Allosteric targeting with antiviral nucleotide analogs allows fine-tuning of SAMHD1 dNTPase activity

TL;DR

This study shows how antiviral drugs can modulate the activity of the SAMHD1 enzyme, offering new ways to control its role in chemotherapy resistance.

Contribution

The study reveals that antiviral nucleotide analogs can fine-tune SAMHD1 activity through allosteric binding, enabling differential dNTPase control.

Findings

Acyclovir- and ganciclovir-triphosphate bind to SAMHD1's AS1 and induce tetramer formation with reduced enzymatic activity.

Activator identity at AS1 fine-tunes dNTPase activity toward different dNTP substrates.

Nucleotide analogs and GTP show synergistic activation, suggesting mixed-occupancy SAMHD1 tetramers.

Abstract

SAM and HD domain–containing protein 1 (SAMHD1) is a dNTP hydrolase that controls intracellular dNTP pools and plays diverse roles in human health and disease. Notably, this enzymatic activity also confers chemotherapy resistance by hydrolyzing the active triphosphate forms of nucleoside analog drugs, thereby reducing their efficacy and contributing to worse treatment outcomes in cancer patients. The dNTPase activity of SAMHD1 is tightly regulated by allosteric activation and oligomerization through binding of (d)NTPs to two allosteric sites (ASs), the first of which—AS1—requires binding of a guanine nucleotide. In the present study, we investigated strategies to pharmacologically modulate SAMHD1 dNTPase activity via AS1. Using a variety of biochemical and biophysical assays, we demonstrate that the antiviral guanine nucleotide analogs, acyclovir- and ganciclovir-triphosphate, are potent AS1 binders that induce the formation of enzymatically competent SAMHD1 tetramers, however with reduced enzymatic activity. Furthermore, we show that AS1 activator identity can fine-tune dNTPase activity toward different dNTP substrates, providing a new avenue to pharmacologically control SAMHD1. This differential activity of acyclovir- and ganciclovir-triphosphate–activated SAMHD1 can be explained by distinct kinetic profiles that deviate from Michaelis–Menten kinetics. Furthermore, based on an apparent synergistic activation between these nucleotide analogs and the physiological AS1 activator GTP, we also propose the existence of mixed-occupancy SAMHD1 tetramers. Our work therefore provides new insights into the allosteric activation and oligomerization process of SAMHD1 and opens new avenues to pharmacologically control the dNTPase activity utilizing non-natural allosteric ligands.

Figures

Full text

SAM and HD domain–containing protein 1 (SAMHD1) is a dNTPase belonging to the HD domain–containing superfamily of phosphohydrolases (1, 2). By hydrolyzing dNTPs, producing deoxyribonucleosides and triphosphates, SAMHD1 is a major regulator of intracellular dNTP pools (3, 4, 5). In addition, SAMHD1 also has noncatalytic functions in DNA repair, localizing to sites of DNA double-strand breaks and stalled replication forks and subsequently recruiting repair factors (6, 7, 8).

To be catalytically active as a dNTPase, SAMHD1 needs to oligomerize into homotetramers (9, 10, 11, 12). The formation of homotetramers is triggered by the binding of allosteric activators to two distinct allosteric sites (ASs): AS1 and AS2. AS1 binds GTP or dGTP, whereas AS2 binds any dNTP (13, 14) (Fig. 1A). While AS1 is thought to be no longer accessible following tetramerization, recent publications suggest that a tetrameric form of SAMHD1 with partially occupied AS2 pockets can exist (15, 16).Figure 1SAMHD1 is an allosterically activated dNTP hydrolase. A, schematic representation of the current model of SAMHD1 allosteric activation and oligomerization. Inactive SAMHD1 monomers dimerize upon binding of (d)GTP to allosteric site 1 (AS1). Upon binding of any dNTP to allosteric site 2 (AS2), the catalytically active homotetramer is formed. B, the guanine nucleotide analogs acyclovir-triphosphate (Ac-TP) and ganciclovir-triphosphate (Gc-TP) closely resemble SAMHD1 AS1 activator GTP. SAMHD1, SAM and HD domain–containing protein 1.

First identified as one of the genetic causes of Aicardi–Goutières syndrome, SAMHD1 is associated with a variety of disease phenotypes. Loss of SAMHD1 function in Aicardi–Goutières syndrome causes cGAS–STING sensing of cytosolic DNA, activating an interferon response (17), which is related to its function in resolving stalled replication forks (7). SAMHD1 also restricts HIV-1 infection (18, 19), but the mechanism is less clear, as the function to deplete dNTPs or oligomerize seems to be disconnected from its viral restriction ability (9, 20, 21, 22, 23). Instead, it has been suggested that post-translational modifications like phosphorylation or acetylation regulate the ability of SAMHD1 to restrict HIV (22, 24, 25, 26).

In certain types of cancer, like lung and colon cancer, SAMHD1 has been shown to act as a tumor suppressor (27, 28, 29). However, SAMHD1 can also act as a chemotherapy resistance factor by protecting cancer cells from nucleoside analogs. This is because, in addition to endogenous dNTPs, SAMHD1 can also hydrolyze the active triphosphate metabolites of commonly used chemotherapies. Clofarabine-triphosphate and cytarabine-triphosphate, the active metabolites of clofarabine and cytarabine (ara-C), are SAMHD1 substrates (30, 31, 32). In preclinical models of acute myeloid leukemia, SAMHD1 dictates response to ara-C therapy, and furthermore, in acute myeloid leukemia patients treated with ara-C, lower SAMHD1 expression correlates with better survival (31, 33). To date, several studies have identified various nucleotide analogs as both substrates and allosteric modulators of SAMHD1 (13, 32, 34, 35, 36).

Due to its roles in various pathologies and being a chemoresistance factor, SAMHD1 is a promising target for the development of small-molecule inhibitors. Nonhydrolyzable dNTP analogs, such as deoxyguanosine-(α-thio)-triphosphate (dGTPαS) or deoxythymidine-(α,β-imido)-triphosphate, are effective inhibitors of SAMHD1 activity, competitively binding to allosteric or active sites and either blocking oligomerization, catalytic activity, or both (14, 37, 38). Due to their inability to cross cell membranes, however, their application in vivo is limited. Egleston et al. (39) have recently developed a small-molecule inhibitor based on deoxyguanosine-monophosphate, which binds to AS1 with higher affinity than the physiological activator GTP, preventing tetramer formation in vitro. Meanwhile, we performed a high-throughput screen, identifying small-molecule inhibitors that also target the allosteric activation process (40). However, neither inhibitor was able to inhibit SAMHD1 activity in cellulo.

