NUDT22 promotes cancer growth through pyrimidine salvage

Abstract

The NUDIX hydrolase NUDT22 converts UDP-glucose into glucose-1-phosphate and the pyrimidine nucleotide uridine monophosphate but a biological significance for this biochemical reaction has not yet been established. Glucose-1-phosphate is an important metabolite for energy and biomass production through glycolysis and nucleotides required for DNA replication are produced through energetically expensive de novo or energy-efficient salvage pathways. Here, we describe p53-regulated pyrimidine salvage through NUDT22-dependent hydrolysis of UDP-glucose to maintain cancer cell growth and to prevent replication stress. NUDT22 expression is consistently elevated in cancer tissues and high NUDT22 expression correlates with worse survival outcomes in patients indicating an increased dependency of cancer cells to NUDT22. Furthermore, we show that NUDT22 transcription is induced after inhibition of glycolysis, MYC-mediated oncogenic stress, and DNA damage directly through p53. NUDT22-deficient cancer cells suffer from growth retardation, S-phase delay, and slower DNA replication fork speed. Uridine supplementation rescues replication fork progression and alleviates replication stress and DNA damage. Conversely, NUDT22 deficiency sensitizes cells to de novo pyrimidine synthesis inhibition in vitro and reduces cancer growth in vivo. In conclusion, NUDT22 maintains pyrimidine supply in cancer cells and depletion of NUDT22 leads to genome instability. Targeting NUDT22 therefore has high potential for therapeutic applications in cancer therapy.

Fig. 1. NUDT22 expression is increased in response to stress.

A Increased NUDT22 expression after inhibition of HK2 with 2-DG measured by qRT-PCR (NUDT22 P = 0.0254; cMYC P = 0.0001; GRP78 P = 0.0016). cMYC and GRP78 are positive controls. B Increased NUDT22 expression after HK2 depletion with siRNA measured by qRT-PCR (HK2 P = 0.006; NUDT22 P < 0.001). C Elevated NUDT22 expression in four independent HA1EB-cMYC clones measured by qRT-PCR (NUDT22 P = 2.4*10⁻⁸; cMYC P = 7.48 *10⁻⁷). D Western blot of cMYC-overexpressing cells with increased NUDT22 protein levels. E cMYC-induced NUDT22 expression measured by qRT-PCR (cMYC P = 0.0006; NUDT22 P = 0.0346). F Transient cMYC overexpression in U2OS (NUDT22 P = 0.019) and hTERT-RPE1 (NUDT22 P = 0.0002) cells. G Western blot of cells fractionated in soluble (cytosolic) and insoluble (nuclei, membranes) of U2OS cells transiently transfected with cMYC. P values were calculated by paired t test. Data are presented as the mean values with SD and all experiments were repeated at least 3 times.

Fig. 2. NUDT22 is a p53 target gene.

A The depletion of p53 abolished the cMYC-mediated activation of NUDT22 in BJ-MYC^ER cells (NUDT22 P = 0.0086; P = 0.0119; P = 0.0046; p53 P = 0.012; P = 0.0146; P = 0.0022), measured by qRT-PCR. B Western blot of stabilized p53 and p21 and increased NUDT22 in BJ-MYC^ER cells. C Relative luciferase levels of the NUDT22 reporter after stabilization of p53 with Nutlin3a in U2OS cells (NUDT22 P = 0.0003, p53 P = 0.0001). D qRT-PCR for the 2 kb CpG 5ʹ region of the NUDT22 gene after ChIP with a p53(DO1) antibody in U2OS cells. GFP served as a transfection control (P = 0.008). P values were calculated by paired t test. Data are presented as the mean values with SD and all experiments were repeated at least 3 times.

Fig. 3. NUDT22 is induced by DNA damage.

A DNA damaging agents induce transcriptional activation of the NUDT22 reporter measured by relative luciferase activity. B The p53-luciferase reporter served as a control. Drug concentrations: doxorubicine (doxo) 5 μM, actinomycin D (actD) 5 nM, hydroxyurea (HU) 2 mM, olaparib 10 μM, camptothecin 10 μM, nutlin3a 2 μM (P values calculated to DMSO control (NUDT22): doxo P = 0.0094; actD P = 0.0036; HU P = 0.0088; olaparib P = 0.0042; CPT P = 0.0021; nutlin3a P = 0.0003. (p53) doxo P = 0.0068; actD P = 0.0008; HU P = 0.0344; olaparib P = ns; CPT P = 0.0003; nutlin3a p < 0.0001). C U2OS and (D) hTERT-RPE1 cells were treated with doxorubicin and actinomycin D for 24 h. NUDT22 expression was increased in both cell lines as measured by qRT-PCR. This is consistent with the stabilization of p53 measured by western blot (C: doxo P = 0.0013; actD P = 0.0013, D: doxo P = 0.0004; actD P = 0.0021). E Increased p53 stability and activity in U2OS cells after NUDT22 knockout measured by western blot. Protein level quantification of NUDT22 and p53 is shown in percent normalised to β-Actin and relative to control cells. F NUDT22 protein levels were significantly reduced after 6 h of translation inhibition with 10 μg/ml CHX detected by western blot and RAD51 served as a positive control. G Western blot of NUDT22 protein levels accumulating after proteasome inhibition with 5 μM MG132. P values are calculated by paired t test in GraphPad Prism. Data are shown as the mean with SD and all experiments were repeated at least 3 times.

