Genomic adaptation of Burkholderia anthina to glyphosate uncovers a novel herbicide resistance mechanism

Abstract Glyphosate (GS) specifically inhibits the 5‐enolpyruvyl‐shikimate‐3‐phosphate (EPSP) synthase that converts phosphoenolpyruvate (PEP) and shikimate‐3‐phosphate to EPSP in the shikimate pathway of bacteria and other organisms. The inhibition of the EPSP synthase depletes the cell of the EPSP‐derived aromatic amino acids as well as of folate and quinones. A variety of mechanisms (e.g., EPSP synthase modification) has been described that confer GS resistance to bacteria. Here, we show that the Burkholderia anthina strain DSM 16086 quickly evolves GS resistance by the acquisition of mutations in the ppsR gene. ppsR codes for the pyruvate/ortho‐Pi dikinase PpsR that physically interacts and regulates the activity of the PEP synthetase PpsA. The mutational inactivation of ppsR causes an increase in the cellular PEP concentration, thereby abolishing the inhibition of the EPSP synthase by GS that competes with PEP for binding to the enzyme. Since the overexpression of the Escherichia coli ppsA gene in Bacillus subtilis and E. coli did not increase GS resistance in these organisms, the mutational inactivation of the ppsR gene resulting in PpsA overactivity is a GS resistance mechanism that is probably unique to B. anthina.


INTRODUCTION
Glyphosate (N-(phosphonomethyl)glycine) (GS) is the most-commonly used herbicide worldwide (Duke & Powles, 2008;Franz, 1979).GS specifically inhibits the 5-enolpyruvyl-shikimate-3-phosphate (EPSP) synthase of the shikimate pathway (Figure 1A) (Herrmann & Weaver, 1999;Steinrücken & Amrhein, 1980).The EPSP synthase converts phosphoenolpyruvate (PEP) and shikimate-3-phosphate to EPSP, a precursor for chorismate biosynthesis.Chorismate, in turn, is required for de novo synthesis of phenylalanine, tyrosine, and tryptophan as well as of folate and quinones (Herrmann & Weaver, 1999;Steinrücken & Amrhein, 1980).The GS-dependent inactivation of the EPSP synthase leads to the depletion of the cellular chorismate levels and thus the death of plants and other organisms possessing the EPSP synthase (Fischer et al., 1986;Gresshoff, 1979;Wicke et al., 2019).The effect of GS on EPSP synthase is well studied and understood.GS targets the PEP binding site and competitively inhibits the EPSP synthase (Schönbrunn et al., 2001).The isolation of GS-insensitive bacterial EPSP synthases and their use to develop herbicide-tolerant crops allowed weed growth control without harming the growth of the desired plants (Duke & Powles, 2008).The application of GS to fields has put strong selective pressure on microbes and plants to evolve resistance mechanisms against the herbicide (Chekan et al., 2016;Hertel et al., 2021;Wicke et al., 2019).For instance, GS toxicity can be relieved by increasing the copy number of the EPSP synthase gene through selective gene amplification (Gaines et al., 2010;Wicke et al., 2019).The overproduction of the GS-sensitive EPSP synthase allows the organism to titrate the herbicide and a small portion of the enzyme to function normally.The evolution of GS resistance due to altered transport has also been observed in various organisms including bacteria, fungi, and plants (Ge et al., 2010;Pan et al., 2021;Wicke et al., 2019).For example, plants can become resistant to GS by accumulating mutations resulting in increased vacuolar GS sequestration or by lowering the cytoplasmic GS level (Ge et al., 2010;Pan et al., 2021).Moreover, the degradation of GS by various bacteria is a protective mechanism against the herbicide (Hertel et al., 2021;Hove-Jensen et al., 2014).GS detoxification by covalent modification also confers resistance against the herbicide (Hertel et al., 2021).For example, the Escherichia coli and Burkholderia pseudomallei hygromycin phosphotransferases inactivate GS by phosphorylation (Penaloza-Vazquez et al., 1995;Rao et al., 1983).