To date, there has been limited investigation of how different AS1 activators modulate SAMHD1 dNTPase activity, and the relationship between AS1 activator chemical structure and the resulting enzymatic activity is so far poorly understood. Structure–function studies have shown that a guanine base is required for nucleotides to bind to AS1 (10, 12, 14). While its intracellular abundance likely makes GTP the physiological AS1 activator, it was recently shown that (d)GDP also induces dNTPase activity in vitro (39). In addition to endogenously occurring nucleotides, Arnold et al. (30, 41) have also identified dideoxy-GTP as well as the triphosphate form of the antiviral guanosine analog acyclovir, acyclovir-triphosphate (Ac-TP), as AS1 activators.

To investigate new avenues for targeting SAMHD1, we were intrigued by these results, especially because the triphosphate metabolite of another close structural analog of GTP and Ac-TP, ganciclovir-triphosphate (Gc-TP) (Fig. 1B), seemingly did not induce enzymatic activity (30). Upon replicating these earlier results, we found that both nucleotide analogs—Ac-TP and Gc-TP, the active metabolites of antiviral agents—are allosteric activators capable of inducing SAMHD1 enzymatic activity, albeit to very different degrees. We therefore wanted to define the mechanism by which these analogs regulate dNTPase activity, thereby helping us to better understand SAMHD1’s allosteric regulation and oligomerization process while also exploring AS1-activator analogs as a potential avenue to pharmacologically control SAMHD1 activity.

Results

Ac-TP and Gc-TP are AS1 activators that differentially activate dNTP hydrolysis

Intrigued by previously published data by Arnold et al. (30), we wanted to determine whether the guanine nucleotide analogs Ac-TP and Gc-TP are allosteric activators or substrates of SAMHD1. Using an improved variant of our enzyme-coupled activity assay (42), we incubated SAMHD1 with potential nucleotide ligands, in this case Ac-TP or Gc-TP, together with different nucleotides established as AS1-/AS2-activators and/or substrates. We subsequently measured released phosphate as an indicator of SAMHD1 dNTP hydrolysis. By this approach, we can deduce at which site the potential nucleotide ligands bind and whether they can be hydrolyzed (Fig. 2A). Validating this setup, in the absence of any additional nucleotides, only the known AS1-/AS2-activator and substrate dGTP activated SAMHD1 dNTP hydrolysis (Fig. 2B), whilst in combination with the AS1-activator GTP, the addition of the known AS2-activator and substrate dATP activated dNTP hydrolysis (Fig. 2B).Figure 2Acyclovir-triphosphate (Ac-TP) and ganciclovir-triphosphate (Gc-TP) are allosteric site 1 (AS1) activators that differentially activate dNTP hydrolysis. A, rationale of the enzymatic assay. SAMHD1 is incubated with test compounds (100 μM, "X") and allosteric activators, substrates, or buffer (indicated by color), allowing determination of where test compounds bind. Control compounds: GTP (AS1-activator), dGTP (AS1-/AS2-activator/substrate), and dATP (AS2-activator/substrate). B, SAMHD1 was incubated with different combinations of activators/substrates/buffer (indicated by color) and test compounds (left) in the enzyme-coupled assay. SAMHD1 activity normalized to buffer + dGTP (100 μM) control. Bars indicate means of three independent experiments, where dots represent individual values of technical duplicates. Error bars indicate SD. C, SAMHD1 was incubated with varying concentrations of GTP, Ac-TP, or Gc-TP and 100 μM of the indicated dNTP in the enzyme-coupled assay. SAMHD1 activity was normalized to dATP (100 μM) + GTP (100 μM) control. Data points indicate the mean of two independent experiments; each performed with technical duplicates; error bars indicate SD. SAMHD1, SAM and HD domain–containing protein 1.

When SAMHD1 was incubated with Ac-TP, dNTP hydrolysis was only observed when combined with dATP, as both the AS2-activator and substrate (Fig. 2B), indicating Ac-TP is an AS1 activator. Consistent with the report by Arnold et al. (30), dATP hydrolysis was reduced 1.7-fold relative to GTP-activated SAMHD1. In the case of Gc-TP, only a slight increase in dNTP hydrolysis was observed with the addition of dATP (Fig. 2B), suggesting this could also be an AS1 activator. Compared with GTP-activated SAMHD1, Gc-TP-induced dATP hydrolysis was reduced by 11.8-fold. No activity was observed when the analogs were combined with ara-CTP, which is a substrate but not an allosteric activator of SAMHD1 (31), indicating that Ac-TP and Gc-TP are not AS2 activators (Fig. S1). For both nucleotide analogs, no activity was observed in combinations with dGTPαS (Fig. 2B), which is an AS1-/AS2-activator but not a substrate of SAMHD1, indicating that Ac-TP and Gc-TP are not hydrolyzed by SAMHD1.

To compare SAMHD1 allosteric activation by Ac-TP and Gc-TP with that of the physiological activator GTP, we next incubated SAMHD1 with varying concentrations of these nucleotides and measured the activity after dNTP addition. This also allowed us to compare the hydrolysis of three endogenous dNTP substrates, dATP, dTTP, and dCTP (Fig. 2C). In all combinations, activation by GTP induced the highest maximum SAMHD1 dNTPase activity. In comparison, Ac-TP-activated hydrolysis of dATP and dTTP was reduced by about 30% and 60%, respectively, whereas no hydrolysis of dCTP was observed. Gc-TP failed to activate both dTTP and dCTP hydrolysis. However, even though absolute activity was very low, dATP hydrolysis was repeatedly observed (Fig. 2C).

Taken together, these data demonstrate that the guanine nucleotide analogs, Ac-TP and Gc-TP, are allosteric activators of SAMHD1 dNTP hydrolysis and indicate that this activation is via AS1 binding. The resulting SAMHD1 activity, however, is reduced compared with the physiological activator GTP. Activity of Ac-TP- and Gc-TP-activated SAMHD1 also depends on the identity of the dNTP substrate and AS2 activator.

