Metabolic flexibilities and vulnerabilities in the pentose phosphate pathway of the zoonotic pathogen Toxoplasma gondii

Metabolic pathways underpin the growth and virulence of intracellular parasites and are therefore promising antiparasitic targets. The pentose phosphate pathway (PPP) is vital in most organisms, providing a reduced form of nicotinamide adenine dinucleotide phosphate (NADPH) and ribose sugar for nucleotide synthesis; however, it has not yet been studied in Toxoplasma gondii, a widespread intracellular pathogen and a model protozoan organism. Herein, we show that T. gondii has a functional PPP distributed in the cytoplasm and nucleus of its acutely-infectious tachyzoite stage. We produced eight parasite mutants disrupting seven enzymes of the PPP in T. gondii. Our data show that of the seven PPP proteins, the two glucose-6-phosphate dehydrogenases (TgG6PDH1, TgG6PDH2), one of the two 6-phosphogluconate dehydrogenases (Tg6PGDH1), ribulose-5-phosphate epimerase (TgRuPE) and transaldolase (TgTAL) are dispensable in vitro as well as in vivo, disclosing substantial metabolic plasticity in T. gondii. Among these, TgG6PDH2 plays a vital role in defense against oxidative stress by the pathogen. Further, we show that Tg6PGDH2 and ribulose-5-phosphate isomerase (TgRPI) are critical for tachyzoite growth. The depletion of TgRPI impairs the flux of glucose in central carbon pathways, and causes decreased expression of ribosomal, microneme and rhoptry proteins. In summary, our results demonstrate the physiological need of the PPP in T. gondii while unraveling metabolic flexibility and antiparasitic targets.

Introduction Toxoplasma gondii, an obligate intracellular parasite capable of infecting virtually all nucleated cells in diverse organisms, is the causative agent of zoonotic toxoplasmosis [1]. It has a complex life cycle, including both asexual and sexual stages [2]. The asexual phase of the parasite can occur in several intermediate hosts, such as humans, pigs, and sheep, whereas sexual development is restricted to the feline intestine [1][2][3]. The parasite is known to flexibly reprogram its metabolism for surviving and proliferating in diverse host cell types and nutritional milieus [4][5][6][7][8][9][10][11][12][13].
Toxoplasma gondii utilizes glucose as a primary carbon source to support the metabolic demands during its asexual reproduction in mammalian cells [5,[14][15][16]. Glucose is imported by facilitative glucose transporter (TgGT1) into the parasite cytoplasm where it is metabolized by glycolysis and/or pentose phosphate pathway (PPP), branching at glucose 6-phosphate (G6P) [5,17,18]. Several studies have suggested that the central carbon metabolism is critical for optimal asexual growth and metabolic flexibility in Toxoplasma [6,16,18,19]. Although PPP is also a key route of glucose catabolism, its contribution to parasite metabolism and pathogenesis remains largely unexplored.
Herein, we performed a functional analysis of the PPP metabolism in T. gondii. We generated eight mutant strains, disrupting seven enzymes, and analyzed their phenotypic and metabolic features. This study suggests that the pentose phosphate pathway supports the parasite metabolism, growth and virulence. Our work also emphasizes the metabolic plasticity of carbon metabolism during the lytic cycle of T. gondii.

The pentose phosphate pathway localizes in the cytoplasm and nucleus of Toxoplasma gondii
The parasite genome encodes eight genes predicted to be the enzymes of PPP (Fig 1A), namely TgG6PDH1, TgG6PDH2, Tg6PGDH1, Tg6PGDH2, TgRPI, TgRuPE, TgTKT, and TgTAL ( Fig  1B). To examine whether all proteins of the pentose phosphate pathway are expressed in the acute (tachyzoite) stage of T. gondii, we performed their 3'-genomic tagging with an HA epitope using CRISPR/Cas9 mediated site-specific integration in the RHΔku80 strain. Immunofluorescent staining of the HA-tagged proteins revealed that TgG6PDH1, TgG6PDH2, Tg6PGDH2, TgRuPE and TgRPI were expressed in the tachyzoite cytoplasm, as shown by colocalization with a cytoplasmic marker, TgALD (Fig 1C) [23]. Especially, to further confirm the cytoplasm localization of the Tg6PGDH2 and TgRPI on the level of the native protein, we produced polyclonal antisera against Tg6PGDH2 and TgRPI which had been recombinantly expressed in E. coli BL21 (S1A, S1B and S2A and S2B Figs). Immunofluorescent staining using these antisera showed that the native Tg6PGDH2 and TgRPI were expressed in the cytoplasm of T. gondii (S1C and S2C Figs). Unexpectedly, TgTKT and TgTAL co-localized with the nucleus marker Hoechst (Fig 1C), and Tg6PGDH1 is not expressed in Toxoplasma tachyzoites, as confirmed by immunofluorescent staining (Fig 1C).

TgG6PDH1/2, Tg6PGDH1, TgRuPE and TgTAL are dispensable for the lytic cycle
To evaluate the physiological significance of each PPP gene to parasite fitness, we attempted to delete the TgG6PDH1, TgG6PDH2, Tg6PGDH1, Tg6PGDH2, TgRPI, TgRuPE, TgTKT and TgTAL genes through CRISPR/Cas9-mediated homologous gene replacement in the RHΔku80 strain (S3A Fig). The direct knockout strains of Δg6pdh1, Δg6pdh2, Δ6pgdh1, Δrupe and Δtal were obtained after pyrimethamine selection, as confirmed by PCR and semi-quantitative RT-PCR screening of the clonal mutants (S3B- S3J Fig). However, no viable Δ6pgdh2, Δrpi and Δtkt mutants could be generated, suggesting a critical function of the corresponding proteins in tachyzoites. The number and size of plaques formed by the Δg6pdh1, Δg6pdh2, Δ6pgdh1, Δrupe and Δtal mutants were similar to that of the parental RHΔku80 strain (Fig 2A  and 2B), indicating that TgG6PDH1, TgG6PDH2, Tg6PGDH1, TgRuPE and TgTAL enzymes are not required during the lytic cycle.
Toxoplasma harbors two TgG6PDH enzymes. We tested their functional redundancy by generating a Δg6pdh1Δg6pdh2 double mutant, in which the Remarkably, the double-deletion strain showed no apparent phenotypic defects in vitro as examined by routine culture, plaque and replication assays (Fig 2C, 2D and 2F), which suggested that tachyzoites can indeed survive with the loss of the two TgG6PDH isoforms.
To test the consequences of gene deletion on the parasite virulence, ICR mice were intraperitoneally injected with the Δg6pdh1, Δg6pdh2, Δg6pdh1Δg6pdh2, Δ6pgdh1, Δrupe and Δtal strains, and their survival was monitored for 30 days. Interestingly, none of them showed an attenuated virulence (Fig 2E), suggesting that TgG6PDH1/2, Tg6PGDH1, TgRuPE and TgTAL , Representative images of immunofluorescence assay. Different enzymes of PPP tagged at the C terminus with smHA epitope by CRISPR-Cas9-mediated homologous recombination in the RHΔku80 strain. Subsequently, the parasites of testing strains were fixed, permeabilized and stained with mouse anti-HA and rabbit anti-TgALD, which were detected by Alexa 488-and Alexa 594-conjugated secondary antibodies, respectively. TgALD and Hoechst were used as the cytoplasm and cell nucleus markers, respectively. Bar = 5 μm. https://doi.org/10.1371/journal.ppat.1010864.g001

