Transketolase regulates sensitivity to APR-246 in p53-null cells independently of oxidative stress modulation

The prevalence and dire implications of mutations in the tumour suppressor, p53, highlight its appeal as a chemotherapeutic target. We recently showed that impairing cellular antioxidant systems via inhibition of SLC7A11, a component of the system xc− cystine-glutamate antiporter, enhances sensitivity to mutant-p53 targeted therapy, APR-246. We investigated whether this synergy extends to other genes, such as those encoding enzymes of the pentose phosphate pathway (PPP). TKT, one of the major enzymes of the PPP, is allegedly regulated by NRF2, which is in turn impaired by accumulated mutant-p53 protein. Therefore, we investigated the relationship between mutant-p53, TKT and sensitivity to APR-246. We found that mutant-p53 does not alter expression of TKT, nor is TKT modulated directly by NRF2, suggesting a more complex mechanism at play. Furthermore, we found that in p53null cells, knockdown of TKT increased sensitivity to APR-246, whilst TKT overexpression conferred resistance to the drug. However, neither permutation elicited any effect on cells overexpressing mutant-p53 protein, despite mediating oxidative stress levels in a similar fashion to that in p53-null cells. In sum, this study has unveiled TKT expression as a determinant for sensitivity to APR-246 in p53-null cells.


Results
TKT is not regulated by mutant-p53 or NRF2 activation. To investigate the effect of mut-p53 on transketolase (TKT) expression, two isogenic cell systems were used. H1299 cells, a non-small cell lung cancer line, harbour a mutation that prevents the formation of full-length active p53 protein 28 . These cells were previously transduced to ectopically overexpress common missense mutant p53 proteins, p53 R273H or p53 R175H16 . JH-EsoAd1 cells are an oesophageal cancer cell line that endogenously express missense mutant p53 G266E29 , and two p53 −/− clones were previously generated using CRISPR-Cas9 16 . RT-qPCR and western blotting showed no consistent relationship between TKT mRNA or protein expression, respectively, and the presence or absence of mut-p53 protein (Fig. 1A,B). To test the effect of NRF2 activation on TKT expression, H1299 cells were treated with hydrogen peroxide (H 2 O 2 ) and mRNA expression analysed by RT-qPCR. Upon H 2 O 2 treatment and subsequent NRF2 activation ( Supplementary Fig. 1A), no change was observed in TKT expression in either H1299 p53 null or p53 R273H cells (Fig. 1C). As a control, expression of SLC7A11, a known NRF2 target gene, was also ana- www.nature.com/scientificreports/ lysed. Consistent with previous findings 16 , SLC7A11 was strongly upregulated upon H 2 O 2 treatment in H1299 p53 null cells (p < 0.001) but not in p53 R273H cells (Fig. 1C). In a complementary approach, the effect of depleting NRF2 on TKT expression was assessed using siRNA to genetically knock down NRF2. TKT mRNA expression was found to be unchanged in response to NRF2 knockdown (Fig. 1D, Supplementary Fig. 1B). In contrast, SLC7A11 expression was downregulated upon NRF2 knockdown in p53 null cells but not p53 R273H H1299 cells, consistent with previous findings 16 . Taken together, these results suggest that TKT expression is not affected by mut-p53 regulation of NRF2 transcriptional activity in H1299 cells.

