Cross-talk between Type Three Secretion System and Metabolism in Yersinia*

Pathogenic yersiniae utilize a type three secretion system (T3SS) to inject Yop proteins into host cells in order to undermine their immune response. YscM1 and YscM2 proteins have been reported to be functionally equivalent regulators of the T3SS in Yersinia enterocolitica. Here, we show by affinity purification, native gel electrophoresis and small angle x-ray scattering that both YscM1 and YscM2 bind to phosphoenolpyruvate carboxylase (PEPC) of Y. enterocolitica. Under in vitro conditions, YscM1, but not YscM2, was found to inhibit PEPC with an apparent IC50 of 4 μm (Ki = 1 μm). To analyze the functional roles of PEPC, YscM1, and YscM2 in Yop-producing bacteria, cultures of Y. enterocolitica wild type and mutants defective in the formation of PEPC, YscM1, or YscM2, respectively, were grown under low calcium conditions in the presence of [U-13C6]glucose. The isotope compositions of secreted Yop proteins and nine amino acids from cellular proteins were analyzed by mass spectrometry. The data indicate that a considerable fraction of oxaloacetate used as precursor for amino acids was derived from [13C3]phosphoenolpyruvate by the catalytic action of PEPC in the wild-type strain but not in the PEPC- mutant. The data imply that PEPC is critically involved in replenishing the oxaloacetate pool in the citrate cycle under virulence conditions. In the YscM1- and YscM2- mutants, increased rates of pyruvate formation via glycolysis or the Entner-Doudoroff pathway, of oxaloacetate formation via the citrate cycle, and of amino acid biosynthesis suggest that both regulators trigger the central metabolism of Y. enterocolitica. We propose a “load-and-shoot cycle” model to account for the cross-talk between T3SS and metabolism in pathogenic yersiniae.

Type three secretion systems (T3SSs) 3 are used by several Gram-negative bacteria as microinjection devices to deliver effector proteins into host cells (1). The translocated effector proteins reprogram the host cell in favor of the microbial invader or symbiont. Pathogenic yersiniae (the enteropathogenic Yersinia enterocolitica and Yersinia pseudotuberculosis and the plague bacillus Yersinia pestis) utilize a plasmid-encoded T3SS to undermine the host primary immune response (2). This is mediated by the injection of a set of effector proteins called Yops (Yersinia outer proteins) into host cells, in particular into cells with innate immune functions, such as macrophages, dendritic cells, and neutrophils (3). The concerted action of Yops, targeting multiple signaling pathways, results in actin cytoskeleton disruption, suppression of proinflammatory signaling, and induction of apoptosis. This strategy enables yersiniae to multiply extracellularly in host tissue.
Expression of the Yersinia T3SS is up-regulated at 37°C, and translocation of Yops across the host cell membrane is triggered by cell contact (4,5). Pathogenic yersiniae cultivated under low calcium conditions at 37°C express a phenotype referred to as "low calcium response" (LCR). The LCR is characterized by growth restriction as well as massive expression and secretion of Yops into the culture medium (6 -9). The allocation of energy and metabolites for the massive synthesis and transport of Yops is demanding, and this burden is believed to be responsible for the observed growth inhibition (10). To give an idea of the metabolic requirements, Yops are secreted to the culture supernatant in 10-mg amounts per liter of culture within 2 h after calcium depletion of the medium. Furthermore, post-translationally secreted substrates need to be unfolded by a T3SS-specific ATPase prior to secretion (11)(12)(13)(14). In addition, T3SS-dependent transport of Yops requires the proton motive force (15). However, there is evidence that growth cessation and Yop expression can be uncoupled (16,17), suggesting a coordinated regulation of metabolism and protein transport rather than the LCR reflecting an inevitable physiological consequence.
What are the candidate proteins that could be involved in such a coordination? YscM1 and YscM2 (57% identical to YscM1) are key candidates, since they act at a major nodal point of the T3SS regulatory network in Y. enterocolitica. In Y. pestis and Y. pseudotuberculosis, only the YscM1 homologue LcrQ exists (99% identical to YscM1). YscM1/LcrQ and YscM2 are secretion substrates of the T3SS that are involved in up-regulation of Yop expression after host cell contact. Upon cell contact, the decrease of intracellular levels of YscM1/LcrQ and YscM2 due to their translocation into host cells results in a derepression of Yop synthesis (18 -21). The two yscM copies of Y. enterocolitica were presumed to be functionally equivalent, since deletion of either gene was found to be phenotypically silent (19,22). Only deletion of both yscM genes could establish the lcrQ phenotype (19,22), distinguished by temperature sensitivity for growth, derepressed Yop expression, and secretion of LcrV and YopD in the presence of calcium ions.
YscM1/LcrQ as well as YscM2 exhibit homology to the N terminus of the effector YopH (19,23,24), a fact that may explain their shared assistance by SycH (specific Yop chaperone) (20,25). It was shown that YscM1/LcrQ and YscM2 exert their influence on Yop expression in concert with the T3SS components SycH, SycD (LcrH in Y. pestis and Y. pseudotuberculosis), and YopD (21, 26 -28). It is further described that YscM1 and/or YscM2 interact with several of the T3SS-specific chaperones, in particular with SycH, SycE, SycD, and SycO (20, 29 -31). This has led to the model that YscM/LcrQ proteins might function as an interface that senses whether chaperones are loaded with Yops and transduces these signals into control of Yop expression (14).
These features of YscM1/LcrQ and YscM2 prompted us to speculate about a key role of these proteins in coordination of metabolism and expression of T3SS components. Using recombinant GST-YscM1 and GST-YscM2 as bait for Y. enterocolitica cytosolic proteins, we identified phosphoenolpyruvate carboxylase (PEPC) as interaction partner of both YscM1 and YscM2. Under in vitro conditions, YscM1 down-regulated PEPC activity and bacterial growth/replication. Isotopologue profiling of Yop proteins and derived amino acids from Y. enterocolitica grown in the presence of [U- 13 C 6 ]glucose showed the functionality of the PEPC reaction under virulence conditions (isotopologues are molecular entities that differ only in isotopic composition (number of isotopic substitutions); e.g. CH 4 , CH 3 D, and CH 2 D 2 ). Moreover, biosynthetic rates of amino acids were increased in mutants defective in YscM1 or YscM2, suggesting a general role of these regulators in the metabolism of Y. enterocolitica. Recently, evidence has been accumulating that the metabolic state contributes to the regulation of T3SSs of diverse pathogens, also including the flagellar T3SS in Pseudomonas and Salmonella (32)(33)(34)(35)(36).

