Although the treatment outcome for children with acute lymphoblastic leukemia (ALL) has improved substantially with the use of risk-directed treatment and improved supportive care, relapse remains a leading cause of mortality among all childhood ALL.1–3Abnormal purine metabolism is associated with the progression of cancers4–7 and thiopurines are among the first line drugs in ALL chemotherapy.1,3 Mutations in phosphoribosyl pyrophosphate synthetase 1 (PRPS1), the first rate-limiting and allosteric enzyme in the purine biosynthesis pathway,8–10 had been identified to drive drug resistance and childhood ALL relapse by reducing nucleotide feedback inhibition.11,12 However, the mechanisms by which purine metabolism regulates ALL relapse are still elusive.
Phosphoribosyl pyrophosphate synthetase 2 (PRPS2) encodes another PRPS isoform in the purine biosynthesis pathway,10,13–16 which shares 95% homology with PRPS1 amino acid sequence.9,17 PRPS2 was identified as a single rate-limiting enzyme coupling protein and nucleotide biosynthesis in Myc-driven tumorigenesis4 and regulated DNA damage18 and cancer stem cell self-renewal.6 Recently, we had implicated that PRPS2 could be important for thiopurine resistance in Burkitt lymphoma19 and ALL.11 PRPS2 forms a complex with PRPS1 and other 2 PRPS-associated proteins (Katashima et al9,20). However, the functions of PRPS2 in cancer metabolism and cancer relapse are still poorly understood.
In this study, integrating sequencing data of total 210 matched diagnosis-relapse samples in 2 independent ALL validation cohorts11,12 with clinical information from our center, Shanghai Children’s Medical Center (SCMC), we identified novel therapy-induced and recurrent relapse-specific mutations in PRPS2. Moreover, the functional PRPS2 mutations specifically regulated drug resistance through influencing PRPS1/2 hexamer stability, leading to reduced nucleotide feedback inhibition of PRPS activity. Our findings demonstrate a novel mechanism by which PRPS2 mutants drive drug resistance and childhood ALL relapse.
2 MATERIALS AND METHODS
2.1. Whole-exome sequencing and analysis
Whole-exome capture libraries were prepared according to standard protocols using SureSelect Human All Exon 50 and 38 Mb kit (Agilent technologies). Whole-exome sequencing was performed by using the Illumina HiSeq2000 instrument. SNVs/indels were detected as we described previously.12
2.2. Cell culture
HEK-293T cells, leukemia Reh, SUP-B15, Jurkat, and Molt4 cell lines were from ATCC (Manassas, Virginia). Reh, SUP-B15, Jurkat, and Molt4 cells were cultured in 10% FBS/RPMI 1640 medium. HEK-293T cells were cultured in 10% FBS/DMEM medium. All cells were incubated at 37 °C in 5% CO2. All cell lines in this study were authenticated using STR DNA fingerprinting, most recently in October 2017 by Shanghai Biowing Applied Biotechnology Co., Ltd (Shanghai, China), and mycoplasma infection was detected using LookOut Mycoplasma PCR Detection kit (Sigma-Aldrich).
2.3. Stable gene knockout using CRISPR/CAS9
Lenti CRISPR/Cas9 vector was a gift from Feng Zhang (Addgene plasmid #49535).32 gsRNAs were designed following the protocol of Zhang laboratory (http://crispr.mit.edu). The sequence targeted by PRPS1 CRISPR is 5′- TTGGTCCTTACCAGGTCTCC-3′ and the sequence targeted by PRPS2 CRISPR is 5′- GGATGATGACGCAATCTTGC-3′.
2.4. Lentivirus production and infection
Human PRPS1 and PRPS2 coding regions were cloned into pGV303 Vector (GeneChem, Shanghai, China) and different mutations were constructed using site-directed mutagenesis and confirmed by DNA sequencing. The constructs were transfected with packaging plasmids psPAX2 and pMD2G into HEK293T cells using the calcium phosphate method to produce replication-defective virus. The supernatant was harvested 48 hours later and concentrated by 100 kDa column (Amicon purification system, MILLIPORE), and Reh cells were virally transduced with supplemented with 8 μg/mL polybrene (Sigma). The medium was changed 24 hours after infection, and GFP-positive cells were sorted using MoFlo XDP (Beckman Coulter, Brea, CA, US).
2.5. Cell viability and apoptosis assays
Cell viability was determined by using Cell Titer-Glo Luminescent kit (Promega) according to the manufacturer’s instructions as we previously described.12 Briefly, cells were seeded in 96-well plated at 10,000 per well and treated with drugs of different serial dilutions for 72 hours. Then, the Cell Titer-Glo Reagents (50 μL) were added to each well and mixed for 10 min before the luminescent signal was measured using a microplate reader (Biotek, Winooski, Vermont, US). Apoptosis was measured using Annexin V-PE and 7-AAD staining (Annexin V-PE Apoptosis Detection kit, BD Biosciencs, Franklin Lakes, NJ, US) followed by flow cytometry analysis using a FACS (Canto II) as we previously described.33
2.6. IP and WB
IP and WB were performed as we previously described.34 Cells were lysed in an IP buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 2 mM Na3VO4, 5 mM NaF, 1%Triton X-100 and protease inhibitor cocktail) at 4°C for 30 minutes. The lysates were centrifuged, and the protein concentrations were determined. Equal amounts of cell lysates were immunoprecipitated with specific antibodies and protein G-agarose beads (Invitrogen, Carlsbad, California). Standard WB was performed with antibodies against γH2AX(S139) (#3522-1, Epitomics, Burlingame, CA, US), PARP (#46D11, Cell Signaling Technology, Danvers, MA ,US), Cleaved PARP (Asp214) (D64E10, Cell Signaling Technology, Danvers, MA ,US), His-Tag (D3I10, Cell Signaling Technology, Danvers, MA ,US), β-actin (I-19, Santa Cruz Biotechnology, Dallas, TX, US), PRPS2 (NBP1-56666, Novus Biologicals), PRPS1 (sc-376440, Santa Cruz Biotechnology, Dallas, TX, US), or Flag (MS2, Sigma-Aldrich, Burlington, MA, US) using the Odyssey system (LI-COR Biosciences, Lincoln, NE, US).
