Glucose Oxidation to Pyruvate Is Not Essential for Brucella suis Biovar 5 Virulence in the Mouse Model

Brucella species cause brucellosis, a worldwide extended zoonosis. The brucellae are related to free-living and plant-associated α2-Proteobacteria and, since they multiply within host cells, their metabolism probably reflects this adaptation. To investigate this, we used the rodent-associated Brucella suis biovar 5, which in contrast to the ruminant-associated Brucella abortus and Brucella melitensis and other B. suis biovars, is fast-growing and conserves the ancestral Entner-Doudoroff pathway (EDP) present in the plant-associated relatives. We constructed mutants in Edd (glucose-6-phosphate dehydratase; first EDP step), PpdK (pyruvate phosphate dikinase; phosphoenolpyruvate ⇌ pyruvate), and Pyk (pyruvate kinase; phosphoenolpyruvate → pyruvate). In a chemically defined medium with glucose as the only C source, the Edd mutant showed reduced growth rates and the triple Edd-PpdK-Pyk mutant did not grow. Moreover, the triple mutant was also unable to grow on ribose or xylose. Therefore, B. suis biovar 5 sugar catabolism proceeds through both the Pentose Phosphate shunt and EDP, and EDP absence and exclusive use of the shunt could explain at least in part the comparatively reduced growth rates of B. melitensis and B. abortus. The triple Edd-PpdK-Pyk mutant was not attenuated in mice. Thus, although an anabolic use is likely, this suggests that hexose/pentose catabolism to pyruvate is not essential for B. suis biovar 5 multiplication within host cells, a hypothesis consistent with the lack of classical glycolysis in all Brucella species and of EDP in B. melitensis and B. abortus. These results and those of previous works suggest that within cells, the brucellae use mostly 3 and 4 C substrates fed into anaplerotic pathways and only a limited supply of 5 and 6 C sugars, thus favoring the EDP loss observed in some species.


INTRODUCTION
Members of the genus Brucella are α2-Proteobacteria that infect a wide range of vertebrates causing brucellosis in mammals (Al Dahouk et al., 2008;Whatmore, 2009;Soler-Lloréns et al., 2016), a zoonosis with a high impact on developing countries worldwide (McDermott et al., 2013). Although the genus includes an increasing number of species (Moreno, 2020), Brucella abortus, Brucella melitensis, and Brucella suis (often referred to as the classical smooth spp.) are by far those having the more severe impact on both livestock and humans and they were divided long ago into biovars following phenotypic criteria (Alton, 1987). Although for a long time thought to be a very homogeneous group (Hoyer and McCullough, 1968;Verger et al., 1987), the identification of new spp. and phylogenomic studies show that B. abortus, B. melitensis, B. suis, Brucella neotomae, Brucella ovis, and Brucella canis plus isolates from sea mammals and the common vole form a relatively heterogeneous core group separated from several early diverging brucellae (Wattam et al., 2014;Soler-Lloréns et al., 2016;Moreno, 2020). These studies also show that, while all B. abortus and B. melitensis biovars group into two clades, the five B. suis biovars show a greater diversity that is inconsistent with their current taxonomic status as a single sp. (Moreno and Moriyón, 2002;Al Dahouk et al., 2008;Scholz et al., 2008a;Whatmore, 2009;Moreno, 2020).
