Discovery of Small Molecule Inhibitors of Human Uridine-Cytidine Kinase 2 by High-Throughput Screening

Abstract

Clinically relevant inhibitors of dihydroorotate dehydrogenase (DHODH), a rate-limiting enzyme in mammalian de novo pyrimidine synthesis, have strong antiviral and anticancer activity in vitro. However, they are ineffective in vivo due to efficient uridine salvage by infected or rapidly dividing cells. The pyrimidine salvage enzyme uridine-cytidine kinase 2 (UCK2), a ~29 kDa protein that forms a tetramer in its active state, is necessary for uridine salvage. Notwithstanding the pharmacological potential of this target, no medicinally tractable inhibitors of the human enzyme have been reported to date. We therefore established and miniaturized an in vitro assay for UCK2 activity and undertook a high-throughput screen against a ~ 40,000-compound library to generate drug-like leads. The structures, activities, and modes of inhibition of the most promising hits are described. Notably, our screen yielded non-competitive UCK2 inhibitors which were able to suppress nucleoside salvage in cells both in the presence and absence of DHODH inhibitors.

Graphical Abstract

In mammalian cells, the biosynthesis of pyrimidine nucleotides is complex, and tightly regulated by multiple mechanisms¹. Briefly, pyrimidine nucleotides are synthesized by a de novo or a salvage pathway, which converge at uridine/cytidine monophosphate (UMP, CMP) (Figure 1). In the de novo biosynthesis pathway, a multifunctional enzyme, CAD, harboring carbamoyl phosphate synthetase, aspartate transcarbamoylase, and dihydroorotase activities, converts L-glutamine, aspartate, and bicarbonate into dihydroorotate (DHO) through a series of reactions. The following enzyme, dihydroorotate dehydrogenase (DHODH), resides on the mitochondrial membrane and coverts DHO into orotic acid while shuttling two electrons into the electron transport chain via Coenzyme Q (CoQ). The remaining steps of de novo pyrimidine biosynthesis involve conversion of orotic acid into UMP by a bifunctional protein, uridine monophosphate synthetase (UMPS) in two steps. Alternatively, UMP can be synthesized by the salvage pathway where extracellular pyrimidines are absorbed from the bloodstream via plasma membrane transporters. In the cytosol, uridine/cytidine kinase (UCK) phosphorylates uridine and cytidine in an ATP-dependent manner. The resulting UMP and CMP from either source is further phosphorylated into UTP and CTP, respectively, through successive action of cytidine monophosphate kinase 1 (CMPK1) and nucleoside diphosphate kinase (NDPK). The nucleotide diphosphate intermediates in these final steps are also utilized by dividing cells for deoxynucleotide biosynthesis.

Figure 1.

Biosynthesis of pyrimidines.

A number of inhibitors of the de novo pyrimidine pathway have been studied in the laboratory and clinic. For example, N-phosphonacetyl-L-aspartate blocks CAD², the first enzyme of the pyrimidine synthesis pathway. Similarly, the FDA-approved drug, teriflunomide, inhibits DHODH, and is used in the treatment of multiple sclerosis and arthritis³. In vitro, DHODH inhibitors have broad-range anti-viral and anti-cancer properties. However, this effect does not translate to in vivo systems, presumably because inhibition of pyrimidine biosynthesis is compensated by the salvage pathway^4–5. Plasma uridine levels are tightly regulated in the low micromolar concentration range through a mechanism that is not fully understood¹, but involves both adipocytes and the bile^6–7.

Our genome-wide functional analysis of GSK983, a potent small-molecule antiviral agent, identified its mode of action as a DHODH inhibitor⁸. The same unbiased growth-based genetic screen revealed that GSK983 activity was synergistic with knock-down of pyrimidine salvage enzyme UCK2 in K562 cells9. UCK2 is differentially expressed in many cancer tissues and in cells subjected to viral infection and is the predominant UCK isozyme expressed in K562 cells^10–12. Together, these facts led us to prioritize UCK2 over UCK1 for this study. Although broad-range nucleoside transport inhibitors, such as the FDA-approved drug diypyridamole (DP), are sometimes used to block nucleoside salvage¹³, the few published UCK2 inhibitors have not been sufficiently characterized for drug development^14–15. We therefore sought to discover lead UCK2 inhibitors by high-throughput screening of a publicly available (PubChem) small molecule library.