2.7. Protein purification
WT or mutants of PRPS1 and PRPS2 genes with an N-terminal hexahistidine (6×His) tag were cloned into the pET-28a expression vector. The plasmids were transformed into and expressed in E. coli BL21 (DE3) strain (Tiangen). Then, harvested cells pellets were suspended in buffer A [50 mM NaH2PO4 (pH 8.0), 1 M NaCl, 15% (Weight/Volume) glycerol, 5 mM 2-mercaptoethanol and 1 mM PMSF] and lysed on ice by sonication before the supernatants were collected by centrifugation. The supernatants were loaded onto a Ni Sepharose FF column (GE Healthcare, Pittsburgh, PA, US) in the AKTA purifier system. The column was washed with buffer A and then eluted with buffer B (Buffer A+500 mM imidazole). We removed imidazole through buffer exchange using G25 desalting columns (GE Healthcare), and assessed protein expression and purity using SDS-PAGE with Coomassie Brilliant Blue R250 staining.
2.8. PRPS1/2 enzymatic activity and ADP/GDP feedback inhibition assays
PRPS1/2 enzyme activity and ADP/GDP feedback inhibition of PRPS enzyme activity were performed using a Kinase-Glo luminescent kinase assay Kit (Promega) according to the manufacturer’s instructions as we previously described.12 In brief, 10 μL purified recombinant WT or mutant PRPS1 or PRPS2 with various concentration was incubated in 100 μL of reaction buffer (50 mM Tris, pH 7.5, 10 mM MgCl2, 1 mM DTT, 500 μM ATP, 500 μM R5P, 2 mM phosphate) at 37°C for 1 hours in a 96-well plate. The reaction was terminated by adding 10 μL Kinase-Glo reagent. In GDP feedback inhibition assay, GDP from 6 to 0.25 μM was added in the reaction buffer.
2.9. Metabolite flux assays
Metabolite flux assays were performed as described previously.12 Cells were cultured in RPMI 1640 media at a density of 5 × 105/mL. Isotope-labeled [13C2, 15N] Glycine (Sigma, Cat#489522) or [13C5, 15N4] Hypoxanthine (Cambridge Isotope Laboratories. Tewksbury, MA, US, Cat#CNLM-7894-PK) was added to cells then cultured for 2 hours. Cells were then harvested, pelleted and quenched in cold 80% methanol, centrifuged at 12,000 rpm for 10 minutes, and the supernatant was applied for metabolite analysis by AB SciexQtrap 5500 coupled with Waters Acquity UPLC. IMP synthesis (flux) through de novo purine synthesis pathway was measured by [13C2, 15N] incorporation into cells (molecular weight peak IMP+3); IMP synthesis (flux) through purine salvage pathway measured by [13C5, 15N4] incorporation into cells (molecular weight peak IMP+9).
For PRPP measurement, cells were cultured in RPMI 1640 media and then labeled with [U-13C6] D-glucose (Cambridge isotope laboratories) for 5 minutes. Cells were then harvested, pelleted and quenched in cold 80% methanol, the newly synthesized PRPP in cells were measured by [13C5] incorporation into cells (molecular weight peak PRPP+5). For ADP and GDP feedback inhibition test, we first treated cell by 2 mM ADP or 1.5 mM GDP for 24 hours, and then harvested the cells and measured the newly synthesized.
2.10. Thiopurine conversion and thiopurine cytotoxic metabolite assays
Cells were cultured in RPMI 1640 media containing 10 μM 6-MP for 4 hours, then harvested and assayed based on a method modified as described previously.12 Intracellular accumulation of TIMP, 6-MP metabolites and their derivatives were determined by LC-MS as described previously.12
2.11. Structural analysis
Structural analysis of various PRPS2 mutations and the 3AA was based on the crystal structure of human PRPS1 (PDB code, 2HCR).17 The figures were prepared using PyMol (www.pymol.org).
2.12. Size-exclusion chromatography
A HiPrep 16/60 Sephacryl S-300 HR column (GE) was used to perform size-exclusion chromatography according to the manufacturer’s recommendation. Briefly, cell lysates were loaded onto the column and collected at a flow rate of 0.3 mL/min. Then, the sample fractions were analyzed using WB. The standard curve of the elution was plotted against LogMW by using a size-exclusion chromatography calibration marker kit (Sigma) according to the manufacturer’s recommendation.
2.13. Immunofluorescence staining
Human PRPS1 or PRPS2 cDNA was cloned into the pcDNA3.1-EGFP or pcDNA3.1-RFP vector, respectively. In pcDNA3.1-EGFP and pcDNA3.1-RFP vectors, the monomeric green fluorescent protein (mEGFP) and the monomeric RFP (mRFP) were derived from hTriGART-mEGFP and phFGAMS-mOFP (gifts from Dr. Stephen J. Benkovic, the Pennsylvania State University).27 Then, PRPS1-mRFP and PRPS2-mEGFP plasmids were transiently co-transfected into Reh cells cultured in purine-rich media (10% FBS/RPMI 1640 medium) using Lipofectamine 2000 (Invitrogen). Cells were maintained in purine-depleted media (RPMI 1640 medium with dialyzed 5% FBS) for 3 days as reported previously.27 Finally, cells were collected, fixed, and images were produced with a confocal microscope (Leica, Buffalo Grove, Illinois) at X 600 magnification.