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma

TgG6PDH1/2 are needed to defend oxidative stress
TgG6PDH enzymes are predicted to catalyze the first step in the oxidative pentose phosphate pathway and provide pentose sugars for the nucleotide synthesis [17]. To assess the functional contribution of TgG6PDH, we performed [ 13 C 6 ]-glucose labeling of extracellular tachyzoites and estimated carbon flux into PPP, glycolysis and TCA cycle intermediates. We found that the Δg6pdh1Δg6pdh2 mutant showed increased 13 C-labeling of glycolytic and TCA cycle intermediates, such as fructose-1,6-bisphosphate, 3-phosphoglyceric acid, 2-phosphoglyceric acid, lactate, fumarate and malate (S5 Fig). Generally, no significant difference was observed in the PPP metabolites. Because NADPH produced by the G6PDH enzyme is thought to be a crucial reductive power to defend against oxidative stress [24][25][26], we also evaluated the growth of RHΔku80, Δg6pdh1, Δg6pdh2 and Δg6pdh1Δg6pdh2 mutants after pre-treatment with 500 μM H 2 O 2 for 3 h. The RHΔku80 and Δg6pdh1 strains exhibited similar replication rates, as determined by the numeration of tachyzoites in the parasitophorous vacuoles. Notably, although the H 2 O 2 also affected the proliferation of wild-type parasites, the pre-treated Δg6pdh2 and pre-treated Δg6pdh1Δg6pdh2 parasites showed a notable defect compared to the parental strain ( Fig 2G), indicating that TgG6PDH2 may be involved in oxidative stress response via NADPH synthesis. Therefore, our further work analyzed the expression differences of TgG6PDH1 and TgG6PDH2, and found that the expression of TgG6PDH2 was significantly higher than that of TgG6PDH1 in T. gondii tachyzoites ( Fig 2H). Subsequently, we measured the abundance of NADPH in the mutants and found that the deletion of TgG6PDH2 significantly affected NADPH production ( Fig 2I). Together, these results suggested that TgG6PDH2 plays a function in maintaining the cytosolic NADP + /NADPH balance and thus plays a vital physiological role in the anti-oxidant response of tachyzoites.

Tg6PGDH2 is critical for the asexual reproduction of tachyzoites
Our further work focused on Tg6PGDH2 that enables decarboxylation of 6PG into Ru5P with concomitant reduction of NADP in tachyzoite cytoplasm. We constructed a rapamycin-inducible Tg6PGDH2 mutant using the DiCre system because a direct knockout by CRISPR/Cas9mediated homologous gene replacement could not be generated. The conditional knockdown mutant was generated via pyrimethamine selection, as shown in the scheme ( Fig 3A) and confirmed by diagnostic PCR for the intended replacement of the Tg6PGDH2 gene by pTUBloxp-Tg6PGDH2-loxp-YFP-DHFR � (Tg6PGDH2-cKD mutant in Fig 3B). Rapamycin-induced knockout of Tg6PGDH2 was confirmed by the appearance of YFP signal in indirect immunofluorescence assays (IFA) (Fig 3C) and further identified by Western blotting (Fig 3D). We cultured the Tg6PGDH2-cKD strain with or without rapamycin for 1 day or 4 days, then tachyzoites of RHΔku80, Δg6pdh1, Δg6pdh2 and Δg6pdh1Δg6pdh2 parasites were allowed to infect HFF monolayers for 1 h, and invaded parasites were cultured at 37˚C with 5% CO 2 for another 24 h. Percentiles of the parasitophorous vacuole (PV) containing 1, 2, 4, 8, 16, or more parasites were determined and plotted. Means ± SEM from three independent experiments (n = 3). (G), Freshly egressed tachyzoites of indicated strains were pretreated with 500 μM H 2 O 2 medium for 3 h, and then subjected to intracellular replication assay (without H 2 O 2 ) for 24 h. Means ± SEM from three independent experiments was graphed. Two-way ANOVA, � , p <0 .05, ��� , P < 0.001. (H), RHΔku80 tachyzoites were purified, total RNA was extracted and reversed into cDNA. Transcript levels for TgG6PDH1 and TgG6PDH2 in each sample were analyzed by quantitative real-time PCR, using β-tubulin as an internal reference. Means ± SEM of four independent experiments (n = 4). Student's t-test, ��� , P<0.001. (I), Parasites (approximately 3×10 7 ) were purified by 3 μm membrane filtration, washed with cold PBS and extracted with extraction buffer. Relative NADPH levels in RHΔku80 (WT), Δg6pdh1 and Δg6pdh2 were determined by the NADPH Assay kit. Means ± SEM from three independent experiments (n = 3). �� , P<0.01; oneway ANOVA. https://doi.org/10.1371/journal.ppat.1010864.g002