As a first step, we confirmed the biological significance of UCK2. To do so, uridine salvage was measured by incorporating exogenous 5-ethynyl-uridine (5-EU) specifically into cellular RNA. A fluorescent probe was used to quantify 5-EU in RNA via click chemistry¹⁶ (Figure 2A). To validate UCK2 as a target for uridine salvage, gene expression was down-regulated using two different shRNA constructs in K562 cells. The resulting cell lines showed a decrease in uridine incorporation into cellular RNA (Figure 2B). The same assay also validated dipyridamole as a control uridine salvage inhibitor (Figure 2C).

Figure 2.

[A] Assay for quantifying uridine salvage into cellular RNA using 5-ethynyl-uridine (5-EU) and a fluorescent probe. [B] Uridine salvage when UCK2 expression was down-regulated using two different shRNA constructs in K562 cells. [C] Dipyridamole inhibits 5-EU uptake in a dose-dependent manner in A549 cells.

We also investigated the effect of DHODH inhibition on the uridine salvage pathway using 5-EU. After K562 cells were treated with GSK983, the 5-EU assay was performed in comparison with untreated K562 cells. Interestingly, GSK983 treatment significantly increased uridine salvage (Figure S1). Thus, when cells were unable to convert L-glutamine into pyrimidines, the salvage pathway was able to compensate for the loss. This finding is consistent with the findings from animal experiments, where DHODH inhibitors were ineffective due to uridine salvage^4–5.

We next conducted enzymatic high-throughput screening (HTS) to enable the discovery of small molecule modulators of UCK2. UCK2 was expressed as a C-terminally His₆-tagged protein. The protein was purified using Ni-NTA affinity chromatography followed by ion-exchange chromatography. Based on the ability of this enzyme to catalyze phosphoryl transfer from ATP to uridine or cytidine, the luciferase-based ADP-Glo™ kinase assay was adopted for HTS. The ADP-Glo™ kinase assay measures ADP formed in the enzymatic reaction by first eliminating all of the unused ATP in the solution. The ADP is then converted into ATP, which is used to generate light in a luciferase reaction. Under conditions where the luminescence generated is proportional to the kinase activity (Figure S2), the assay is well suited for measuring compound activity with reasonable Z`-factors with linear response over 30 min (Figure S3A).

We first optimized the ADP-Glo™ kinase assay for a 1536-well plate format by adjusting the conditions of the ATP removal step (see Experimental Procedures for details). Under these conditions, UCK2 was stable up to 6 h on ice in the presence of 0.5 mg/mL BSA (Figure S3B); activity decreased over 9 h, but a linear dose-response was preserved (Figure S3C). The enzyme was unaffected by DMSO up to 3% (Figure S3D). Under steady state conditions, the following kinetic parameters were measured by varying uridine: K_M = 62 ± 5 µM and k_cat / K_M = 2.0 × 10⁵ s⁻¹·M⁻¹. When ATP was varied, the analogous kinetic parameters were K_M = 61 ± 1 µM and k_cat/K_M = 2.5 × 105 s⁻¹·M⁻¹, consistent with values reported previously¹⁷ (Figure S4).

Using the above 1536-well assay format, a library of 40,000 small molecules (the “PubChem set”) was screened at 33 μM and showed acceptable performance (average Z´ = 0.6). Using 3 standard deviations of the negative control as a cut-off for hit selection, 128 inhibitory hits were identified in the primary screen and 21 compounds showed apparent activation with up to 10-fold higher signal than DMSO controls. These compounds were retested at 8 different concentrations with half-log dilutions ranging from 0.05–167 µM, which validated 59 inhibitors and 6 activators of UCK2. All activators had AC₅₀ values higher than 166 µM and were therefore not followed up in this study (data not shown).

A continuous UCK2 enzyme assay was developed for detailed analysis of some promising inhibitors and their analogs identified via HTS. In this assay, UCK2 activity was coupled to the pyruvate kinase (PK) reaction, which in turn was coupled to lactate dehydrogenase (LDH). Overall reaction progress could be followed in a continuous assay mode by monitoring the consumption of NADH via UV spectroscopy at 340 nm.