Ac-TP and Gc-TP bind to AS1 with higher affinity than the physiological activator GTP

The ability of Ac-TP or Gc-TP to activate dNTP hydrolysis would be consistent with AS1 binding. To directly test if reduced AS1 binding of Ac-TP and Gc-TP is responsible for the reduced dNTPase activity, we used the environmentally sensitive 2′/3′-O-(N-methyl-anthraniloyl)-GTP (mant-GTP) probe, which has enhanced fluorescence emission upon binding to hydrophobic regions of a protein, for example, SAMHD1 (43). Previously, this probe has been employed by Orris et al. (44) to study AS1 binding of single-stranded DNA. We adapted the assay to measure displacement of the probe by Ac-TP and Gc-TP compared with GTP (Fig. 3A). We first confirmed that mant-GTP could effectively activate SAMHD1 dNTP hydrolysis by comparing it with the endogenous AS1 activator GTP (Fig. S2A). The resulting maximal activity is comparable to GTP, consistent with AS1 binding.Figure 3Acyclovir-triphosphate (Ac-TP) and ganciclovir-triphosphate (Gc-TP) bind to allosteric site 1 (AS1) and induce tetramerization. A, schematic of the AS1 competitive binding assay using the environmentally sensitive mant-GTP probe. As mant-GTP is displaced from AS1 by competitors, fluorescence signal decreases. B, SAMHD1 (2 μM) and mant-GTP (0.5 μM) were incubated with varying concentrations of AS1 competitors. Data points indicate means of three independent experiments; each performed with technical duplicates. No-compound controls for each competitor dilution series were considered as shared technical replicates (N = 6). Error bars indicate SD. pKi values ± SD were calculated from a competitive binding equation (see the Experimental procedures section). ANOVA shows significant differences between pKi values for the three AS1 competitors (p = 0.0007). Results of a post hoc Tukey’s multiple comparisons are indicated by asterisks, where ∗∗∗ and ∗∗ indicate adjusted p values ≤0.001 and ≤0.01, respectively. GTP versus Ac-TP: adjusted p = 0.0006; GTP versus Gc-TP: adjusted p = 0.0041; and Ac-TP versus Gc-TP: adjusted p = 0.1331. C, mass photometry quantification of SAMHD1 oligomeric states (monomer: M, dimer: D, and tetramer: T) after incubation with AS2 activator dATP (250 μM) and increasing concentrations of either GTP, Ac-TP, or Gc-TP between 0 and 1200 μM. Data shown are representative of three independent experiments. D, quantification of relative abundance of monomer, dimer, and tetramer oligomeric states as seen in C. Intensities are grouped by oligomeric state, and colors indicate different AS1 activators. Data points indicate means of three independent experiments; error bars indicate SD. mant-GTP, 2'/3′-O-N-methyl-anthraniloyl)-GTP; SAMHD1, SAM and HD domain–containing protein 1.

Incubating mant-GTP alone with an increasing concentration of recombinant SAMHD1 results in increased fluorescence (Fig. S2B). A KD of 1 μM in the absence of competitors was calculated using a saturation binding equation (see the Experimental procedures section) and was consistent with the previous results by Orris et al. (44), although we observed an apparent binding cooperativity with a Hill slope of 2.2 that had not been reported before. Upon addition of an AS1 competitor, mant-GTP would be displaced from the AS, and the fluorescence decrease would be proportional to the relative affinities for the binding site. As expected, the addition of GTP, Ac-TP, or Gc-TP strongly reduced the SAMHD1-dependent increase in fluorescence, consistent with competition for the AS (Fig. S2B).

In a reversed experimental setup, varying the concentrations of AS1 competitors, equilibrium dissociation constants (Ki) can be calculated using a competitive binding equation (see the Experimental procedures section). The physiological AS1 activator GTP completely displaces mant-GTP at concentrations of >100 μM, with a Ki of 6.47 μM. Surprisingly, Ac-TP and Gc-TP display significantly higher affinities, with Ki values of 1.17 μM and 1.95 μM, respectively (Fig. 3B).

Altogether, the data demonstrate that both Ac-TP and Gc-TP bind to AS1 and do so with a higher affinity compared with the physiological activator GTP. Considering the observed differences in enzymatic activity between these activators, this implies that another effect, rather than AS affinity, is responsible.

Ac-TP and Gc-TP induce tetramerization

As the mant-GTP competitive binding assay showed stronger AS1-binding affinities for both Ac-TP and Gc-TP when compared with the physiological activator GTP, we next asked if the reduced dNTPase activity of Ac-TP- and especially Gc-TP-activated SAMHD1 was a result of reduced oligomerization. The SAMHD1 oligomerization process requires allosteric activation through sequential (d)NTP binding and results in the formation of an enzymatically active homotetramer (Fig. 1A).

Using a mass photometry oligomerization assay, we measured the changes in relative abundance of monomer, dimer, and tetramer fractions of recombinant SAMHD1 incubated with increasing concentrations of GTP, Ac-TP, or Gc-TP in the presence of AS2-activator dATP. In the assay, all three nucleotides—GTP, Ac-TP, and Gc-TP—effectively induced tetramerization of the recombinant SAMHD1 protein in a dose-dependent manner, starting from as low as 5 μM (Fig. 3, C and D). Nevertheless, at the highest concentration used (1200 μM), whilst GTP sustained a high level of SAMHD1 tetramers at 83% of the total protein population, Ac-TP had a lower level of tetramer induction at 66%. The tetramer fraction in Gc-TP-activated SAMHD1 even drops from 52% at 400 μM to only 32% at 1200 μM (Fig. 3D).

We independently confirmed the ability of Ac-TP and Gc-TP to induce SAMHD1 tetramers using a disuccinimidyl glutarate (DSG) chemical crosslinking assay (Fig. S3A). In this orthogonal assay, we observed no discernible difference between the fraction of SAMHD1 tetramers induced by GTP, Ac-TP, and Gc-TP at the tested concentrations, and the concentration of activators required to induce tetramerization was comparable to that observed with mass photometry (>1 μM) (Fig. S3E). The two assays differed, however, in the baseline and maximal fraction of SAMHD1 tetramers. Virtually, no tetramers are observed in the mass photometry assay at activator concentrations below 1 μM, whereas chemical crosslinking shows tetramers even in the absence of any activators (Fig. S3, D and E). On the other hand, tetramer fractions top out at 60% to 70% at the highest tested activator concentrations in the crosslinking assay (Fig. S3, D and E), whereas such a plateau is not observed when assessing oligomerization using mass photometry (Fig. 3D). Despite these differences, both assays indicate that oligomerization patterns are similar between physiological activator GTP and the Ac-TP and Gc-TP analogs.