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma performed competition, plaque and replication assays under standard culture condition. A knockout of Tg6PGDH2 after rapamycin treatment led to significant phenotypic defects in vitro as seen by competition assay (Fig 3E), plaque size (Fig 3F and 3G) and vacuole size distribution ( Fig 3H). To further check the importance of Tg6PGDH2 in vivo, a Tg6PGDH2-knockout (Δ6pgdh2) clonal mutant was produced from rapamycin-treated Tg6PGDH2-cKD strain and ascertained by diagnostic PCR (S6A Fig) and Western blotting (S6B Fig). The Δ6pgdh2 parasites were viable despite their growth significantly slowed down. Subsequently, purified days formed smaller plaques than their parental strain, and untreated Tg6PGDH2-cKD parasites formed plaques similarly to DiCre. (H), Intracellular replication assay comparing parasite growth in vitro. Tg6PGDH2-cKD parasites were treated with rapamycin for 4 days, and then they were allowed to infect fresh HFF cells and grown for 24 h, subsequently, the number of parasites in each PV was checked by IFA. Results are means ± SEM for n = 3 independent experiments. (I), Survival curves of mice infected with tachyzoites of indicated strains. Tg6PGDH2-cKD and Δ6pgdh2 mutants were used to infect ICR mice (100 parasites/mouse, n = 10 mice for each strain) by intraperitoneal injection, and the survival of mice was followed for 30 days. � , p <0 .05, Gehan-Breslow-Wilcoxon tests.

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma Δ6pgdh2 tachyzoites were used to infect ICR mice by intraperitoneal injection, and the survival of mice was monitored for 30 days. Typical results of such virulence tests indicated that Tg6PGDH2 deletion led to attenuated virulence ( Fig 3I).
To confirm the specificity of the observed growth defect, we complemented the Tg6PGDH2-cKD strain with a Tg6PGDH2-expressing cassette into the UPRT locus (S7A Fig). The complemented strain (compTg6PGDH2) was confirmed by diagnostic PCRs (S7B Fig). As expected, the growth defect was fully restored in the compTg6PGDH2 strain as shown by plaque (S7C Fig) and replication assays (S7D Fig). These results suggest that Tg6PGDH2 plays a vital role during the lytic cycle.

TgRPI is vital for in vitro and in vivo growth of tachyzoites
We next investigated ribulose 5-phosphate isomerase (RPI), which serves the non-oxidative branch of the PPP. As described above, a conditional knockdown mutant of TgRPI was generated and confirmed by PCR screening, immunofluorescence assay and Western blotting ( Fig  4A-4D). The rapamycin-induced TgRPI knockout parasites displayed an impaired phenotype in vitro as seen by competition ( Fig 4E), plaque (Fig 4F and 4G) and replication assays ( Fig  4H). To further determine the importance of TgRPI, a TgRPI-knockout (Δrpi) clonal mutant was produced from rapamycin-treated TgRPI-cKD strain (S6C Fig). Western blotting confirmed the loss of TgRPI expression in the Δrpi mutant (S6D Fig). The Δrpi parasites were viable and could be maintained in vitro but their ability to invade host-cell was significantly reduced (Fig 4I). The TgRPI-cKD and Δrpi strains were used to infect ICR mice, and the parasite load in the peritoneal fluid was quantified by qPCR of the β-tubulin transcript ( Fig 4J). The results suggested that TgRPI plays a critical role in parasite propagation in vivo. To test the parasite virulence, we infected mice with TgRPI-cKD and Δrpi tachyzoites, and monitored their survival for 30 days (Fig 4K). Consistent with in vitro data, knockout of TgRPI resulted in attenuated virulence in mice. To decipher the catalytic function of TgRPI, we complemented the Δrpi strain by expressing hemagglutinin (HA)-tagged Trypanosoma brucei RPI (TbRPI) because the latter enzyme has been well-characterized [27]. The complemented strain comp-TbRPI was obtained after drug selection and clonal dilution ( Fig 5A). The PCR and IFA ( Fig  5B and 5C) confirmed the integration and expression of TbRPI, respectively. Indeed, the growth and invasion defects were partly restored in the compTbRPI strain, as shown by plaque ( Fig 5D and 5E), replication ( Fig 5F) and invasion assays ( Fig 5G). These data suggest that tachyzoites depend on RPI activity for the lytic cycle.

TgRPI deletion perturbs the proteome of tachyzoites
To further enhance our understanding of the consequences of Δrpi and provide an integrative and global picture of the roles of ribose-5-phosphate in Toxoplasma gondii physiology, we also examined the effect of TgRPI deletion on protein synthesis by 4D label-free mass spectrometry-based proteomics, which allowed identification and relative quantification of 3895 proteins. In total, 464 proteins were upregulated while 384 were downregulated in the Δrpi parasites. Significant repression of TgRPI in the mutant validated the obtained data ( Fig 6A). Subsequently, we analyzed the localization of proteins and found that many nuclear proteins were impacted upon depletion of TgRPI ( Fig 6B). The pathway enrichment analysis highlighted perturbation of proteins mainly involved in the peptide biosynthesis and ribosome assembly ( Fig 6C). We observed a striking decrease in several ribosome constituent proteins including RPLs, RPPs and RPSs (Fig 6D), which are vital for protein synthesis. Other proteins affected in the Δrpi mutant included the micronemal and rhoptry proteins, some markedly decreased (Fig 6E), supporting our host-cell invasion data. Not least, we also noted a few bradyzoite-specific SRS increased, while tachyzoite-specific SRS were repressed. To validate these data, quantitative real-time PCR was used to analyze some differentially expressed genes between the wild type and Δrpi strain. As expected, the expression of TgMAG1, TgSRS12B, TgSRS35A and TgSRS53F was up-regulated in the Δrpi mutants ( Fig 6F). Collectively, these results are indicative of slowed protein synthesis coupled with stage transition upon impairment of ribose 5-phosphate synthesis.

