibition of nucleotide synthesis has been commonly used in treatment of cancer, as well as infectious and immune-mediated diseases [3]. Inhibiting DNA precursor synthesis or incorporation of nucleoside analogs into DNA results in DNA damage and stalled replication forks, followed by activation of the S-phase checkpoint, which may lead to cell death by apoptosis [3] (Fig. 1). Methotrexate and similar compounds that inhibit dihydrofolate reductase required for purine biosynthesis [34], are widely used in chemotherapy for solid and lymphocytic tumors. On the other hand, purine and pyrimidine analogs such as forodesine and gemcitabine respectively, can disrupt DNA replication [35]. Moreover, 5-fluorouracil is a pyrimidine adduct that is misincorporated into DNA and inhibits nucleotide synthesis [34]. Most of clinically available nucleoside analogs act by incorporation into DNA, which relies on balanced dNTP pool for high fidelity of incorporation during replication. Excess intracellular dNTPs may out-compete these nucleoside analogs, thus conferring chemotherapeutic resistance [36]. For this reason, these drugs may be used in combination with other therapies, such as inhibitors of RNR that reduce dNTP pool. Efficient RNR inhibition is accomplished by hydroxyurea, a free radical scavenger and iron chelator that inactivates the catalytic capacity of the R2 subunit of RNR [3]. Current clinically used inhibitors of RNR, such as hydroxyurea or gemcitabine, have limitations, including severe cytotoxicity, short half-life, and drug resistance when used as a single therapy. New horizons for nucleotide synthesis-directed therapy include small molecule targeting subunit activity modulation, such as anti-sense oligonucleotides and gene therapy targeting RNR [3, 26].

It is currently unclear whether SAMHD1 can degrade these nucleoside analogs in cells and reduce their anti-cancer potency. A study indicates that reduction of SAMHD1 levels in T-cells significantly decreases HIV-1 sensitivity to thymidine analogs (such as zidovudine or stavudine), but not other tested nucleotide analogs of HIV-1 reverse transcriptase inhibitors [37]. These results suggest that SAMHD1 may have a differential effect over the different dNTPs. Owing to its dNTPase activity, SAMHD1 may serve as an intriguing target of cancer therapy as reduction of dNTP pool may prevent tumorigenesis or an increase in efficacy of anti-nucleotide chemotherapeutics (Fig. 1).

SAMHD1 was identified in human myeloid cells as a novel HIV-1 restriction factor [5, 6]. SAMHD1 is responsible for maintaining low levels of intracellular dNTPs in non-dividing cells [9, 38, 39]. In non-transformed fibroblasts, knockdown of SAMHD1 results in loss of cell-cycle regulation, and cells accumulate in G₁ with oversized dNTP pool [40]. Therefore, SAMHD1 influences cell cycle progression and DNA replication by degrading dNTPs [40] (Fig. 1). SAMHD1 siRNA-mediated knockdown has differential effects on the dNTP pools of proliferating cells and quiescent cells [40]. In proliferating lung or skin fibroblasts, siRNA-mediated knockdown of SAMHD1 leads to loss of the cell-cycle regulation of dNTP pool sizes and thus resulting in dNTP imbalance [40]. On the other hand, in cells made quiescent via serum starvation, SAMHD1 down-regulation leads to a marked expansion of dNTP pools [40]. The function of SAMHD1 in sensing intracellular dNTPs is critical for maintaining pool balance in cells at the appropriate phase in cell cycle.

The opposite role of increasing the intracellular dNTP pool is achieved by RNR, which catalyzes the rate-limiting step in de novo nucleotide synthesis by converting ribonucleotides to deoxyribonucleotides [3, 24]. RNR heterotetramer is present at consistent levels throughout the cell cycle and functions in de novo nucleotide synthesis even in non-dividing cells [27]. The gene encoding the active subunit, R1, is transcriptionally regulated, responding to DNA replication stress and progression through the cell cycle [31]. RNR transcription increases by 10–20 fold in S phase over the levels in G1 phase [27, 31]. RNR can also be recruited to sites of DNA damage, by the virtue of p53-inducible subunit R2, to supply the necessary dNTPs for the damage repair machinery [27]. Cancer cells require RNR for de novo synthesis of dNTPs and elevated RNR expression is a characteristic of many cancers, which contribute to increased dNTP levels and uncontrolled cellular proliferation [27].