2.14. Tumorigenesis studies
White severe combined immunodeficiency (SCID) female mice aged 6–8 weeks (SLAC, Shanghai, China) were used. Mice were randomly divided into 5 per group. In total, 1 × 106 clinical ALL cells were injected into recipients through the tail vein as previously described.35 After 7 days, the mice were treated with the vehicle (PBS) or 0.6 mg/kg 6-MP per day for 10 days. Mice were euthanized when ALL symptoms developed. All animal experiments were performed in accordance with a protocol approved by Shanghai Jiao Tong University Institutional Animal Care and Use Committee (IACUC).
2.15. Statistics analysis
GraphPad Prism version 5.0 for Windows (GraphPad Software Inc., San Diego, California) was used to perform one-way analysis of variance (ANOVA) with Newman–Keuls post hoc test or an unpaired, 2-tailed Student t-test. Relapsed ALL analysis was carried out by Kaplan–Meier analysis and was compared with Newman–Keuls post-test as we previously described.34 A P value of less than .05 was considered statistically significant. All data represent the mean ± SD of 3 independent experiments/samples unless specifically indicated.
3.1. PRPS2 mutations are closely associated with drug resistance and childhood ALL relapse
To determine how genetic lesions contribute to the relapse in childhood ALL, we screened our deep sequencing data of 210 paired diagnosis-relapse bone marrow samples in 2 independent ALL validation cohorts with clinical information11,12 from our center, SCMC and found 7 relapse-specific PRPS2 mutations in 6 patients (n = 6; 6/210, 2.9%) causing A134T, S106I, V48M, R22S, P173R, P173Y, or A175T mutation, respectively (Fig. 1A and Supplemental Table 1, https://links.lww.com/BS/A50). Since human PRPS1 exists as a hexamer comprising 3 homodimers,17 we made a simulated hexamer crystal structure of human PRPS2 based on the reported crystal structure of human PRPS117,21 and then mapped PRPS2 mutations. As shown in Figure 1B, the S106, A134, and A175, but not V48 and R22, residues are at trimer–trimer interface, and the P173 residue is in a turn motif and may be critical to maintain interface helix conformation. Since PRPS2 functions in purine biosynthesis and its mutations are associated with on-treatment relapse, we determined whether mutations in PRPS2 allow for resistance to nucleotide analogs in ALL thiopurine chemotherapy. Ectopic expression of PRPS2 P173R, P173Y, or A175T mutant markedly increased cell resistance to thiopurines (6-MP and 6-TG)3 in Reh ALL cells compared to the empty vector (EV) or PRPS2 wild type (WT) (Fig. 1C and D) whereas the expression of PRPS2 WT, A134T, S106I, V48M, or R22S mutant had minimal or no effects on drug resistance (Fig. 1C and D). In addition, ectopic expression of PRPS2 WT or these mutants had minimal effects on sensitivity to other chemotherapeutics used in clinical ALL treatment (Supplemental Figure 1A, https://links.lww.com/BS/A50), such as methotrexate (MTX), L-asparaginase (L-ASP), daunorubicin (DNR), or cytosine arabinoside (Ara-C)22 compared with the EV control. There were no significant effects on cell proliferation expressing PRPS2 WT or mutants compared to the control (Supplemental Figure 1B, https://links.lww.com/BS/A50). These results suggest that PRPS2 mutations at P173 and A175 residues are specific for thiopurine resistance.
Thiopurines exert their cytotoxicity primarily through mismatch repair pathway-mediated DNA damage response (DDR) and apoptosis.23 Drug-resistant PRPS2 P173R, P173Y, and A175T mutations reduced 6-MP-induced cell apoptosis compared to the EV control (Fig. 1E), whereas PRPS2 WT, S106I, and R22S mutations, but not the A134T and V48M mutations, increased cell apoptosis (Fig. 1E). This observation was validated by assessing expression levels of the apoptosis biomarker cleaved poly (ADP-ribose) polymerase (PARP) and the DDR biomarker γ-H2AX (Fig. 1F). We further tested the effects of PRPS2 WT and mutations on cell viability and cell apoptosis in other leukemia cell lines, including SUP-B15, Jurkat, and Molt4, and found the similar results. Overexpression of the drug-resistant PRPS2 P173R, P173Y, or A175T mutant significantly increased cell viability (Supplemental Figure 2A and B, https://links.lww.com/BS/A50) and reduced cell apoptosis (Supplemental Figure 2C, https://links.lww.com/BS/A50) after treatment with 6-MP and 6-TG compared with WT PRPS2 and the EV control in the 3 leukemia cell lines. These data support that drug-resistant PRPS2 mutations promote thiopurine resistance by impairing thiopurine-induced DDR and cell apoptosis.
Finally, we checked the treatment and survival of ALL patients with PRPS2 mutations. Interestingly, all functional PRPS2 mutations (P173 and A175) were identified in patients with continuous thiopurine (6-MP and/or 6-TG) treatment (Supplemental Figure 3, https://links.lww.com/BS/A50), whereas other PRPS2 mutations were detected in patients after stopping thiopurine treatment. We further performed ultra-deep sequencing of matched samples obtained at diagnosis, remission, and relapse from 2 patients with PRPS2 A175 or P173 mutation, and found that PRPS2 A175 or P173 mutation was only identified in the relapse specimen (Supplemental Table 2, https://links.lww.com/BS/A50). Then, we examined the relationship of PRPS2 functional mutations and ALL relapse by the Kaplan–Meier analysis in the 2 cohorts and found a statistically significant shorter relapse time for ALL patients with PRPS2 mutations (mut) compared with those with WT PRPS2 and WT PRPS1 (WT) with a median relapse time 17.47 and 35.10 months, respectively (P < .05, Fig. 1G). There was no survival difference between those with PRPS2 mut and PRPS1 mut, with a median relapse time 17.47 and 15.29 months, respectively (Fig. 1G). Taken together, our data strongly indicate that PRPS2 mutations are closely associated with drug resistance and childhood ALL relapse.