In the past years, several GS-resistant and -degrading Burkholderia strains have been isolated (Dotson et al., 1996;Hertel et al., 2022;Manogaran, Ahmad, et al., 2018;Manogaran, Shukor, et al., 2018;Penaloza-Vazquez et al., 1995;Shahid & Khan, 2018).Moreover, it has been observed that the repeated application of GS to soil led to an increase in the abundance of Firmicutes and Burkholderia species (Lancaster et al., 2010;Ramirez-Villacis et al., 2020).Recently, we have identified Burkholdera anthina and Burkholderia cenocepacia that are resistant to high amounts of GS (Hertel et al., 2022).Here, we evaluated whether GS resistance could be a basic property of the members of the genus Burkholderia.For this purpose, we assessed the effect of GS on the B. anthina strain DSM 16086.We found that the strain has no intrinsic GS resistance, but quickly develops resistance to the herbicide.Characterization of the evolved strains revealed that GS-resistance can be acquired by mutations enhancing PEP synthase-dependent synthesis of PEP, the EPSP synthase co-substrate that competes with GS for binding to the enzyme.Thus, the overproduction of PEP could be a novel mechanism conferring GS resistance to Burkholderia species.

EXPERIMENTAL PROCEDURES
Bacterial strains, growth media, culture conditions and chemicals Bacterial strains are listed in Table S1.The B. anthina strain DSM 16086 was obtained from the German Collection of Microorganisms and Cell Cultures GmbH (DSMZ, www.dsmz.de).Chemicals and media were purchased from Sigma-Aldrich (Germany), Carl Roth (Karlsruhe, Germany) and Becton Dickinson (Heidelberg, Germany).B. subtilis, B. anthina and E. coli strains were cultivated in lysogeny broth (LB) and C-Glc medium.C-Glc is a minimal medium containing glucose and ammonium as sources of carbon and nitrogen, respectively (Commichau et al., 2007).Pyruvate was used as carbon source as previously described (van den Esker et al., 2017).Agar plates were prepared with 15 g agar/l (Roth, Germany).E. coli transformants were selected on LB plates containing kanamycin (50 μg/mL) or ampicillin (100 μg/mL).Growth in liquid medium was monitored using 96-well plates (Microtest Plate 96-Well, F Sarstedt, Germany) at 37 C and medium orbital shaking at 237 cpm (4 mm) in a Synergy H1 plate reader (Agilent, USA) equipped with the Gen5 software, and the OD 600 was measured in 10-15 min intervals.Single colonies were used to inoculate 5 mL overnight LB cultures that were incubated at 220 rpm and 30 C. The OD 600 was adjusted to 0.1 and 150 μL of the cell suspensions were transferred into 96-well plates.Bacteria were cultivated in the Synergy H1 plate reader as described above.

Genome sequencing, assembly, and annotation
Genome sequencing was performed by the G2L (Göttingen, Germany) and GENEWIZ GmbH (Leipzig, Germany).Sequencing libraries were constructed using the Nextera XT DNA sample preparation kit (Illumina, San Diego, CA, USA) and NEB Next Ultra II FS DNA Library Prep 136 (New England Biolabs GmbH, Frankfurt, Germany).Sequencing was realized on a MiSeq instrument using the 2 Â 300 bp pairedend protocol, as recommended by the manufacturer (Illumina, San Diego, CA, USA) and on the Nova-Seq6000 using the 2 Â 150 bp paired-end protocol.The quality of the reads was assessed with FastQC version 0.11.9 (Andrews, 2010) and quality was processed using Trimmomatic v.0.36 (Bolger et al., 2014).The Genome of B. anthina DSM16086 was assembled with SPAdes version 3.14.0(Bankevich et al., 2012) with the options-only-assembler and-isolate.Read coverage was determined with QualiMap v.2.2.1 (Okonechnikov et al., 2016)  Contigs < 500 bp and with less than 10% of the average coverage were discarded.The Prokaryotic Genome Annotation Pipeline (PGAP) (Tatusova et al., 2016) was used for automated annotation during genome submission to GenBank.Single-nucleotide polymorphism (SNP) analyses of the mutant genomes were performed using the breseq pipeline (Deatherage & Barrick, 2014).The annotated draft genome of B. anthina DSM16086 is available at Gen-Bank with the accession numbers JAFCIQ010000000.