The weight of the evidence shows that the brucellae have evolved from environmental α2-Proteobacteria (Moreno, 2020). Since they are facultative intracellular pathogens unable to persist in nature outside their hosts, this origin implies that they have probably adapted their metabolism to the peculiarities of the Brucella containing vacuoles (BCV) where they multiply. Undoubtedly because of their early identification and greater impact on domestic livestock and humans, metabolism has been investigated almost exclusively in B. abortus, B. melitensis, and biovars 1 and 3 of B. suis. These spp. and biovars, although auxotrophic only for a few vitamins and occasionally for a few amino acids (Gerhardt and Wilson, 1948;Plommet, 1991), are often described as fastidious because of their complex requirements for primary isolation (peptone-yeast extract media, often supplemented with serum) and slow growth. However, B. suis biovar 5 and Brucella microti, both rodent-associated brucellae, display much faster growth (Scholz et al., 2008b;Zúñiga-Ripa et al., 2018), which suggests a more ancestral metabolism. Consistent with this, we have recently found (Machelart et al., 2020) that, like the environmental α2-Proteobacteria neighbors, the Entner-Doudoroff pathway (EDP) is fully active in those rodent-associated spp. so that glucose is fueled into the tricarboxylic acid cycle (TCA) mostly through EDP with little contribution of the Pentose Phosphate Pathway (PPP). Thus, whereas B. suis biovar 5 mutants in edd (coding for the 6-phosphogluconate dehydratase involved in the first step of EDP) have a severe growth defect, mutation of gnd (6-phosphogluconate dehydrogenase of the first step of PPP) has only a reduced effect. On the other hand, B. abortus, B. melitensis, and B. suis other than biovar 5 rely exclusively on PPP because all carry a disabling mutation in edd. This shows the dispensability of EDD in the spp. that cause disease in livestock and, consistent with the lack of the phosphofructokinase of the Embden-Meyerhof-Parnas (classical glycolysis) in all brucellae , supports the hypothesis that glucose fueling into TCA is not essential in BCVs and was thus lost in some clades. Although B. suis biovar 5 and B. microti represent a suitable mode to test this hypothesis, we have consistently failed to obtain a double edd-gnd mutant, in all likelihood because PPP becomes essential when EDD is not functional (Machelart et al., 2020). However, as that hypothesis can be tested by blocking pyruvate synthesis at other levels of the central C pathways, in this work we applied this approach by deleting edd in B. suis biovar 5 and, instead of gnd, the genes putatively coding for pyruvate phosphate dikinase (ppdK) and pyruvate kinase (pyk) (Figure 1). Here, we present experiments that confirm the corresponding predicted metabolic phenotypes as well as the results of an assessment of virulence in the mouse model of brucellosis.
incubator (Lab Systems) using wells with sterile medium as the blank. All experiments were repeated at least three times. was excised using BamHI-NotI and cloned into a pJQKm suicide vector (Scupham and Triplett, 1997). The resulting pLZI-1 mutator plasmid (Supplementary Table 1) was transformed into E. coli strains TOP10F' and S17λpir and transferred to B. suis 513 by conjugation. Integration of the suicide vector in the chromosome was selected by polymyxin (B. suis 513 is intrinsically resistant) and kanamycin resistance, and excision of pLZI-1 (producing Bs5 pyk by allelic exchange) by polymyxin and sucrose resistance and kanamycin sensitivity. The resulting colonies were screened by PCR with primers Pyk-F1 and Pyk-R4, which amplified a fragment of 578 bp and 1917 bp in the mutant and parental strain, respectively. Bs5 ppdK pyk was constructed introducing pLZI-1 into Bs5 ppdK by conjugation and selection by polymyxin and sucrose resistance and kanamycin sensitivity, and confirmed by PCR using oligonucleotides Pyk-F1 and Pyk-R4. To check for both mutations, the internal primer Pyk-R5 (5 -TTTTCCGTCATCGATCAACA -3 ) hybridizing in the deleted region was used.

DNA Manipulations
To construct mutant Bs5 edd, the suicide mutator plasmid pNPTS edd (Supplementary Table 1), carrying the edd deleted allele (Machelart et al., 2020) was introduced into E. coli S17λpir by transformation. Then, pNPTS edd was introduced into B. suis 513 by conjugation. Following the protocol described above, colonies from the second recombination were screened by PCR with primers Edd-F1 (5 -GGCACGATTTCATCAGCGCA-3 ) and Edd-R4 (5 -CCGCCATTCATGGCATTCTGG-3 ), which amplified a fragment of 1,447 bp in the mutant and a fragment of 3,271 bp in the parental strain. The deletion removed 56% of the ORF and was identified using the internal primer Edd-R5 (5 -TCCTGAATGCGTTTATGTGC-3 ) which hybridized in the deleted region.