For the similar functions of A175 and the P173 residue in a hexamer and drug resistance to 6-MP/6-TG, we selected P173R did the next experiments.
3.2. PRPS2 P173R mutation regulates PRPS activities
We previously reported that ectopic expression of PRPS1 WT had some partial thiopurine resistance effect, while PRPS1 A190T mutant, with constitutive high enzyme activity, had enhanced thiopurine resistance.12 However, in our current study, the overexpression of PRPS2 WT did not affect drug resistance compared with the EV control, whereas PRPS2 P173R mutation significantly enhanced thiopurine drug resistance (Fig. 2A). This observation suggests that PRPS2 might play different roles from PRPS1 in purine biosynthesis and thiopurine resistance.
PRPS2 was thought to have lower enzymatic activity than PRPS1 and not subject to feedback inhibition.9 We purified WT and mutant of PRPS1 and PRPS2 and performed enzymatic activity assays. As shown in Figure 2B, the enzymatic activity of WT PRPS2 was significantly lower than that of PRPS1 WT or A190T mutant, and the activity of PRPS2 P173R mutant had no marked difference from that of PRPS2 WT. This result suggests that the drug resistance of PRPS2 P173R mutant is not directly correlated with PRPS2 enzymatic activity.
As we and others have demonstrated that PRPS activities are best measured using cell-based assays,12,24 we further detected PRPS1/2 downstream metabolites specific for the de novo and salvage purine pathways (Fig. 2C) by LC-MS with isotope-labeled substrates with or without 6-MP treatment in Reh cells. Consistent with our earlier results,12 compared to the EV control, ectopic expression of PRPS1 WT or A190T mutant increased the levels of purine nucleotides in the de novo purine pathway, IMP (+3), IMP (+9), HX, ADP, PRPP, and GDP with or without 6-MP treatment and decreased the levels of cytotoxic molecules (TIMP and TGMP) in the salvage purine pathway under 6-MP treatment (Fig. 2D). In contrast, overexpression of WT PRPS2 had no significant influence on all purine biosynthesis of intermediates compared with the EV control with or without 6-MP treatment (Fig. 2D). However, only on 6-MP treatment, ectopic expression of PRPS2 P173R mutant markedly increased the levels of HX, ADP, and GDP and moderately decreased TGMP and TIMP levels compared with the EV control (Fig. 2D). These data suggest again that PRPS2 P173R mutation specifically mediates therapy-induced purine metabolism and its functions may be different from PRPS1 WT and A190T mutant.12
We further performed nucleotide feedback inhibition of PRPS activity with ADP or GDP treatment in Reh cells.12,24 As shown in Figure 2E, ADP/GDP treatment inhibited the labeled PRPP production in the cells expressing PRPS1 WT but not the A190T mutant as we previously reported that PRPS1 A190T mutation impairs the nucleotide feedback inhibition of PRPS activities.12 PRPS2 WT overexpression had no effect on the labeled PRPP production after ADP or GDP treatment, whereas ectopic expression of PRPS2 P173R mutant significantly increased PRPP production (Fig. 2E), suggesting that PRPS2 P173R mutation also affects the nucleotide feedback inhibition of cellular PRPS enzymatic activity. This is further supported by effects of ectopic expression of PRPS2 P173R mutant on purine derivative hypoxanthine (HX)12 and GART inhibitor lometrexol12,25 treatment-induced 6-MP resistance. HX treatment enhanced 6-MP resistance in all the indicated cells, including the cells expressing an EV control (Fig. 2F). However, as in WT or A190T mutant PRPS1 cells, lometrexol treatment enhanced 6-MP resistance in the cells expressing PRPS2 P173R mutant but not WT PRPS2 or the EV control (Fig. 2G). These data suggest that PRPS2 P173R mutation resembles PRPS1 A190T mutation to cause defects in the nucleotide feedback inhibition of PRPS activities in thiopurine resistance.
3.3. PRPS2 is critical for nucleotide feedback inhibition of PRPS activity
To demonstrate the function of PRPS2 in nucleotide feedback inhibition of PRPS activity, we established PRPS2 KO and PRPS1 KO cell lines using CRISPR/Cas9 technology. Knockout of PRPS1 or PRPS2 significantly caused Reh cell sensitivity to 6-MP treatment (Fig. 3A and B) and promoted cell apoptosis (Fig. 3C), suggesting that both PRPS1 and PRPS2 are important for thiopurine resistance in childhood ALL.
Next, we determined nucleotide feedback inhibition of PRPS activity with ADP or GDP treatment in Reh cells with PRPS1 KO or PRPS2 KO as we described previously.12,24 Both PRPS1 KO and PRPS2 KO markedly reduced the PRPS activity (Fig. 3D). However, ADP and GDP treatment significantly reduced PRPP production in PRPS2 KO and control cells but not PRPS1 KO cells (Fig. 3D). This shows that PRPS2 is critical for nucleotide feedback inhibition of PRPS activity.
The dNTP pools are affected by the metabolites of the purine synthesis and regulate genomic stability.5,12 To support our above observation, we measured the dNTP pools by metabolite flux in the indicated Reh cells (Fig. 3E). Deletion of PRPS1 or PRPS2 markedly reduced dNTP pools whereas there was no difference between PRPS1 KO and PRPS2 KO (Fig. 3E). Taken together, these results demonstrate that PRPS1 and PRPS2 have different functions in regulating PRPS activity.