EXPERIMENTAL PROCEDURES
Cloning and Mutagenesis-For details on construction of plasmids and mutants, see below. Overproduction of YscM1 and YscM2 was based on plasmid pWS (37) and controlled by an isopropyl ␤-D-thiogalactoside-inducible tac promoter. For purification of recombinant Yersinia PEPC, a GST fusion construct based on pGEX-4T3 (Amersham Biosciences) was used. Plasmids for expression of GST-YscM1 and GST-YscM2 fusions have been described (38). The ppc mutant was generated as described (15) by replacement of the ppc gene with a kanamycin resistance cassette mediated by homologous recombination of a transformed PCR product. The primers used were 5Ј-CTAACAACCCTGCGGCGTCAAGGGCGAA-GGGGATATGGGTCAGGGGTCATTCACTGACACCCTC-ATCAGTG-3Ј and 5Ј-ACGGGCGCCTGGCCCGTTTTGTC-TTTTATTCACGATAATTCTACTGGCAACGTCAAGTC-AGCGTAATGCTC-3Ј. Construction of yscM1 and yscM2 deletion strains was previously described (31). After cleavage of pWS with NdeI and SalI, appropriate inserts with PCR-introduced NdeI and SalI sites, respectively, were ligated. As template, DNA from Y. enterocolitica WA-314 was used. For construction of pWS-YscM1 primers, 5Ј-CATATGAAAATCAA-TACTCTTCAATCG-3Ј and 5Ј-GTCGACTCAGCCGTCAG-CCG-3Ј were used; for pWS-YscM2 construction, primers 5Ј-CATATGGGGAGCATTATGAAAAATAAACG-3Ј and 5Ј-GTCGACTTAAAGCTTTTGCATTTTCCGTGTC-3Ј were used. Primer design for YscM1 and YscM2 constructs was based on sequences with accession number AY150843.
Protein Expression and Purification-YscM1 and YscM2 were fused to GST based on two vectors, pGEX-4T3 and pGEX-6P3 (Amersham Biosciences). GST fusions based on pGEX-4T3, which harbor a thrombin cleavage site, were used for initial PEPC binding studies (affinity purification and native gel electrophoresis). For production of recombinant YscM1/ YscM2 without GST, pGEX-4T3-based constructs cannot be used, since YscM1/YscM2 harbor an intrinsic thrombin cleavage site (38). Therefore, recombinant YscM1 and YscM2 were produced as described using pGEX-6P3 as vector (38). Note, however, that GST-YscM1 and GST-YscM2 based on pGEX-6P3 were found to be incompetent to bind PEPC.
Affinity Purification and Identification of PEPC-Glutathione-Sepharose beads preloaded with GST fusions or GST as a control were incubated with lysates of Y. enterocolitica cells (WA-314, harboring the virulence plasmid pYV, and WA-C, a WA-314 derivative cured of pYV). Yersiniae were therefore cultured in brain heart infusion (BHI) medium at 27°C (no expression of the pYV-encoded T3SS) and in parallel at 37°C in the presence of 5 mM EGTA and 10 mM MgCl 2 (expression and secretion of Yops). Harvested cells were resuspended in PBS and disrupted by French press treatment. Soluble fractions after centrifugation (20,000 ϫ g, 30 min, 4°C) were incubated with beads preloaded as described above at 4°C for 1 h, and then beads were washed twice with phosphate-buffered saline and subjected to SDS-PAGE analysis. Coomassie-stained bands of interest were excised from the gel. Matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry (Reflex III; Bruker Daltonics GmbH, Bremen, Germany) was conducted as described recently (39) to identify PEPC as the ligand interacting with GST-YscM1 and GST-YscM2.
Native Gel Electrophoresis-Native gel electrophoresis was performed as described (40).
PEPC Activity Assay-PEPC activity was measured in a coupled reaction with malate dehydrogenase, as reported earlier (41). Assay components were 100 mM Tris-HCl, pH 8.0, 10 mM NaHCO 3 , 10 mM MgCl 2 , 0.2 mM NADH, 5 mM PEP, 1 unit of malate dehydrogenase, and 5 milliunits of PEPC. The reaction was initialized by the addition of PEP, and assay mixtures were incubated at 27°C. The decrease of NADH absorption was monitored at 340 nm in a 96-well Nunc plate using a microplate reader (Biomolecular Devices; SpectraMax software). ⌬A 340 was determined by linear regression (typically, the correlation coefficient was 0.98 or above). We ruled out an effect of YscM1 on activity of malate dehydrogenase in the coupled reaction.
Growth under PEPC-requiring Conditions-Yersiniae (Y. enterocolitica WA-314) harboring plasmids pWS-YscM1, pWS-YscM2, and pWS were grown in rich medium overnight (BHI medium supplemented with 1% glucose) 1 day after selection of transformants. The next day, bacteria were pelleted, washed, and resuspended in M9 minimal salts supplemented with 1% glucose, 0.01% casamino acids, and 10 g/ml thiamine. Under these conditions, the ppc mutant was unable to grow. 1 mM isopropyl ␤-D-thiogalactoside was added to all cultures at time 0 to induce overproduction of YscM1/YscM2 or as a control (WA-314 pWS). All experiments were performed in triplicate from independent cultures. Growth curves were determined by measuring optical density at 600 nm. Expression of the respective protein was controlled by Western blotting using polyclonal antiserum. It appeared crucial that these growth studies were performed the day after selection of transformants and that catabolite repression was maintained by the addition of glucose to selection plates and to overnight cultures to repress the tac promoter at best. Later analyses frequently resulted in a loss of YscM1 and YscM2 expression, suggesting a strong selection pressure of protein production. 13 C Labeling and Mass Spectrometry Analysis of Secreted Yops-Yersinae overnight cultures grown at 27°C in BHI medium were centrifuged, and the pellet was washed twice in F-12 cell culture medium (Invitrogen; composition given in Table S1). Pellets were resuspended in F-12 medium, and suspensions were adjusted to an optical density (600 nm) of 0.15. F-12 medium was then supplemented with 2% [U-13 C 6 ]glucose or alternatively with 2% unlabeled glucose. To initially suppress secretion of Yops, the medium was supplemented with 0.2 mM CaCl 2 . Cultures (3 ml) were grown for 2 h at 37°C to stimulate expression of the T3SS, and then cultures were stimulated to secrete Yops by the addition of 5 mM EGTA and 10 mM MgCl 2 . After 2 h of additional incubation at 37°C, supernatants were collected by centrifugation, and the secreted proteins were precipitated with 10% tricarboxylic acid. Following separation on denaturing SDS gels, the Coomassie-stained protein bands corresponding to YopM, YopE, and YopH were excised from the gel, and the proteins were subjected to in-gel tryptic digestion and analyzed by MALDI-TOF mass spectrometry. To this end, Coomassie-stained protein bands from SDS gels were washed twice with water and twice with 40 mM NH 4 HCO 3 buffer, pH 8.0. After 2-fold treatment with 50% acetonitrile for 5 min, trypsin (sequencing grade modified; Promega) was added, and proteins were digested overnight in 40 mM NH 4 HCO 3 buffer, pH 8.0, at 37°C while shaking. Probes were directly used for MALDI-TOF experiments. 10 l of each sample were first purified and concentrated on a C18 reversed phase pipette tip (Zip-Tip; Millipore). The peptides were eluted with 1 l of ␣-cyano-4-hydroxycinnamic acid (Sigma) and then directly spotted on a MALDI sample plate (Applied Biosystems). MALDI-TOF measurements were performed on a Voyager-DE STR TOF mass spectrometer (Applied Biosystems). The resulting spectra where displayed using the Data Explorer Software (Applied Biosystems).