The inhibitors were grouped into three sets based on structural similarity (Figure 3). The first set (represented by compounds 135546734 and 135416439) contained a pyrazolo[3,4-d]- or triazolo[4,5-d]-pyrimidine core (Figure 3A). Compound 135416439, which had an IC₅₀ of 16.4 µM upon retesting in the ADP-Glo assay, was non-competitive with uridine (K_i = 13 ±3 µM) and ATP (K_i = 12 ±2 µM) (Figure S5, Table 1). The structurally-related compound 135546734 had an IC₅₀ of 48.0 µM in the ADP-Glo assay when retested (Figure S5). Compound 135546734 showed no inhibition in an analogous ADP-Glo screen for CMPK1.

Figure 3.

Selected UCK2 inhibitors from high-throughput validation screen, grouped according to structures.

Table 1.

Activities of compounds identified from the HTS screen and structurally-related analogs

PubChem ID (for primary hits)	Qualified IC50 obtained from the HTS (µM)	IC50 obtained from validation experiments	Ki (Uridine) (µM)	Ki (ATP) (µM)	%Inhibition at highest tested concentration (EU assay)
135416439	9.2	16.6	13 ±3	12 ± 2	31.3%, 50 µM
135546734	18.9	48.0	50 ± 7	31 ± 10	41.5%, 50 µM
135546734–1	--	137.2	--	--	26.3%, 50 µM
135546734–2	--	inactive	--	--	19.3%, 50 µM
135546734–3		inactive	--	--	No inhibition
20874830	52	3.8	3.5 ± 0.3	1.0 ± 0.1	51.7%, 50 µM
20874830–1	--	26.8	--	--	No inhibition
20874830–2	--	4.7	--	--	49.6%, 50 µM
20874830–3	--	11.8	--	--	14.9%, 50 µM

Kinetic analysis revealed that it was similarly non-competitive with both uridine (K_i = 50 ± 7 µM) and ATP (K_i = 31 ± 10 µM) (Table 1). To verify its ability to inhibit uridine salvage in intact cells, the 5-EU uptake assay was performed on K562 cells using 20 µM dipyridamole as a control. At 50 µM, compound 135546734 inhibited 5-EU salvage by ~40%. At a concentration of 50 µM, this compound inhibited 5-EU salvage by K562 cells by ~30% (Table 1). Encouraged by these results, we tested three commercially available analogs of 135546734, as 135416439 analogs were not commercially available (Figure S6). While methyl-shifted 135546734–1 had an IC₅₀ value of 137.2 µM, both fluoro-135546734–2 and methoxy-135546734–3 were completely inactive highlighting the importance of the fine structural details of this class of leads to their recognition by UCK2 (Figure S6). The EU assay confirmed that these analogs were largely ineffective at inhibiting uridine salvage (Figure S6, Table 1).

The second set of inhibitors had a tricyclic chromeno[2,3-d]pyrimidine core with three distinct substituents (Figure 3B). Modification of the 2-phenyl substituent by methyl, hydroxy, or fluoro groups at the para position resulted in IC₅₀ values of 52 µM, 166 µM, and 166 µM, respectively, in the ADP-Glo assay. Upon retesting, the most active compound (20874830) had an IC₅₀ value of 3.8 µM and was also primarily non-competitive with both ATP and uridine, with Ki values of 1.0 ± 0.1 µM and 3.5 ± 0.3 µM, respectively (Figure S7, Figure 4). This compound showed 52% inhibition of uridine salvage in K562 cells at a concentration of 50 µM (Figure S7). Four commercially available analogs of 20874830 were retested in our kinetic assay to enable higher throughput at multiple time points (Figure 4); however, none showed improved activity (Table 1). Promisingly, we observed that 20874830 successfully prevented the increase in EU uptake normally observed upon co-administration of GSK983 (Figure 4E).

Figure 4.

Kinetic and biological characterization of 20874830 and analogs. [A]Uridine titration of UCK2 activity with 20874830 at a concentration of 0 µM (blue), 0.5 µM (red), 1.5 µM (green) or 4.5 µM (purple). [B]ATP titration with 20874830 at a concentration of 0 µM (blue), 0.5 µM (red), 1.5 µM (green) or 4.5 µM (purple). [C] Compound 20874830 analogs. [D] Kinetic analysis yielded IC₅₀ values of 3.8 µM (20874830), 26.8 µM (20874830–1), 4.7 µM (20874830–2) and 11.8 µM (20874830–3) [E] EU uptake assay conducted in K562 cells using 50 µM compound and 50 µM EU. [F] EU uptake assay conducted in K562 cells using 50 µM compound and 50 µM EU in the presence of DHODH inhibitor GSK983.