3.4. The 3-amino acid V103-G104-E105 insertion in PRPS2 significantly decreases its PRPS activity
To determine the functional difference, we compared the amino acid sequences of human PRPS1 and PRPS2 and found that a sequence difference is the 3-amino acid V103-G104-E105 (3AA, VGE) of PRPS2 (Fig. 4A and Supplemental Figure 4A, https://links.lww.com/BS/A50). The PRPS1 hexamer crystal structure shows that the loop of 98-DKKDKSRAPISAK-110 is critical for the compact hexamer formation in PRPS1 (Supplemental Figure, 4B https://links.lww.com/BS/A50) and the 3AA (VGE) insertion in PRPS2 caused severe steric clash to the interface of 2 trimers of the hexamer formed by a PRPS1 trimer and a simulated PRPS2 trimer (Fig. 4B).
To test our hypothesis, we swapped the 3AA between PRPS2 and PRPS1: inserting the nucleotides encoding the 3AA into the full length of PRPS1 (named PRPS1+3AA), or deleting the nucleotides from PRPS2 (named PRPS2−3AA) (Fig. 4C). PRPS1+3AA mutation significantly attenuated cell resistance to 6-MP compared to PRPS1 WT, whereas PRPS2-3AA mutant increased cell resistance compared with PRPS2 WT. We purified PRPS1+3AA and PRPS2−3AA mutant proteins and found that PRPS1+3AA mutant markedly reduced PRPS enzyme activity in comparison with PRPS1 WT, whereas PRPS2-3AA mutant has higher activity than PRPS2 WT (Fig. 4E). These data support that the 3AA is critical for PRPS activity.
Next, we detected the effects of the 3AA on PRPS downstream metabolites in Reh cells. As shown in Figure 4F, compared to WT PRPS1, ectopic expression ofPRPS1+3AA mutant significantly reduced levels of PRPP, ADP, GDP, and HX and increased TGMP and TIMP levels in Reh cells with or without 6-MP treatment. On the contrast, overexpression of PRPS2−3AA mutant significantly increased HX levels in cells without 6-MP treatment and ADP, GDP levels in 6-MP-treated cells compared with PRPS2 WT (Fig. 4F). PRPS2−3AA mutation also reduced TGMP and TIMP levels in cells with 6-MP treatment (Fig. 4F). These data support that the 3AA in PRPS2 is critical for its functional difference from PRPS1.
We further investigated the effects of 2 mutants on nucleotide feedback inhibition of PRPS activity with ADP or GDP treatment. As shown in Figure 4G, ectopic expression of PRPS1+3AA mutant markedly inhibited PRPP production compared with PRPS1 WT, whereas PRPS2−3AA overexpression significantly increased PRPP production compared with PRPS2 WT. ADP and GDP treatment significantly inhibited PRPP production in Reh cells expressing WT PRPS1 or PRPS2−3AA mutant but not in the cells expressing WT PRPS2 or PRPS1+3AA mutant (Fig. 4G). These data further suggests that the 3AA is critical for nucleotide feedback regulation of PRPS activity.
We also tested the effects of insertion of the 3AA into PRPS1 A190T mutant or deletion of the 3AA in PRPS2 P173R mutant on 6-MP resistance and PRPS enzyme activity. Insertion of the 3AA into PRPS1 A190T mutant reversed A190T mutant’s drug resistance (Fig. 4H). In contrast, deletion of the 3AA in PRPS2 P173R mutant further enhanced P173R mutant’s drug resistance (Fig. 4H). On the other hand, insertion of the 3AA decreased PRPS1 A190T mutant’s activity, while remained insensitive to GDP feedback inhibition (Fig. 4I). Deletion of the 3AA in PRPS2 P173R mutant increased P173R mutant’s activity, while gaining sensitivity to GDP feedback inhibition (Fig. 4I). Taken together, these data demonstrate that the 3AA in PRPS2 is critical for its activity and feedback regulation of PRPS activity.
3.5. Thiopurine resistance of PRPS2 P173R mutation depends on PRPS1 expression
Since PRPS1 and PRPS2 are required for purine biosynthesis in ALL cells and they might form complex with other 2 PRPS-associated proteins inside the cell,20,26 we hypothesized that higher PRPS cellular activity and the drug resistance of PRPS2 P173R mutant might be related to its interaction with PRPS1 under pathological conditions. To test our hypothesis, we first determined the drug resistance of PRPS1 KO Reh cells expressing PRPS2 WT or P173R mutant and PRPS2 KO Reh cells expressing PRPS1 WT or A190T mutant (Fig. 5A). Compared to the control in Reh cells, PRPS1 KO markedly decreased 6-MP resistance of the cells expressing PRPS2 WT, P173R mutant, or an EV control, whereas PRPS2 KO significantly increased 6-MP resistance of the cells expressing PRPS1 WT and A190T mutant (Fig. 5B). These data suggest that PRPS1 is required for thiopurine resistance of PRPS2 P173R mutation.
Next, we assessed the effects of PRPS2 or PRPS1 KO on PRPS1 and PRPS2 mutants-mediated PRPS downstream metabolites by LC-MS with isotope-labeled substrates. Compared with the control, PRPS1 KO inhibited upregulation of ADP and GDP and downregulation of TIMP and TGMP by PRPS2 P173R mutant but not PRPS2 WT on 6-MP treatment (Fig. 5C). PRPS2 KO further enhanced upregulation of PRPP, ADP, GDP, HX, and downregulation of TGMP and TIMP by PRPS1 WT and P173R mutant with or without 6-MP treatment. These data further support that thiopurine resistance of PRPS2 P173R mutation depends on PRPS1 expression.