Transient knockdown of TKT with siRNA slows cell growth and induces cell cycle arrest in H1299 cells independent of mutant-p53 expression. Having established that p53 status and NRF2
activation had no apparent bearing on TKT expression, we sought to uncover whether cell dependence on TKT varied with respect to p53 status. To examine this, TKT was transiently knocked down in H1299 and JHEso-Ad1 cells with the use of siRNA. Despite a highly efficient and rapid knockdown of TKT at the mRNA level across all cell lines ( Fig. 2A), TKT protein appeared remarkably stable and did not noticeably decline until 96 h after transfection with siRNA (Fig. 2B). This rapid mRNA knockdown with lagging decrease in protein levels was common to all tested cell lines. Depletion of TKT reduced cell proliferation in H1299 cells but not JHEso-Ad1, irrespective of p53 status ( Fig. 2C; Supplementary Fig. 2A,B). This diminished cell growth in H1299 cells could be completely rescued by the addition of N-acetyl-cysteine (NAC), a potent antioxidant (Fig. 2D), suggesting that the impact of TKT on cellular redox balance may be tantamount to its effect on cell growth. Because maintaining an optimal redox state is necessary for proper cell replication 30 , the impact of TKT on cell cycle progression with respect to p53 status was also investigated. Knockdown of TKT induced arrest in the G0/G1 phase in H1299 cells ( Fig. 2E; Supplementary Fig. 2C-F). Moreover, this effect was seen in both H1299 p53 null and p53 R273H cell lines. Altogether, these data support the notion that the role of TKT in moderating oxidative stress may be required for cell proliferation, p53 status notwithstanding, but is cell context dependent.  www.nature.com/scientificreports/ Transient knockdown of TKT with siRNA sensitises p53 null cells to APR-246. We next investigated the relationship between TKT and the antioxidant-depleting drug APR-246. Analysis of the NCI-60 human tumour cell line screen found that TKT expression correlated with sensitivity to PRIMA-1, an unmethylated analogue of APR-246 ( Fig. 3A) 31,32 . Increased expression of TKT mRNA was positively correlated with an increased growth inhibitory concentration (GI50) of APR-246. This suggests a potential protective effect of TKT against APR-246. Transient knockdown of TKT did not alter the response of cells expressing mutant p53 protein, nor when the endogenous mutant TP53 was knocked out in JH-EsoAd1 ( Supplementary Fig. 3A-E). Interestingly, we noticed that knockdown of TKT increased sensitivity of H1299 p53 null cells to low doses of APR-246, with a coefficient of interaction (CI) of 0.66 ± 0.06 indicative of synergistic activity ( Fig. 3B-D, Supplementary Fig. 3F). Synergistic interactions were also observed in Cas9-expressing H1299 p53 null cells transduced with two independent synthetic guide RNAs (sgRNAs) targeting TKT (Fig. 3E, Supplementary Fig. 3F). To investigate this further, we utilised another cell line with an endogenous TP53 mutation that results in a p53 null status (OACP4C, Fig. 3F) and HCT116 cells (Fig. 3G), which express wild-type p53 protein. Similar to H1299 p53 null cells, OACP4C p53 null cells showed synergistic activity between TKT knockdown and APR-246 ( Supplementary Fig. 3A), as well as a significant decrease in the GI50 of APR-246 in TKT depleted cells. In contrast, the maximal CI in HCT116 cells was 0.78 ± 0.12 ( Supplementary Fig. 3A) indicative of an additive/borderline-synergistic effect. Thus, knockdown of TKT is synergistic with APR-246 in cancer cell lines with an endogenous p53 null phenotype.
Transient knockdown of TKT augments oxidative stress levels, resulting in increased cell death. We next investigated whether the sensitisation of H1299 p53 null cells to APR-246 through TKT knockdown was a product of the combined effect of these conditions on oxidative stress levels. Treatment with APR-246 (25 µM) for a relatively short period (18 h) did not significantly alter the levels of NADPH (Fig. 4A) nor mitochondrial superoxides in H1299 cells ( Fig. 4B; Supplementary Fig. 3G). In contrast, TKT knockdown significantly decreased NADPH levels ( Fig. 4A) and increased mitochondrial ROS in H1299 p53 null cells, which was further augmented by APR-246 (p < 0.05) ( Fig. 4B; Supplementary Fig. 3G). Treatment with NAC entirely rescued the combination effect of TKT knockdown and APR-246 on cell viability (p < 0.001) (Fig. 4C). Altogether, these results indicate that the combination of TKT knockdown and APR-246 treatment induces a greater oxidative stress environment than either condition alone. The fact that the sensitisation of H1299 p53 null cells to APR-246 by TKT knockdown was rescued by antioxidant treatment suggests that greater intracellular ROS levels are the cause of the increased cell death. However, transient knockdown of TKT did not sensitise H1299 cells to Erastin, a known inducer of oxidative stress (Fig. 4D, Supplementary Fig. 3I), suggesting a more complex mechanism at play in the context of APR-246.
Overexpression of TKT reduces sensitivity to APR-246 in H1299 p53 null cells. We next deter-  Supplementary Fig. 4D). However, the protective effect of TKT overexpression in these cells did not extend to long-term survival as assessed in clonogenic assays ( Supplementary Fig. 4E). Combined, these data support the notion that overexpression of TKT delays the onset of APR-246-induced cell death in H1299 p53 null cells.
