Evaluating the Therapeutic Potential of a Non-Natural Nucleotide that Inhibits Human Ribonucleotide Reductase

Abstract

Human ribonucleotide reductase (hRR) is the key enzyme involved in de novo dNTP synthesis and thus represents an important therapeutic target against hyperproliferative diseases, most notably cancer. The purpose of this study was to evaluate the ability of non-natural indolyl-2’-deoxynucleoside triphosphates to inhibit the activity of hRR. The structural similarities of these analogs with dATP predicted that they would inhibit hRR activity by binding to its allosteric sites. In silico analysis and in vitro characterization identified one particular analog designated as 5-nitro-indolyl-2'-deoxyribose triphosphate (5-NITP) that inhibits hRR. 5-NITP binding to hRR was determined by isothermal titration calorimetry. X-ray crystal structure of 5-NITP bound to RR1 was determined. Cell-based studies demonstrated the anti-cancer effects of the corresponding non-natural nucleoside against leukemia cells. 5-NITP binds to hRR with micromolar affinity. Binding does not induce hexamerization of hRR1 like dATP, the native allosteric inhibitor of hRR that binds with high affinity to the A-site. The X-ray crystal structure of S. cerevisiae RR1-5-NITP (ScRR1-5-NITP) complex determined to 2.3 Å resolution shows that 5-NITP does not bind to the A-site but rather at the S-site. Regardless, 5-NIdR produces cytostatic and cytotoxic effects against human leukemia cells by altering cell-cycle progression. Our studies provide useful insights towards developing new inhibitors with improved potency and efficacy against hRR.

Introduction

Ribonucleotide reductase (RR) is the sole enzyme that catalyzes the reduction of ribonucleoside diphosphates of adenine, guanine, uridine and cytosine to their corresponding deoxyribose form. RR is a multi-subunit enzyme consisting of a large α (RR1) and a small β (RR2) subunit (1). RR1 contains two allosteric sites and a catalytic site, while the RR2 subunit houses a free radical required for catalysis (1). There are four classes of RR that are classified based on free-radical chemistry (2). All eukaryotic RRs belong to class I that use a tyrosyl free-radical (3, 4). RR1 contains a specificity site (S-site) that upon binding of the nucleoside triphosphates ATP/dATP, TTP, or dGTP dictates the selection of CDP/UDP, GDP and ADP substrates, respectively, for their conversion to the corresponding deoxyribose forms at the catalytic C-site (5) (Fig. 1A). In addition, the N-terminus of RR1 contains a four helical bundle ATP binding cone called the activity site (A-site) (6). Binding of ATP at the A-site activates RR while the binding of dATP at this site inhibits RR activity (5).

Figure 1.

The structure of ScRR1-5-NITP. (A) Cartoon diagram of ScRR1 with the monomers colored in salmon and cyan. P-site (orange), 5-NITP (red), A-site (yellow), C-site (blue), Loop1 (green), and disordered Loop2 (dashed black). (B) 2F_o–F_c electron density map of 5-NITP contoured at 1.5 σ. (C) Superposition of main chain ScRR1-5-NITP (cyan), hRR1- dATP (magenta), and ScRR1-AMPPNP (orange). (D) Cartoon depicts the interactions of 5-NITP (green) with the S-site (the monomers at the dimer interface composing the S-site are shown in yellow and cyan, respectively). (E) Structures and electron density surface potentials of natural and non-natural nucleotides that were screened, in silico, against hRR1. For ease, only the nucleobase portion of the corresponding deoxynucleotide is provided. Models were generated using Spartan '04 software (40). Electronegative regions are indicated in red, neutral region are in green, and the electropositive regions are in blue as previously described (36).

While the above-mentioned selection rules (5) are important for maintaining a balanced nucleotide pool, the molecular basis for this selection remained undefined for many years. However, insights into the intricate mechanism were made clear by several crystallographic studies (7–10). These studies show that a polypeptide chain called loop 2 (residues 285–296) connects the specificity site to the catalytic site (7–10). In the AMPPNP structure of ScRR1, loop 2 acts as a steric gate that blocks substrate binding at the C-site. This is in agreement with biochemical data showing that hRR activity is only 10% in the absence of nucleotide binding at the S-site (11, 12). When a nucleoside triphosphate binds at the S-site, loop 2 shifts away from the C-site towards the S-site to create space for substrate binding at the C-site (7). The nucleoside triphosphate that binds at this S-site show extensive interactions with residues 255–270 of Loop 1 (Fig. 1A).