When GTP, Ac-TP, or Gc-TP is titrated in the absence of AS2-activator dATP, a pronounced rise in dimer fraction is observed in both the mass photometry and DSG crosslinking assays (Fig. S3, B and C and Fig. S4), which was not observed in the presence of AS2-activator dATP. Surprisingly, both assays also show tetramer formation at GTP/Ac-TP/Gc-TP-activator concentrations as low as ∼10 μM (Fig. S3, B and C and Fig. S4).

To summarize, two orthogonal assays show that Ac-TP and Gc-TP can support SAMHD1 oligomerization. Importantly, SAMHD1 showed similar oligomerization behavior in the presence of either GTP or Ac-TP/Gc-TP analogs. This implies that another factor, rather than a defect in oligomerization, contributes to the observed differences in enzymatic activity between the activators.

Ac-TP and Gc-TP modulate the SAMHD1 kinetic profile

Both Ac-TP and Gc-TP bind to AS1 and induce tetramerization to a similar degree as the physiological activator GTP but activate dNTP hydrolysis to vastly different extents. To understand this discrepancy, we next aimed to investigate the kinetic properties of analog- versus GTP-activated SAMHD1. We used an ^1^H-NMR–based continuous dNTPase assay, first employed by Bhattacharya et al. (20), which allows us to follow the depletion of dNTP substrates and buildup of deoxyribonucleoside products based on the relative ratio of integrated NMR peaks (Fig. 4A).Figure 4Acyclovir-triphosphate (Ac-TP) and ganciclovir-triphosphate (Gc-TP) modulate SAMHD1 activity by changing the kinetic profile. A, an overlay of five NMR spectra, showing peaks for SAMHD1 substrate and activator dATP, as well as reaction product dA. Different measurement timepoints (2, 5, 10, 15, and 20 min after reaction start) show reaction progress over time. B, Michaelis–Menten (GTP) or allosteric sigmoidal (Ac-TP/Gc-TP) equation parameters fitted to experimental data by least squares regression for the depletion of 500 μM dATP in the presence of 250 μM GTP, Ac-TP, or Gc-TP AS1 activator by SAMHD1 (1 μM). Data are shown representative of three independent experiments. NMR spectra were recorded in 30 s intervals for GTP- and Ac-TP-activated SAMHD1 and 5- or 7-min intervals for Gc-TP-activated SAMHD1. Estimated Michaelis–Menten equation parameters for GTP-activated dATP depletion (numbers in brackets indicate SD for three experimental replicates): kcat: 0.88 s^-1^ (±0.54 s^-1^), KM: 140 μM (±84 μM), kcat/KM: 0.006 μM^-1^ s^-1^ (±0.001 μM^-1^ s^-1^). Estimated allosteric sigmoidal equation parameters for Ac-TP-activated dATP depletion (numbers in brackets indicate SD for three experimental replicates): kcat: 0.99 s^-1^ (±0.58 s^-1^), Khalf: 244 μM (±46 μM), h: 2.1 (±0.2), kcat/Khalf: 0.004 μM^-1^ s^-1^ (±0.002 μM^-1^ s^-1^). Estimated allosteric sigmoidal equation parameters for Gc-TP-activated dATP depletion (numbers in brackets indicate SD for three replicates): kcat: 0.03 s^-1^ (±0.01 s^-1^), Khalf: 211 μM (±25 μM), h: 3.1 (±0.6), kcat/Khalf: 0.13 nM^-1^ s^-1^ (±0.08 nM^-1^ s^-1^). SAMHD1, SAM and HD domain–containing protein 1.

Michaelis–Menten equation parameters were fit to the experimental data, resulting in high-confidence fits for GTP-activated hydrolysis of dATP (R^2^ > 0.99 for each replicate experiment) (Fig. 4B, left). Although kinetic parameter estimates varied between replicate experiments, averages (kcat: 0.9 s^-1^, KM: 140 μM) are comparable to those previously reported. In contrast, trying to fit the Michaelis–Menten equation parameters to the dATP depletion data obtained with Ac-TP-activated SAMHD1 produced low-confidence parameter estimates. This is due to a clear plateau in dATP hydrolysis at low substrate concentrations, where Ac-TP-activated dATP hydrolysis slows down considerably (Fig. 4B, left). Fitting the parameters of an allosteric sigmoidal equation instead resulted in a high-confidence fit (R^2^ > 0.99 for each replicate experiment) (Fig. 4B, left). Parameter estimates from experimental replicates resulted in averages with slightly reduced catalytic efficiency (kcat: 1 s^-1^, Khalf: 244 μM), whereas a Hill coefficient (h) of 2.1 indicates dATP binding cooperativity.

Even though Gc-TP-activated SAMHD1 showed very low catalytic activity in the enzyme-coupled assay, dATP depletion was clearly observed in the NMR-based assay. Remarkably, Gc-TP-activated SAMHD1 was able to deplete >80% of dATP substrate over a period of 12 h, emphasizing the ability of Gc-TP to trigger the formation of stable and enzymatically competent SAMHD1 tetramers. While estimates for h (3.1) and Khalf (211 μM) are similar to those for Ac-TP-activated SAMHD1, the substrate turnover is drastically reduced with an estimated kcat of 0.03 s^-1^ (Fig. 4B, right).

To summarize, while GTP-activated SAMHD1 follows Michaelis–Menten kinetics, we show that Ac-TP- and Gc-TP-activated SAMHD1 follow a sigmoidal kinetic profile with positive dATP binding cooperativity and, in the case of Gc-TP, further reduced substrate turnover.

Ac-TP and Gc-TP cooperate with endogenous AS1 activator GTP to form functional tetramers

Thus far, we have shown that guanine nucleotide analogs, Ac-TP and Gc-TP, bind to AS1 and induce the formation of catalytically competent SAMHD1 tetramers. Depending on the activator–substrate combination, dNTPase activity is strongly reduced. This is especially apparent for Gc-TP-activated SAMHD1. We therefore wanted to test whether the reduced activity of guanine nucleotide analog–activated SAMHD1 would translate into inhibition through competition with the endogenous AS1-activator GTP in the enzyme-coupled SAMHD1 activity assay.