To further demonstrate whether PRPS2 P173R mutation affects the activity of the PRPS1/2 complex, we co-expressed PRPS2 WT or P173R mutant and PRPS1 WT to the same levels in Escherichia coli cells and purified PRPS1/2 and PRPS1/PRPS2-P173R complex, respectively (Fig. 5D). The enzymatic activity of the PRPS1/2 complex is lower than that of PRPS1 alone but markedly higher than that of PRPS2. On the other hand, the activity of the PRPS1/PRPS2-P173R complex is similar to that of PRPS1/2 (Fig. 5D), supporting that thiopurine resistance of PRPS2 P173R mutation is not due to its enzyme activity. We further investigated the effect of P173R mutation on the nucleotide feedback inhibition of PRPS activity with GDP treatment. Consistent with our previous report,12 GDP treatment significantly increased the nucleotide feedback inhibition of PRPS activity of purified PRPS1 (Fig. 5E). While there were no effects on the enzyme activities of purified PRPS2 and PRPS1/2 with GDP treatment, the GDP treatment significantly reduced the nucleotide feedback inhibition of PRPS activity of the purified complex of PRPS1/PRPS2-P173R (Fig. 5E). Taken together, our data suggest that thiopurine resistance of PRPS2 P173R mutation depends on PRPS1 in ALL cells.
3.6. PRPS2 P173R mutation influences PRPS hexamer stability
Our simulated hexamer crystal structure of human PRPS2 indicated that the P173 residue in a turn motif may be critical to maintain interface helix conformation and hexamer stability. To determine this observation, we performed immunoprecipitation (IP)-WB analysis and immunofluorescence (IF) staining to examine the effects of the P173R mutant on the interaction and co-localization of PRPS1 and PRPS2 in Reh cells. Compared to PRPS2 WT, nonfunctional S106I and R22S mutants, P173R mutation markedly inhibited its association with PRPS1 in Reh cells (Fig. 6A).
It has been previously demonstrated that PRPS1 and PRPS2 form clusters with other 4 enzymes in the purine de novo pathway in cells.27 To assess whether P173R mutation influences the cluster formation of PRPS enzymes, red fluorescent protein (RFP)-fused PRPS1 WT was transiently co-infected with EGFP-fused PRPS2 WT or P173R mutant into Reh cells, and then cells were cultured in purine-depleted medium for 3 days as described previously.27 As shown in Figure 6B, while PRPS2 WT co-localized with PRPS1 within clusters, the PRPS2 P173R mutant was only partially co-localized with PRPS1 within clusters, supporting that P173R mutation influences the stability of its complex. This observation was further validated by IP and IB analyses in normal peripheral blood mononuclear cells (PBMC) and clinical ALL specimens with PRPS2 mutations. As shown in Figure 6C, compared to normal PBMC control, the functional P173R mutation of PRPS2 but not the nonfunctional S106I and R22S mutations decreased PRPS2 protein binding to PRPS1.
Next, to further reveal how PRPS2 P173R mutation influences the stability of the PRPS1/2 complex, we co-expressed His-tagged PRPS1 WT with Flag-tagged PRPS2 WT or P173R mutant in Reh cells, and performed the size-exclusion chromatography assay as described previously.5 As PRPS hexamers and monomers have been detected in human cancer cells, we collected all the fractions eluted as previously described.5 As shown in Figure 6D, Flag-tagged PRPS2 WT or the P173R mutant with His-tagged PRPS1 was detected only in fractions 75-103 that co-eluted with a molecular weight around 200 kDa (according to the molecular weight calibration standard), suggesting that PRPS1 and PRPS2 WT or P173R mutant mainly form hexamers in Reh cells. Quantitative western blot analysis of PRPS1 and PRPS2 relative amounts suggested that the composition of WT PRPS1 and PRPS2 formed stable hexamer complex at a (PRPS2)3:(PRPS1)3 ratio, but PRPS2 P173R mutant formed hexamers with WT PRPS1 at ratios of 4:2, 3:3, and 2:4 (Fig. 6D), implicating changes in complex equilibrium and potential weakening of PRPS1 and PRPS2 interaction within hexamer complex. We further collected 3 fractions with PRPS2-WT/PRPS1-WT or PRPS2-P173R/PRPS1-WT and performed enzyme activity assays. As shown in Figure 6E, all the fractions with the same ratio of PRPS2 WT versus PRPS1 WT had similar enzymatic kinetic curves and there was no difference for their enzyme activity. Compared to the complex fraction 86 with a 3:3 ratios of PRPS2 P173R versus PRPS1 WT, the complex fraction 80 with a 4:2 ratio showed lower enzyme activity but the complex fraction 96 with a 2:4 ratio had higher enzyme activity. Taken together, these data demonstrate that PRPSR P173R mutation weakens PRPS hexamer stability and increases thiopurine resistance.
To evaluate the in vivo drug resistance effect of PRPS2 P173R mutation, we employed an ALL xenograft model. Clinical ALL specimens, ALL-114 with nonfunctional R22S mutation and ALL-127 with P173R mutation were separately implanted into the immunocompromised mice with intravenous tail injection. The effects of P173R mutation on ALL tumorigenesis with or without 6-MP treatment were assessed. As shown in Figure 6F, 6-MP treatment significantly increased the survival of animals bearing R22S mutation xenografts (ALL-114) but did not affect the survival of those bearing P173R mutant xenografts (ALL-127), providing in vivo evidence supporting the role of PRPS2 P173R mutation in ALL drug resistance and tumor relapse. Taken together, these data demonstrate that PRPS2 P173R mutation influences PRPS1/2 hexamer stability and reduces nucleotide feedback inhibition of PRPS activity, leading to abnormal drug resistance and ALL relapse.
Metabolism is a hallmark for cancer,28 yet the mechanisms by which abnormal metabolism causes drug resistance and tumor relapse are still unclear. Herein, we identified thiopurine therapy-induced PRPS2 mutations as new drivers of drug resistance and ALL relapse. Our study not only demonstrate a novel mechanism by which PRPS2 mutations induce ALL relapse, but also reveals previously unknown regulation of PRPS2 enzymatic activity.