Tg6PGDH2 and TgRPI depletion impairs the flux of 13 C-glucose in central carbon pathways
To interrogate whether Tg6PGDH2 depletion affected pentose sugars synthesis, the intracellular Tg6PGDH2-cKD (WT) and Δ6pgdh2 parasites were cultured with 1,2-13 C 2 -glucose for 12 hours, followed by assessment by liquid chromatography-MS (LC-MS). The tracer labeling of Means ± SD of more than 60 plaques for each strain was graphed. (H), Intracellular replication assay comparing parasite growth in vitro. TgRPI-cKD parasites were treated with rapamycin for 4 days, and then they were allowed to infect fresh HFF cells for 24 h, subsequently, the number of parasites in each PV was checked by IFA. Results are means ± SEM for n = 3 independent experiments. (I), Invasion assay where freshly egressed parasites were used to invade HFF monolayers for 20 min. Efficiencies of invasion, as determined by two-color staining to distinguish invaded vs non-invaded tachyzoites (means ± SEM, n = 3 assays), ��� , P<0.001; Student's t-test. (J), Parasite loads in the peritoneal fluids of ICR mice. ICR mice were infected with TgRPI-cKD and Δrpi tachyzoites (10 4 tachyzoites/mouse) by intraperitoneal injection (n = 5 for each group), and parasite loads in peritoneal fluids 5 days post-infection were estimated by qPCR. (K), Virulence tests of indicated strains in ICR mice (100 parasites per mouse, n = 10 mice for TgRPI-cKD strain, n = 20 mice for Δrpi mutants), ��� , P<0.001; Gehan-Breslow-Wilcoxon tests. https://doi.org/10.1371/journal.ppat.1010864.g004

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma fresh intracellular parasites showed that the M+1 labeled R5P in Δ6pgdh2 mutants was significantly less than WT parasites (Fig 7A). We also observed that the M+1 labeled Xu5P and Ru5P containing one 13 C were reduced in Δ6pgdh2 mutants (Fig 7B and 7C). These data likely reflected an impaired pentose sugars synthesis upon Tg6PGDH2 depletion.
To investigate whether TgRPI deletion affected the metabolite abundances of PPP intermediates, the TgRPI-cKD (WT) and Δrpi parasites were cultured in HFF cells and metabolite abundances were determined by LC-MS. Interestingly, although TgRPI deletion resulted in a marked reduction in growth rate, it did not significantly affect the metabolite abundances of several PPP intermediates (S8 Fig). To further reveal the metabolic consequences of genetic lesions, metabolic labeling of intracellular TgRPI-cKD (WT) and Δrpi parasites was performed using 1,2-13 C 2 -glucose. We found reduced incorporation of 13 C into R5P upon TgRPI depletion ( Fig 7D). However, it should be noted that there was no significant change in the M+1

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma labeled R5P by TgRPI disruption, which may be due to the metabolic complexity and the fact that R5P can also be produced by non-oxidized PPP. Because a previous study has demonstrated that the pentose phosphate cycle is non-negligible in the extracellular stage [4], we also performed [ 13 C 6 ]-glucose labeling in fresh extracellular parasites to analyze the incorporation

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma of 13 C into Ru5P and Xu5P. We found that the Δrpi parasites showed increased 13 C labeling of Ru5P and Xu5P (Fig 7E). Similarly, the TgRPI-cKD strain was cultured with or without rapamycin for 4 days, and fresh extracellular tachyzoites were labeled with [ 13 C 6 ]-glucose, followed by evaluation of selected metabolites using LC-MS. As expected, the inclusion of 13 C into 6-phosphogluconate (6PG, an intermediate of oxidative branch) was significantly increased upon depletion of TgRPI (Fig 7F). A reduction in the incorporation of 13 C into sedoheptulose-7-phosphate (S7P, an intermediate of non-oxidative branch) was also observed (Fig 7F). In our extended work, we determined the glycolysis flux alterations upon depletion of TgRPI. We found that the incorporation of 13 C into fructose-1,6-bisphosphate and glyceraldehyde-3-phosphate was increased (Fig 7F). These data together indicated that a loss of TgRPI impairs the operation of PPP and leads to an increase in glucose-fueled glycolysis.

Discussion
This study shows that Toxoplasma gondii encodes a functional pentose phosphate pathway distributed in the cytoplasm and nucleus. TgG6PDH1/2, Tg6PGDH1, TgRuPE and TgTAL are dispensable for parasite growth and virulence. Our work also reveals that TgG6PDH-knockout tachyzoites usually grow but are susceptible to oxidative stress. By contrast, Tg6PGDH2 and TgRPI are needed for the optimal growth of tachyzoites. In particular, depletion of TgRPI impairs the metabolic flux of 13 C-glucose and perturbs the parasite proteome. Our results suggest that PPP supports the core carbon metabolism by producing metabolic intermediates and NADPH in T. gondii (Fig 8A and 8B); However, the parasite can reprogram its carbon flux to maximize its survival upon genetic perturbation (Fig 8C-8G).
G6PDH as the first and rate-limiting enzyme of the PPP has an important function in various cells [28][29][30][31]. It has been reported previously that the knockdown of G6PDH inhibited the growth of cancer cells, such as leukemia THP-1 and human melanoma A375 cells [28,29]. Loss of G6PDH in Trypanosoma brucei indicated no obvious phenotype in the procyclic stage but was lethal for the bloodstream form [32]. Interestingly, G6PDH deletion in Saccharomyces cerevisiae appeared to grow normally [33]. We found that none of the two TgG6PDHs is required for parasite development. Consistent with this finding, two TgG6PDH genes have been assigned a positive phenotype score based on the genome-wide CRISPR-Cas9 screen (TgG6PDH1 phenotype score = 0.69, TgG6PDH2 phenotype score = 1.49) [34]. We speculate that ribose 5-phosphate in the Δg6pdh1Δg6pdh2 mutant is provided through the transketolasetransaldolase (TgTKT-TgTAL) pathway, which can bridge the PPP and glycolysis by sharing GA3P and F6P (Fig 8C). Consistent with this, our attempts to delete TgTKT by CRISPR/ Cas9-assisted homologous gene replacement were futile. On the other hand, disruption of TgG6PDHs will block NADPH synthesis but maintain the intact downstream pathway. A previous study has reported that Toxoplasma expresses a phosphate transporter that plays a crucial role in the phosphate import [35]. There may be transporters in the parasite membrane to import 6-phosphogluconate and possibly other host-derived PPP intermediates that can sustain (low-level) synthesis of NADPH in the Δg6pdh1Δg6pdh2 strain. In addition, there remains a possibility of the remaining-G6PDH activity is not encoded by TgG6PDH1 and TgG6PDH2. Previous studies determined that tracing deuterium isotope from [3-2 H] glucose could reveal the contribution of the oxidative PPP enzymes to the cellular NADPH pool [36]. Therefore, using [3-2 H] glucose to label NADPH would be an excellent way to assess NADPH production by TgG6PDH or other sources. It is plausible that other NADPH-generating enzymes (such as isocitrate dehydrogenase, malic enzyme and transhydrogenase) may contribute synergistically in T. gondii [37,38]. NADPH is required to protect mammalian cells against the oxidative stress [24][25][26]. Likewise, we observed that the replication of the H 2 O 2 -treated double mutant was notably impaired, which indicates a physiological role of G6PDH in NADPH homeostasis and under oxidative stress.
Our bioinformatic search did not find annotation for the second enzyme of the PPP (6-phosphogluconolactonase or 6PGL) in T. gondii. The related apicomplexan parasite Plasmodium falciparum harbors a bifunctional G6PDH-6PGL catalyzing the first two reactions [39]. It is, therefore, possible that TgG6PDH1/2 encodes 6PGL activity that requires further investigation. This study also reports that two Tg6PGDH enzymes in T. gondii, one residing in the cytoplasm (Tg6PGDH2) play a crucial role in the parasite growth. In contrast, the other protein (Tg6PGDH1) is not expressed during the lytic cycle. Our data indicated that tachyzoites require Tg6PGDH2 to produce ribulose-5-phosphate (Ru5P), which in turn yields ribose 5-phosphate (R5P). Upon deletion of Tg6PGDH2, the TgTKT-TgTAL pathway may