DNA manipulation, sequencing, and cloning
Primers were purchased from Sigma-Aldrich (Germany) and are listed in Table S1.Chromosomal DNAs from B. subtilis, B. anthina and E. coli strains were isolated using the peqGOLD Bacterial DNA Kit (Peqlab, Germany).Plasmids were isolated from E. coli using the Nucleospin Extract Kit (Macherey and Nagel, Germany).PCR products were purified using the PCR Purification Kit (Qiagen, Germany).Phusion DNA polymerase, restriction enzymes and T4 DNA ligase were purchased from Thermo Scientific (Germany) and used according to the manufacturer's instructions.DNA sequencing was performed at the Microsynth Laboratories (Göttingen, Germany).The E. coli strain XL1-Blue served as the host for plasmid construction.The plasmid pBP1225 for overproduction of N-terminally Strep-tagged PpsA from E. coli was constructed by generating a PCR fragment using the primer pair IS49/IS50.The PCR product was digested with SacI/BamHI and ligated to pGP172 (Merzbacher et al., 2004) that was cut with the same enzymes.The plasmids pBP1226 and pBP1227 for the overproduction of N-terminally 6 Â His-tagged PpsR and PpsR Q173P variants, respectively, from E. coli were constructed as follows.The ppsR wild type allele was amplified with the primer pair IS51/IS52, the PCR product was digested with PstI/HindIII and ligated to pWH844 (Schirmer et al., 1997) that was cut with the same enzymes.The ppsR A518C allele was generated via the fusion of two PCR products that were generated with the primer pairs IS51/IS63 and IS52/IS64.The resulting ppsR A518C allele was introduced via the restriction sites PstI/HindIII into the plasmid pWH844.The primer pair IS55/IS56 was used for sequencing the ppsR gene in the GS-resistant B. anthina mutants.The plasmid pBP1224 contains the E. coli ppsA gene that was amplified with the primer pair IS47/IS48 and ligated to pGP380 that was cut with PstI/HindIII (Herzberg et al., 2007).

Bacterial two-hybrid (B2H) assay
The primary protein-protein interactions were analysed using the bacterial two-hybrid (B2H) system that is based on the reconstitution of the adenylate cyclase from Bordetella pertussis in the E. coli cya mutant BTH101 (Karimova et al., 1998).The plasmid pairs pUT18C/pKT25 and pUT18/pKNT25 were used for the expression of proteins fused to the C and N termini, respectively, of the T18 and T25 fragments of CyaA from B. pertussis.The plasmids constructed for the B2H analysis are listed in Table S1.The genes were amplified using the primers listed in Table S1 and cloned between the XbaI and KpnI sites of the plasmids pUT18C, pKT25, pUT18 and pKNT25.The ppsA and ppsR/ppsR A518C alleles from E. coli DH5α were amplified using the primer pairs IS59/IS60 and IS61/ IS62, respectively.The plasmids pBP78-81 and pBP82-pBP85/pBP1228-1231 carry the E. coli DH5α ppsA and ppsR/ppsR A518C alleles, respectively (Table S1).The ppsA and ppsR alleles from the B. anthina strain DSM 16086 were amplified using the primer pairs CM1/CM2 and CM3/CM4, respectively.The plasmids pBP1211-pBP1214 and pBP1207-pBP1210 carry the B. anthina ppsA and ppsR alleles, respectively (Table S1).pUT18C-zip and pKT25-zip served as controls.The generated plasmids were used to transform E. coli BTH101, and the protein-protein interactions were analysed by plating the cells on LB plates containing 100 μg/mL ampicillin, 50 μg/mL kanamycin, 100 μg/mL X-Gal and 0.5 mM IPTG.The plates were incubated for 36-72 h at 30 C.

Protein overproduction and purification
The E. coli strains BL21(DE) and DH5α were used for protein overproduction based on the plasmids pGP172 and pWH844, respectively.The strains were cultivated in 1 L LB medium at 37 C. Protein overproduction was induced by the addition of 1 mM IPTG when the cultures had reached an OD 600 of about 0.6.The cultures were further incubated for 2-3 h at 37 C.The cells were collected by centrifugation and disrupted in lysis buffer (100 mM Tris-HCl pH 8.0, 150 mM NaCl) using a FrenchPress.The soluble protein fraction was separated from the cell debris by centrifugation for 30 min at 10,000 g.The N-terminally Strep-tagged PpsA and 6 Â His-tagged PpsR/PpsR Q173P proteins from E. coli were purified by Strep-tag/Streptactin and Ni 2+ -NTA affinity purification, respectively, as described previously (Commichau et al., 2007;Rosenberg et al., 2015).Histagged PpsR Q173P could not be overproduced and purified using the standard procedure.To facilitate the purification of PpsR Q173P, the main culture of the E. coli strain DH5α carrying the plasmid pBP1227 was grown in 1 L LB medium supplemented with ampicillin (100 μg/mL) and ethanol (1% (v/v)) at 37 C until an OD 600 of 0.5.The culture was cooled down, supplemented with 0.1 mM IPTG and incubated overnight at room temperature.After elution, the fractions were tested for the desired protein using 12.5% SDS PAGE gels.Protein concentration was determined using the BioRad dye-binding assay (BioRad, Germany).