When we became aware of the seemingly strong selection pressure relating to the YscM1 and YscM2 expression constructs (see above), we reasoned that the yscM1 and yscM2 deletion mutants may rapidly change genetically. On that account, we isolated the virulence plasmid of the yscM1 and yscM2 mutant, respectively, and transformed it into the plasmidcured derivative WA-C of the parental strain WA-314, as recently described for Yop mutants (42).

C Labeling and Gas Chromatography-Mass Spectrometry
Analysis of the Protein-derived Amino Acids-Yersinae cultures (200 ml) grown at 27°C in BHI medium overnight were centrifuged, and the pellet was washed twice in F-12 cell culture medium (Invitrogen). Pellets were resuspended in F-12 medium, and 5-liter suspensions were adjusted to an optical density (600 nm) of 0.15. F-12 medium was supplemented with 0.2% [U-13 C 6 ]glucose or alternatively with 0.2% unlabeled glucose. The medium was supplemented with 0.2 mM CaCl 2 to suppress secretion of Yops initially. 5-liter cultures were grown in a pH-controlled BIOSTAT B fermenter for 2 h at 37°C, and then cultures were stimulated to secrete Yops by the addition of 5 mM EGTA and 10 mM MgCl 2 . After 2 h of additional incubation at 37°C, bacterial cells were collected by centrifugation, frozen, and lyophilized. For gas chromatography/mass spectrometry (GC/MS) analysis, 5 mg of bacterial cell mass were hydrolyzed in 6 M HCl. The hydrolysate was dried and treated with N-(tert-butyldimethylsilyl)-N-methyl-trifluoroacetamide containing 1% tert-butyldimethylsilylchloride. The resulting tert-butyldimethylsilyl derivatives of amino acids were subjected to GC/MS, as described earlier (43). Samples were analyzed at least three times. Data were processed as described before (44), affording the molar excess of carbon isotopomer groups (M ϩ 1, M ϩ 2, . . . M ϩ n, with M being the mass ion of the fragment under study and n being the number of carbon atoms of the respective fragment). The excess of multiple labeled isotopologues was calculated according to the equation,

JOURNAL OF BIOLOGICAL CHEMISTRY 12167
Small Angle X-ray Scattering Experiments and Data Analysis-The synchrotron radiation x-ray scattering data of PEPC alone, PEPC in complex with YscM1, and PEPC in complex with YscM2 were collected following standard procedures on the small angle scattering beamline X33 (45) of the EMBL Hamburg on the storage ring DORIS III of the Deutsches Elektronen Synchrotron. Details of data recording and processing are described in the supplemental material. The PEPC scattering was recorded at 2.4 mg/ml. PEPC-YscM1 and PEPC-YscM2 complexes were measured at a PEPC concentration of 0.6 and 1.6 mg/ml of the respective YscM protein. Protein solutions were equilibrated to phosphate-buffered saline using PD-10 cartridges for buffer exchange (GE Healthcare).

YscM1 and YscM2
Interact with PEPC-In order to identify interaction partners of YscM1 and YscM2, we incubated glutathione S-transferase (GST) fusions GST-YscM1 and GST-YscM2 immobilized on glutathione-Sepharose beads with different yersiniae lysates. To this end, Y. enterocolitica strain WA-314 (9), harboring the virulence plasmid pYV, and a plasmid-cured derivative (WA-C) were cultivated under conditions either permissive or restrictive for Yop expression and secretion (46). Beads were analyzed by denaturing gel electrophoresis (Fig. S1). We detected one band of ϳ95 kDa that was associated with GST-YscM1 and GST-YscM2 samples but not with GST alone. The appearance of this band did not depend on expression of the T3SS or the presence of the virulence plasmid, indicating a chromosomally encoded protein bound to YscM1 and YscM2. This protein was identified by mass spectrometry to be PEPC (data not shown).
To prove the relevance of this finding, we recombinantly expressed Y. enterocolitica PEPC in Escherichia coli and purified the protein to facilitate interaction studies. Subsequently, we performed native gel electrophoresis to test binding of recombinant PEPC to recombinant YscM1 and YscM2, respectively. Fig. 1A displays a Coomassie-stained native gel, demonstrating that GST-YscM1 causes a mobility shift of PEPC. PEPC mobility was also shifted in the presence of YscM1 released from GST (Fig. 1B). Electrophoretic mobility of PEPC was only slightly retarded when it was incubated with YscM2 prior to gel loading. Therefore, the native gel was electroblotted to verify the YscM2/PEPC interaction (Fig. 1C). The same blot was developed with anti-PEPC and anti-YscM2 sera successively. YscM2, when loaded alone, could not be detected, because YscM2 did not enter the gel under the conditions applied (38). This is due to the basic isoelectric point of YscM2 (9.79) and a pH of 7.4 of the native gel system. However, when a mixture of PEPC and YscM2 was applied on the gel, YscM2 co-localized with PEPC and could be detected after immunoblotting. Taken together, recombinant PEPC interacts with recombinant YscM1 and YscM2, respectively. PEPC Activity Is Inhibited by YscM1 but Not by YscM2-PEPC catalyzes the carboxylation of phosphoenolpyruvate to form oxaloacetate in many microorganisms (45). This so-called anaplerotic reaction replenishes the tricarboxylic acid cycle intermediate oxaloacetate. Tricarboxylic acid cycle intermediates (oxaloacetate and ␣-ketoglutarate) are required for the bio-synthesis of amino acids of the aspartate and glutamate families.