This study identified and validated the first purine biosynthesis rate-limiting enzyme PRPS2 mutants as new regulators of thiopurine resistance and childhood ALL relapse. PSPR2 was early characterized as a purine biosynthesis enzyme.15 Later PRPS2 was found as a rate-limiting enzyme of coupling protein and nucleotide biosynthesis.4 PRPS2-mediated purine metabolism was also related with tumor glucose deprivation or hypoxia,5 innate immune response,18 and maintenance of brain tumor initiating cells.6 However, its roles in drug resistance and tumor relapse are still unknown. Here we integrated the sequencing data of total 210 matched diagnosis-relapse samples in 2 independent ALL validation cohorts with clinical information from our center and identified PRPS2 mutations (2.8%, 3/107 in 2015 SCMC cohort, and 2.9%, 3/103 in 2020 SCMC cohort) as new drivers of therapy resistance and ALL relapse. The overall frequency was lower than PRPS1 (13%, 18/138 in 2015 SCMC cohort, and 3.9%, 4/103 in 2020 SCMC cohort). In addition, there were no patient with both PRPS1 and PRP2 mutation. So, it suggested that the functional PRPS2 mutations were prognostic factor for ALL relapse. Thus, we reveal a new function of PRPS2, providing a rationale for developing therapeutic strategies to overcome thiopurine resistance in the clinic.
Although PRPS1 and PRPS2 are 2 isoforms in human PRPS family with 95% amino acid sequence identity and can form a complex,9,17,29,30 their functional difference is poorly characterized. As far as we know, we were the first to identify that the 3-amino acid V103-G104-E105 in PRPS2 is critical for its PRPS activity. Using genetic approaches, we demonstrated that exchanging the 3AA could shift the enzymatic activity, allosteric regulation, the nucleotide feedback inhibition of PRPS activities and drug resistance between PRPS1 and PRPS2. The 3AA (VGE) insertion in PRPS2 caused severe steric clash to the interface of PRPS hexamer, leading to its low enzyme activity. We found only PRPS2 P173R mutation without PRPS1 was sensitive to thiopurine. That means the main function of PRPS2 P173R mutation was regulation of the nucleotide feedback inhibition of PRPS activities. The PRPS2 P173 residue was in a turn motif and may be critical to maintain interface helix conformation. PRPS2 P173R mutation caused the hexamer unstability, reduced the nucleotide feedback inhibition of PRPS activities to induce drug resistance. The molecular insight of this unique observation warrants further investigation.
Our finding demonstrates that PRPS hexamer stability defect could cause drug resistance and tumor relapse. PRPS2 and PRPS1 form complexes with other 2 phosphoribosyl pyrophosphate synthetase associated proteins (PAP39 and PAP41),26,27,31 and the complex stability is critical for PRPS enzymatic activity.5 For example, PRPS1 and PRPS2 formed hexamer and mediates nucleotide synthesis to maintain glioma tumor cell growth and survival.5 AMPK-dependent phosphorylation of PRPS1 leaded to conversion of PRPS1/2 hexamers to monomers, thereby inhibiting PRPS1/2 activity and nucleic acid synthesis in response to energy stress.5 As we know, we are the first to identify that PRPS2 mutation regulates PRPS1/PRPS2 hexamer stability and the P173 mutation leads to thiopurine resistance and ALL relapse in clinic. Our data are in line of the previous findings and further support the importance of PRPS hexamer.
Our work identified new drivers in drug resistance and ALL relapse and demonstrate a novel mechanism by PRPS2 mutation impairs PRPS hexamer stability, leading to reduced nucleotide feedback inhibition of PRPS activity. Furthermore, our study identified PRPS2 mutations as new clinical diagnosis markers and potential therapeutic targets in childhood ALL relapse.
This work was supported in part by National Natural Science Foundation of China (81972341, 81900158, 81470315, 81874078, 82072896); Shanghai Municipal Science and Technology Commission (201409002700,19JC1413500, 21XD1403100) and Shanghai Municipal Education Commission-Gaofeng Clinical Medicine Grant Support (20161310), Pudong New Area Science &Technology Development Fund (PKJ2018-Y47), Collaborative Innovation Center for Translational Medicine at Shanghai Jiao Tong University School of Medicine fund TM201502.
We thank Dr. Stephen J. Benkovic in the Pennsylvania State University for kindly providing hTriGART-mEGFP and phFGAMS-mOFP.
1. Hunger SP, Mullighan CG. Acute lymphoblastic leukemia in children. N Engl J Med 2015;373:1541–1552.
2. Inaba H, Greaves M, Mullighan CG. Acute lymphoblastic leukaemia. Lancet 2013;381:1943–1955.
3. Pui CH, Carroll WL, Meshinchi S, Arceci RJ. Biology, risk stratification, and therapy of pediatric acute leukemias: an update. J Clin Oncol 2011;29:551–565.
4. Cunningham JT, Moreno MV, Lodi A, Ronen SM, Ruggero D. Protein and nucleotide biosynthesis are coupled by a single rate-limiting enzyme, PRPS2, to drive cancer. Cell 2014;157:1088–1103.
5. Qian X, Li X, Tan L, et al. Conversion of PRPS hexamer to monomer by AMPK-mediated phosphorylation inhibits nucleotide synthesis in response to energy stress. Cancer Disc 2018;8:94–107.
6. Wang X, Yang K, Xie Q, et al. Purine synthesis promotes maintenance of brain tumor initiating cells in glioma. Nat Neurosci 2017;20:661–673.
7. Zhou W, Yao Y, Scott AJ, et al. Purine metabolism regulates DNA repair and therapy resistance in glioblastoma. Nat Commun 2020;11:3811.