Several studies have shown that oligomerization of eukaryotic RR1 is required for regulation by the activator, ATP, and the inactivator, dATP (11–15). RR1 forms hexamers at physiological concentrations of 3 mM ATP and 20–50 µM dATP. The dATP induced hexamers are shown to be inactive while dimers retain activity (15). The latter was demonstrated by the discovery of the D16R human and yeast RR1 mutants that form dATP induced dimers but not hexamers which retain wild-type-like activity (15). The dATP bound ScRR1 structure revealed that the first 18 residues of the A-site are at the hexamer interface providing an elegant model for dATP induced hexamerization. It appears that ATP-induced hexamers have a different packing arrangement from the dATP-induced hexamers explaining how one can be active while the other inactive (15). A recent study by Stubbe and co-workers showed that the anti-cancer drug Clofarabine, a dATP analogue, also causes hRR1 to hexamerize and further highlight the importance of oligerimerzation as a way to modulate hRR activity (16).

The crucial role of RR during replication makes it an important anti-viral and anti-cancer target (17–19). RR has four druggable sites on hRR1. These include the specificity site (S-site), the activity site (A-site), the catalytic site (C-site), and a peptide-binding site (P-site) (see Fig. 1A) (20, 21). Nucleoside analogs such as Clofarabine and gemcitabine are important anti-cancer agents that can inhibit hRR activity by targeting the allosteric and catalytic sites of the enzyme (16, 22, 23, 24,). In addition Fludarabine and cladribine are clinically used drugs and their metabolites target the hRR activity (25–27). In our previous structural study, we defined the molecular interactions between gemcitabine diphosphate interactions and the ScRR1 C-site (23). However, gemcitabine can also inhibit hRR through a mechanism that involves altering the oligomeric state of hRR. In this case, recent biochemical studies demonstrated that gemcitabine diphosphate targets the hRR1 hexamer and inactivates it through a covalent modification (24).

In this current study, we investigate the ability of non-natural nucleotides to function as hRR1 inhibitors. We chose 5-substituted indolyl-2’-deoxynucleoside triphosphates since these analogs mimic the size and shape of dATP (Fig. 1E). Additionally, previous studies demonstrated that these analogs function as effective surrogates for dATP during the misreplication of damaged DNA (28). In this study, we examined if these non-natural nucleotides can interact with the various druggable sites on hRR1 to generate anti-cancer effects. In silico screening identified 5-NITP as a potential lead candidate that can interact favorably with the A- and S-site of hRR1. We provide biochemical evidence that 5-NITP is a moderate inhibitor of hRR1. In addition, the corresponding non-natural nucleoside produces cytostatic effects against Jurkat cells, consistent with a mechanism involving the inhibition of hRR activity inside a cancer cell. The data from combined functional and structural studies illustrate how non-natural nucleotides can be rationally designed to inhibit key chemotherapeutic targets. Furthermore, a structural study provides insight into the design of additional non-natural analogs that possess improved selectivity and affinity as inhibitors of RR.

Materials and methods

Compound Synthesis

5-NIdR and 5-NITP were synthesized and characterized as previously described (28).

Docking non-natural nucleotides into hRR1

In silico docking of the non-natural nucleotide library was performed using Surflex dock module (29) integrated in Sybyl8.1.1. Non-natural nucleotides were docked against the crystal structure of hRR1 in complex with TTP and dATP bound at the S- and A-allosteric sites, respectively (15). The docked hits were scored using docking function and a consensus scoring function that averages score from many scoring functions (C-score). The docking function takes into account a linear combination of non-linear functions of atomic surface distances between proteins and ligand, steric, polar, entropic and solvation effects (29).

Expression and purification of hRR1, hRR2 and ScRR1

hRR1 and hRR2 were expressed and purified as described in Fairman et al. (15). ScRR1 was expressed and purified as described previously (7, 23). Briefly, the RR1 subunit of both hRR and ScRR are purified using peptide-affinity chromatography. The small subunits of hRR and ScRR were purified using Ni-affinity chromatography. The iron was loaded onto the small subunit of hRR and ScRR using the procedures outlined previously (15).