We analyzed the activity of PEPC in the presence of YscM1 or YscM2. PEPC enzymatic activity was determined by coupling the PEPC reaction to malate dehydrogenase and monitoring the rate of NADH oxidation spectrophotometrically at 340 nm (41). We observed a titratable inhibitory effect of YscM1 on PEPC activity (Fig. 2) and determined an IC 50 value of 4 M for YscM1. Determination of the K i (1 M) further revealed a competition between YscM1 and the substrate PEP (Fig. S2). By comparison, PEP analogues are described that inhibit homologous E. coli and plant PEPC with K i values ranging from 18 to 85 M (48 -50), and allosteric inhibitors of PEPC, such as aspartate, inhibit PEPC in the millimolar range (51) (data not shown). To further evaluate the biological impact of the YscM1 inhibitory effect, we examined whether the PEPC inhibitor YscM1 FIGURE 1. Native gel electrophoresis reveals binding of PEPC to YscM1 and YscM2, respectively. Recombinantly expressed and purified proteins were mixed, incubated at room temperature for 5 min, and loaded on 6% HEPES-buffered native polyacrylamide gels, as indicated. A and B, Coomassiestained native gels; C, Western blot analysis after native gel electrophoresis. A, GST-YscM1 (dimeric) was in 2-fold excess over PEPC (tetrameric). The arrowhead indicates formation of GST-YscM1-PEPC complexes. The excess load of GST demonstrates that GST on its own exhibits no PEPC binding activity. B, YscM1-PEPC complex formation is indicated by an arrowhead. YscM1 (dimeric) was in 3-fold excess over PEPC (tetrameric). C, Western blot analysis of a native gel to demonstrate the YscM2/PEPC interaction. Following electroblotting, the same blot was developed with anti-PEPC and with anti-YscM2 sera successively. Data are representative of 3-5 independent experiments. FIGURE 2. Effect of YscM1 on PEPC activity. PEPC activity was monitored spectrophotometrically by coupling the PEPC reaction with the malate dehydrogenase (MDH) reaction recording NADH oxidation at 340 nm. Relative PEPC activities ⌬A i /⌬A 0 in the presence of varying concentrations of YscM1 were plotted. All proteins were recombinantly expressed and purified to homogeneity. The inhibitory effect of YscM1 on PEPC activity was invariably demonstrated by four experimenters independently. could antagonize acetyl-CoA, an allosteric activator of PEPC. In the presence of 50 M acetyl-CoA, PEPC activity was stimulated by a factor of 2 relative to the PEPC control (100%), but only 10 M of YscM1 was sufficient to fully compensate this stimulation (Fig. S3).
In contrast to YscM1, no significant influence of YscM2 on PEPC activity could be observed. Also, in the presence of YscM1, aspartate, or acetyl-CoA, we could not observe effects of YscM2 on PEPC activity (data not shown). In conclusion, our data suggest that YscM1 is a biologically relevant inhibitor of PEPC. Interestingly, although YscM2 clearly interacts with PEPC, we could not prove any significant influence of YscM2 on PEPC activity.
Small Angle X-ray Scattering Data Suggest That YscM1 and YscM2 Trigger Different Conformational Transitions of PEPC-Small angle x-ray scattering (SAXS) provides information about the overall shape of macromolecules in solution. Assuming that the interaction of YscM proteins with the allosteric enzyme PEPC could result in measurable conformational changes, we applied SAXS to study these interactions in solution. Fig. 3A shows three different views of the PEPC homotetramer calculated from the scattering data with the constraint of a p22 symmetry. This assumption is based on the results for the radius of gyration, molecular weight analysis, and ab initio shape reconstructions without any symmetry constraint. Further, this assumption is in line with the crystal structure of E. coli PEPC (52), which exhibits 80% identity with the Yersinia PEPC. This crystal structure shows that the homotetramer consists of two dimerized dimers, with the two dimers twisted relative to each other. The twist of the two dimers visible in the E. coli high resolution structure is also represented in our SAXS-based model of the Yersinia PEPC and is best illustrated in the middle and right panels of Fig.  3A. Fig. 3, B and C, shows the model of PEPC in the presence of YscM1 and in the presence of YscM2, respectively. YscM1 and YscM2 both cause dramatic conformational changes that are clearly distinct from one another. Similarly, however, YscM1 and YscM2 both induce conformational changes that reduce the twisting of the two dimers, leading to a more planar arrangement of the four subunits. These conformational changes may represent transitions from a relaxed "R" state to a tense "T" state, which are assumed to control the activity of allosteric enzymes. Due to the dramatic extent of the conformational changes induced by the YscM proteins and due to their small size relative to PEPC, it was not possible to draw any solid conclusion about the binding site of either YscM protein. Overall, the allosteric enzyme PEPC undergoes significant conformational transitions in the presence of either YscM1 or YscM2, further underscoring the biological relevance of these interactions. In particular, these data suggest that YscM2 could modulate PEPC activity under certain conditions that were not covered by our PEPC activity assay.