In a two-dimensional concentration gradient, we first preincubated SAMHD1 with a range of GTP concentrations before adding Ac-TP/Gc-TP competitors (or a GTP control) and starting the dNTPase reaction with the addition of a fixed concentration of AS2-activator and substrate dATP (see schematic in Fig. 5A). In the absence of competitors, increasing GTP preincubation concentrations led to increasing activity (Fig. 5B, left, 0 μM preincubation concentration curves). GTP added at the reaction start further increased SAMHD1 activity. At preincubation concentrations of 50 μM GTP, no further increase in activity was observed when supplying additional GTP (Fig. 5B). In the absence of GTP in the preincubation step, Ac-TP and Gc-TP produced intermediate or unobservable activity, respectively (Fig. 5, C and D, left, 0 μM preincubation concentration curves), as seen in previous experiments (see Fig. 2C for comparison).Figure 5Combination of GTP and acyclovir-triphosphate (Ac-TP)/ganciclovir-triphosphate (Gc-TP) allosteric activators promotes SAMHD1 activity. A, a schematic representation of the experiment flow. SAMHD1 was preincubated with varying concentrations of GTP (0–50 μM, indicated by color gradient in B–D), followed by the simultaneous addition of varying concentrations of AS1-activators GTP (B), Ac-TP (C), or Gc-TP (D) (0–50 μM) and a fixed concentration of AS2-activator and substrate dATP (50 μM). B–D, dATP substrate turnover was quantified using the enzyme-coupled activity assay, and activity was normalized to GTP (50 μM, preincubation) + GTP (50 μM, added) control. Select experimental conditions are highlighted as bar charts for easier comparison. Left, data points indicate means of three independent experiments; each performed with technical duplicates; error bars indicate SD. Right, bars indicate means of three independent experiments, where dots represent individual values of technical duplicates. Error bars indicate SD. SAMHD1, SAM and HD domain–containing protein 1.

Surprisingly, combinations of low GTP preincubation concentrations together with Ac-TP/Gc-TP shifted the enzyme activity to higher levels than would be naively expected. Gc-TP only induces 1% SAMHD1 activity when normalized to the maximal activity in the GTP-incubation control, even at the highest tested concentration of 50 μM (Fig. 5D, right). Likewise, a GTP concentration of 6.25 μM only induces 6% dNTPase activity when used as the sole activator (Fig. 5D, right). Combined, however, SAMHD1 dNTPase activity is elevated up to 50% in a Gc-TP concentration–dependent manner (Fig. 5D, right). While still below the maximal activity seen with GTP-activated SAMHD1, the resulting activity (50%) is much greater than the sum of the activities observed with each activator alone (1% and 6%). The same cooperative effect is observed with Ac-TP, where the activity resulting from a combination of high Ac-TP (50 μM) and low GTP (6.25 μM) is much greater (68%) than the sum of activities observed with each activator alone (32% and 5%) (Fig. 5C, right). In contrast, a combination of high GTP addition (50 μM) with low GTP preincubation (6.25 μM) results in 69% normalized activity, which is very close to the sum of activities observed with each activator alone (60% and 5%). Altogether, these data show that Ac-TP and Gc-TP can complement GTP in allosterically activating SAMHD1, especially at low GTP concentrations.

Discussion

SAMHD1 is a dNTPase that controls intracellular dNTP pools and plays diverse roles in human health and disease, including promotion of chemoresistance by hydrolyzing active triphosphate metabolites of nucleoside analog drugs (31, 32, 33, 34, 35, 36). The dNTPase activity of SAMHD1 is allosterically regulated by nucleotide binding at two distinct sites—AS1 and AS2—that, when occupied, promote formation of the catalytically competent homotetramer (Fig. 1A) (10, 11, 12, 14). To explore alternative strategies to pharmacologically modulate SAMHD1 activity and probe the untapped potential of allosteric drugs, we investigated AS1 ligands. Specifically, we characterized the interactions between SAMHD1 and GTP analogs, Ac-TP and Gc-TP—the active metabolites of antiviral therapies acyclovir and ganciclovir, respectively—to understand the relationship between AS1-activator structure and enzyme activity.

Following an earlier study by Arnold et al. (30) we were curious about the mechanism(s) behind the differences in dNTPase activity between GTP-, Ac-TP-, and Gc-TP-activated SAMHD1. In replicating these results, we found that Ac-TP (as previously reported) but also Gc-TP can act as AS1 activators and induce SAMHD1 activity. McCown et al. (16) have recently characterized the influence of AS2-activator identity on the substrate preference of SAMHD1. Based on our results, we suggest that a similar mechanism exists for the AS1-activator identity. We found that the activity of Ac-TP- and Gc-TP-activated SAMHD1 varies drastically, depending on the identity of AS2-activator and substrate. Gc-TP-activated SAMHD1 showed detectable activity only toward a dATP substrate, which agrees with the lack of activity toward dTTP observed by Arnold et al. (30). Strikingly, dCTP hydrolysis was completely absent with both Ac-TP- and Gc-TP-activated SAMHD1.

Seeking to understand the mechanism for differential SAMHD1 activity, we first ruled out a lack of AS affinity and an inability to induce oligomerization as explanations for the reduced dNTPase activity. In fact, Ac-TP and Gc-TP bind AS1 with significantly higher affinity than the physiological activator GTP.