Pentose phosphate pathway in Toxoplasma
In the non-oxidative branch of the PPP, RPI supports ribose-5-phosphate synthesis. Its inactivation severely inhibits the tachyzoite growth, suggesting that the TgTKT-TgTAL pathway cannot adequately compensate for its loss in the TgRPI-cKD mutant (Fig 8E). TgRuPE was dispensable for yielding X5P, which consistent with the CRISPR/Cas9 screen (phenotype score = -0.31) [34]. We reasoned that the Δrupe tachyzoites might supply X5P by the TgTKT-TgTAL pathway (Fig 8F). Moreover, TgTAL was not needed for the parasite growth and virulence, as the Δtal mutant may obtain S7P by TgSBPase, which removes one phosphate group from sedoheptulose-1,7-bisphosphate (SBP) to produce S7P (Fig 8G). Previous studies showed that deletion of TgSBPase reduced the fitness of T. gondii [17].
Ribose 5-phosphate (R5P) is a vital source for de novo synthesis of nucleotide and amino acid [41]. In this study, we reasoned that TgRPI depletion impairs carbon metabolic homeostasis and hinders R5P production, which may result in transcribing and translating tardily, affecting regular protein biosynthesis. This hypothesis was further supported by our quantitative proteomic data which displayed that the biosynthesis of ribosomal proteins, micronemal proteins, rhoptry proteins and other proteins were significantly affected (Fig 6). Our quantitative proteomic analysis also showed that long-term TgRPI-depleted caused a few bradyzoitespecific SRS increases while tachyzoite-specific SRS were repressed, which perhaps hinting that impairing ribose 5-phosphate synthesis caused slow growth and triggered stage conversion. Consistent with quantitative proteomic data, the up-regulated expression of TgMAG1, TgSRS12B, TgSRS35A and TgSRS53F in the Δrpi mutants was further validated by quantitative real-time PCR. Type I T. gondii strain (RH) has less tendency for conversion to bradyzoites. Because Δrpi mutants were generated in the RH strain, it would be very insightful to analyze the bradyzoite conversion of TgRPI deletion mutants in more cystogenic type II parasites (ME49) in further study.
Based on bioinformatics analysis, combined with phenotypic features of Tg6PGDH2 and TgRPI knockout strains in vitro and in vivo, we speculate that Tg6PGDH2 and TgRPI are potential targets for drug designs against Toxoplasma infections. Our bioinformatics analysis found that although the amino acid sequence of Tg6PGDH2 and homo sapiens 6PGDH (Hs6PGDH) share about 51% identity, its C-terminal sequences diverge from that of Hs6PGDH. The amino acid sequence of TgRPI was more similar to that of Plasmodium falciparum RPI, with only 42% identity with homo sapiens RPI. Especially, encouraged by selective inhibition of T. brucei 6PGDH by hydroxamic derivatives [42], we thought it would be interesting to screen inhibitors of Tg6PGDH2 in future studies. In addition, the crystal structure of TgRPI has been identified [22], which provided reference information for the design of inhibitors against TgRPI.
In conclusion, our findings delineate the pentose phosphate pathway in T. gondii, and establish its physiological importance in the parasite metabolism, reproduction and virulence. Moreover, Tg6PGDH2 and TgRPI emerge as the two potential therapeutic targets against toxoplasmosis. Not least, we further highlight the metabolic plasticity in tachyzoites that enables them to survive.

Ethics statement
All animal experiments were approved by the Ethical Committee of South China Agricultural University (permit no.2021f146).

Mice and parasite strains
Seven-week-old female ICR mice were purchased from the Guangdong Medical Experimental Animal Center in Guangdong Province. They were maintained under standard conditions

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma according to the regulations of the Administration of Affairs Concerning Experimental Animals.
The RHΔku80 and DiCre strains of T. gondii used in this study were grown within human foreskin fibroblast (HFF) cells (ATCC, Manassas, VA, USA). Cultures were done in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (Gibco Life Technologies, Inc., Rockville, MD, USA), 10 U/ml penicillin and 100 μg/ml streptomycin. All genetically modified strains were generated from the parental strains and maintained in vitro under the same growth conditions as parental strains.

Plasmid construction
All primers used for each reaction and plasmids used in the mutant constructions are listed in S1 and S2 Tables, respectively. Locus-specific CRISPR/Cas9 plasmids were generated using the Q5 site-directed mutagenesis kit (New England Biolabs Inc., USA) as described previously [43]. Other plasmids were constructed by multifragment ligation using the ClonExpress II one-step cloning kit (Vazyme Biotech Co., Ltd, Nanjing, China). All plasmids were verified by DNA sequencing before use.