Inhibition of NADPH production slows cell growth in H1299 p53 null cells but does not affect sensitivity to APR-246 or induce cell cycle arrest. A major role of TKT in the PPP is the shuttling of intermediates back into glycolysis, from where they return to the oxidative arm of the PPP and contribute to the production of NADPH 33 . Given this, we next investigated whether treatment with 6-aminonicotinamide (6AN), an inhibitor of NADPH production ( Supplementary Fig. 5A), would elicit the same effects as TKT knockdown. 6AN treatment elicited a significantly greater reduction in growth rate than that caused by TKT knockdown, but this effect was not enhanced when TKT knockdown and 6AN treatment were administered concomitantly (Fig. 6A). Further, 6AN treatment did not sensitise H1299 p53 null cells to APR-246 (Fig. 6B), and no significant shift in the GI50 of APR-246 was observed in these cells (p = 0.14), despite significantly increased mitochondrial superoxides following treatment with 6AN, indicating drug activity ( Fig. 6C; Supplementary Fig. 5B). The impact of 6AN on cell cycle progression was also assessed. After 24 h of treatment with 6AN, neither H1299 p53 null nor p53 R273H cells showed any sign of cell cycle arrest ( Fig. 6D; Supplementary Fig. 5C-F). This is in stark contrast to the clear interference with cell cycling caused by TKT knockdown. Taken together, these results suggest some similarities between the consequences of TKT knockdown and inhibition of NADPH production by 6AN, but 6AN treatment does not entirely phenocopy the effects of TKT gene knockdown.

Discussion
Despite the clear parallels between these factors relating to antioxidant response mechanisms, the relationship between mut-p53 and the role of the PPP and TKT in the antioxidant response remains poorly understood. This work aimed to delineate the relationship between TKT and mut-p53, as well as TKT and the drug APR-246, which has both mutant-p53 reactivating and oxidative stress-inducing mechanisms. Using isogenic models of p53 null or mut-p53 cells and additional cell lines with varying p53 status, we examined the effects of perturbations to TKT on oxidative stress and sensitivity to APR-246, relative to p53 status. Our findings indicate a potential role for TKT in modulating sensitivity to APR-246, but suggest that this is likely not via altering redox balance. www.nature.com/scientificreports/ Previous studies dictate that accumulated mut-p53 inhibits the transcriptional ability of NRF2 16,19,20 whilst NRF2 has been found to transcriptionally activate TKT 25 . From this, it may be inferred that through inhibition of NRF2, accumulated mut-p53 may impact transcription of TKT. In contrast, our work determined that neither manipulation of mut-p53 nor NRF2 had any bearing on expression of TKT in H1299 or JH-EsoAd1 cells (Fig. 1A-D). This suggests that mechanisms of TKT regulation may differ among cancer cell types or those with altered dependency on NRF2. For example, TKT was found to be a transcriptional target of NRF2 in A549 cells which harbour endogenous KEAP1 mutation, leading to constitutively active NRF2 in these cells 26 . Given that neither cell line used in this study contain KEAP1 mutations, is it possible that this difference may account for the discrepancy between our work and previous findings.
Consistent with previous findings 25 , we found that TKT depletion hinders cell growth by arresting the cell cycle, and that this occurred irrespective of p53 status (Fig. 2C,E). That this could be rescued with antioxidant treatment (i. e. NAC; Fig. 2D) is in congruence with previous studies involving TKT and other PPP enzymes, such as ribose-5-phosphate isomerase (RPIA) 34 . RPIA converts ribulose-5-phosphate (Ru5P) into ribose-5-phosphate (R5P) in the PPP and thus plays a major role in ribonucleotide biosynthesis 35 . This similarity strongly suggests an ability of NAC to counteract the decline in ribonucleotide synthesis that is likely caused by depletion of these PPP enzymes.
