Inhibition of the de novo pyrimidine biosynthesis pathway limits ribosomal RNA transcription causing nucleolar stress in glioblastoma cells

Abstract

Glioblastoma is the most common and aggressive type of cancer in the brain; its poor prognosis is often marked by reoccurrence due to resistance to the chemotherapeutic agent temozolomide, which is triggered by an increase in the expression of DNA repair enzymes such as MGMT. The poor prognosis and limited therapeutic options led to studies targeted at understanding specific vulnerabilities of glioblastoma cells. Metabolic adaptations leading to increased synthesis of nucleotides by de novo biosynthesis pathways are emerging as key alterations driving glioblastoma growth. In this study, we show that enzymes necessary for the de novo biosynthesis of pyrimidines, DHODH and UMPS, are elevated in high grade gliomas and in glioblastoma cell lines. We demonstrate that DHODH's activity is necessary to maintain ribosomal DNA transcription (rDNA). Pharmacological inhibition of DHODH with the specific inhibitors brequinar or ML390 effectively depleted the pool of pyrimidines in glioblastoma cells grown in vitro and in vivo and impaired rDNA transcription, leading to nucleolar stress. Nucleolar stress was visualized by the aberrant redistribution of the transcription factor UBF and the nucleolar organizer nucleophosmin 1 (NPM1), as well as the stabilization of the transcription factor p53. Moreover, DHODH inhibition decreased the proliferation of glioblastoma cells, including temozolomide-resistant cells. Importantly, the addition of exogenous uridine, which reconstitutes the cellular pool of pyrimidine by the salvage pathway, to the culture media recovered the impaired rDNA transcription, nucleolar morphology, p53 levels, and proliferation of glioblastoma cells caused by the DHODH inhibitors. Our in vivo data indicate that while inhibition of DHODH caused a dramatic reduction in pyrimidines in tumor cells, it did not affect the overall pyrimidine levels in normal brain and liver tissues, suggesting that pyrimidine production by the salvage pathway may play an important role in maintaining these nucleotides in normal cells. Our study demonstrates that glioblastoma cells heavily rely on the de novo pyrimidine biosynthesis pathway to generate ribosomal RNA (rRNA) and thus, we identified an approach to inhibit ribosome production and consequently the proliferation of glioblastoma cells through the specific inhibition of the de novo pyrimidine biosynthesis pathway.

Fig 1. De novo pyrimidine biosynthesis is necessary to maintain efficient ribosomal RNA transcription in glioblastoma.

(A) CAD, DHODH and UMPS mRNA levels in lower (II-III) to higher grade (IV/GBM) glioma patients from TCGA database. (B) Western blot of p14 ARF-/- human astrocytes, LN229, GBM9, and SF188. Fold change in expression as indicated was normalized to actin levels and compared to the expression in astrocytes by densitometry analysis with Image J. (C) Representation of de novo and salvage pyrimidine biosynthesis pathways and brequinar. (D) Representation of the high rDNA transcription rate (left panel) and the potential effects of blocking the de novo pyrimidines biosynthesis (right panel). (E-F) UMP, UDP, UTP and uridine measured by LC-MS/MS in LN229 after incubation with brequinar, 4 replicates, (E) and in GMB9 after incubation with ML390 (F), 3 replicates. (G) qPCR of 47S pre-rRNA with or without brequinar and uridine for 24 h in LN229, GBM9, and SF188. N = 2–5. (H) qPCR of 47S pre-rRNA with or without ML390 and uridine for 48 h in LN229, GBM9, and SF188. N = 2–3. (I-J) qPCR of ACTIN with or without brequinar for 24 h (I) or ML390 (J) for 48 h and uridine in LN229, GBM9, and SF188. N = 2–4. For G-J, individual technical replicates values of each biological replicate are represented by diamonds. * indicates p-values ≦0.05. Numerical values for each of the experiments represented are available in S1 Data.

Fig 2. Inhibition of the de novo pyrimidine biosynthesis pathway decreases proliferation of temozolomide-sensitive and -resistant glioblasomta cells.

(A) Relative proliferation of p14 ARF-/- human astrocytes, LN229, GBM9 and SF188 cells with increasing amounts of brequinar. Media with drugs was replaced every 2 days for 6 days. N = 3. (B) Relative proliferation of LN229, GBM9 and SF188 cells with or without brequinar and uridine. Media with drugs was replaced every 2 days for 6 days. N = 3–4. (C) Representation of the effects of temozolomide (TMZ) and the depletion of nucleotides causing DNA double-strand breaks and the phosphorylation of H2AX (γ-H2AX). (D, E) Relative proliferation of LN229 and GBM9 in the presence of TMZ, brequinar (Breq.) or brequinar + TMZ. Media with drugs was replaced the day after seeding, and cell proliferation was measured 4 days later. N = 3–4. (F, G) Western blot of LN229 and GBM9 for γ-H2AX and p53. Also see Supplementary S2F and S2G Fig. (H) Representation of the in vitro generation of SF188 TMZ-resistant cells. (I) Relative proliferation of SF188 TMZ-sensitive and -resistant cells with or without TMZ. (J) Relative proliferation of SF188 TMZ-sensitive or -resistant cells with or without TMZ, brequinar or brequinar + TMZ with or without uridine normalize to each DMSO condition. Also see Supplementary S2H Fig. Media with drugs was replaced the day after seeding, and cell proliferation was measured 4 days later. (K) Western blot of SF188 TMZ-sensitive or -resistant cells with or without TMZ, brequinar or brequinar + TMZ with or without uridine for γ-H2AX, p53 and, MGMT. Also see Supplementary S2I and S2J Fig. (L) Cell cycle analysis of LN229 treated with 0.1 μM brequinar, 2 μM ML390 in the presence or absence of 100 μM uridine, and 100 μM TMZ for 24 h. Media with drugs was replaced the day after seeding and cells harvested after 24 h. Also see Supplementary S2K Fig. This experiment was done three times with similar results. (M) Western blot of ARPE, LN229, SF188 and GBM9 for p53, p21 and cleaved caspase 3 after 72 h of treatment with 0.1 μM brequinar in the presence or absence of 100 μM uridine. (N) Western blot of ARPE, LN229, and GBM9 for p53, p21 and cleaved caspase 3 after 72 h of treatment with 2 μM ML390 in the presence or absence of uridine 100 μM. (O) Western blot of ARPE, LN229, SF188 and GBM9 for p53, p21 and cleaved caspase 3 after 48 h of treatment with 1 μM brequinar or 4 μM ML390. For all the Western blot experiments, media with drugs and metabolites were replaced the day after seeding and cells harvested at the indicated time points. Anti-phosphorylated H2AX antibody shows non-ubiquitinated (~ 15 KDa) and ubiquitinated (~ 25 KDa) γ-H2AX. * indicates p-values ≦0.05. Numerical values for each of the experiments represented are available in S2 Data.