Virulence Assays in Mice
Seven-week-old female BALB/c mice (Envigo-Harlan Laboratories, Barcelona, Spain) were accommodated under BSL-3 biosafety containment conditions in the facility of Centro de Investigación y Tecnología Agroalimentaria de Aragón Frontiers in Microbiology | www.frontiersin.org (CITA; Registration code ES502970012025) with water and food ad libitum. The animal handling and procedures were in accordance with the current European (directive 86/609/EEC) and Spanish (RD/53/2013) legislation and authorized by the Animal Welfare Committee of the institution. For each strain, inoculum was prepared from a 24 h culture on Blood Agar Base No. 2 (see above "Bacterial Strains and Growth Conditions") at 37 • C. Bacterial suspensions in sterile phosphate buffered saline (0.85% NaCl, 0.1% KH 2 PO 4 , 0.2% K 2 HPO 4 ; pH 6.85) were spectrophotometrically adjusted to 1 × 10 9 colony forming units (CFU) and diluted to the required concentration. Mice (n = 5) were inoculated intraperitoneally with approximately 1 × 10 5 CFU in 0.1 mL of the corresponding inoculum (exact doses were retrospectively assessed by CFU accounts on BAB plates) and then euthanized 2 and 8 weeks after inoculation. Spleens were aseptically removed, individually weighed, homogenized in nine volumes of sterile saline buffer and serial 10-fold dilutions plated by triplicate on BAB plates for CFU accounts. The identity of the isolates was confirmed by PCR. Individual data (mean CFU/spleen) were normalized by logarithmic transformation and the mean log CFU/spleen values and the standard deviation (n = 5) calculated for statistical comparisons by one-way ANOVA followed by the Dunnett's test.

RESULTS
The Simultaneous Deletion of edd, ppdK, and pyk Abolishes B. suis 513 Growth on 5 and 6 C Sugars Whereas the oxidative PPP yields phosphoenolpyruvate (PEP) that is then converted into pyruvate, the EDP produces PEP and directly pyruvate (Figure 1). Since pyruvate can be converted directly into acetyl-CoA to feed the TCA, when glucose is the only C source it can be predicted: (i) that the steps connecting PEP and pyruvate should be dispensable for growth if EDP is active; and (ii) that an edd mutant defective in these steps should not grow on glucose.
According to genomic predictions, the B. suis 513 (the reference strain of the B. suis biovar 5) enzymes involved in PEP-pyruvate conversions would be a (putative) pyruvate phosphate dikinase (PpdK) and a (putative) pyruvate kinase (Pyk) (Figure 1). Therefore, as a first test for those predictions, we examined Bs5 ppdK and Bs5 pyk for growth on glucose and on peptone-glucose as a control. As can be seen in the upper left panel of Figure 2, Bs5 ppdK grew on glucose reaching the level of the parental strain with a short delay and similar generation times (about 8 h during the exponential phase). On glucose, although Bs5 pyk displayed a longer generation time (about 11 h), it also reached a stationary phase level like that of the parental strain (Figure 2, upper left panel). Both mutants grew normally in peptone-glucose (Figure 2, lower left panel). As a control, we included a mutant in edd. Whereas this Bs5 edd mutant grew normally on peptoneglucose (Figure 2, lower left panel), we observed that it grew less and more slowly than its parental B. suis 513 strain on glucose (Figure 2, upper left panel). This result confirms the functionally of the EDP and, since growth was not abrogated, it also shows a minor activity of a complementary glucose oxidative route, which should be the oxidative PPP because of the lack of phosphofructokinase and hence classical glycolysis in all brucellae.