Parasite culturing and transfection
The corresponding locus-specific CRISPR/Cas9 plasmids and homologous donor templates (S2 Table) were co-transfected into freshly tachyzoites of the RHΔku80, Dicre [44,45] or derivative strains, selected with 1 μM pyrimethamine (Sigma-Aldrich, USA) or 30 μM chloramphenicol (Sigma-Aldrich, USA), and single cloned by limiting dilution. The TbRPI complement strain was constructed by cotransfecting the UPRT-specific CRISPR plasmids along with the TbRPI-expressing cassettes into the Δrpi mutants and selected with 10 μM 5-flurodeoxyuracil (FUDR). The compTg6PGDH2 strain was constructed by inserting a Tg6PGDH2 expressing cassette into the UPRT locus of the Tg6PGDH2-cKD strain and selected with FUDR. Positive clones were identified by diagnostic PCRs (primers listed in S1 Table). All transgenic parasites used in this study are listed in S3 Table.

Polyclonal antibodies production and immunofluorescence assays
The Tg6PGDH2 and TgRPI were amplified from the cDNA of the RHΔku80 strain and cloned into the vector pCold. Subsequently, recombinant proteins TgRPI and Tg6PGDH2 were purified from E. coli BL21 (DE3). The expression of proteins was tested by SDS-PAGE and western blot analyses. The purified recombinant proteins TgG6PDH2 and TgRPI were used to immunize 7-week-old female mice respectively. Positive antiserum was collected and stored at -80˚C. To check the localization of native TgG6PDH2 and TgRPI, the intracellular RHΔku80 tachyzoites were purified and used to infect HFF cells. Then, the cells were fixed, permeabilized, and blocked. Next, the coverslips were incubated with rabbit anti-TgALD (a cytoplasm marker) and mouse anti-TgG6PDH2 or mouse anti-TgRPI, stained with Alexa Fluor 488 goat anti-mouse IgG, Alexa Fluor 594 goat anti-rabbit IgG secondary antibodies and Hoechst and then imaged under fluorescence microscopy on a BX53 microscope (Olympus, Tokyo, Japan).

Western blot assay
The Tg6PGDH2-cKD and TgRPI-cKD parasites were grown in a DMEM medium with or without rapamycin treatment for 4 days. The parasites of Δrpi and Δ6pgdh2 were cultured under standard tissue culture conditions. Then all parasites were purified and collected respectively, and the whole proteins of each mutant were extracted with lysis buffer. Finally, the

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma corresponding protein expression was detected by Western blot using mouse anti-Tg6PGDH2 or mouse anti-TgRPI. The rabbit anti-TgALD was used as an internal reference. Finally, the signals were visualized with the super ECL detection reagent on UVP ChemStudio (Analytik Jena, Germany).

Plaque assay
HFF monolayers seeded on 6-well plates were infected with freshly egressed parasites (100 tachyzoites/well, three wells per strain). Subsequently, the plates were cultured at 37˚C with 5% CO 2 for 7 or 9 days. Then they were fixed with 4% paraformaldehyde, stained with 0.1% crystal violet for 20 minutes and imaged on a scanner (Microtek Scan Marker i600, MICRO-TEK, China) to analyze the relative sizes and number of plaques, as described previously [6].

Intracellular replication assay
Parasites were used to infect fresh HFF cells seeded on coverslips (two wells per strain) for 1 h. Subsequently, the extracellular parasites were washed away using PBS, and the rest of the cultures were kept at 37˚C with 5% CO 2 for another 24 h. Cells were fixed with 4% paraformaldehyde, incubated with rabbit anti-TgALD for 20 min, permeabilized with 0.1% Triton X-100 for 15 min, incubated with mouse anti-Tg IgG for 20 min and stained with Hoechst, Alexa Fluor 488-conjugated goat anti-mouse lgG secondary antibodies and Alexa Fluor 594-conjugated goat anti-rabbit lgG secondary antibodies for 20 min. Following fluorescence staining, a minimum of 150 vacuoles were examined for each sample by fluorescence microscopy to determine the number of parasites in each parasitophorous vacuole.

Analysis of H 2 O 2 resistance
Freshly egressed RHΔku80, Δg6pdh1, Δg6pdh2 and Δg6pdh1Δg6pdh2 tachyzoites were incubated for 3 h in DMEM containing 500 μM H 2 O 2 as described previously [46,47]. Then parasites were used to infect fresh HFF cells for 2 h, and invaded parasites were grown under standard growth conditions for another 24 h. Subsequently, the samples were fixed, stained and analyzed according to the protocols specified in the above intracellular replication assay.

NADPH assay
All parasites (approximately 3x10 7 cells/strain) were first purified and collected. Strains were then washed with cold PBS and collected by rotating at low speed for 5 minutes. Subsequently, the NADPH was detected using the NADP/NADPH assay kit (Abcam, Cambridge, UK). Briefly, 400 μl NADP/NADPH lysis buffer was added to extract parasites by carrying out two freeze cycles (20 min on dry ice followed by 10 min at room temperature). Then the extractions were centrifuged, and 200 μl supernatants of each sample were heated at 60˚C for 30 min and cooled down on ice immediately to decompose the NADP + . Added 100μl of the reaction mix to each standard, sample well, and incubated the plate at room temperature for 5min. At last, 10 μl of NADPH developer was added to each well and mixed. After 4 hours, the optical density of samples was measured at OD450 nm by a SYNERGY multi-mode reader (BioTek Instruments, Inc, USA).

Competition assay
The Tg6PGDH2-cKD and TgRPI-cKD parasites were pretreated with or without 50 nM rapamycin for 24 h. Then pretreated and untreated parasites were harvested, mixed in approximately 1:1 ratio, and YFP positive parasites were monitored every 2 days by flow cytometry.

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma About 10, 000 events per sample were acquired on a CytoFLEX (Beckman Coulter, Inc., USA) and data analysis was proceeded using CytExpert software.