Metabolomics
The B. anthina strains were grown overnight in 4 mL LB medium at 37 C and 160 rpm.The cells were harvested by centrifugation, washed twice in C-Glc medium, and used to inoculate 100 mL shake flasks containing 10 mL C-Glc medium to an OD 600 of 0.1.The cultures were incubated at 37 C and 160 rpm until an OD 600 of about 0.5.The cultures were further incubated for 30 min, the OD 600 was determined and the cells in 5 mL of the cultures were passed over a PVDF filter (0.45 μm pore site) in a glass frit.The cells on the filters were resuspended in 1 mL ice-cold extraction solution (acetonitrile/methanol/ultrapure water, 40%/40%/20%) and incubated for 1 h at À20 C. The cell extracts were centrifuged for 15 min and 20,000 g at À9 C, and stored at À80 C until further processing.
Cell extracts were pooled 1:1 with a 13 C internal standard as described previously (Guder et al., 2017).
PEP concentrations in cell extracts were measured on an Agilent 6495 triple quadrupole mass quadrupole mass spectrometer equipped with an ESI ion source and coupled to an Agilent 1290 Infinity II UHPLC (both Agilent Technologies) using an iHILIC-Fusion (P) (50 Â 2.1 mm, HILICON AB) column.LC-solvents were: Solvent A: water with ammonium carbonate (10 mM) and ammonium hydroxide (0.2%).Solvent B: acetonitrile.The LC-Gradient was: 0 min 90% B, 1.30 min 40% B, 1.5 min 40% B, 1.7 min 90% B, 2.3 min 90% B (flow rate 0.4 mL/min).3 μL was injected per sample.The settings of the ESI source were: 200 C source gas temperature, 0.4 L/min drying gas and 24 psi nebulizer pressure.The sheath gas temperature was at 300 C and flow at 11 L/min.The electrospray nozzle was set to 500 V and capillary voltage to 2500 V. PEP was analysed in negative ionization mode with a transition from 167 m/z ( 12 C) or 170 m/z ( 13 C) to 79 ( 12 C) or 79 m/z ( 13 C) with a collision energy of 29 keV and a dwell time of 12 ms.Raw data were converted into text-files using MSConvert (Chambers et al., 2012).Data analysis was performed with a customized Matlab script.Relative PEP concentrations are displayed as the ratio between 12 C and 13 C signal intensity.

Burkholderia anthina DSM 16086 quickly evolves GS resistance
Recently, we isolated B. anthina and B. cenocepacia strains that are highly resistant to GS (Hertel et al., 2022).In contrast to Bacillus subtilis and Escherichia coli strains that genomically adapted to GS and grew with 10 mM GS, the Burkholderia isolates grew in the presence of up to 60 mM GS (Hertel et al., 2022;Wicke et al., 2019).To assess whether it is a general property of B. anthina to quickly evolve GS resistance, we propagated the B. anthina strain DSM 16086 on C-Glc minimal agar plates supplemented with 10 mM GS. Pure C-Glc agar plates served as a control.As shown in Figure 1B, the growth of B. anthina was inhibited by 10 mM GS.However, already after 1 day of incubation several GS-resistant colonies emerged.After 2 days of incubation of the plates, the number of mutants doubled and only a few more mutants emerged after incubation for a total of 3 days (Figure 1C).To verify whether the mutants had acquired stable mutations conferring GS resistance, we randomly isolated three mutants (S1-S3) and performed growth experiment with C-Glc plates and liquid medium.As shown in Figure 1D,E, the three mutants could grow both on plates and in liquid medium containing 10 mM GS.At a GS concentration of 20 mM, the mutants were no longer able to grow (Figure 1E).To conclude, although the B. anthina strain DSM 16086 is not intrinsically resistant to GS, the bacteria quickly develop resistance against GS.