YscM1 Overproduction Causes Growth Restriction under Conditions Requiring PEPC-We reasoned that overproduction of YscM1 in Y. enterocolitica under conditions requiring PEPC activity should influence growth rates if YscM1 were to be an inhibitor of PEPC. It is known that in E. coli, PEPC is required when cells are grown on glucose as the sole carbon source (47,53). Applying the Red recombinase technique (15), we generated a Y. enterocolitica mutant deleted of the PEPCencoding gene ppc. The ppc mutant exhibited no significant proliferation on M9 minimal salts supplemented with 1% glucose, 0.01% casamino acids, and 10 g/ml thiamine over a period of 7 h. By contrast, the parental strain and the trans- complemented ppc mutant grew well in the same medium, defining these conditions as PEPC-requiring (Fig. S4). The same conditions were chosen to dissect the effect of YscM1 and YscM2 overproduction on the growth rate of Y. enterocolitica. We constructed plasmids to overproduce YscM1 and YscM2 in yersiniae. Y. enterocolitica WA-314 was transformed with plasmid pWS-YscM1 or pWS-YscM2 or the control plasmid pWS and grown overnight in rich medium. The next day, bacteria were washed and resuspended in minimal medium composed as above and grown for several h at 27°C, thus avoiding thermal stimulation of the T3SS regulon. Fig. 4A shows that yersiniae overproducing YscM1 were subjected to massive growth restriction, whereas overproduction of YscM2 had no significant effect on growth rates. Overproduction of YscM1 and YscM2 was confirmed by immunoblotting (Fig. 4B), and growth restriction was controlled by plating culture aliquots on solid media to determine colony forming units (data not shown). These results are in accordance with the in vitro analyses presented above and suggest that YscM1 also functions as an inhibitor of PEPC activity in vivo. As a further conclusion, the growth restriction mediated by YscM1 overproduction does not require the presence of additional components of the Yersinia T3SS, since growth was analyzed at 27°C, which suppresses expression of the T3SS. In support of this conclusion, overproduction of YscM1 also repressed growth in E. coli BL21 (Fig. S5).
Mass Spectrometry Reveals that PEPC Reaction Affects Yop Biosynthesis-Assuming that YscM1 and possibly YscM2 manipulate amino acid biosynthesis pathways (particularly that of the aspartate family) via the PEPC reaction, the origin of amino acids incorporated into Yops should differ in the ppc, yscM1, and yscM2 mutant strains compared with the wild type strain. To challenge this model, we cultured the strains in F-12 cell culture medium supplemented with 2% [U-13 C 6 ]glucose, induced Yop secretion by chelating calcium ions in the medium, and analyzed secreted Yops by MALDI-TOF mass spectrometry. The rationale for the choice of the medium was that it should contain sufficient amounts of amino acids to allow growth and T3 secretion of the ppc mutant but that it should be meager enough for the PEPC reaction to become limiting. This was indicated by slightly subdued growth of the ppc mutant in F-12 medium compared with the parental strain (data not shown). As expected, supplementation of the F-12 medium with [U-13 C 6 ]glucose led to a characteristic right shift and broadening of the mass spectra of YopE, YopH, and YopM proteins, which were secreted in large amounts and could be purified by SDS-PAGE ( Fig. 5 and Figs. S6 and S7, compare A and B). The shift toward higher masses and the broadening in the mass patterns of a given fragment reflect that a fraction of amino acids used for Yop biosynthesis was made de novo using precursors derived from the proffered [U-13 C 6 ]glucose. However, the mass shift of less than 50 mass units also shows that the major fraction of Yop amino acids was still unlabeled and either synthesized from unlabeled precursors or recruited from unlabeled amino acids present in the medium.
As shown in Fig. 5C, the MS traces of 13 C-labeled YopM from the ppc mutant clearly differed from the corresponding signals of 13 C-labeled YopM from the parental strain grown under the same conditions. Specifically, a shift of the mass/charge ratios of peptide fragments toward lower values was observed. The same left shift was detected for YopH and YopE fragments from the ppc mutant (Figs. S6 and S7, compare B and C). We conclude that the PEPC reaction contributed to the biosynthesis of oxaloacetate-derived amino acids in the parental strain and that loss of this reaction in the ppc mutant was compensated by an increased uptake of unlabeled amino acids from the medium.
Using the same experimental approach, a possible influence of yscM1 or yscM2 deletion on PEPC activity and on Yop biosynthesis was analyzed. To rule out mutations in the chromosome of the yscM1 and yscM2 mutant strains, the mutated plasmids were transformed into the WA-C strain cured of its virulence plasmid. Transformants were subjected to the labeling experiment as described above, and secreted Yops were evaluated by mass spectrometry. The spectra of Yop peptides obtained from the yscM1 and yscM2 mutants were similar (Figs. 5, D and E, for YopM), but clearly differed from those of the parental strain (Fig. 5B) and the ppc mutant (Fig. 5C). Specifically, broadened mass patterns resembling the corresponding patterns in the spectra of the parental strain were observed. Yersiniae harboring plasmids pWS-YscM1, pWS-YscM2, and pWS were grown in rich medium overnight. The next day, bacteria were resuspended in M9 minimal salts supplemented with 1% glucose, 0.01% casamino acids, and 10 g/ml thiamine. 1 mM isopropyl ␤-D-thiogalactoside was added to all cultures at time 0 to induce overproduction of YscM1/YscM2 or as control (pWS). The optical density was determined at 600 nm (A), and bacteriostasis was checked by plating and colony counting. All experiments in triplicate from independent cultures (error bars, Ϯ one S.D.). B, to control overproduction of YscM proteins, whole cell lysates from all cultures were loaded on denaturing SDS gels. After electroblotting, the membranes were developed with a serum raised against YscM1 or with a serum raised against YscM2, respectively. The experiment as represented here was repeated more than five times with similar results.
However, a small but significant shift toward lower masses indicated that both YscM1 and YscM2 influenced carbon fluxes from exogenous [U-13 C 6 ]glucose to amino acids used for Yop formation. The same conclusion could be drawn from observations in the MS spectra of YopE and YopH fragments (Figs. S6 and S7, compare B-E). 13

C Incorporation into Bacterial Amino Acids When [U-13 C 6 ]Glucose Was Provided during Yop Biosynthesis-In
order to analyze the role of PEPC and YscM proteins in the synthesis of specific amino acids in cellular proteins, we used isotopologue profiling monitored by gas chromatography/mass spectrometry (43,54). More specifically, Y. enterocolitica WA-314 and mutants thereof defective in PEPC, YscM1, or YscM2, respectively, were grown in F-12 cell culture medium containing 0.2% [U- 13 C 6 ]glucose in a fermenter. After 2 h of growth, EGTA was added in analogy to the experiment designed for the MALDI-TOF analysis of Yop fragments. After 4 h, the bacteria were harvested in the late logarithmic state. Whole cell protein was hydrolyzed under acidic conditions, and the resulting amino acids were converted into tert-butyldimethylsilyl derivatives, which were subjected to GC/MS. The 13 C excess values of nine amino acids were determined from the mass ratios of selected ions (55). The data are summarized graphically in Fig. 6. The data are also shown numerically in Table S2.