Fig 3. Blocking DHODH with brequinar reduced reduces glioblastoma tumor xenografts growth in vivo.

(A) Representation of the subcutaneous xenograft experiment using LN229 cells. Mice were treated with 10 mg/kg brequinar with daily intraperitoneal injections (IP). (B) Tumor volume measurements of LN229 xenografts once the tumors reached 100 mm³. (C) Tumor weight of LN229 xenografts at day 60 (end of experiment). See also Supplementary S5B and S5C Fig. (D) Representative xenograft tumors from control and brequinar-treated mice. See also Supplementary S5B Fig. (E) Mouse weights at day 60 before the tumors were harvested. (F) Three representative control and brequinar-treated LN229 xenografts tumors showing increased blood vascularity in the control group (left panel) and qPCR for VEGFA in the LN229 xenografts. (G) Western blot of 5 control and brequinar-treated LN229 xenograft tumors for DHODH, p53, acetyl-tubulin, and HIF1α. * indicates p-values ≦ 0.05. Numerical values for each of the experiments represented are available in S3 Data.

Fig 4. Blocking DHODH with brequinar reduces the production of ribosomal RNA in glioblastoma tumors xenograft in vivo.

(A) Amounts of brequinar, UMP, UDP, UTP and uridine in the LN229 xenografts measured by LC-MS/MS. (B) qPCR of 47S pre-rRNA and 28S and 18S rRNAs in the LN229 xenografts. 47S pre-rRNA, 28S and 18S levels were normalized by ACTIN mRNA levels. (C) Amounts of brequinar, UMP, UDP, UTP and uridine measured by LC-MS/MS in the brain tissues of mice used for xenografts. (D) qPCR of 47S pre-rRNA and 28S and 18S rRNAs in the brain tissues of the mice used for LN229 xenografts. 47S pre-rRNA, 28S and 18S levels were normalized by ACTIN mRNA levels. (E) Amounts of brequinar, UMP, UDP, UTP and uridine measured by LC-MS/MS in the liver tissues of mice used for xenografts. (F) Amounts of brequinar, UMP, and uridine measured by LC-MS/MS in the serum of mice used for xenografts. Last brequinar injection was 3 h before harvesting the tissues. Numerical values for each of the experiments represented are available in S4 Data.

Fig 5. The inhibition of de novo pyrimidine biosynthesis causes nucleolar stress in glioblastoma cells.

(A) IF of UBF in LN229 with or without brequinar and uridine for 24 h. See also Supplementary S4A Fig. (B) IF of NPM1 in LN229 cells with or without brequinar and uridine for 24 h. See also Supplementary S4C Fig. Lower panels in (A) and (B) show UBF and NPM1 3D representation from Z-stack images. (C) IF of p53 in LN229 (upper panels) and GBM9 (lower panels) in the presence of brequinar with or without uridine for 24 h. (D) Quantification of (C) with Image J. (E) IF of rRNA in LN229 cells with the anti-rRNA Y10b antibody with or without brequinar and uridine for 24 h. See also Supplementary S4F Fig. (F) Quantification of (E) with Image J. (G) Ribosome profiling of LN229 treated with or without brequinar and uridine for 72 h. This experiment was done twice with similar results. For all experiments, media with drugs and metabolites were replaced the day after seeding, and cells harvested at the indicated time points. Scale bars = 5 μm. Numerical values for each of the experiments represented are available in S5 Data.

Fig 6. Inhibition of de novo pyrimidine biosynthesis reduces ribosomal RNA production and the ability to repair DNA in glioblastoma.

Inhibition of the de novo pyrimidine biosynthesis by blocking the activity of DHODH leads to reduced pyrimidine nucleotide availability and thus affects the synthesis of pre-rRNA and rRNA. This in turns induces nucleolar stress leading to changes in nucleolar morphology, the stabilization of p53, cell cycle arrest and cell death. Furthermore, brequinar and TMZ treatment leads to decreased expression of the DNA damage repair enzyme MGMT. The decrease in MGMT levels leads to an increase in H2AX phosphorylation and thus enhances the sensitivity of glioblastoma cells to DNA damage-induced death.