The third group of inhibitors, represented by compounds 135546812 and 135546817 had a [(5-Cyano-6-oxo-4-phenyl-1,6-dihydropyrimidin-2-yl)thio]acetamide core. Compounds 135546812 and 135546817 were both relatively weak inhibitors of UCK2, with IC₅₀ values of 80 µM and 138 µM, respectively. Therefore, we elected to not investigate the kinetic properties of this group of compounds any further. Interestingly, this group of inhibitors has a nitrile group that may be a potential site for covalent inhibition.

In summary, a high-throughput screen against a publicly available small molecule library of ~ 40,000 small molecules was undertaken with human UCK2, an enzyme that plays a critical role in uridine salvage from the extracellular environment. Three structurally distinct sets of inhibitors were identified, and their mechanism of inhibition were analyzed. The biological activities of some of these inhibitors were also investigated using an exogenously added alkynyl-tagged uridine analog, 5-EU, in K562 cells. Notwithstanding the availability of a crystal structure of human UCK2¹⁸, we were unable to reliably predict the precise binding site of either family of inhibitors. Nonetheless based on kinetic observations, the two most promising inhibitor classes were non-competitive with both substrates of this bisubstrate enzyme. Due to the large number of kinases that utilize ATP as a phosphate donor, finding selective inhibitors targeted at the active sites of individual kinases has been a major challenge. Non-competitive inhibitors bind to more structurally diverse sites on kinases, and thus have inherent value as drug leads.

The first class of inhibitors has a pyrazolo[3,4-d]pyrimidine core with fused heterocyclic ring, and can be considered as bioisosteres of purines. This class of compounds has been extensively investigated as adenosine receptor antagonists and also in anti-viral, anti-cancer, tuberculostatic, anti-inflammatory and cardiovascular assays^19–22. Some of these molecules have inhibitory activity toward certain protein kinases^23–25. Compound 135546734 was identified as a hit in an HTS against DNA polymerase beta, eta, and iota at IC₅₀ values of 40, 79, and 79 µM, respectively, while compound 135416439 was active in assays of DNA polymerase eta and kappa at IC₅₀ values of 56 µM and 16 µM, respectively (PubChem). These DNA polymerases replicate through unrepaired DNA lesions by a mechanism called translesion synthesis (TLS)²⁶. The synthesis of pyrazolopyrimidines is well established^{23, 27–28}. Therefore, this structure offers unexplored potential to discover more potent and selective UCK2 inhibitors. The second class of inhibitors has a chromeno[2,3-d]pyrimidine core. Here too, alternative synthesis schemes are available^29–33, and analogs have been studied as antibacterial and antifungal reagents^{31, 34–35}.

Uridine plays a crucial role in RNA/DNA building, protein glycosylation, central nervous system, lipid metabolism, and glycogen synthesis³⁶. A recent study suggested that the plasma uridine affects glucose and insulin sensitivity by controlling dietary responses and modulating leptin signaling^6–7. Tools that interfere with uridine processing will help us to understand its physiological role and potentially treat metabolic diseases that arise due to adiposity, impaired glucose tolerance, and hypercholesterolemia^6–7.

In preliminary experiments, we have shown that inhibition of DHODH increases uridine salvage (Figure S1). We have also demonstrated that UCK2 inhibitor 20874830 can suppress this increase (Figure 4E). Understanding the dynamic between these two pathways and how they affect uridine plasma levels could be critical for use of this strategy as therapy.

As a result of our screen we identified compounds that inhibit UCK2 in the micromolar range. Ideally, in vivo studies require inhibitory compounds that are in the nanomolar range for clear pharmacological activity with minimal toxicity. Therefore, the inhibition profile of the three classes of UCK2 inhibitors identified herein can be further improved using different chemical modifications with careful examination of drug-enzyme crystal structure. In the future, in vivo studies could be performed using clinically useful DHODH inhibitors such as teriflunomide or brequinar and the UCK2 inhibitors identified from this screen. The findings of this study will form the basis for examining the dynamics between de novo synthesis and salvage pathway for pyrimidine synthesis.

Abbreviations

UCK2

uridine-cytidine kinase 2

DP

dipyridamole

DHODH

dihydroorotate dehydrogenase

CMPK1

cytidine monophosphate kinase 1