In the experiment with the parental strain, the highest 13 C excess (2-3% 13 C) was found in Ala and Ser, followed by Asp, Gly, Ile, and Val (1-2%). Lower levels (0.6%) were observed in Phe. No incorporation was detected for Lys and Glu (Ͻ0.2%). The 13 C enrichments document that the transfer of label from the proffered 13 C-labeled glucose into the amino acids under study varied over a wide range under the experimental conditions. Probably, this is a consequence of the complex medium containing unlabeled amino acids (Table S1) that are incorporated into bacterial amino acids at different rates. With the exception of Lys and Glu, all amino acids under study acquired some label indicating that a fraction of Ala, Ser, Asp, Gly, Ile, Val, and Phe had been made de novo from exogenous [U-13 C 6 ]glucose.
In the experiment with the ppc mutant, Asp, Lys, and Glu were unlabeled (Fig. 6). The lack of 13 C enrichment of Asp is in sharp contrast to the experiment with the parental strain, FIGURE 5. MALDI-TOF mass spectrometry of peptide fragments derived from secreted YopM after feeding of yersiniae cultures with uniformly 13 C-labeled glucose. Y. enterocolitica WA-314 and mutant strains ppc, yscM1, and yscM2, respectively, were grown in BHI medium overnight. The next day, bacteria were resuspended in F-12 medium supplemented with 0.2 mM CaCl 2 and 2% [U-13 C 6 ]glucose ( 13 C) or 2% unlabeled glucose ( 12 C). Cultures were incubated at 37°C for 2 h, and then Yop secretion was induced, and cultivation was continued for 2 h at 37°C. The supernatant of the cultures was collected, and proteins were precipitated by tricarboxylic acid and subjected to SDS-PAGE. The gel was Coomassie-stained, and the protein band corresponding to YopM was excised to perform MALDI-TOF mass spectrometry analysis. A, spectra of two YopM fragments from the parental strain WA-314 grown without labeling. where significant 13 C label was detected in Asp. The 13 C excess in Ile from the ppc mutant was also lower than in Ile from the parental strain, whereas the values of the other amino acids under study were similar or even slightly higher than the corresponding values in amino acids from the parental strain. Obviously, the de novo biosynthesis of Asp using 13 C-labeled building blocks was stopped when the PEPC reaction was defective. Under these conditions, unlabeled Asp obtained from the culture medium appears to be used for protein biosynthesis. Moreover, the de novo biosynthesis of Ile was significantly reduced in the mutant, and a higher fraction of unlabeled Ile was taken from the medium. The fact that Asp but not Ala and Ser were unlabeled in the ppc mutant reveals that there is no detectable carbon flux to oxaloacetate and its downstream product Asp from 13 C-labeled C 3 metabolites (such as pyruvate and phosphoglycerate) serving as precursors for Ala and Ser. Similarly, no 13 C flux is observed into oxaloacetate/Asp via the citrate cycle using 13 C-labeled acetyl-CoA as a precursor (compare Figs. 8 and 9). Thus, the PEPC reaction is the predominant or even exclusive route leading to oxaloacetate and its downstream products under the experimental conditions.
In the experiment with mutants defective in the formation of YscM1 or YscM2, respectively, the 13 C excess values of all amino acids under study were similar (Fig. 6). This is in line with the results from the mass traces of Yop fragments (see above). In comparison with the labeling values observed in the experiments with the ppc mutant and the parental strain, all amino acids under study acquired more 13 C label. This also holds true for Asp and related amino acids where no or lower values were detected in the ppc mutant (see above). This modulation could reflect that some inhibitory effect of YscM1 or YscM2 on the PEPC reaction is lost with the deletion of YscM1 or YscM2, respectively; however, the increased de novo biosynthetic rates of many amino acids under study point to an additional and/or more general regulatory role of YscM1 and YscM2 on the central metabolic network.
More information was obtained by a detailed analysis of the isotopomer distributions in tert-butyldimethylsilyl-amino acids (for a detailed description of the method, see Refs. 43 and 44). Numerical values of 13 C excess of isotopomer groups are given in Table S3. The ratios of 13 C-labeled isotopomers carrying either two (M ϩ 2) or three (M ϩ 3) 13 C atoms in Ala, Asp, Ser, and Phe from the parental strain and the mutants are displayed in Fig. 7.
Within the limits of experimental accuracy, the ratios were identical in amino acids from YscM1 and YscM2 mutants, respectively. Again, this indicates that YscM1 and YscM2 serve similar roles in the modulation of the metabolism. In all experiments, Ser was characterized by the (M ϩ 3) isotopologue (Fig. 7C), which can be explained by conversion of [U-13 C 6 ]glucose into [U-13 C 3 ]Ser via [U-13 C 3 ]triose phosphate and [U-13 C 3 ]phosphoglycerate, which can be formed by reactions of glycolysis (Fig. 8) and/or the Entner-Doudoroff pathway. Indeed, all genes required for these pathways have been identified in Y. enterocolitica (56) (KEGG data base). The fact that (M ϩ 2)-Ser is apparently absent shows that the alternative pathway of Ser formation (i.e. hydroxymethylation of [U-13 C 2 ]Gly) did not contribute to Ser biosynthesis. As shown in Fig. 7C, the isotopomer profile of Ser was very similar in the parental strain and all mutants under study. This suggests that the isotopomer distribution was also similar in the Ser precursors, such as triose phosphate and phosphoglycerate, and it can be concluded that carbon flux from the labeled glucose precursor to triose phosphate and phosphoglycerate was not affected by deletion of PEPC, YscM1, or YscM2, respectively (compare Figs. 8 -10).
The isotopomer distribution of Phe reflects the composition of the precursors, phosphoenol pyruvate and erythrose 4-phosphate, via the shikimate pathway. As shown in Fig. 7D, (M ϩ 3)and (M ϩ 2)-isotopomer groups were present in Phe from all experiments. Notably, the abundances of both isotopomer groups were higher in Phe from YscM1 and YscM2 mutants. Since the labeling pattern of Ser suggested similar patterns in phosphoglycerate and its downstream product, phosphoenol pyruvate, it can be inferred that the increased abundances must be due to differences in the erythrose 4-phosphate precursor. On this basis, deletion of YscM1 and YscM2 appears to increase carbon flux from the proffered glucose into the tetrose phos- phate via the nonoxidative branch of the pentose phosphate pathway (Fig. 10).