IC₅₀ determination

The activity of hRR was determined using in vitro ¹⁴C-ADP reduction assays as previously described.(14, 23) Briefly, the buffer solution (50 mM HEPES at pH 7.6, 5% (v/v) glycerol, 0.1M KCl) and the hRR2 protein solution were brought inside a glove box under deoxygenated conditions. A total of 5 equivalents of Fe(II) per hRR2 dimer from FeNH₄SO₄ based on Ferozine assay was added to the protein solution and incubated at 4°C in the glove box. The protein solution was removed from the glove box and the O₂-saturated buffer was added. Excess iron was removed by S200 10/300 size exclusion chromatography. To determine the specific activity of hRR1, the reaction mixture contained 0.3 µM hRR1 and 2.1 µM hRR2 in an activity assay buffer of 50 mM HEPES pH 7.6, 15 mM MgCl₂1 mM EDTA, 100 mM KCl, 5 mM DTT, 3 mM ATP, 100 µM dGTP and 1 mM ¹⁴C-ADP (~3000 cpm/nmol)). The reaction mixture was pre-incubated for 3 min at 37°C, and 30 µL aliquots were sampled at fixed time intervals after reaction initiation. Reactions were quenched by immersion in a boiling water bath, cooling, and treatment with alkaline phosphatase. Product ¹⁴C-dADP that formed during the reaction was separated from substrate ¹⁴C-ADP using boronate affinity chromatography (24). The amount of ¹⁴C-dADP formed was quantified by liquid scintillation counting using a Beckman LS6500 liquid scintillation counter. The IC₅₀ was determined by putting the specific activity of hRR1 at varying 5-NITP concentrations and determining the concentration of 5-NITP at 50% activity.

Multi-angle light scattering analysis of hRR1 bound with 5-NITP

Multi-angle light scattering (MALS) experiments was performed immediately following size-exclusion chromatography (SEC) by online measurement of static light scattering (mini DAWN TREOS, Wyatt Technology), differential refractive index (dRI, Optilab rEX, Wyatt Technology) at a wavelength of 658 nm and ultraviolet absorbance at a wavelength of 280 nm (Dionex ultimate 3000 variable wavelength detector). A ProSEC 300S, 250 × 4.6 mm SEC column was connected upstream of the MALS-RI detectors and used to fractionate the injected sample. The SEC-MALS-RI system as a whole was validated using BSA (Sigma-Aldrich). 20 µl samples were injected onto an analytical SEC column ProSEC 300S, 250 × 4.6 mm (Varian, Inc). Prior to sample injection, the column was equilibrated at a flow rate of 0.3 ml/min in 50 mM Tris pH 7.6, 5 mM MgCl₂100 mM KCl, 20 µM 5 NITP. The chromatograms and resultant molecular weight data were analyzed using the Astra 5.3 software from the Wyatt Corporation.

Isothermal titration calorimetry (ITC)

5-NITP binding to hRR1 subunit was measured by ITC using a VP-ITC200 instrument (Microcal 432 Inc., Northampton, MA USA). 1.0 µl aliquots (except the first injection used 0.5 µl) of 1.0 mM 5-NITP in buffer A (50 mM Tris buffer pH 7.9 containing 5 mM MgCl₂5 mM DTT, 5% (v/v) glycerol) were injected into the cell containing 13 µM hRR1 in buffer A at 25 °C. The blank titration for hRR1 was conducted in buffer A at 25 °C. The integrated heat of injection after correcting for the blank buffer was used to fit the two site sequential binding model using Microcal Origin 7.0 (Microcal Inc., Northampton, MA USA). The observed binding constants were used to calculate the Gibbs free energy relationship (ΔG) using ΔG = −RTlnK_obs. ΔS and −TΔS were calculated from ΔG using the Gibbs free energy equation, ΔG = ΔH - TΔS.

Crystallization

ScRR1 was crystallized as previously described (7). The ScRR1-NITP complex was obtained by powder soaking for two hours. We have found through experience with the ScRR1 system that the powder soaking method is the least harmful to crystals rather than soaking them in a different buffer from the mother liquor. After transferring the soaked crystals into the cryo-protectant solution (0.1 M sodium acetate at pH 6.5, 25% PEG 3350, 0.2 M ammonium sulfate, 20% (v/v) glycerol), the crystals were flash frozen in liquid nitrogen for data collection.

Data Collection

Data for the ScRR1-NITP crystal at 2.3 Å resolution belonging to space groups P2₁2₁2 were collected at the GMCA 23ID-D beam line at the Advanced Photon Source at 100°K and were processed using HKL2000 (30). We also screened hRR1-5-NITP crystals at NE-CAT 24ID-E beam line.