Based on these results, we then constructed and tested the double Bs5 ppdK pyk and triple Bs5 ppdK pyk edd mutants. We found that, while growth of the double mutant Bs5 ppdK pyk was delayed but not arrested on glucose (Figure 2, upper central panel), edd became essential for growth when both ppdK and pyk were mutated (Figure 2, upper right  panel). On the other hand, these mutants grew normally in complex medium (Figure 2, lower central right panels). These results strongly suggest that PpdK and Pyk are functional, confirm that B. suis 513 has an operative ED route and are consistent with our predictions.
In the experiments presented thus far, we noticed that Bs5 ppdK grew faster than Bs5 pyk, which implies that ppdK cannot fully replace pyk when the bacteria are growing on glucose (Figure 2, upper left panel). This prompted us to investigate whether the reactions catalyzed by these two enzymes are similarly effective when combined with the EDP. For this, we compared mutants Bs5 ppdK edd and Bs5 pyk edd on glucose. Unexpectedly, the results (Figure 2, upper central panel) showed no growth for Bs5 ppdK edd and generation times for Bs5 pyk edd not very different from those of the Bs5 edd single mutant, suggesting a major and not dispensable role for PpdK. No growth defect was apparent on peptone-glucose (Figure 2, lower central panel).
We also noticed that Bs5 edd showed longer generation times and stationary phase yields lower than those of Bs5 ppdK or Bs5 pyk (Figure 2, left panel). As indicated above, growth of Bs5 edd under these conditions should occur only through the oxidative PPP, being in this regard similar to the three classical smooth Brucella spp. The PPP yields PEP through glyceraldehyde-3-P (GAP), and then PEP yields pyruvate through the Pyk and PpdK catalyzed reactions (Figure 1). Therefore, a direct proof of PPP as the only sugar catabolic route remaining in mutant Bs5 edd would be that Pyk and Ppdk become essential for growth on pentoses when EDP is not functional. We confirmed this prediction (Figure 3) by taking advantage of the ability of B. suis 513 to grow on xylose and ribose as the only C source .
The Simultaneous Deletion of edd, ppdK, and pyk Does Not Affect B. suis 513 Virulence in Mice The phenotype of the Bs5 ppdK pyk edd provided a tool to investigate whether the catabolism of 6 (and 5 C) sugars was essential during infection. To investigate this, we inoculated BALB/c mice with Bs5 ppdK pyk edd and, as controls, Bs5 ppdK and B. suis 513, and determined the CFU/spleen after 2 or 8 weeks (acute and chronic phase of infection, respectively). We found that the triple mutant Bs5 ppdK pyk edd was not attenuated in this virulence model (Figure 4).

DISCUSSION
In this work, we confirm and extend our previous results supporting the existence of an active ED route in B. suis 513 (biovar 5) that together with the oxidative PPP sustains growth of this biovar on glucose as the only C source in vitro (Machelart et al., 2020). In keeping with the prediction that these routes produce PEP/pyruvate, we also found that growth requires PpdK and Pyk, the former apparently playing a major and nondispensable role in vitro. Indeed, simultaneous dysfunction of Edd, PpdK, and Pyk also abolished the ability of B. suis 513 to grow on xylose or ribose. Indirectly, the data also confirm the lack of an active EMP pathway, consistent with the absence of phosphofructokinase in all brucellae .