By performing enzyme kinetics experiments, we find that the activity of GTP-activated SAMHD1 can be explained accurately by Michaelis–Menten kinetics, as had been previously reported (14, 16, 20, 30). In contrast, fitting kinetic parameters to data collected from Ac-TP- and Gc-TP-activated SAMHD1 required a sigmoidal (instead of hyperbolic) model that also takes into account dATP binding cooperativity. Importantly, both Ac-TP- and Gc-TP-activated SAMHD1 display positive dATP-binding cooperativity. We suggest that this cooperativity, together with the overall reduced enzymatic turnover of Gc-TP-activated SAMHD1, plays a role in modulating SAMHD1 activity toward different dNTP substrates (Fig. 6A). The mechanism of this cooperativity is not clear, however, because dATP (and dNTPs in general) binds both AS2 and the catalytic site. Acton et al. (15) recently showed that reduced dNTP concentrations can result in partially depleted AS2 pockets. As nucleotides bound at AS1 and AS2 directly communicate, one could imagine that the identity of the AS1-activator influences not only substrate binding but also AS2-activator affinity to the AS pocket.Figure 6Antiviral nucleotide analogs fine-tune SAMHD1 dNTPase activity. A, in the presence of guanine nucleotide analogs, Ac-TP and Gc-TP, SAMHD1 forms homotetramers that are catalytically competent but display reduced catalytic activity with distinct kinetic profiles compared with those containing the endogenous activator GTP. B, due to the cooperativity between Ac-TP/Gc-TP and GTP in allosterically activating SAMHD1, we suggest the existence of mixed SAMHD1 tetramers where a mixture of GTP- and analog-bound monomers make up a homotetramer. The resulting mixed tetramers can display enzymatic activity comparable with GTP-only-activated SAMHD1. Ac-TP, acyclovir-triphosphate; Gc-TP, ganciclovir-triphosphate; SAMHD1, SAM and HD domain–containing protein 1.

Knowing that Ac-TP and especially Gc-TP can induce the formation of a homotetramer with reduced catalytic activity, and that the nucleotide analogs' affinity to the AS was higher than the physiological activator, we envisioned an enzymatic inhibition mechanism of "trapping" SAMHD1 in an inactive tetramer, where the ubiquitous physiological activator GTP is replaced by Ac-TP/Gc-TP. To our knowledge, catalytically active tetramers with heterogeneously occupied AS1 sites of individual monomers have not been observed, so we were surprised to see an apparent synergistic effect between GTP and Ac-TP/Gc-TP in allosterically activating SAMHD1. We hypothesized that this apparent synergy can only be explained by the formation of mixed-occupancy tetramers consisting of both GTP- and analog-bound monomers (Fig. 6B). While these hypothetical tetramers had not been observed before, they would be analogous to mixed-occupancy tetramers consisting of dATP and dCTP bound to AS2 of different monomers that have been observed already in early structural studies of SAMHD1 (45). A different type of mixed-occupancy SAMHD1 tetramer was identified by Yu et al. (46), who showed that single-stranded oligonucleotides can bind the ASs of one dimer, whereas another dimer is activated by (d)NTP binding, and that these mixed-occupancy tetramers might retain enzymatic activity. At low GTP concentrations, when SAMHD1 is present predominantly in the monomeric and dimeric forms, Ac-TP and Gc-TP might help to overcome an allosteric activation bottleneck by promoting the formation of tetramers. Supporting this hypothesis is Ac-TP/Gc-TP's higher affinity for AS1 observed in the competitive binding assay. It is unclear if all monomers in Ac-TP-/Gc-TP-bound mixed-occupancy tetramers would be active, or if only the GTP-bound monomers could show full catalytic activity. The existence of SAMHD1 tetramer subunits acting independently was recently shown by Acton et al. By resolving catalytically active tetramers using cryo-EM, the authors showed that the predominant catalytically active tetramer species of SAMHD1 consists of two dimers with distinct conformations, where reaction product formation was only observed in one of them (15).

In this work, we developed a suite of biochemical and biophysical assays to holistically examine SAMHD1 tetramerization, catalytic activity, and the intricate interplay between these two protein statuses that are largely considered synonymous. With these assays, unexpectedly, we also observed that SAMHD1 tetramerization could be induced by AS1 activators alone (Fig. S3, B and C and Fig. S4). Since it is well established that SAMHD1 requires dNTP binding to AS2 to form a homotetramer, we hypothesize that these observations are likely explained by the formation of short-lived tetramers, which quickly dissociate in the absence of AS2-activators. In the presence of chemical crosslinking agents, these transiently formed tetramers are further stabilized, leading to the surprisingly high fraction of tetramers in this assay (Fig. S3, D and E). The formation of tetramers in the presence of only GTP was observed by others as well, using glutaraldehyde as a chemical crosslinker (39, 47, 48, 49). Together, this highlights that caution should be used in interpreting the results of chemical crosslinking assays with SAMHD1.

Clinical data support that SAMHD1 confers chemoresistance to deoxycytidine analogs, cytarabine and decitabine (31, 33, 50) and thus inhibition of SAMHD1 could be a promising strategy to improve patient outcomes. Several direct SAMHD1 inhibitors have been identified, often blocking dNTPase activity via competitive inhibition; however, translation into cell-active inhibitors has been challenging (14, 37, 38). Compared with orthosteric drugs, allosteric drugs are suggested to have several benefits, including a generally increased specificity, as ASs are usually less conserved, which also strengthens their ability to differentiate between the target protein and potential homologs (51, 52). High concentrations of physiological activators are also generally associated with lower affinity, making it easier to outcompete them with optimized non-natural allosteric binders (51).

Such an allosteric inhibitor—although lacking cell activity—was recently reported by Egleston et al. (47), who performed a high-throughput fragment-based screen to target SAMHD1's AS. These inhibitors are based on acyclovir connected to a dibromonaphthol moiety via linkers of variable length. Whilst they bind AS1 and induce dimerization of SAMHD1, their mode of inhibition is thought to result from blocking dNTP access to the AS2 binding pocket, thereby preventing formation of the catalytically competent tetramer. However, the effects of total SAMHD1 inhibition are difficult to predict. Outright inhibition of SAMHD1's dNTPase activity would result in increased cellular dNTP pools, which can have a number of deleterious effects, including increased mutagenesis and replication stress (reviewed in Refs. (53, 54). The AS1-mediated fine-tuning mechanism reported here could potentially circumvent these problems by specifically preventing hydrolysis of certain SAMHD1 substrates while not globally affecting dNTP levels. To our knowledge, the allosteric fine-tuning through formation of an enzymatically competent homotetramer described here is the first example of non-natural allosteric activators modulating enzymatic activity toward only a subset of substrates.