Structure determination, refinement, and analysis

The structure of the ScRR1-NITP crystal was solved by the difference Fourier method. The model used was PDB ID 2CVV without any ligand, ion, or water. To decrease model bias, high temperature (8000 °K) simulated annealing was performed at the beginning of the refinement procedure using the Phenix software suite (31). Model building was interspersed with refinement using Coot (32) and Phenix, respectively. Figures were generated using Pymol (33). Contacts between 5-NITP and ScRR1 were analyzed using Contact in CCP4i (34). Solvent accessible areas of the substrates in ScRR1 and hRR1 were analyzed using Areamol (35) incorporated in CCP4i.

Cell culture procedures

Jurkat cells (ATCC) were cultured in a humidified atmosphere of 5% CO₂ at 37° C. Cells were maintained in Cellgro^® formulated RPMI-1640 supplemented with 10% heat-inactivated FBS, 5% L-glutamine, and 2.5% penicillin/streptomycin antibiotic. Cells were routinely propagated and used for experiments in logarithmic phase.

Cell Proliferation Assays

All the cell lines were authenticated by ATCC. Jurkat cells were obtained from the American Type Culture Collection (Manassas, VA, USA). Jurkat cells were maintained in RPMI-1640 media supplemented with 100 U/ml penicillin, 100 µg/ml streptomycin, 0.25 µg/ml amphotericin B, and 10% fetal bovine serum and incubated at 37° C with 5% CO₂.Cells were seeded at a population density of ~200,000 cell/ml and treated with variable concentrations of non-natural nucleoside (0.1–100 µg/ml) for up to 72 hours. Viability was assessed via trypan blue staining and counting the number of viable (clear) versus non-viable (blue) cells under a microscopy using a hemocytometer.

Cell Cycle Analyses

Cells were grown at a density of 200,000/ml. 5-NIdR (100 mg/mL) was added for time periods varying from 1 to 3 days. Cells were harvested by centrifugation. The supernatant was removed and then washed with PBS. After aspiration of PBS, 500 µl of 70% ethanol was added and cells were incubated on ice for 15 minutes followed by centrifugation and the removal of ethanol. One ml of PI staining solution [(10 ml of 0.1 Triton X-100/PBS, 0.4 ml of 500 µg/ml of PI, and 2 mg/ml of DNase-free RNase)] was added to the cell suspension, placed on ice for 30 minutes, and then analyzed using a Beckman Coulter XL flow cytometer with a red filter.

Accession Numbers

Atomic coordinates and structures factors have been deposited in the Protein Data Bank with the ID code: 3RSR.

Results

In silico screening of non-natural nucleosides

Several non-natural nucleoside analogs illustrated in Figure 1E the library reported in (36) were screened using the in silico docking programme, surflex doc. Based on these studies show that 5-NITP is an excellent candidate that could bind to both the S and A site of human ribonucleotide reductase.

ITC study of 5-NITP binding and IC50 determination to hRR1

To experimentally validate in silico findings, we used ITC to measure the dissociation constants for 5-NITP binding to hRR1. When 5-NITP was injected into the buffer (50 mM HEPES at pH 7.0 containing 5 mM MgCl₂5 mM DTT, 5% (v/v) Glycerol) alone, a relatively small exothermic heat change was observed (data not shown). A similar titration experiment with hRR1 resulted in large endothermic heat changes exhibiting characteristic binding isotherms (Fig. 2). The heat changes at various molar ratios of 5-NITP added to hRR1 can be best fit to a two binding site model that yields a K_d1 of 44 µM and a K_d2 of 5 mM. These results indicate that 5-NITP binds hRR1 at a high and low affinity binding site. The isotherm profile of 5-NITP binding to wild type hRR1 shows a high enthalpy change and a positive entropic change suggesting that binding of this non-natural nucleotide is driven primarily by enthalpy with small entropic contributions.

Figure 2.

ITC profile of 5-NITP binding to the hRR1 subunit. The binding isotherm was obtained as described in methods with correction for heat of dilution at 25°C. 5-NITP binding was derived from the non-linear least square fit of the isotherm. The isotherm profile of 5-NITP could be best fitted to the two site sequential binding model.

The IC₅₀ value for 5-NITP was determined by an in vitro activity assay using ¹⁴C-ADP or ³H-CDP as the substrates (15, 24). The wild type hRR activity under hRR1 limiting conditions was as previously reported (15). Using ADP as the diphosphate substrate, the concentration of 5-NITP required to inhibit 50% hRR activity is 170 ± 5 µM. Surprisingly, 5-NITP did not inhibit hRR activity when CDP was used as the substrate (data not shown). This difference in inhibitory effects is consistent with the higher binding affinity of 5-NITP for the S-site as opposed to the A-site.