Although these early data did not identify any relationship between TKT and mut-p53, it remains that, according to NCI-60 data, a correlation exists between TKT expression and sensitivity to PRIMA-1 (Fig. 3A). Critically, we found that transient knockdown or knockout of TKT sensitised only p53 null cells, and not cells expressing mut-p53, to low doses of APR-246 ( Fig. 3B-F, Supplementary Fig. 3A-F), whilst overexpression of TKT conferred resistance to the drug, again only in p53 null cells (Fig. 5C). Notably, in contrast to cells with an endogenous p53 null mutation, TKT knockdown did not increase sensitivity in JH-EsoAd1 cells in which the www.nature.com/scientificreports/ endogenous missense mut-p53 had been knocked out using CRISPR/Cas9. While the reason for this remains unclear, we speculate that this may reflect differences between tumour cells that have evolved without the ability to express any p53 protein (H1299 and OACP4C cells) compared to cells that evolved expressing a missense mut-p53 that was then knocked out.
Our results in mut-p53 expressing cells were surprising, given previous findings that mut-p53 expressing cells are highly sensitive to further assaults on cellular redox balance (particularly the antioxidant-depleting effects of APR-246) due to inherent impaired antioxidant function 11,16 . However, mut-p53 has been shown to transactivate a range of anti-apoptotic and growth enhancing factors 8,9,36,37 as well as alter tumour cell metabolism to promote survival 7,38,39 . In fact, it has been found that the antioxidant thioredoxin is upregulated in mut-p53 expressing cells 11 , which alludes to a thioredoxin-dependent mechanism that allows mut-53 cells to compensate for APR-246-induced glutathione depletion. As such, it is possible that the mut-p53 expressing cells were unaffected by the consequences of TKT knockdown due to upregulation of pro-survival measures or prioritisation of alternative metabolic pathways.
Upon further investigation of oxidative balance, and consistent with the findings of Xu et al. 25 , we found that in all cell types, knockdown (Fig. 4A,B) and overexpression (Fig. 5E) of TKT increased and decreased intracellular ROS burden, respectively. These data suggest that modulations to TKT influence flux through the oxidative arm of the PPP and subsequent production of NADPH. Intriguingly, this occurred in both p53 null and mut-p53 expressing cells, leading to the conundrum of why TKT depletion only induced sensitivity to APR-246 in p53 null cells, whilst the effect on ROS levels was also seen in mut-p53 expressing cells. Cancer cells natively exhibit higher levels of oxidative stress than untransformed cells 30,40,41 . Basal ROS levels in mut-p53 expressing cells were found to be higher still, in comparison to p53 null cells ( Supplementary Fig. 3H; compare with Fig. 4B), congruent with previous findings 16,21 , likely due to the suppressing effect of mut-p53 on NRF2-dependent antioxidant pathways. Despite higher levels of oxidative stress, mut-p53 cancers have been found to show improved survival and augmented proliferation in response to oxidative stress 21 . Mut-p53 is also thought to be able to commandeer additional pro-survival pathways 8,9,36,37 . Thus, it is possible that due to these gains-of-function, mut-p53 expressing cells were able to escape the effects of TKT knockdown that increased APR-246-induced cell death in p53 null cells. www.nature.com/scientificreports/ TKT has a dualistic role in the PPP and its activity modulates depending on the metabolic needs of the cell 33 . It accomplishes this by either shuttling intermediates back towards the oxidative arm of the PPP to produce NADPH in response to oxidative demand, or producing R5P to promote production of ribonucleotides 33 . In order to separate these effects and their potential role in altering sensitivity to APR-246, 6-aminonicotinamide (6AN), a compound that impedes NADPH production but is not known to effect production of ribonucleotides, was utilised. 6AN inhibits the two major NADPH-producers in the PPP, 6-phosphogluconoate (6PGD) and glucose-6-phosphate dehydrogenase (G6PD) 42,43 . In this way, the isolated effect of NADPH depletion on APR-246 sensitivity was able to be investigated, without the alternative function of TKT (i. e. ribonucleotide production) confounding the results. Intriguingly, 6AN slowed cell growth (Fig. 6A) despite eliciting no effect on cell cycle progression (Fig. 6D), irrespective of p53 status. In contrast to transient TKT knockdown, 6AN treatment did not significantly alter sensitivity to APR-246 in H1299 p53 null cells (Fig. 6B), despite eliciting a similar effect on intracellular ROS levels (Fig. 6C). This suggests that impaired R5P production may be the driving force behind TKT knockdown-induced APR-246 sensitivity in p53 null cells, rather than increased levels of ROS due to inadequate antioxidant mechanisms. Additionally, TKT knockdown did not increase sensitivity to erastin, a known oxidative stress inducer, indicating that loss of TKT expression is not broadly sensitising cells to oxidative stressmediated cell death. Imperatively, this suggests that the mechanism of action of APR-246 is more complex than previously described and may incorporate aspects of macromolecule biosynthesis that have not been explored previously in this context. Future work is therefore required to examine the effect of TKT knockdown on R5P production and the impact of this on APR-246 sensitivity and how this is related to p53 status.