Identification of the mutations in the GSresistant B. anthina mutants
Genome sequencing analyses of the three GSresistant suppressors (S1-S3) revealed that all mutants had acquired mutations in the ppsR gene encoding PpsR (Figure 2A,B) (Table 1).In E. coli, PpsR regulates the activity of the PEP synthetase PpsA (Burnell, 2010).PpsR activates and inhibits PpsA in a P i -and ATP/ADP-dependent manner, respectively (Burnell, 2010).PpsR belongs to the DUF299 protein family that also contains the pyruvate/ortho-P i dikinase regulatory protein from maize and Arabidopsis (Burnell, 2010;Jiang et al., 2016).The suppressors S1 and S3 carry the same mutations in ppsR (Table 1).The single nucleotide exchanges in the ppsR gene of the suppressors S1/S3 and S2 would cause the amino acids substitutions L261P and Q167P, respectively.Moreover, in the suppressor mutant S1, a 11 bp-long tandem repeat upstream of the 3-deoxy-D-arabino-heptulosonate-7-phosphate (DAHP) synthase gene was expanded by one repeat unit (Table 1).The DAHP synthase catalyses the conversion of PEP and erythrose-4-phosphate to DAHP in the shikimate pathway.It is tempting to speculate that the expansion of the tandem repeat affects the cellular concentration of the DAHP synthase (see Section 3).The suppressors S1 and S3 had also acquired the mutation T886G in Δ 1 -pyrroline-2-carboxylate reductase gene (Table 1).
Δ 1 -Pyrroline-2-carboxylate reductase is involved in arginine and proline metabolism and thus not linked to the shikimate pathway (Meister et al., 1957).Furthermore, the suppressor S2 carries single nucleotide exchanges in genes that are predicted to code for a sensor histidine kinase and a fimbria biogenesis protein (Table 1).
Since the ppsR gene was mutated in all GSresistant suppressor mutants, we assumed that this genomic alteration causes GS resistance in B. anthina.Moreover, in E. coli, PpsR regulates synthesis of PEP, which is a co-substrate of the AroA EPSP synthase (Figures 1A and 2A) (Bartlett et al., 2012;Burnell, 2010).Thus, an increased cellular PEP concentration possibly abolishes the GS-dependent inhibition of the EPSP synthase (see below).Based on the structure of the pyruvate/ortho-P i dikinase regulatory protein from maize (30.6% overall sequence identify), we generated a structural model for the B. anthina PpsR protein (Figure 2C).We also generated a sequence alignment based on sequences of PpsR homologues from bacteria and the maize pyruvate/ ortho-P i dikinase regulatory protein (Figure 2D).As shown in Figure 2C, the residues Q167 and L261 are part of α-helices that are probably required for the formation of a stable and functional PpsR dimer.Moreover, the sequence alignment revealed that the residues Q167 and L261 are highly conserved among bacterial PpsR homologues and are even present in the maize pyruvate/ortho-P i dikinase regulator (Figure 2D).Proline has an exceptional conformational rigidity and acts as a structural disruptor in secondary structure elements such as α-helices and β-sheets.It is therefore very likely that the Q167P and L261P substitutions in PpsR in the GS-resistant B. anthina suppressors S1-S3 indeed impair the function of the PEP synthetase regulatory protein and thus affect the cellular PEP concentration.
Glutamine 167 is a critical residue in the PEP synthetase regulatory protein Next, we analysed the interaction between PpsR and PpsA from B. anthina.For this purpose, we made use of a bacterial two-hybrid (B2H) system that is based on the functional reconstitution of a split adenylate cyclase from Bordetella pertussis in an E. coli cya mutant strain (see Section 4).As a control, we analysed the interaction between PpsR and PpsA from E. coli.As shown in Figure 3A,B, the PpsR and PpsA homologues from B. anthina and E. coli showed self-interaction, which is consistent with previous reports indicating that the proteins form dimers (Jiang et al., 2016;Narindrasorasak & Bridger, 1977).The B2H assay also revealed that B. anthina and E. coli PpsR and PpsA interact with each other (Figure 3A,B).However, to observe the interaction among the B. anthina proteins, the B2H plates indicator strain had to be incubated 25 times longer.Since the genome of B. anthina has a GC content of about 66% (https://www.ncbi.nlm.nih.gov/genome/browse/#!/prokaryotes/42805/), the weak interaction between PpsR and PpsA is likely due to the inefficient protein translation.