Cell-Based Studies to Evaluate the Anti-Cancer Effects of 5-Nitroindolyl-2'-Deoxynucleoside

We measured the cytostatic and/or cytotoxic effects of the corresponding nucleoside, 5-nitroindolyl-2'-deoxynucleoside (5-NIdR), against Jurkat cells. Jurkat cells are an attractive model for testing new anti-cancer agents since they display resistance to many existing chemotherapeutic agents (37). Exponentially growing Jurkat cells were treated with DMSO (vehicle control) or two fixed concentrations of 5-NIdR (50 or 100 µg/mL) for time periods of up to 3 days. Fig. 3A provides representative time courses for the number of viable (green) versus non-viable (black) cells in the absence and presence of either 50 µg/ml or 100 µg/ml 5-NIdR. Direct comparison of these time courses shows that treatment with 100 µg/ml of 5-NIdR produces a cytostatic effect as the number of viable cells is reduced by ~65% compared to cells treated with DMSO. In addition, 5-NIdR generates a weak cytotoxic effect due to the small yet reproducible increase in the number of non-viable cells.

Figure 3.

Anti-Cancer Effects of 5-Nitroindolyl-2'-Deoxyriboside. (A) Time courses in the number of viable (left) and non-viable (right) Jurkat cells in the absence and presence of non- natural nucleoside. (B) Flow cytometry of Jurkat cells treated with (i) DMSO or (ii) 100 µg/ml 5- NIdR. (C) Cell cycle analyses of Jurkat cells treated with (A) DMSO or (B) 100 µg/ml 5-NIdR after 72 hours.

We next analyzed the effects of 5-NIdR on cell-cycle progression after 3 days post-treatment using PI staining to measure cellular DNA content. The histogram of Jurkat cells treated with DMSO (Fig. 3B) shows a standard cell-cycle distribution for asynchronous cells as the vast majority of cells exist at G1 (59 +/− 2%) and S-phase (34 +/− 2%) while a significantly smaller population exists at G2/M (7 +/− 1%). Treatment with 100 µg/ml of 5-NIdR produces two important effects (Fig. 3C). First, there is an increase in the percentage of sub-G1 cells compared to those treated with DMSO. This increase is consistent with the cytotoxic effects of the non-natural nucleoside described above for the viability studies. Secondly, treatment with 5-NIdR also produces significant alterations in cell-cycle progression. In particular, cells treated with the non-natural nucleoside show significant accumulation at S-phase (45 +/− 2%) with a concomitant decrease at G1 (50 +/ 1%) while a minimal decrease in populations at G2/M (5 +/− 1%, Fig. 3C). The increase is S-phase cells combined with a decrease in G1 is consistent with a mechanism for inhibiting DNA synthesis; perhaps by reducing the availability of dNTPs caused by RR inhibition by 5-NidR or its various phosphorylated metabolites.

Oligomeric state of hRR in the absence and presence of 5-NITP

The results of the docking studies combined with ITC data suggest that 5-NITP binds both at the S- and A-sites of hRR1. This is possible since 5-NITP is a dATP analog. To further investigate if 5-NITP binds to the A-site of hRR1, we performed multi angle light scattering (MALS) experiments to determine the oligomeric state of hRR1 in the presence of 5-NITP. Since 5-NITP is a dATP analog, it may bind at the A-site and thus induce hexamer formation which has been documented with dATP. MALS determines the molecular weight of molecules via a method that is independent of molecular mass reference standards, column calibration and assumptions of molecular shape (38). In addition, SEC-MALS separates mixtures of oligomers and measures the absolute molecular weight of an oligomer in elution fractions. As such, it provides an ideal technique to define the oligomeric state of hRR in the absence and presence of different nucleotide analogs. The results of this study reveal that the eluted fraction in the presence of 5-NITP corresponds to the dimer of hRR1 with a molecular weight of 176 KDa (Fig. 4A). This contrasts the hexamer which is observed using dATP (14). It is important to mention here that our studies used 20 µM and 40 µM NITP, concentrations that correspond to the K_d of 5-NITP, only dimers were observed at both the concentrations (Fig. 4B).

Figure 4.

(A) Multi-angle light scattering (MALS) analysis of 5 NITP interactions with hRR1. Oligomeric status of hRR1 was determined using Wyatt TREOS multi-angle light scattering as described in methods. The chromatograms and resultant molecular weight data were analyzed using the Astra 5.3 software from the Wyatt Corporation. (B) Size exclusion chromatography analysis of hRR1 in complex with 5-NITP. The calibration with molecular weight standards are shown in the inset.