Outside the context of SAMHD1, there are additional examples of allosteric inhibitors with similar, but distinct, mechanisms. For example, multiple inhibitors of ribonucleotide reductase, a functional antagonist to SAMHD1 in the regulation of dNTP pools (54), are structural analogs of its physiological allosteric regulators and substrates. Clofarabine-, cladribine-, and fludarabine-di- and triphosphate metabolites are adenosine analogs that all induce the formation of an inactive oligomeric form of the enzyme. Specifically, they trigger the formation of an enzymatically inactive hexameric form of the alpha subunit (55, 56). Despite these similarities, however, the differences are also apparent: first, Ac-TP/Gc-TP induce an enzymatically competent oligomer that can, depending on the activator, retain activity comparable to the GTP-activated enzyme. Second, Ac-TP/Gc-TP-activated SAMHD1 has the same homotetrameric makeup as the GTP-activated enzyme. A different type of allosteric inhibition was described by Ostrem et al., targeting the oncogenic mutant K-Ras(G12C). This mutant shows altered substrate affinity, resulting in increased activity. A binding site found only in this mutant was exploited with allosteric inhibitors to revert the phenotype. This example shows that non-natural allosteric regulators can influence enzyme substrate preference without preventing activity. SAMHD1 shows a similar fine-tuned substrate preference when allosterically activated with Ac-TP. The main difference here is that SAMHD1 activity is modulated with an analog of a physiological activator, binding to the same site as its physiological counterpart, whereas K-Ras is not canonically allosterically regulated.

In summary, here we characterize a mechanism by which SAMHD1 dNTPase activity is regulated via the identity of its AS1 activator. Ac-TP and Gc-TP, analogs of the physiological AS1 activator GTP and active metabolites of antiviral drugs, can bind to AS1 and induce the formation of catalytically active homotetramers with a distinct kinetic profile that follows a sigmoidal model instead of hyperbolic Michaelis–Menten kinetics. Importantly, we find that the identity of the AS1 activator can differentially affect SAMHD1's dNTPase activity toward different physiological substrates. While Ac-TP-activated SAMHD1 hydrolyzes dATP and dTTP (albeit with reduced efficiency compared with GTP-activated SAMHD1), dCTP hydrolysis is completely abolished, and future studies will identify AS1 activators that fine-tune dNTPase activity in different ways. This work therefore provides a foundation for targeting AS1 to selectively control SAMHD1 dNTPase activity, both as chemical biology tools to study SAMHD1 and therapeutically to reduce the detoxification of certain nucleotide analogs while minimizing disturbances in the cellular dNTP pool.

Experimental procedures

Chemicals

GTP (HY-12695) was obtained from MedChemExpress. Ac-TP (NU-877), Gc-TP (NU-275), dGTPαS (NU-424), mant-GTP (NU-206), and ara-CTP (NU-117) were obtained from Jena Bioscience. dATP (27-1850-04), dCTP (27-1850-04), and dTTP (27-1880-04) were obtained from GE Healthcare. DSG (A35392) was purchased from Thermo Fisher Scientific.

Recombinant protein production and purification

Recombinant His-tagged SAMHD1 and pyrophosphatase (PPase) were expressed and purified by the Karolinska Institutet Protein Science Facility, as described previously (31).

Enzyme-coupled SAMHD1 activity assay

An enzyme-coupled assay to test the effect of small molecules on SAMHD1’s dNTPase activity was described previously in a dedicated methods article (42). Here, an expanded variant of this assay is used to discern the role of GTP nucleotide analogs in activating SAMHD1 activity and to quantify SAMHD1 dNTPase activity with different AS1 activators.

To discern the roles of nucleotide analogs as AS1- or AS2-activators or substrates, SAMHD1 was incubated with nucleotide analogs or known (deoxy)nucleotide activator–substrate controls (final concentration: 100 μM), together with either buffer, GTP, dGTPαS, ara-CTP, or dATP, final concentrations 12.5, 12.5, 200, or 100 μM, respectively (see Fig. 2A for a schematic representation of the experimental setup).

To compare SAMHD1 dNTPase activity toward different substrates and in the presence of different AS1 activators, SAMHD1 was incubated with varying concentrations of GTP, Ac-TP, or Gc-TP, together with either dATP, dCTP, or dTTP AS2-activators and substrates (final concentration: 100 μM).

To assess SAMHD1 dNTPase activity of GTP-activated SAMHD1 in the presence of potential nucleotide analog AS1-binding competitors, SAMHD1 was preincubated with varying concentrations of GTP for 10 min before the simultaneous addition of varying concentrations of nucleotide analog competitor (or GTP control) and dATP substrate (final concentration: 50 μM) (see Fig. 5A for a schematic representation of the experimental setup).

All reactions took place in 25 mM Tris acetate buffer at pH 8.0 containing 40 mM NaCl, 1 mM MgCl_2_, 0.005% Tween-20, 0.3 mM Tris(2-carboxyethyl)phosphine (TCEP), 12.5 U/ml Escherichia coli PPase, and 0.35 μM SAMHD1 in 384-microwell plates.

dNTP substrates are hydrolyzed by SAMHD1 into deoxynucleosides and inorganic triphosphates, which is subsequently broken down into three molecules of inorganic monophosphate by the coupled PPase enzyme. After incubation for 20 min (10 min in the case of AS1 activator competition experiments) at room temperature, EDTA stop solution was added (final concentration: 3.95 mM). Reaction progress was quantified by the addition of a malachite green solution (final concentrations: 0.5 mM malachite green, 2.58 mM ammonium molybdate, and 0.036% Tween-20). The color change resulting from complex formation between malachite green, inorganic phosphate, and molybdate was quantified using a Hidex Sense microplate reader (Hidex Oy) to measure the absorbance at a wavelength of 630 nm. The linear range of the assay was determined using a sodium phosphate standard curve.

Mant-GTP AS1 competitive binding assay

The environmentally sensitive fluorophore mant-GTP was used to assess AS1 binding of GTP analogs (see Fig. 3A for a schematic representation of the experimental setup). In one experimental setup, mant-GTP and AS1 competitors were held constant (final concentrations: 0.5 μM and 250 μM, respectively), whereas SAMHD1 concentrations were varied between 0 and 10 μM.

The binding affinity for AS1 binding of mant-GTP in the absence of competitors was determined by fitting the parameters of the "Specific binding with Hill slope" saturation binding equation in GraphPad Prism (version 10.4.1; GraphPad Software) (Equation 1) to the experimental data using least squares regression. Bmax is the maximum fluorescence, KD is the equilibrium dissociation constant, and h is the Hill slope. A positive Hill slope indicates binding cooperativity.

$$[eqn]$$

To test for binding equilibrium, fluorescence was recorded for 90 min, with no apparent change in fluorescence after 10 min (data not shown). To confirm that the KD is not affected by titration, the trace binding partner (mant-GTP) concentration was varied between 0.1 and 10 μM, with no change in apparent KD observed below 2.5 μM (data not shown).