The isotopomer compositions of Ala were characterized by (M ϩ 3) and (M ϩ 2) species at similar abundances (Fig. 7A). Ala from the ppc mutant showed a similar if not identical (M ϩ 3)/(M ϩ 2) ratio as in Ala from the YscM mutants. On the other hand, Ala from the parental strain had a different ratio with a lower value for the (M ϩ 3)-isotopologue (Fig. 7A). The data document that a significant fraction of Ala was derived via [U-13 C 3 ]PEP and [U-13 C 3 ]pyruvate, giving rise to the (M ϩ 3)-isotopologue. Again, the formation of [U-13 C 3 ]pyruvate from proffered [U-13 C 6 ]glucose can be explained by glycolysis or the Entner-Doudoroff pathway. The doubly labeled isotopomer group characterized by (M ϩ 2)-Ala had a more complex biosynthetic history, where pyruvate must have been assembled from a 13 C 2 fragment and a 12 C 1 fragment. This process can be explained by the action of malic enzyme (Fig. 8). The fact that (M ϩ 3) and (M ϩ 2) species can be detected in Ala from all experiments indicates that all of these processes are operative in the parental strain as well as in the mutants. However, the different ratios (Fig. 7A) demonstrate that the efficiencies of the processes are modulated by the mutations. More specifically, the higher (M ϩ 3) values in the PEPC mutants and especially in the YscM mutants document that more carbon from glucose is transferred to pyruvate and alanine, respectively, via the direct pathways by glycolysis and/or the Entner-Doudoroff pathway ( Figs. 9 and 10).
The distribution of (M ϩ 3)-and (M ϩ 2)-isotopomers was also different in Asp from the various experiments (Fig. 7B). Since Asp is derived from oxaloacetate by transamination, the (M ϩ 3)-isotopologue can be considered as a marker for the PEPC reaction, where a 13 C 3 -labeled phosphoenol pyruvate and unlabeled CO 2 are converted into [ 13 C 3 ]oxaloacetate. Since, on the basis of the Ser profile, phosphoenol pyruvate appeared to comprise only (M ϩ 3) species at significant amounts, the presence of (M ϩ 2)-isotopologues does not reflect the PEPC reaction but rather indicates formation of oxaloacetate by the citrate cycle, where 13 C 2 -labeled acetyl-CoA (derived from [ 13 C 3 ]pyruvate by pyruvate dehydrogenase) is introduced (Fig. 8). As shown in Fig. 7B, the YscM1 and YscM2 mutants were characterized by higher values for the (M ϩ 2)-isotopomer groups, indicating that flux via the citrate cycle was increased by deletion of either YscM1 or YscM2 (Fig. 10). On the other hand, the abundances of the (M ϩ 3)-isotopologues appeared to be lower in the mutants (Fig. 7B), suggesting that carbon flux via the PEPC reaction was slightly lower under the deficiency of YscM1 or YscM2, respectively.
Taken together, isotopologue profiling confirmed that the PEPC reaction contributed to amino acid biosynthesis under conditions of Yop secretion and that yscM1 and yscM2 were involved in control of carbon fluxes through the central metabolic network of Y. enterocolitica. In contrast to the in vitro studies with recombinant YscM proteins, yscM1 and yscM2 mutants exhibited very similar isotopologue profiles.

DISCUSSION
Bifunctionality of the YscM Proteins-The Yersinia T3SS components YscM1/LcrQ and YscM2 are involved in regulation of Yop expression (18 -20, 23). This regulatory function is mediated in concert with YopD, LcrH, and SycH (19,26,28). Stainer et al. (19) described the phenomenon that deletion of   either yscM1 or yscM2 alone had no influence on the level of Yop expression and secretion, whereas the yscM1/yscM2 double mutant exhibited the lcrQ phenotype known from Y. pestis and Y. pseudotuberculosis. Consequently, a functional redundancy of the YscM proteins was assumed. In addition to this described function, we show here that YscM1 and YscM2 interact with the metabolic enzyme PEPC. The interaction of YscM1 and YscM2 with PEPC has been demonstrated unequivocally by applying affinity purification, native gel electrophoresis, and SAXS experiments. The influence of YscM1 on PEPC activity is of inhibitory nature as revealed by PEPC activity assays and overproduction experiments. The role of YscM2 remains elusive, since we were unable to attribute any influence of YscM2 on PEPC activity in vitro. Further, its overproduction had no effect on growth of yersiniae under conditions requiring PEPC activity. In marked contrast, mass spectrometry following stable isotope labeling revealed an influence of yscM2 deletion on biosynthesis of Yops similar to that associated with the yscM1 deletion, and, more specifically, isotopologue profiling revealed similarly increased carbon flux from glucose into amino acid for both yscM deletion strains. This discrepancy could be attributed to an additional component, proteinaceous or low molecular weight compound, required for YscM2 to modulate PEPC activity that is missing in our in vitro assays and under the conditions chosen for YscM2 overexpression.
Taken together, with respect to this newly discovered function, YscM1 and YscM2 do not display complete redundancy. From a Darwinian point of view, this is as expected, since a functional equivalence of the YscM proteins cannot explain their stable coexistence. The acquisition of a second copy of the yscM gene in Y. enterocolitica points to a more sophisticated regulation of metabolism in this species compared with Y. pestis and Y. pseudotuberculosis, which harbor only the yscM1 orthologue lcrQ.
The YscM1/PEPC Interaction, a Case of Molecular Coadaptation?-It is worthwhile to mentioning that we have determined the K m (or more precisely, the S 0,5 value) for the substrate PEP to be 4.1 mM for Y. enterocolitica PEPC, whereas the reported K m /S 0,5 (PEP) values for the 80% identical E. coli PEPC range from 0.1 to 0.2 mM (51, 52). Since we found that YscM1 competes for the PEP binding site, the unexpectedly high S 0,5 (PEP) may indicate that Yersinia PEPC has evolutionarily co-adapted to the interaction with YscM1/LcrQ. This hypothesis might be revisable; Y. enterocolitica biotype 1A does not harbor the virulence plasmid pYV, which encodes the T3SS. Given that the biotype 1A lineage diverged from the Yersinia lineage either before acquisition of the virulence plasmid or early after its acquisition, the catalytic properties of Yersinia biotype 1A PEPC and its response to YscM1/LcrQ interaction should differ from that of PEPC of pYV-carrying yersiniae. Moreover, if this molecular co-adaptation hypothesis holds, changed PEPC properties may have triggered further adaptation processes, resulting in divergence of the metabolism of T3SS-harboring yersiniae compared with biotype 1A strains. This scenario might also explain why the phenotypes of the yscM deletion mutants, although clearly linking T3S to metabolism, are not as clear cut as expected.