It has to be noted that but for the ppdK one , the mutants investigated were not complemented despite several attempts (Lázaro-Antón, Moriyón and Zúñiga-Ripa; unpublished results). We have already detected this experimental difficulty with some Brucella mutants affected in intermediary C pathways, and this could be due to the intricacies of metabolic regulatory loops, plasmid stability, and other factors (Zúñiga-Ripa et al., 2014). However, while strict proof would require such experiments, it has to be stressed that the phenotype of the mutants studied here fully correspond with the predicted ones, which strongly suggests that the conclusions that can be drawn are valid. Also worth commenting is that, while further research is necessary to ascertain the metabolic peculiarities of the slow-and fast-growing brucellae, the observation that deletion FIGURE 4 | The triple mutant Bs5 ppdk pyk edd is not attenuated in the mouse model. Each point is the mean ± standard deviation (n = 5) of the logs of CFUs per spleen in technical triplicates. There were no statistical differences at any of the two times tested (one-way ANOVA followed by Dunnett's test; p > 0.5).
of edd considerably reduces the growth rates of B. suis 513 suggests that the shift from EDD to PPP as the major route for sugar metabolism could be one of the reasons for these different phenotypes.
On the connection virulence-metabolism in brucellae, here we examined whether glucose oxidation beyond pyruvate is necessary for B. suis 513 virulence in the mouse model, and we obtained a negative answer. Considering the in vitro phenotypes of B. suis 513 and its Bs5 ppdK pyk edd mutant, this conclusion can be extended to at least xylose and ribose, two pentoses feeding into the PPP. As discussed below, these conclusions do not exclude other uses of hexoses and/or pentoses by the brucellae.
Several works offer insight on the role of hexose/pentose metabolism in Brucella virulence. A signature-tagged mutagenesis screening in mice identified a gluP [glucose/galactose transporter; (Essenberg et al., 1997)] mutant of B. abortus 2308 among those attenuated 8 weeks after infection but not among those identified as attenuated at 2 weeks post-infection (Hong et al., 2000). This mutant, however, was not critically compromised [virulent/gluP mutant co-infection competitive index at week 8 23.4 as compared to 72.4 for a gltD (glutamate synthase) mutant tested in parallel]. In subsequent work with gluP, Xavier et al. (2013) proposed that an increased glucose availability mediated by peroxisome proliferatoractivated receptor γ (PPARγ) facilitates B. abortus 2308 survival during the chronic phase in alternative activated macrophages. Although other authors have interpreted these results to mean that glycolysis may play an important role in metabolism and virulence of intracellular Brucella (Gao et al., 2016), the multiplication of B. abortus in mouse spleens occurs early during infection [when no role for gluP was observed (Hong et al., 2000)] before the numbers of bacteria reach a short plateau after which it decreases progressively (Grilló et al., 2012). Therefore, the gluP studies suggest that glucose or galactose are used after the acute phase of infection for purposes other than major C/energy sources for multiplication in at least those laboratory models. Similar considerations can explain the mild attenuation of B. suis 1330 ribose kinase (rbsk) and 6-phosphogluconate dehydrogenase (gnd) Tn5 mutants observed in macrophage-like human THP-1 cells 48 h after infection (log CFU reduction for both mutants of 1.8 vs. 5 for genes involved in amino acid synthesis) (Köhler et al., 2002). Moreover, in our hands a B. suis 1330 mutant in gnd is severely attenuated (Machelart et al., 2020). On the other hand, other works that provide results on how virulence is affected by mutation of enzymes of hexose/pentose metabolism cannot be unequivocally interpreted in terms of metabolism. For example, a 3 log CFU attenuation was found for a B. suis 1330 P-glucose isomerase (pgi) Tn5 mutant (Foulongne et al., 2000) but, indeed, this mutation has pleiotropic effects, including that on the synthesis of mannose and hexosamine, two sugars required for lipopolysaccharide building. Similarly, a glucose-6-P dehydrogenase (zwf ) mutant of B. abortus 544 was described as completely unable to multiply in Hela cells but surprisingly the mutant was severely hampered in invasiveness (Kim et al., 2003), a phenotype that strongly suggests defects not related to its ability to multiply within cells.