Fig. 4. Loss of NUDT22 leads to replication stress.

A Model of NUDT22 in UDP-glucose hydrolysis as a UMP salvage pathway. B U2OS NUDT22 knockout (KO) cells. Western blot of two independent clones are shown. C hTERT-RPE1 NUDT22 knockout (KO) cells. Western blot of two independent clones are shown. D LC-MS nucleotide pool measurement of U2OS and (E) hTERT-RPE1 cells. F Replication fork speed (IdU incorporation) in ctrl and NUDT22 KO cells. G Quantification of the percentage of EdU-positive cells by high content microscopy (ctrl::KO1 P = 0.0109; ctrl::KO2 P = 0.0003; KO1 DMSO::KO1 pyrazofurin P = 0.0227; KO2 DMSO::KO2 pyrazofurin P = 0.0109). P values were calculated by unpaired t test. Errors as the mean with SD. H Growth rates of U2OS and (I) hTERT-RPE1 cells determined by resazurin fluorescence. All experiments were repeated at least 3 times.

Fig. 5. Loss of NUDT22 potentiates inhibition of nucleotide metabolism.

Dose response curves of ctrl and NUDT22 KO U2OS cells exposed to (A) pyrazofurin, (C) hydroxyurea and (E) after glutamine starvation. Dose response curves of ctrl and NUDT22 KO hTERT-RPE1 cells exposed to (B) pyrazofurin, (D) hydroxyurea and (F) after glutamine starvation. G Expression levels of pyrimidine synthesis genes after 24 h of glutamine starvation relative to β-actin measured by qRT-PCR. Statistical analysis between hTERT-RPE1 and U2OS cells (NUDT22 P = 0.0135; TK1 P = 0.0071; DCK P = 0.0132; DHODH P = ns; UMPS P = 0.004; TYMS P = 0.01; RNR P = 0.0026). P values were calculated by paired t test. Data are presented as the mean values with SD. H Growth rate comparison between U2OS and hTERT-RPE1 cells. All experiments were repeated at least 3 times.

Fig. 6. Loss of NUDT22 activates DNA damage response.

A, B Cell cycle checkpoint activation in NUDT22 KO U2OS cells measured by western blot. C No significant cell cycle checkpoint activation in NUDT22 KO hTERT-RPE1 cells measured by western blot. D Increased single stranded DNA (nuclear RPA intensity) and DNA damage (nuclear γH2A.X (E) and 53BP1 (F) intensity) quantified by high content immunofluorescence microscopy. G Quantification of γH2A.X DNA damage foci in ctrl and NUDT22 KO U2OS cells with and without pyrazofurin by high content immunofluorescence microscopy (P < 0.0001). H Uridine supplementation reverses the DNA damage in NUDT22 KO U2OS cells exposed to brequinar or pyrazofurin. I Confocal microscopy reveals strong colocalization of nuclear RPA and γH2A.X in NUDT22 KO U2OS cells (scale bar:10 μM). P values were calculated by the Mann–Whitney test. Data are presented as the mean values with SEM. All experiments were repeated at least 3 times.

Fig. 7. NUDT22 expression is increased in cancer and loss of NUDT22 reduces cancer growth in vivo.

A RSEM expected count analysis of NUDT22 expression in the panCancer TCGA and GTEx datasets (P < 0.0001; Mann–Whitney test, mean with SD). B RSEM expected count analysis of pyrimidine metabolism gene expression in breast cancer TCGA and GTEx datasets (P < 0.0001; Mann–Whitney test, mean with SD). C Overall survival (OS) of breast cancer patients with NUDT22 gene alterations (TCGA PanCancer). Altered group is defined as patients with at least one alteration in the NUDT22 gene. D Cell cycle checkpoint activation of ctrl and NUDT22 KO MCF7 cells was rescued by uridine supplementation detected by western blot. E NUDT22 KO MCF7 cells have reduced EdU incorporation, which is further reduced by pyrazofurin exposure quantified by high content microscopy (mean with SD). F Reduced replication fork speed in NUDT22 KO MCF7 cells can be rescued by uridine supplementation (mean with SD). G NUDT22 KO MCF7 cells have increased gH2A.X foci formation, which was further increased with brequinar, leflunomide and pyrazofurin quantified by high content immunofluorescence microscopy (P < 0.0001). H uridine supplementation rescues DNA damage caused by NUDT22KO (P < 0.0001) (mean with SEM). P values were calculated by the Mann–Whitney t test. I Dose-response curves of ctrl and NUDT22 KO MCF7 cells treated with pyrazofurin, (J) brequinar and (K) leflunomide. L In vivo mammary cancer xenograft model with ctrl and NUDT22 KO MCF7 cells. Luc2⁺ cells were injected into mammary fat pads and imaged weekly by IVIS imaging (P < 0.0023; Mann–Whitney test, mean with SEM). All experiments were repeated at least 3 times.