Cross-talk between T3SS and Metabolism-It was outlined above that assembly and operation of the T3SS are metabolically demanding. It therefore seems plausible that functional T3 secretion requires a coordinated adaptation of metabolic pathways. Here, we propose that modulation of Yersinia PEPC activity could be necessary to adapt to different metabolic requirements during infection. During those stages of infection relying on T3SS activity, two principle situations might be distinguishable. One phase is characterized by the massive synthesis of Yops and other T3SS components in preparation of phagocytic attack ("loading"). Under these conditions, the provision of building blocks for amino acid synthesis should require an active PEPC enzyme to replenish the tricarboxylic acid cycle. The second phase is initiated by contact with phagocytic cells and requires the rapid injection of the presynthesized pool of Yops ("shooting"). The latter conditions might force yersiniae to prioritize maintenance of energy charge over biosynthetic processes and propagation. Shutdown of anaplerosis via inhibition of PEPC could push the balance in that direction. After injection of the presynthesized Yop pool, biosynthesis of new Yops should be prioritized again and so on. This "load-andshoot cycle" model states that functioning of the T3SS in vivo requires a reprogramming of metabolic pathways and that it is the bifunctionality of YscM/LcrQ that allows coordination of Yop expression and metabolism. YscM/LcrQ interact with a multitude of T3SS proteins, mostly T3SS chaperones, which prompted us to speculate that YscM/LcrQ proteins function as a molecular interface that senses if T3SS chaperones are loaded with Yops and transduces this information into repression or derepression of Yop synthesis in concert with YopD, LcrH, and SycH (14,31). In parallel, this integrated information is also ideally suited to coordinate the metabolism via modulation of PEPC activity and possibly other metabolic enzymes.
The YscM1/PEPC Interaction, a Missing Link to Low Calcium Response?-It has long been known that virulence of Y. pestis is lost upon cultivation at 37°C and that avirulent variants are rapidly enriched. It was found that virulence could be maintained by the addition of calcium ions (6), glutamate, aspartate, or bicarbonate (57). This prompted Baugh et al. (58) to compare carbon dioxide fixation into oxaloacetate in virulent and avirulent Y. pestis. However, they were unable to demonstrate any difference between virulent and avirulent strains. Later, Fowler and Brubaker (59) emphasized the importance of CO 2 fixation into oxaloacetate for Y. pestis, especially in the context of the lack of aspartase activity in this species (60). In contrast, degradation of L-aspartate via aspartase, fumarase, and malic dehydrogenase enables the regeneration of oxaloacetate in Y. pseudotuberculosis (60). Recently, Lee et al. (61) could demonstrate that T3 secretion in Yersinia is activated by glutamate, glutamine, aspartate, and asparagine. Although these pioneering studies pointed at an interrelationship between virulence and metabolism in Yersinia, particularly emphasizing the PEPC reaction, such a conjunction has never been demonstrated on a molecular level. An inhibitory effect of LcrQ on PEPC activity fits with these previous observations of restricted growth in Y. pestis. However, the lcrQ mutant is characterized by growth restriction at 37°C in the presence of Ca 2ϩ ions, which means that, provided that PEPC is crucially involved in growth cessa-tion of Y. pestis, another factor takes over inhibition of PEPC in the absence of LcrQ. YopH could be such a candidate, since the first 128 amino acids of YopH show 42% identity to LcrQ (23).
Perspectives-The YscM/LcrQ proteins interact with several components of the T3SS and represent major nodal points of the T3SS regulatory network (14,31). A complete T3SS interactome is therefore required to elucidate the significance of the PEPC/YscM interactions. The lack of understanding of the YscM2/PEPC interaction might be due to cooperative interaction partners that are missing under the conditions tested. Also, the gap in our understanding of growth restriction in the absence of LcrQ, as discussed above, might be bridged by increasing knowledge of more protein/protein interactions. SAXS experiments have been proven to be a powerful tool to analyze PEPC/YscM interactions in solution. In continuation, we will compare the conformational transitions of PEPC caused by YscM1 and YscM2 with those induced by known allosteric effectors to add to the understanding of the mode of action.
We have demonstrated here that the introduction of 13 C labeling via [U-13 C 6 ]glucose, followed by MALDI-TOF and GC/MS analysis, respectively, enables to distinguish the ppc, yscM1, and yscM2 mutants from the parental strain with respect to Yop synthesis and amino acid biosynthesis. GC/MSbased analysis of 13 C enrichment and isotopologue patterns indicates that next to a modulation of the PEPC reaction, flux through the central metabolic network, including the citrate cycle and the pentose phosphate pathway, appears to be triggered by YscM, but more sophisticated methods, such as NMRbased isotopologue profiling (62), will be required to gain quantitative insights into the positional redistribution of carbon fluxes caused by the deletions. In addition, a more direct approach to monitor the contribution of carbon dioxide fixation by PEPC would be to introduce 13 C labeling via bicarbonate. Time-resolved metabolic profiling will be required to challenge the load-and-shoot cycle model.
Y. enterocolitica strains deleted of yscM1, yscM2, and ppc, respectively, have to be further characterized in cell culture and mouse infection models to shed light on the cross-talk between T3SS and metabolism in its biological context. Another step will be the characterization of the temperature-sensitive yscM1/yscM2 double mutant and the lcrQ mutants of Y. pseudotuberculosis and Y. pestis. The interpretation of these data, however, will be more complicated because of the bifunctionality of the YscM/LcrQ proteins. Since the yscM genes are redundant with respect to their Yop regulatory function (19), deletions of either yscM1 or yscM2 alone are ideally suited to study the PEPC regulatory effect without disturbing the Yop regulatory network.
Further, deletion of ppc in Y. pseudotuberculosis and Y. pestis is expected to have more dramatic effects on T3 secretion, since the LCR-related growth restriction of both species is more pronounced compared with Y. enterocolitica (7). Provided that PEPC is the key enzyme involved in growth arrest of Y. pestis, it should be an ideal candidate for drug targeting.