X-ray crystallography of 5-NITP binding to RR1

To further understand the molecular basis for hRR1 inhibition by 5-NITP, we attempted to cocrystallize the hRR1-5-NITP complex. Unfortunately, we were unable to obtain an atomic resolution structure since co-crystals of the hRR1-5-NITP complex diffracted to a low resolution of 8Å. However, we previously demonstrated that the enzyme for S. cerevisiae, ScRR1, has the most conserved structure compared to hRR1 (15). Human and yeast RR1 share 66% sequence identity and 83% sequence similarity. Furthermore, hRR1 shares structural homology with ScRR1, with an r.m.s. deviation of 0.8 Å. As such, we performed soaking experiments using the previously reported ScRR1 crystals (7). Although previous attempts to co-crystallize other complexes have failed, it should be noted that ScRR1 co-crystallizes with TTP which can be then subjected to soaking experiments. Here, we show that the ScRR1-5-NITP complex crystals diffracted to 2.3 Å resolutions which allowed for structural determinations. The structure was refined to an acceptable range of R and R_free values with good geometrical parameters (Table 1). The 2F_o–F_c difference map (Fig. 1B) clearly shows electron density for 5-NITP bound at the S-site. In addition, the F_o-F_c omit maps show electron density for 5-NITP binding at the S- site (data not shown). In this structure, 5-NITP adopts a 2’-endo conformation as previously observed with the deoxynucleotides bound at the S-site (8, 14). The ScRR1-5NITP complex structure superimposes with ScRR1-AMPPNP and hRR1-dATP complexed structures with RMSD of C_α atoms of 0.45 Å and 0.57 Å, respectively.

Table 1.

Data collection and refinement statistics

Data Collection
Space group	P21212
Cell dimensions a,b,c (Å)	108.4, 118.2, 63.9
Monomer per asymmetric unit	1
Wavelength (Å)	1.0
Resolution (Å)	2.3
Unique reflections	37089
Redundancy	6.2 (5.7)a
Completeness (%)	99.9 (99.0)a
Rsym,b I/σ(I)	6.8 (49.6)a
I/σ(I)	30.1 (3.5)a
Refinement
Rwork/Rfreec	18.8/23.3
R.M.S.D. from ideal geometry	0.008/1.303
No. of atoms	5039
Protein	4796
Mg2+	1
Water	242
Ligand/ion	32
Mean B-value (Å2)
Main chain/Side chain	43.4/45.2
Ligand/Ion/Water	37.8/54.3/44.3
Ramachandran Plot statistics (%)
Most favorable region	97.99
Allowed region	1.68
Disallowed region	0.34

^aNumbers in parentheses represent values in the highest resolution shell.

^bR_sym =Σ_hklΣ_i/I_i(hkl)−〈I(hkl)〉/Σ_hklΣ_iI_i(hkl), where I_i(hkl) is the ith observation of reflection hkl and 〈I(hkl)〉i is the weighted average intensity for all observations i of reflection hkl.

^cR_work and R_free=Σ||F_o|−|F_c||/Σ|F_o|, where F_o and F_c are the observed and calculated structure factor amplitudes, respectively. For the calculation of Rfree, 10% of the reflection data were selected and omitted from refinement for cross-validation purpose.

The interactions of 5-NITP bound at the S-site of ScRR1 are shown in Fig. 1D. 5-NITP interacts mainly with Loop1 (residues 255–270) and to a lesser extent with residue 285 of Loop2 (residue 285–296). Although the nitro group of 5-NITP does not make any H-bonds, it is involved in an ion-pair interaction with K243. This likely accounts for the enthalpic contributions determined in the ITC analysis (vide supra). In addition, the indole ring of 5-NITP makes van der Waals contacts with V286, K243, and T265. The 3’ OH of deoxyribose forms a strong H- bond (2.5Å) with the carboxylic side chain of D226. As previously reported, we observe an Mg²⁺ ion that coordinates the α- and γ-phosphate of 5-NITP. The presence of the Mg²⁺ versus water was confirmed by comparing B-factors after refinement. Mg²⁺ always refined with the lower B- factor suggesting that the electron density peak corresponded to Mg²⁺ rather than water. The Mg²⁺ ion is octahedrally coordinated by three solvent molecules and the α- and γ-phosphates. The Nδ atom of K243 forms bifurcated salt bridges with the α- and β-phosphates. The α-phosphate forms three H-bonds with three solvent molecules. Two of these solvent molecules form several H-bonds with hRR1 at the S-site (Fig. 1D). The guanidinium side chain of R256 forms two salt bridges with the γ-phosphate. Overall, 5-NITP forms 12 H-bonds, 3 salt-bridges, and 360 van der Waals contacts with residues and bound water molecules composing the S-site (Fig. 1D and supplementary Table 1).