8. Becker MA, Kim M. Regulation of purine synthesis de novo in human fibroblasts by purine nucleotides and phosphoribosylpyrophosphate. J Biol Chem 1987;262:14531–14537.
9. Nosal JM, Switzer RL, Becker MA. Overexpression, purification, and characterization of recombinant human 5-phosphoribosyl-1-pyrophosphate synthetase isozymes I and II. J Biol Chem 1993;268:10168–10175.
10. Wyngaarden JB. Regulation of purine biosynthesis and turnover. Adv Enzyme Regul 1976;14:25–42.
11. Li B, Brady SW, Ma X, et al. a) Therapy-induced mutations drive the genomic landscape of relapsed acute lymphoblastic leukemia. Blood 2020;135:41–55.
12. Li B, Li H, Bai Y, et al. Negative feedback-defective PRPS1 mutants drive thiopurine resistance in relapsed childhood ALL. Nat Med 2015;21:563–571.
13. Gerhart J. From feedback inhibition to allostery: the enduring example of aspartate transcarbamoylase. FEBS J 2014;281:612–620.
14. Hove-Jensen B, Andersen KR, Kilstrup M, Martinussen J, Switzer RL, Willemoes M. Phosphoribosyl diphosphate (PRPP): biosynthesis, enzymology, utilization, and metabolic significance. Microbiolo Mol Biol Rev 2017;81:e00040-16.
15. Taira M, Ishijima S, Kita K, Yamada K, Iizasa T, Tatibana M. Nucleotide and deduced amino acid sequences of two distinct cDNAs for rat phosphoribosylpyrophosphate synthetase. J Biol Chem 1987;262:14867–14870.
16. Tatibana M, Kita K, Taira M, et al. Mammalian phosphoribosyl-pyrophosphate synthetase. Adv Enzyme Regul 1995;35:229–249.
17. Li S, Lu Y, Peng B, Ding J. Crystal structure of human phosphoribosylpyrophosphate synthetase 1 reveals a novel allosteric site. Biochem J 2007;401:39–47.
18. Liu R, Li J, Shao J, et al. Innate immune response orchestrates phosphoribosyl pyrophosphate synthetases to support DNA repair. Cell Metab 2021;33:2076–2089.e9.
19. Li T, Song L, Zhang Y, et al. b) Molecular mechanism of c-Myc and PRPS1/2 against thiopurine resistance in Burkitt’s lymphoma. J Cell Mol Med 2020;24:6704–6715.
20. Kita K, Ishizuka T, Ishijima S, Sonoda T, Tatibana M. A novel 39-kDa phosphoribosylpyrophosphate synthetase-associated protein of rat liver. Cloning, high sequence similarity to the catalytic subunits, and a negative regulatory role. J Biol Chem 1994;269:8334–8340.
21. Tang W, Li X, Zhu Z, et al. Expression, purification, crystallization and preliminary X-ray diffraction analysis of human phosphoribosyl pyrophosphate synthetase 1 (PRS1). Acta Crystallogr Sect F Struct Biol Cryst Commun 2006;62:432–434.
22. Pui CH, Relling MV, Downing JR. Acute lymphoblastic leukemia. N Engl J Med 2004;350:1535–1548.
23. Karran P, Attard N. Thiopurines in current medical practice: molecular mechanisms and contributions to therapy-related cancer. Nat Rev Cancer 2008;8:24–36.
24. Zoref E, De Vries A, Sperling O. Mutant feedback-resistant phosphoribosylpyrophosphate synthetase associated with purine overproduction and gout. Phosphoribosylpyrophosphate and purine metabolism in cultured fibroblasts. J Clin Invest 1975;56:1093–1099.
25. Christopherson RI, Lyons SD, Wilson PK. Inhibitors of de novo nucleotide biosynthesis as drugs. Acc Chem Res 2002;35:961–971.
26. Katashima R, Iwahana H, Fujimura M, et al. Molecular cloning of a human cDNA for the 41-kDa phosphoribosylpyrophosphate synthetase-associated protein. Biochim Biophys Acta 1998;1396:245–250.
27. An S, Kumar R, Sheets ED, Benkovic SJ. Reversible compartmentalization of de novo purine biosynthetic complexes in living cells. Science 2008;320:103–106.
28. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144:646–674.
29. Becker MA, Heidler SA, Bell GI, et al. Cloning of cDNAs for human phosphoribosylpyrophosphate synthetases 1 and 2 and X chromosome localization of PRPS1 and PRPS2 genes. Genomics 1990;8:555–561.
30. de Brouwer AP, van Bokhoven H, Nabuurs SB, Arts WF, Christodoulou J, Duley J. PRPS1 mutations: four distinct syndromes and potential treatment. Am J Hum Genet 2010;86:506–518.
31. Eriksen TA, Kadziola A, Bentsen AK, Harlow KW, Larsen S. Structural basis for the function of Bacillus subtilis phosphoribosyl-pyrophosphate synthetase. Nat Struct Biol 2000;7:303–308.
32. Shalem O, Sanjana NE, Hartenian E, et al. Genome-scale CRISPR-Cas9 knockout screening in human cells. Science 2014;343:84–87.
33. Li Y, Feng H, Gu H, et al. The p53-PUMA axis suppresses iPSC generation. Nat Commun 2013;4:2174.
34. Lv D, Li Y, Zhang W, et al. TRIM24 is an oncogenic transcriptional co-activator of STAT3 in glioblastoma. Nat Commun 2017;8:1454.
35. Gao Y, Yang P, Shen H, et al. Small-molecule inhibitors targeting INK4 protein p18(INK4C) enhance ex vivo expansion of haematopoietic stem cells. Nat Commun 2015;6:6328.