Next, we aimed to assess the effect of the Q167P substitution on the activity of the PpsR PEP synthases regulatory protein.As shown above, the protein sequence alignment revealed that the residue Q167 is conserved among bacterial PpsR homologues and the maize pyruvate/ortho-P i dikinase regulator (Figure 2D).Due to the low-level synthesis of the B. anthina proteins in E. coli, we thought to study the effect of the Q167P substitution on the activity of the E. coli PpsR homologue.For this purpose, we generated plasmids for the overproduction and purification of Strep-PpsA and His-PpsR and His-PpsR Q173P by Strep-tag/Streptactinand nickel NTA-affinity purification (see Section 4).The residue Q167 in PpsR from B. anthina corresponds to Q173 in E. coli PpsR.The purification of the wild type  1).(C) Localization of the amino substitutions that likely affect the function of the PEP synthetase regulatory protein PpsR.The structure model was generated using the SWISS-MODEL server for homology modelling of protein structures (Waterhouse et al., 2018) and a model of the dimer structure of the maize pyruvate orthophosphate dikinase regulatory protein (PDBid: 5D0N) (Jiang et al., 2016) be due to misfolding of PpsR Q173P.To conclude, Q173 (Q167 in B. anthina PpsR) seems to be a critical residue for the proper folding and function of the E. coli PEP synthetase regulatory protein PpsR.

Elevated cellular PEP concentration due to inactivation to PpsR confers B. anthina resistance to GS
To further verify that mutations in the ppsR gene confer GS resistance to B. anthina, we randomly isolated six additional suppressor mutants (S4-S9) from C-Glc plates supplemented with 10 mM GS. Sanger sequencing revealed that the mutants had also acquired mutations in the ppsR gene (Table 1).The single nucleotide exchanges in the ppsR gene of the suppressors S4/S5, S6/S9 and S8 would cause the amino acids substitutions E97K, T68A and G109D, respectively (Figure 2C).The residues T68 and G109 in PpsR are less conserved than the residues at the positions 167 and 261 (Figure 2D).However, the function of the PEP synthetase regulator is certainly affected in the suppressors S4-S8 and S9 because T68, E97 and G109 are substituted by amino acids with different biochemical properties (Table 1).In the GS-resistant suppressor S7, a 99 bp-long deletion in the ppsR gene leads to a frameshift, which would change the sequence from position 146 and truncate PpsR by 32 amino acids.The loss-of-function mutation in ppsR of suppressor S7 suggests that the PpsA-dependent overproduction of PEP likely confers resistance to GS in B. anthina.To test this idea, we performed metabolome analyses and determined the relative cellular PEP concentrations in the parental strain and the suppressors S1, S2, S4, S6, S7 and S8 with different amino acid exchanges in PpsR.As shown in Figure 4, the relative amount of PEP was significantly higher in all GSresistant suppressor mutants.The variation in PEP levels in the suppressors could be due to the PpsR variants regulating PEP synthetase to different extents.However, the increased synthesis of PEP by PpsA indeed appears to prevent EPSP synthase from being inhibited by GS, thereby allowing the bacteria to grow in the presence of the herbicide.
F I G U R E 3 Interaction analysis between PpsR and PpsA homologues from Burkholderia anthina and Escherichia coli.(A) B2H assay to assess the interaction between PpsR and PpsA from B. anthina.(B) B2H assay to assess the interaction between PpsR and PpsA from E. coli.The agar plates were incubated for 36 h at 30 C. (C) B2H assay to assess the interaction between PpsR Q173P and PpsA from E. coli.The ppsR and ppsA alleles were introduced into the plasmids pUT18, pUT18C, pKNT25 and pKT25.Plasmids pUT18 and pUT18C allow the expression of the proteins fused to the N-and C-terminus of the T18 domain of the Bordetella pertussis adenylate cyclase, respectively.Plasmids pKNT25 and pKT25 allow the expression of the proteins fused to the N-and C-terminus of the T25 domain of the adenylate cyclase.The E. coli transformants were spotted onto LB plates supplemented with X-Gal, IPTG, ampicillin and kanamycin.The agar plate shown in (A) was incubated for 48 h at 30 C, followed by 40 days at 4 C.The agar plates shown in (B) and (C) were incubated for 36 h at 30 C.