Despite similar TKT expression levels across cells in this model with varying p53 status, a distinct change in sensitivity to APR-246 was observed when TKT was ectopically overexpressed or transiently depleted in p53 null cells. This, combined with correlative data from the NCI-60 human tumour cell line screen (Fig. 3A), suggests that TKT expression regulates sensitivity to APR-246, at least in p53 null tumours. Of particular interest is the evidence that sensitivity of mut-p53 cells to APR-246 is not regulated in this same way in this model. We propose a model to explain this observation based on the induction of wild-type-p53-like function in mut-p53 cells when treated with APR-246 ( Supplementary Fig. 6). Wild-type-p53 has been found to inhibit activity of G6PD, the www.nature.com/scientificreports/ rate-limiting enzyme of the PPP, upstream of TKT, by preventing dimerization of the enzyme 44 . Because treatment with APR-246 induces wild-type-p53-like conformation and activity in mut-p53 cells, mut-p53 cells treated with APR-246 sustain impeded function of the PPP via G6PD inhibition. Therefore, due to upstream interference and consequential limited flux through the PPP, modulations of downstream TKT may have no effect on APR-246 activity. Treatment of p53 null cells with APR-246 would not confer this same wild-type-p53-like effect. Thus, modulations of TKT in p53 null cells would strongly interfere with the previously intact PPP, rendering the cells more susceptible to APR-246 treatment. This is consistent with our finding that upstream interference with the PPP via G6PD inhibition by 6AN had no synergistic or additive effect in combination with TKT modification in any cell lines, irrespective of p53 status. In summary, this study bridges the gaps in current knowledge surrounding the relationship between the PPP, mut-p53 cancers and therapies. The findings indicate that whilst mut-p53 is known to interfere with NRF2, it did not affect TKT in the investigated models. Furthermore, it is proposed that regulation of TKT is not as simple as per the previously suggested mechanism of NRF2 activation 25 , and that other factors must certainly be at play. This study also addressed the growing complexity of mut-p53 gain-offunction capabilities. Given the failure of NRF2 modulation to affect TKT expression in this study, future research is needed to explore the role of mut-p53 in regulating the antioxidant response and its influence on the PPP. The present findings also mar the current predilection in research for targeting cellular redox balance as an avenue for chemotherapeutics, particularly for mut-p53 cancers. This is due, in part, to growing evidence of the complex nature of mut-p53 in the antioxidant response, as well as the uncovering of the potential influence of external factors. Finally, this study has identified TKT as a determinant for APR-246 sensitivity in p53 null cells, which may open a new avenue for exploration and potentially allow for personalised, effective treatment for p53 null cancers. Gene expression using RT-qPCR. Cells were seeded in 6 cm dishes and allowed to adhere overnight.

Materials and methods
Following the relevant experiments, extraction and purification of RNA was conducted using Nucleospin RNA kit (Macherey-Nagel) before reverse transcription to cDNA using Transcriptor First Strand cDNA Synthesis Kit (Roche) as per manufacturer's instructions, including the optional denaturing step. Quantitative PCR (qPCR) was conducted using LightCycler 480 SYBR-Green qPCR (Roche) as per manufacturer's protocol, with gene expression normalised within each sample to GAPDH and analysed using the ΔΔCt method. Primer sequences were obtained using the NCBI Primer Blast application and are detailed in Supplementary www.nature.com/scientificreports/ phosphatase (PhosphoSTOP, Roche) and protease (Complete ULTRA, Roche) inhibitors. Protein concentrations were quantified using the DC Protein assay (Bio-Rad). Equivalent amounts of protein lysates were boiled, resolved by SDS-PAGE using 10% w/v acrylamide gels, and transferred to polyvinylidene difluoride membranes. Membranes were incubated for 1 h in blocking buffer (5% w/v skim milk, 0.05% v/v Tween-20 in Tris-buffered saline; TBS-T) and probed overnight at 4 °C with the primary antibody. Blots were washed thrice in wash buffer (0.05% v/v TBS-T) for 10 min, followed by incubation with peroxidase-conjugated secondary antibody (Dako) for 1 h at room temperature. Proteins were visualised by Amersham ECL western Blotting Detection Reagents (GE Life Sciences) or ECL Plus western blotting substrate kit (Thermo Scientific). Anti-β-actin or anti-GAPDH antibodies were used to assess protein loading. Antibodies are detailed in Supplementary Table 3.