Whereas our results are not in open conflict with those summarized in the previous paragraph, the lack of a role in virulence of pyk and ppdK manifested in the phenotype in mice of Bs5 ppdK pyk edd apparently contradicts conclusions obtained in other studies. Gao et al. (2016) constructed a B. abortus S2308 pyk mutant that, in contrast to the parental strain, was impaired for growth on glucose but not on pyruvate. This B. abortus pyk mutant did not multiply in RAW 264.7 macrophages and was attenuated in BALB/c mice (approximately 2.5 log CFU less than the parental strain 1 and 5 weeks after infection). More recently, Pitzer et al. (2018) reported that a B. abortus 2308 pyk mutant proved to be defective in the activity of Pyk displayed reduced ability to metabolize glucose, fructose, and galactose but not ribose, xylose, arabinose or erythritol, and was attenuated in C57BL/6 mice. The reasons for the discrepancies in both attenuation and the range of substrates used in vitro are not obvious. For the attenuation, a plausible explanation would be that these studies have been conducted in B. abortus 2308 and ours with B. suis 513. As emphasized in the Introduction, B. suis 513 (but not B. abortus 2308) is fast-growing and can use a wider menu of substrates as the only source of C and energy including lactate and glutamate, which by themselves do not support growth of B. abortus 2308W . In the host, these abilities of B. suis 513 could provide a way to circumvent the PEP → pyruvate conversion as lactate can provide pyruvate, and the TCA cycle can also be fed by glutamate . Discrepancies in the use of C sources in vitro by B. abortus in different works could be explained by subtle differences between strains 2308 and 2308W, as they are not genetically identical (Suárez-Esquivel et al., 2016) and/or by the experimental conditions. Gao et al. (2016) tested 2308 growth in a medium containing glucose or pyruvate and mineral salts but also 0.1% yeast extract, which makes the medium nonminimal and could thus account for the differences in growth on glucose of 2308 and 2308W. Also, the utilization of hexoses and pentoses by 2308 in the work of Pitzer et al. (2018) was tested in a Biolog system, which implies an undefined medium and, therefore, provides no information on the use of those substrates as the only C/energy sources. Like in the study of Gao et al. (2016) the minimal medium used by Pitzer et al. (2018) contained 0.1% yeast extract.
Regarding ppdK, we reported in a previous work that this gene is necessary for B. abortus 2308W virulence (Zúñiga-Ripa et al., 2014), and this was confirmed in the study of Pitzer et al. (2018). Recently, we showed that the homologous ppdK mutant in B. suis 513 was not attenuated in mice and we elucidated the reasons for this discrepancy: this B. suis biovar 5 strain can use PpdK and PEP carboxykinase (PckA) for PEP synthesis in vitro (Figure 1), PckA catalyzing oxaloacetate conversion into PEP, while B. abortus 2308W uses only PpdK, that catalyzes the PEP-pyruvate interconversion . Indeed, we showed that B. suis 513 attenuation occurs in the double PckA-PpdK mutant .
Consistent with the absence of phosphofructokinase (Pfk) in all brucellae and the dispensability of the EDP, our results are in line with the hypothesis that hexose/pentose catabolism through the TCA is not necessary for growth in BCVs, which may account for the loss of EDP in B. abortus, B. melitensis, and some B. suis biovars. As PpdK catalyzes an amphibolic reaction that can support the synthesis of 3 to 6 C biosynthetic precursors and Pyk is catabolic, this could explain why the former seems to have a more important role for growth of at least B. suis 513. The data presented here together with those of previous works are consistent with a model (Zúñiga-Ripa et al., 2014 in which the brucellae thrive intracellularly by using 3 and 4 C substrates with a limited supply of 5 and 6 C sugars that are devoted to biosynthesis.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by the Animal Welfare Committee of CITA.

AUTHOR CONTRIBUTIONS
AZ-R, JL, MI, and IM conceived and coordinated the study. LL-A and AZ-R carried out the genomic analyses, mutants construction, and metabolic tests. MM, TB, RC-Á, and PM contributed in mutant construction, growth measurements, and experiments in mice. AZ-R, LL-A, and IM wrote the manuscript. All authors analyzed the results and approved the manuscript content.