To assess whether the enhanced synthesis of PEP leads to GS resistance in the B. subtilis and E. coli strains SP1 and W3110, respectively, we constructed the shuttle vector pBP1224 that carries the E. coli ppsA gene.Expression of the ppsA gene is driven by the constitutively active P degQ promoter (Herzberg et al., 2007).Next, the plasmids pGP380 (empty) and pBP1224 ( ppsA) were introduced into B. subtilis and E. coli.The expression of the native ppsA gene in E. coli slightly enhanced growth of the bacteria on C agar plates containing glucose or pyruvate as carbon sources in the absence and in the presence of 2.5 mM GS (Figure S2).By contrast, the overexpression of the ppsA gene in B. subtilis did not confer GS resistance.Therefore, we only examined the influence of PpsA overexpression on growth and GS resistance in E. coli in liquid culture.As shown in Figure 5, the overexpression of PpsA also improved the growth of E. coli in C liquid medium containing either glucose or succinate in the absence and in the presence of GS.Thus, the apparently higher GS resistance of E. coli is more likely due to a general growth-promoting effect caused by PpsA.Indeed, metabolome analyses revealed that the relative cellular PEP concentration was not affected by the overexpression of the ppsA gene (Figure 4B).The fact that glutamate alleviated the negative effect of GS on the growth of E. coli indicates that the growth inhibition was due to the presence of the herbicide that is  transported via glutamate uptake systems (Wicke et al., 2019).The unexpected observation that the overexpression of ppsA did not increase GS resistance in B. subtilis and E. coli indicates that the central carbon metabolism is operating in a fundamentally different way in B. anthina.
A deletion in the phosphoglycerate mutase gene increases GS resistance of a B. anthina ppsR mutant As demonstrated above, the B. anthina strain DSM 16086 quickly develops GS resistance by acquiring mutations in the ppsR gene (Figure 1B,C).However, the isolated suppressor mutants could not withstand GS concentrations of 20 mM.To assess the potential of a B. anthina ppsR mutant to evolve increased GS resistance, we propagated the arbitrarily chosen suppressor S2 on C-Glc plates supplemented with 35 mM GS.After 3 days of incubation, we identified a single colony, which after its isolation and purification was named suppressor 2.1.As shown in Figure 6A,B, the isolated mutant S2.1 grew on plates and in liquid medium supplemented with 35 mM GS. Genome sequencing analysis revealed that the suppressor S2.1 carries had acquired eight additional mutations in addition to the three already present mutations (Table 1).Like in the suppressor S1, a 11 bp-long tandem repeat upstream of the 3-deoxy-D-arabino-heptulosonate-7-phosphate (DAHP) synthase gene was expanded by one repeat unit and the Δ 1 -pyrroline-2-carboxylate reductase gene carries the T886G mutation (Table 1).Moreover, the mutations upstream and downstream of glxA and a gene encoding the DUF488 protein of unknown function, respectively, probably affect the expression of the respective genes.Other mutations affect the essential β-ketoacyl-[acyl-carrier-protein] synthase, a sigma 54-interacting transcriptional regulator and a hypothetical protein (Table 1).Finally, we identified a 9 bp-long in-frame deletion in the gpmA gene resulting in the loss of three amino acids (Δ71RMD73) in the encoded a P i -glycerate mutase.
The P i -glycerate mutase catalyses the reversible conversion of 2-P i -glycerate to 3-P i -glycerate in glycolysis and gluconeogenesis (Figure 6C).Growth experiments revealed that the strain S2.1 had lost the ability to grow with the gluconeogenic carbon source succinate (Figure 6D).The deletion of three amino acids in GpmA likely prevents the enzyme from participating in gluconeogenesis.We suspect that the mutation in the gpmA gene is responsible for the increased GS resistance of the suppressor S2.1 since an altered activity of the P iglycerate mutase likely affects the PEP pool (Figure 6C).Indeed, metabolome analyses revealed that the relative cellular PEP concentration was increased as compared to the parental strain S2 (Figure 4A).To conclude, a further increase in the cellular PEP level by altered GpmA activity prevents EPSP synthase from being inhibited by elevated GS concentrations (see Section 3).