Both SAMHD1 and RNR have similar regulation mechanisms of their expression and activity, which are cell-cycle dependent and tightly controlled in cells. Both enzymes are active oligomers with two types of allosteric sites that sense the concentration of nucleotides in the cell and control enzymatic activity [23, 24, 41]. Regulation of SAMHD1 dNTPase activity is achieved by allosteric binding of dGTP or GTP to the first allosteric site, followed by binding of any dNTP to the second allosteric site, causing a conformational change in the substrate-binding pocket [23, 42]. SAMHD1 has sensory activity that detects the cellular dNTP pool composition, and allows a feedback system to achieve a balance with the opposite catalytic activity of RNR [23, 41]. The most sensitive regulator of dNTP pool is likely dATP, as it has higher affinity for the second allosteric site of SAMHD1, weak affinity for the catalytic site of SAMHD1, and is produced least efficiently by RNR [26, 43]. The dATP:ATP ratio is likewise an important regulator of RNR activity [26]. Elevations in dATP concentrations are inhibitory to the anabolic activity of RNR, whereas increased ATP concentration is stimulatory [26]. Therefore, dATP plays a fine-tuned regulatory role in providing negative feedback from SAMHD1 to RNR and vice versa [23, 26, 41, 43]. Allosteric binding of dNTPs is responsible for tetramerization of SAMHD1 to its functional catalytic state. Tetramer-disrupting mutations abolish dNTPase activity as well as the ability of SAMHD1 to restrict HIV-1 infection [23, 42].

SAMHD1 inhibits retroviral replication in non-dividing cells by depleting the intracellular dNTP pool to a level that is limiting for viral reverse transcription [9, 38, 39]. In terminally differentiated macrophages and dendritic cells, or resting CD4⁺ T-cells with arrested cell cycle (G₀/G₁), dNTP levels are significantly decreased by SAMHD1, and therefore HIV-1 infection is restricted [9, 38, 39]. Vpx, a protein encoded by HIV-2 or certain simian immunodeficiency viruses, causes proteasomal degradation of SAMHD1, which is associated with an increase in dNTP pool, accelerated proviral DNA synthesis, and enhanced viral infectivity [9, 38, 39]. Low cellular dNTP serves as a common mechanism of SAMHD1-mediated retroviral restriction. In addition to HIV-1, SAMHD1 also restricts many other retroviruses and DNA viruses (such as herpes simplex virus 1 and hepatitis B virus) in non-dividing cells [44‐46]. However, it remains unclear how these viral restriction studies provide any clues that may link with cell cycle dysregulation in cancer cells, particularly in the context of viral infection.

Phosphorylation at threonine (T) 592 by cyclin dependent kinases (CDK) regulates dNTPase activity and HIV-1 restriction activity of SAMHD1 [47‐49]. CDK1 as well as CDK2 in complex with cyclin A can bind to and phosphorylate this site [25, 47‐49]. CDK2 interaction with SAMHD1 is regulated by CDK6 in T-cells and macrophages [50]. Type I IFN reduces phosphorylation of SAMHD1 at the T592 residue [47], likely by inducing CDK inhibitors, such as p21, resulting in HIV-1 restriction and decreased dNTP pool [50, 51]. For proper maintenance of dNTP levels, SAMHD1 protein must be reduced in the cell during S-phase to facilitate DNA replication [40]. In differentiated human macrophages, SAMHD1 interacts with cyclin L2, forming an ubiquitin ligase complex and resulting in its proteasomal degradation [52]. However, it is unclear whether the degradation process of endogenous SAMHD1 occurs in other cell types or in dividing cells. Phosphorylation of SAMHD1 by CDKs may be the signal for its proteasomal degradation in a cell type- or cell cycle-dependent manner. With an understanding of SAMHD1 dNTPase function and dNTP regulation in cancer, we discuss the physiological consequences of alterations in SAMHD1 expression.

SAMHD1 is a negative regulator of IFN-induced innate immune responses [53]. Aicardi-Goutieres syndrome (AGS) is an hereditary neurodegenerative autoimmune disorder that is characterized by progressive encephalopathy and is accompanied by increased IFN-α production [54]. AGS is attributed to defective clearance of excessive intracellular (self) nucleic acids, which trigger immune responses resulting in activation of IFN pathway in a manner that mimics viral infection [10, 55]. SAMHD1 homozygous mutations have been identified in 17 % of patients with AGS, suggesting a pivotal role of SAMHD1 in preventing activation of innate immune response to self-nucleic acids [10]. Further, a homozygous deletion (~9 Kb) in the SAMHD1 gene was identified in patients with atypical AGS, characterized by genomic instability [56]. Mutations in SAMHD1 result in increased dNTP pools in fibroblasts from AGS patients [25], suggesting that the dNTPase function of SAMHD1 prevents autoimmunity by maintaining genome stability. Together, these studies implicate SAMHD1 in regulation of self-nucleic acid stimulated-autoimmune responses.

One of the most critical mechanisms by which cancer cells grow uncontrollably is by inhibiting the function of tumor suppressor genes [57]. This is achieved mainly through mutation (leading to loss of function) and/or downregulation of tumor suppressor genes that normally arrest cell growth [57]. Studies using CLL patient samples have identified SAMHD1 mutations in 4 out of 160 cases (2.5 %) [13]. Acquired SAMHD1 mutations were reported at a frequency of 11 % in relapsed or chemotherapy refractory CLL patients and 3 % in the pretreatment group [11]. These findings suggest that SAMHD1 mutation can be an important driving factor contributing to CLL progression, and can potentially be utilized as a biomarker for CLL prognosis.