In a reversed setup, varying concentrations of competitors (0–312.5 μM GTP, Ac-TP, and Gc-TP) are incubated with a fixed concentration of SAMHD1 and mant-GTP (final concentrations: 2 μM and 0.5 μM, respectively). pKi values (-log_10_ Ki) were determined by fitting the parameters of the “One site—Fit Ki” competitive binding equation in GraphPad Prism (version 10.4.1) (Equations (2), (3)) to the experimental data using least squares regression. Ki is the equilibrium dissociation constant of the competitor, top and bottom are Y-axis plateaus, KD,mantGTP is the equilibrium dissociation constant of mant-GTP, and [mantGTP] is the concentration of mant-GTP.

$$[eqn]$$ $$[eqn]$$

All mant-GTP binding reactions were incubated in 50 mM Hepes assay buffer at pH 7.5 containing 50 mM KCl, 1 mM EDTA, and 0.5 mM TCEP for 30 min at room temperature before fluorescence was quantified at 330/436 nm excitation/emission wavelength using a Hidex Sense microplate reader.

DSG crosslinking oligomerization assay

To assess oligomerization in the presence of different allosteric activators, SAMHD1 (final concentration: 5 μM) was incubated with varying concentrations of AS1 activators, either in the presence or the absence of AS2-activator dATP (final concentration: 250 μM) for 10 min at room temperature (see Fig. S3A for a schematic representation of the experimental setup). Crosslinking was started by the addition of DSG (final concentration: 1 mM), followed by 30 min of incubation at room temperature. Crosslinking was stopped by the addition of 1 M Tris. All oligomerization reactions took place in 50 mM Hepes buffer at pH 7.5 containing 50 mM KCl, 5 mM MgCl_2_, and 0.5 mM TCEP. Crosslinked protein samples were combined with sample loading buffer (Thermo Fisher, NP0008) and reducing agent (Thermo Fisher, NP0009) before boiling at 95 °C for 5 min. Samples were run on a 4% to 15% gradient polyacrylamide gel (Bio-Rad, #4561085) at 200 V for 40 min in a Tris–glycine–SDS running buffer (Bio-Rad, #1610732) using a Mini-PROTEAN Tetra system. Afterward, the gel was stained with Coomassie Blue R250 solution (0.1% w/v) for 15 min and destained overnight in a shaking water bath. Gels were imaged with an Odyssey Fc imager (Li-Cor) on the 700 nm channel for 2 min. Monomer, dimer, and tetramer bands were quantified using the analysis function in Image Studio 6.0 (LI-COR Biosciences).

Mass photometry oligomerization assay

SAMHD1 oligomerization was assessed using a Two-MP mass photometer (Refeyn) in the presence of different allosteric activators. SAMHD1 (2 μM) was incubated with varying concentrations of AS1 activators (GTP, Ac-TP, or Gc-TP), either in the presence or the absence of 250 μM AS2-activator dATP in 25 mM Tris acetate buffer at pH 8.0 containing 40 mM NaCl, 1 mM MgCl_2_, and 0.3 mM TCEP. Prior to the mass photometry measurement, SAMHD1 was diluted to 0.2 μM in buffer while keeping the activator concentration constant. A droplet of buffer containing allosteric activators was placed on the glass slide, and the diluted oligomerization mix containing 0.2 μM SAMHD1 was added to the droplet, resulting in a final concentration of 20 nM. Data were typically acquired over 60 s using the AcquireMP software (version 2024 R1.1) and analyzed using the DiscoverMP software (version 2024 R1) by fitting Gaussian curves to histogram peaks corresponding to SAMHD1 monomer, dimer, or tetramer. Oligomeric state ratios were calculated from the integrated counts under the respective curve. Molecular mass values were calculated from an IgG standard in the assay buffer described above.

In experiments where total binding events were very low, pairs of low-mass antibinding/binding peaks were observed more prominently. Stemming from background surface noise instead of SAMHD1 protein binding events, these were not included in the analysis.

NMR enzyme kinetics assay

^1^H NMR spectra were recorded on either a Bruker Avance III 600 MHz or Bruker Avance III 700 MHz spectrometer using a room temperature triple resonance probe or cryogenic triple resonance inverse probe, respectively, with temperatures set to 298.2 K. Samples were prepared in a 5% D_2_O buffer (50 mM Tris–HCl [pH 8], 150 mM NaCl, 5 mM MgCl_2_, and 5 mM DTT) in the presence of 1 μM SAMHD1 and 250 μM GTP, Ac-TP, or Gc-TP. The spectrometer and experimental settings were calibrated using samples containing only protein and activators. Reactions were started by the addition of dATP (final concentration: 500 μM) into the NMR tube, and the samples were quickly mixed. Spectra were recorded by collecting four dummy scans followed by eight scans with an acquisition time of 2.0 s per scan and a recycle delay of 0.5 s between scans, resulting in spectra at 30 s intervals. The data were analyzed using TopSpin 4.5.0 (Bruker). dATP substrate and deoxyadenosine product peaks were quantified by peak integration followed by background correction. Due to a difference in relaxation rates between the protons in substrate and product, in combination with the short recycle delay between scans, a peak integral correction factor was needed to keep the sum of the substrate's and product's peak areas constant over time (we have good reason to believe that the sum of concentrations for each substrate–product pair stays constant over time). Concentrations of substrate and product were determined by calculating ratios between the corrected peak integrals and their sum and normalizing to the known total concentration of the substrate–product pair. Substrate hydrolysis rates were calculated by fitting parameters of a Michaelis–Menten (Equation 4) or allosteric sigmoidal model (Equation 5) to the dATP depletion data in GraphPad Prism (version 10.4.1) using custom differential equations where [dATP] is the concentration of dATP measured at each timepoint, d[dATP]/dt is the reaction velocity, kcat is the catalytic rate constant, KM is the Michaelis constant, Khalf is the analogous parameter in the allosteric sigmoidal model, equal to the substrate concentration at half-maximal reaction rate, and h is the Hill coefficient, an indicator for cooperativity.

$$[eqn]$$ $$[eqn]$$

Data availability

The data that support the findings of this study are available in the supporting information.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.