Discussion

This report describes the in vitro and in vivo characterization of a unique non-natural nucleoside analog as an anti-cancer agent that inhibits the activity of hRR. Results from in silico docking studies of a library of non-natural nucleotide analogs identified one analog, 5-NITP that was predicted to bind both at the S and A-site. ITC data shows 5-NITP binds one binding site on hRR1 with micromolar affinity and at another site with low millimolar affinity. In a recent study, we showed that a key to hRR inhibition by the negative regulator dATP is the ability of the natural nucleotide to induce the protein into an inactive hexamer (15). In an independent study, the dATP analog, Clofarabine, was shown to inhibit hRR by inducing hexamer formation with hRR1 (16). This result further underscores the importance of developing chemical entities that inhibit hRR activity by modulating its oligomerization state (16). The results of our MALS studies reveal that 5-NITP, unlike dATP, does not induce hexamer formation in hRR1 at 20–40 µM concentrations. We assume that the A site of hRR1 is low affinity site for 5-NITP binding as the non-natural nucleotide is unable to hexamerize hRR1 when its concentration is maintained at 40 µM (concentration for high-affinity binding site). In contrast, dATP was shown to hexamerize hRR1 at a concentration approximately equal to the K_d for the A site (15). The inability of 5-NITP to induce hexamerization of hRR is likely due to the extremely poor affinity (K_d ~5 mM) for 5-NITP binding at the A-site. Instead, the non-natural nucleotide induces dimer formation (Fig. 4) which is consistent with previous reports indicating that nucleotide triphosphates that bind at the S-site located on the dimer interface (10). Finally, the crystal structure of the ScRR1-5-NITP complex confirms that the non-natural nucleotide binds at the S-site.

Previous mechanistic studies of 5-NITP have demonstrated that the nitro moiety and the indole scaffold play important roles toward modulating the binding to various biological targets. For example, 5-NITP functions as an excellent surrogate for dATP as a polymerase substrate during translesion DNA synthesis. In this case, the hydrophobic nature of the nitro moiety coupled with its extensive pi-electron density enhances its base-stacking potential and allows for optimal insertion opposite DNA lesions such as the non-instructional abasic site (29). In addition, 5-NITP can function as a surrogate for ATP by inhibiting the activity of the bacteriophage T4 clamp loading complex, gp44/62, needed for assembly of the DNA replication complex. In this case, the nitro moiety plays multiple roles in initial ground-state binding to the active site of the clamp loader. While the hydrophobic nature of nitro group is important, more favorable electrostatic interactions are formed between this moiety and an active site arginine residue. In addition, the indole ring interacts with an active site phenylalanine through of π-π electron stacking interactions. The structural data obtained here with 5-NITP bound to ScRR1 also highlight the importance of π-π stacking interactions. In particular, the conformation of the indole ring can be superposed onto the adenine ring of dATP bound to hRR1 by an approximate 90° rotation (Fig. 5A, (15)). It should be noted that we were able to do such comparison confidently, as previously we have shown that the S site of ScRR1 is very similar to hRR1 (15). The relative orientation of the indole ring with respect to the adenine is forced to adopt this conformation as to avoid steric clashes with Y285 present on Loop2 (Fig. 5B). Similar clashes with Loop2 are observed when comparing the structures of ScRR1-5-NITP with ScRR1-AMPPNP (data not shown). In fact, the nitro group is likely to clash with residues 287–290 of Loop2 which results in the disorder of Loop2 in the electron density map of the ScRR1-5-NITP. Modeling studies of 5-NITP bound to hRR1 shows that the interactions made to the S-site of ScRR1 are conserved (Figure 5C).

Figure 5.

Comparison of 5-NITP binding with dATP and AMPPNP. (A) Stereo figure of the mode binding of 5-NITP (cyan) with dATP (yellow). (B) Stereo figure of the modeling of 5-NITP (cyan) bound at the hRR1 S-site (The model of 5-NITP adopting the dATP conformation is shown in green). (C) Stereo figure shows the negative impact of the nitrate group (cyan) on Loop2 (yellow).