Invasion assay
Freshly egressed parasites were purified, counted and then used to infect HFF cells (10 6 tachyzoites/strain) for 20 min at 37˚C with 5% CO 2 . Subsequently, the samples were washed with PBS, fixed with 4% paraformaldehyde, incubated with rabbit anti-TgALD for 20 min, permeabilized with 0.1% Triton X-100 for 15 min and blocked with 10% FBS. The TgRPI-cKD strain was incubated with mouse anti-Tg IgG for 20 min while Δrpi and compTbRPI strains with YFP-positive omitted this step. Subsequently, the TgRPI-cKD strain was stained with Hoechst, Alexa Fluor 488 goat anti-mouse lgG secondary antibodies and Alexa Fluor 594 goat anti-rabbit lgG secondary antibodies, while Δrpi and compTbRPI strains were stained with Hoechst and Alexa Fluor, 594-conjugated goat anti-rabbit lgG secondary antibodies for 20 min. After fluorescence staining, the number of parasites invading cells in each field was counted by fluorescence microscope. At least 150 fields were observed in each sample.

Virulence testing
Seven-week-old female ICR mice were injected intraperitoneally with 100 freshly egressed tachyzoites. The symptoms and survival of mice were monitored for 30 days, and blood samples from mice that survived were collected afterwards. Mice seronegative by IFA or enzymelinked immunosorbent assay were not included in the analysis [19]. Cumulative mortality was plotted as Kaplan-Meier survival plots and analyzed using Prism 5 (GraphPad Software Inc., La Jolla, CA, USA).

Determination of parasite burden in peritoneal fluids of mice
Freshly egressed 10 4 tachyzoites were allowed to infect female ICR mice (7 weeks old) by intraperitoneal injection. Five days post-infection, the mice were euthanized, their peritoneal fluids were collected and genomic DNA was extracted using the TIANamp Blood DNA Kit (Tiangen Biotech Co. Ltd, Beijing, China). Parasite burden in peritoneal fluids was determined by quantitative PCR (primers listed in S1 Table), as described previously [19].

Whole genome sequencing
Freshly egressed Δg6pdh1Δg6pdh2 tachyzoites were purified and genomic DNA from the parasites was extracted using the TIANamp Blood DNA Kit (Tiangen Biotech Co. Ltd, Beijing, China). Subsequently, purified genomic DNA was subject to genome sequencing as described previously [48]. Clean reads were mapped to the reference genome of the T. gondii GT1 strain. The mapping results were visualized by the Integrative Genomics Viewer (https://software. broadinstitute.org/software/igv/).

Semi-quantitative RT-PCR and quantitative real-time PCR
All freshly egressed tachyzoites (RHΔku80; Δg6pdh1; Δg6pdh2; Δg6pdh1Δg6pdh2; Δrupe; Δtal; TgRPI-cKD; Δrpi) were purified using 3 μm polycarbonate membranes and collected firstly. Then total RNA was extracted from each mutant by Eastep super total RNA extraction kit (Promega Biotech Co. Ltd, Beijing, China) and reversely transcribed to cDNA referring to the method of cDNA synthesis kit (Yeasen Biotechnology Co., Ltd., Shanghai, China). To validate the successful gene knock-out of each mutant, semi-quantitative PCR was performed using equivalent cDNA. Besides, Transcript levels for TgG6PDH1, TgG6PDH2, TgMAG1,

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma TgSRS12B, TgSRS35A and TgSRS53F in each sample were analyzed by quantitative real-time PCR with SYBR Green PCR mix (TOYOBO, Osaka, Japan) in a Light Cycler 480 (Roche, Basel, Switzerland), using β-tubulin as an internal reference. Primers used for semi-quantitative RT-PCR and real-time PCR are listed in S1 Table in the supplemental material.

Transcriptome analysis
The tachyzoites of RHΔku80 and Δg6pdh1Δg6pdh2 were collected and purified. Then total RNA from each sample was extracted using Transzol UP Reagent (TransGen Biotech Co., Ltd, Beijing, China) according to the manufacturer's instructions. Subsequently, RNA sequencing was performed as described previously [49]. Briefly, mRNA was purified and captured by magnetic beads with Oligo (OT) and then was fragmented. Random primers were used for reverse transcription to synthesize the first strand of cDNA, and second-strand synthesis was combined with A-tailing. The quality of libraries was assessed by qPCR, and the qualified libraries were sequenced by the Illumina platform with the PE150 strategy. Then clean reads were obtained by removing low-quality sequences and were analyzed using three analytical processes: sequencing data quality control, data comparison analysis, and transcriptome deep analysis. In the end, classification and feature analysis were performed according to different genomic annotation information. Then the expression levels of each mutant were calculated, and differential expression analysis was performed. The transcriptome data analysis was performed using the online platform of Majorbio Cloud Platform (www.majorbio.com).

Proteomic analyses
All parasites (3x10 7 parasites/strain) were sonicated using an ultrasonic processor in lysis buffer (1% SDS,1% protease inhibitor cocktail) and centrifuged at 4˚C at 12000 g for 10 min. Then the supernatant was collected and the protein concentration was determined with a BCA kit (Beyotime Biotechnology, Shanghai, China) according to the manufacturer's instructions. Subsequently, the supernatants were precipitated with cold acetone at -20˚C for 2 h. After centrifuging at 4500 g at 4˚C for 5 min, the remaining precipitates were washed with cold acetone (Zhejiang Hannuo Chemical Technology Co., Ltd., Lanxi City, Zhejiang Province, China) twice and redissolved in 200 mM tetraethylammonium bromide (Sigma-Aldrich, St. Louis, Missouri, USA). Then the protein was hydrolyzed overnight with trypsin (at a 1:50 mass ratio of trypsin to protein). The next day, the protein solution was reduced with 5 mM DL-dithiothreitol (Sigma-Aldrich) for 30 min at 56˚C and alkylated with 11 mM iodoacetamide (Sigma-Aldrich) at room temperature in the dark for 15 min. The peptides obtained by enzymolysis were first dissolved in solvent A (aqueous solution of 0.1% formic acid and 2% acetonitrile) and then were transferred into a home-made reversed-phase analytical column with a length of 25 cm and an inner diameter of 75/100 μm. At last, the peptides were separated with solvent B (aqueous solution containing 0.1% formic acid and 99.9% acetonitrile) by using UltiMate 3000 UHPLC system (liquid gradient setting: from 6% to 24% solvent B in 70 min, from 24% to 35% in 14 min, and went up to 80% in 3 min then stayed at 80% for 3min; flow rate: constant current, maintained at 450 nL/min). The peptides were subjected to a capillary source followed by the timsTOF Pro mass spectrometry (Bruker Daltonics Inc., USA) with the mode of parallel accumulation serial fragmentation (PASEF). Fragments and precursors were analyzed utilizing a TOF detector and the MS/MS scan ranged from 100 to 1700 m/ z. Precursors were selected for fragmentation with charge states 0 to 5, and PASEF-MS/MS scans were performed 10 times per cycle. Besides, set the dynamic exclusion time to 30s. Subsequently, a sample-specific protein database was constructed based on the samples, and the quality control analysis was performed on the peptide and protein levels. Then, the GO,

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma KEGG and other databases were used to annotate standard functions of identified proteins and quantitative analysis of proteins was performed. According to the quantitative results, differential screening and functional classification statistical analysis of differential proteins were conducted. Finally, fisher's exact test was used to analyze the statistical results of rich cluster analysis to compare the functional relationship of differential proteins under different experimental conditions.