Cell cycle analysis. Cells were seeded in 24-well plates and exposed to siRNA TKT knockdown for 96 h,

ROS detection.
MitoSOX Red reagent (ThermoFisher Scientific) was utilised to detect mitochondrialspecific ROS accumulation, using DAPI as a viability marker counter-stain. H1299 cells were seeded in 24-well plates and exposed to siRNA TKT knockdown for 78 h before treatment with APR-246 (25 µM) 18 h prior to ROS detection. Complete media containing 5 µM MitoSOX Red reagent was applied to cells and incubated at 37 °C for 30 min prior to analysis. Media was removed and replaced with media containing 1 µg/mL DAPI and incubated in the dark for 15 min at room temperature. Cells were then washed three times, trypsinised and resuspended in PBS supplemented with 1% v/v FBS and 25 mM EDTA before ROS-induced fluorescence was quantified using the FACSymphony (BD Biosciences).
Cellular proliferation and viability assays. Cellular proliferation was assessed using a microscopybased live cell imaging system (Incucyte FLR, Essen BioScience). Cells were seeded in 96-well plates and treated with indicated doses of APR-246 or vehicle for 72 h. Cells were imaged every 24 h. The confluency of APR-246 treated wells was normalised to pre-treatment reading and compared with vehicle treatment. To assess cell viability, AlamarBlue (Life Technologies) fluorometric assays were conducted as per established protocols 45 . Following treatments, 20 µL of 2% v/v AlamarBlue was added to each well and incubated for 2 h at 37 °C. Fluorescence was measured using a FLUOstar OPTIMA microplate reader (BMG Labtech) at an excitation of 540 nm and an emission of 590 nm. Results were normalised to vehicle-treated wells.
APR-246 dose responses, antioxidant rescue and NAPDH inhibition. Cells were seeded at 2 × 10 3 cells/well in 96-well plates, exposed to siRNA TKT knockdown or non-targeting control and allowed to adhere overnight. The following day, media was replaced with fresh media supplemented with APR-246 or vehicle control. To determine GI50 values, cells were exposed to a range of concentrations of APR-246 (1 to 100 µM) for 72 h and assayed using AlamarBlue. For antioxidant rescue, after 24 h of siRNA TKT knockdown, cells were treated with 5 mM N-acetylcysteine (NAC) or 15 μM APR-246, individually or in combination, for 72 h before assaying with AlamarBlue. For NADPH inhibition, after 24 h of siRNA TKT knockdown, cells were treated with 10 µM 6-aminonicotinamide (6-AN) or vehicle control in combination with a range of APR-246 concentrations for 72 h before assaying with AlamarBlue. APR-246 dose responses both with and without NADPH inhibition were repeated in cells overexpressing TKT or RFP control, in the aforementioned manner, without exposure to siRNA.
Colony formation assay. Cells in six-well plates were treated with 50 µM APR-246 for 5, 10, 15, 20 or 24 h, then typsinised and reseeded in six-well plates, 2 × 10 3 cells per well and grown for 7 days. Cell colonies were stained with crystal violet (0.5% w/v) containing methanol (11% v/v) for fixing, rinsed in water, air-dried and digitally scanned. Discrete colonies of > 50 cells were counted using MetaMorph software (Molecular Devices) and expressed as a percentage of vehicle treatment.
Data analysis and statistics. Data were analysed using ANOVA with Dunnet's multiple comparison post-test or Student's t-test as indicated. Correlation between two groups was evaluated by the Pearson's test as indicated. GI50 concentration of APR-246 was determined using the Hill equation. Coefficient of interaction (CI) between drugs (APR-246 or erastin) and TKT knockdown (TKT siRNA) or knockout (TKT sgRNA) was