The affinity of 5-NITP binding to the S-site of hRR1 compared to dATP is 75 times weaker (Fig. 2 and (9)). One of the reasons for the poor affinity of 5-NITP is due to unfavorable electrostatic interactions between the nitro group and the carbonyl oxygen of K243. It appears that these unfavorable contacts are not accommodated well by the rearrangement of the S-site. 5-NITP forms less H-bonds and less ion-pair interactions as compared to dATP bound at the S- site (Supplementary Table 1). 5-NITP has more surface accessible area (96 Ų) compared to dATP (76 Ų) when bound at the S-site. We attribute the loss of affinity of 5-NITP compared to dATP due to unfavorable electrostatic interactions, less H-bonds, less ion-pair interactions, and less surface area buried at the S-site.

Our data support a mechanism in which the inhibition of hRR1 by 5-NITP involves binding at the S-site as opposed to the A-site. This results in the disruption of the allosteric regulation of hRR as an effector nucleotide triphosphate binding at the S-site which is required for substrate selection. This is clear as our data shows that 5-NITP inhibits ADP reduction but does not inhibit CDP reduction. This difference in inhibitory effects indicates that 5-NITP must compete with the effector nucleotide triphosphate, dGTP, to disrupt the selection rules (5). It should be emphasized that adenine analogues binding at the S-site will not select for the ADP substrate. However, it will select for pyrimidine substrates. This also explains why 5-NITP will promote CDP reduction. Finally, 5-NITP is a weak competitive inhibitor at the S-site as it has a K_d of 44 µM that is higher than dATP, dGTP, and TTP which have K_d values in the low micromolar range (0.5 – 20 µM).

The anti-cancer activity of 5-NIdR correlated with expression levels of terminal deoxynucleotidyl transferase, a unique polymerase involved in the etiology of this form of leukemia. Despite this correlation, however, it was evident that 5-NIdR displayed cytostatic effects in cell lines that did not show elevated levels of terminal deoxynucleotidyl transferase. As such, the anti-cancer effects of 5-NIdR in these published experiments could be explained by to the inhibition of ribonucleotide reductase (39). 5-NITP displays cytostatic and weak cytotoxic effects against human leukemia cells. These anti-cancer effects are caused by the accumulation of cells at S- phase, a result that is consistent with the inhibition of DNA synthesis caused by the inhibition of cellular hRR activity. The structure shows the negative impact of the nitro group on the indole ring binding at the S-site and provides insights on how to improve affinity by the addition of the positively charged moiety such as two amino groups. Although 5-NITP does not inhibit hRR with exceedingly high potency, it is important to note that this analog was initially developed as an inhibitor of other proteins involved in DNA replication, namely ATP-dependent clamp loader proteins and polymerases. Hence, it is possible that 5-NITP inhibits secondary targets that are involved in DNA synthesis which may explain its efficacy against cancer. Future studies will characterize compounds with amino substitutions on the indole ring. Moreover, it is now becoming evident that a good hRR inhibitor must have the ability to hexamerize hRR1 similar to dATP or clofarabine. Clearly 5-NITP is unable to hexamerize hRR1 at 20–40 µM, the cellular concentration of dATP, hence explaining its moderate inhibitory potency. We propose the ability to induce the formation of hRR1 hexamers can be used as a good indicator of a highly potent hRR inhibitor, a factor that should be implemented in inhibitor design.

Abbreviations

RR

Ribonucleotide reductase

hRR

human ribonucleotide reductase

hRR1

human ribonucleotide reductase 1

hRR2

human ribonucleotide reductase 2

ScRR

S. cerevisiae ribonucleotide reductase

ScRR1

S. cerevisiae ribonucleotide reductase 1 and

5-NITP

5-nitro-indolyl-2'-deoxyribose triphosphate

AMPPNP

Adenosine 5′-(β,γ-imido)-triphosphate

5-NIdR

5-nitro-indolyl-2'-deoxynucleoside

IndTP

indolyl-2'-deoxyribose triphosphate

5-FITP

5-fluoroindolyl-2'-deoxyribose triphosphate

5-MeITP

5-methylindolyl-2'-deoxyribose triphosphate

5EtITP

5-ethylindolyl-2'-deoxyribose triphosphate

5-CHITP

5-cyclohexylindolyl-2'-deoxyribose triphosphate

5-CEITP

5-cyclohexenylindolyl-2'-deoxyribose triphosphate

5-PhITP

5-phenylindolyl-2'-deoxyribose triphosphate