SAMHD1 somatic mutations have been identified in several human cancers, including chronic lymphocytic leukemia (CLL) [11, 13, 16], myeloma [18], breast cancer [17], lung carcinoma [14], colon and rectal cancer [17, 19, 58], pancreatic cancer [12], and glioblastoma [15]. All the identified SAMHD1 alterations in cancer are summarized in Table 1. Of note, several of these mutations are located in the catalytic region of HD domain of SAMHD1 that is responsible for its dNTPase activity (Table 1) [7, 8]. However, it is important to note that it is still unknown whether these specific mutations in the HD domain of SAMHD1 can directly affect its dNTPase function. Generation and characterization of SAMHD1 variants harboring these specific mutations can help define their activity changes. Importantly, physiological significance and functional contributions of these SAMHD1 mutations to progression of cancer also needs to be investigated, especially since the SAMHD1 knockout in mice does not result in spontaneous tumorigenesis [53, 59]. Further functional analyses of SAMHD1 mutations identified in human cancer cells would aid in answering these important questions.

SAMHD1 mRNA levels were significantly lower in CLL patient B-cells with SAMHD1 mutation relative to normal B-cells [11]. Although this study suggests that these mutations result in reduction of SAMHD1 mRNA levels, the exact mechanism involved in this process is unknown. It is important to determine whether these mutations lead to reduced transcription of SAMHD1 or if they cause a post-transcriptional change such as nonsense-mediated decay of SAMHD1 mRNA. SAMHD1 expression is also significantly lower in other cancers, including cutaneous T-cell lymphoma (CTCL) [20], lung carcinoma [21], breast carcinomas and various tumor cell lines [11].

CTCL is a subset of non-Hodgkin’s lymphomas that is characterized by infiltration and proliferation of malignant CD4⁺ T-lymphocytes into the skin [60]. Decreased mRNA and protein levels of SAMHD1 were identified in peripheral blood mononuclear cells (PBMCs) of CTCL patients relative to healthy donors [20]. However, it is unknown whether decreased SAMHD1 expression has an effect on CTCL lymphomagenesis, if restored SAMHD1 expression can rescue CTCL phenotype, or whether SAMHD1 mRNA/protein expression correlates with patient prognosis. Lung carcinoma patients have reduced SAMHD1 mRNA and protein [21], but a direct correlation to disease progression or outcome has not been established. Overexpression of SAMHD1 in a lung cancer cell line results in decreased proliferation in vitro, with a concomitant increase in intracellular dNTP levels [21]. Similarly, overexpression of SAMHD1, but not the dNTPase-defective mutant, can significantly reduce HeLa cell proliferation [11].

Epigenetic mechanisms play a major role in cancer development and progression by modulating gene expression [61]. DNA methylation and chromatin remodeling via histone acetylation or deacetylation are the most important epigenetic mechanisms through which cancer cells achieve increased expression of oncogenes or decreased expression of tumor suppressor genes [61]. Thus, many therapeutic strategies target epigenetic mechanisms in cancer [62].

Studies in lung cancer and CTCL patients have revealed the role of epigenetic mechanisms in downregulation of SAMHD1 expression. The SAMHD1 promoter in PBMCs of CTCL patients is highly methylated, while it is unmethylated in PBMCs from healthy donors [20]. This was also observed in human lymphoma/leukemia CD4⁺ T-cell lines relative to primary CD4⁺ T-cells that have high endogenous SAMHD1 expression [63]. Importantly, treatment with DNA methyltransferase inhibitors in CD4⁺ T-cell lines rescues SAMHD1 mRNA and protein expression, validating that DNA methylation has a direct inhibitory effect on SAMHD1 transcription [63]. Similar effects of SAMHD1 promoter methylation on its expression were identified in lung cancer tissues [21]. Interestingly, inhibition of histone deacetylation leads to induction of SAMHD1 mRNA and protein levels in CD4⁺ T-cells, suggesting that histone deacetylation also contributes to repression of SAMHD1 expression [63]. Despite implications for epigenetic downregulation of SAMHD1 in cancer, the exact mechanisms are not known. Better understanding of these pathways may aid in development of therapeutic strategies targeting epigenetic modifications.

Gene expression is also regulated by microRNAs and transcription factors. Although, transcription factors that can regulate SAMHD1 expression are yet to be identified, one study has indicated that microRNA-181 binds directly to the 3’ untranslated region of SAMHD1 mRNA and downregulates its expression [64]. Overexpression of microRNA-181 in monocytic THP-1 cells reduced endogenous SAMHD1 mRNA and protein levels, while microRNA-181 downregulation in T-cell leukemic-derived Jurkat cells lead to an increase in SAMHD1 mRNA and protein [64]. It is unclear whether microRNA-181 downregulates SAMHD1 expression in cancer patients. Together these findings support the hypothesis that SAMHD1, through its dNTPase function, could potentially act as a tumor suppressor (Fig. 2).