Metabolic analysis
The TgRPI-cKD parasites were grown in the corresponding medium with or without rapamycin treatment for 4 days. The tachyzoites of RHΔku80, Δg6pdh1Δg6pdh2 and Δrpi were grown in vitro under standard tissue culture conditions. The freshly egressed parasites were filtered using 3 μm polycarbonate membranes. Then 3x10 7 parasites were cultured in a glucose-free DMEM medium supplemented with 8 mM 13 C 6 -glucose for 4 h. After that, the parasites were washed with PBS and lysed in 1 ml of ice-cold methyl alcohol, as described previously [6,19]. Then the samples were treated with 5 cycles of "1 min ultrasound and 1 min interval" on an ice bath and placed at -20˚C for 30 min. After that, the supernatant centrifuged was evaporated with nitrogen and redissolved in 50% aqueous acetonitrile. Then, chromatographic separation was performed on the Ultimate 3000 UHPLC system (Thermo Fisher Scientific, USA) with a Waters BEH Amide column (2.1 mm × 150 mm, 1.7 μm). The mobile phase consisted of (A) water containing 15 mM ammonium acetate (pH = 8.5) and (B) acetonitrile/water (90:10, volume ratio) and the flow rate was set at 0.35 mL/min. Subsequently, the metabolites were eluted in a linear gradient mode with the following program: 0-2 min, 90% B; 14 min, 75% B; 15 min, 65% B; 15.2-16.9 min, 50% B; 17-20 min, 90% B. At last, the eluents were analyzed in heated electrospray ionization negative (HESI-) mode using Q Exactive Hybrid Quadrupole-Orbitrap Mass Spectrometry (Thermo Fisher Scientific, USA) and the parameters were set as follows: Spray voltage: 3500 V; Scan rage: mass/charge ratio: 70-1050 and AGC:1× 10 6 . At last, the data was analyzed with Xcalibur software, and the raw mass spectrometry data was corrected using IsoCor v2 as described previously [50].
Quantification of PPP metabolites by UHPLC-MS. The TgRPI-cKD (WT) and Δrpi parasites were processed as described above, except for the evaporated residues redissolved in 40 μL of 50% aqueous acetonitrile with 5 μg/mL of 13 C 6 -F1,6P to UHPLC-HRMS analysis. The raw mass spectral data were acquired using Xcalibur software, and relative abundances of PPP intermediates were obtained using the added 13 C 6 -F1,6P as a reference.
Labeling of PPP intermediates from 1,2-13 C 2 -glucose in intracellular parasites. The intracellular Tg6PGPDH2-cKD, Δ6pgdh2, TgRPI-cKD and Δrpi were grown in a glucose-free DMEM medium containing 8 mM 1,2-13 C 2 -glucose (Sigma-Aldrich, USA) for 12 h and approximately 3x10 7 parasites were collected. Then the samples of each strain were prepared as described previously [6,19,51] and redissolved in 50 μL of 10% methanol with 0.1% formic acid. Subsequently, the LC-MS/MS analysis was performed on an Agilent 1290 Infinity II UHPLC system coupled to a 6470A triple quadrupole mass spectrometry (Santa Clara, CA, United States) as described previously [51]. Chromatography was performed using a Waters HSS T3 column (2.1 mm × 100 mm, 1.7 μm). The instrument setting parameters of LC-MS/MS was almost the same as those above, except that the flow rate was changed to 0.25 mL/min, and the chromatographic separation gradient elution procedure was changed to 0-1 min, 1% B; 2 min, 10% B; 7 min, 10% B; 9 min, 99% B. 11 min, 99% B; 11.1 min, 1% B and kept until 13 min. While the main parameters of the ion source were changed as follows: the spray voltage was 3000 V. The capillary temperature and the Probe heater temperature were 300˚C and 350˚C respectively. The sheath gas flow rate was changed to 11 L/min. Finally, data were acquired

PLOS PATHOGENS
Pentose phosphate pathway in Toxoplasma using the MassHunter software (version B.08.00, Agilent), and the raw MS data was corrected by IsoCor v2 as described previously [50].

Statistical analysis
All statistical analyses were performed in GraphPad Prism 5 (GraphPad Software Inc., La Jolla, CA, USA) using Student's t-tests, Gehan-Breslow-Wilcoxon tests, one-way analysis of variance (ANOVA) with Bonferroni post-tests, or two-way ANOVA as indicated in the figure legends.  deletion mutants. (A), Tg6PGDH2-cKD strain was treated using rapamycin for 24 h and then one clean Tg6PGDH2-knockout clone (Δ6pgdh2) was produced through limiting dilution. Diagnostic PCR on a selected Δ6pgdh2 clone confirming the deletion of Tg6PGDH2. (B), Loss of Tg6PGDH2 expression in the Δ6pgdh2 mutant as checked by Western blotting using mouse anti-Tg6PGDH2 and rabbit anti-TgALD. TgALD was included as a loading control. (C), Diagnostic PCR on a selected Δrpi clone confirming the deletion of TgRPI. Primers used for diagnostic PCR validation were listed in S1 Table Δrpi parasites were collected and purified using a 3 μm pore size filter. Then 3x10 7 parasites were lysed with ice-cold methyl alcohol and metabolite abundances were measured by LC-MS. Relative concentrations of indicated metabolites were calculated using the added 13 C 6 -F1,6P as the internal standard. Means ±SEM from five independent experiments (n = 5) were graphed. Student's t-test. (TIF) S1