DISCUSSION
Unlike the previously characterized B. anthina and B. cenocepacia isolates that grew in the presence of 60 mM GS (Hertel et al., 2021), the B. anthina strain DSM 16086 shows no intrinsic GS resistance.However, B. anthina DSM 16086 quickly acquires resistance to GS during growth on minimal medium agar plates supplemented with 10 mM GS.Sequencing analyses revealed that all GS-resistant B. anthina mutants had acquired mutations in the ppsR gene encoding PpsR, which shares 45.8% overall sequence identity with the homologue of E. coli (Figure 2D).In E. coli it has been demonstrated that PpsR regulates the activity of PpsA in a P i -and ATP/ADP-dependent manner (Bartlett et al., 2012;Burnell, 2010).The PEP synthetase PpsA is a gluconeogenic enzyme that is required for the growth of E. coli on three-carbon substrates like pyruvate (Antonovsky et al., 2016;Cooper & Kornberg, 1965, 1967;Cooper & Kornberg, 1969;Herz et al., 2017;Niersbach et al., 1992).Metabolome analysis revealed that the cellular concentration of PEP, the co-substrate of the EPSP synthase of the shikimate pathway, was elevated in the B. anthina GS-resistant mutants.Since GS competes with PEP for binding to the EPSP synthase, we suspect that the increased cellular concentration of PEP due to the mutational inactivation of the ppsR gene allows the B. anthina mutants to grow in the presence of GS (Figure 7).A previous study also identified residues in the E. coli PpsR protein that are critical for enzyme folding and catalysis (Bartlett et al., 2012).However, here, we identified different amino acid residues that are crucial for proper enzyme function.It is interesting to note that despite the presence of a futile cycle that involves the ATP-generating pyruvate kinase and the ATP-consuming PEP synthetase the de-regulation resulted in a significant increase in the cellular PEP concentration in the GS-resistant B. anthina mutants (Figure 6C).In E. coli, it has indeed been shown that the overexpression of the ppsA gene stimulates oxygen and sugar consumption (Patnaik et al., 1992).However, we observed that the overexpression of the E. coli ppsA gene in B. subtilis and E. coli did not confer GS resistance.Recently, it has been shown that the overexpression of the native ppsA gene only slightly improved metabolites of the PEP-consuming shikimate pathway (Chen et al., 2014).Thus, the pyruvate kinase activity probably prevents the accumulation of PEP in E. coli and probably also in B. subtilis.This unexpected result might be due to fundamental differences in the regulation of central carbon metabolism in B. anthina as compared B. subtilis and E. coli.Thus, the genomic adaptation uncovered a novel mechanism conferring GS resistance that is probably unique to B. anthina.In the present study, we also found that further adaptation of B. anthina resulted in the acquisition of a mutation in the gpmA gene encoding the P i -glycerate mutase that catalyses the conversion of 2-P i -glycerate to 3-P i -glycerate in glycolysis and gluconeogenesis (Figure 6C).The altered activity of the P i -glycerate mutase also affects the PEP pool, thereby preventing the inhibition of the EPSP synthase by elevated GS concentrations.Previously, it has been shown that also B. subtilis and E. coli rapidly adapt to GS at the genome level (Wicke et al., 2019).In B. subtilis, the mutational inactivation of the gltT and gltP genes, encoding the promiscuous high-and low-affinity glutamate transporters GltT and GltP, respectively, confers resistance against GS as well as against glufosinate that inhibits glutamine biosynthesis (Wicke et al., 2019) (Figure 7).In E. coli mutations that affect the activity or the cellular concentration of the EPSP synthase by an amino acid substitution in the GS target and by promoter-up as well as selective gene amplification, respectively, allow the bacteria to grow in the presence of 10 mM GS (Figure 7) (Wicke et al., 2019).Moreover, altered transport of GS via the transporter proteins was also associated with variations in GS resistance in fungi and plants (Figure 7) (Pan et al., 2021;Staub et al., 2012;Tao et al., 2017).Furthermore, bacteria can detoxify GS by covalent modification and degrade the herbicide (Figure 7) (Hertel et al., 2021;Hove-Jensen et al., 2014;Penaloza-Vazquez et al., 1995;Rao et al., 1983).Thus, the isolation and characterization of GS-resistant bacteria revealed various mechanisms conferring GS resistance.

F
I G U R E 1 Effect of GS on growth of Burkholderia anthina, emergence of GS-resistant mutants and their characterization.(A) GS inhibits the enolpyruvylshikimate 3-phosphate (EPSP) synthase AroA.Dashes files indicate that multiple reactions are involved in the biosynthesis of aromatic amino acids, quinones and folates.(B) Emergence of GS-resistant B. anthina mutants.B. anthina was grown in C-Glc medium at 37 C until an OD 600 of about 2.0. 100 μL containing 10 7 colony forming units (CFU) were propagated on C-Glc minimal medium plates that were incubated for 24 h at 37 C. (C) Time-dependent emergence of GS-resistant B. anthina mutants.B. anthina was grown in C-Glc medium at 37 C until an OD 600 of about 2.0. 10 7 CFU were propagated on C-Glc plates supplemented with 10 mM GS.The plates were incubated for up to 3 days at 37 C and the emerging mutants were counted.Dots indicate biologically independent replicates and bars indicate mean values.(D) Evaluation of growth of the parental strain and three isolated GS-resistant mutants on C-Glc plates in the absence and in the presence of the herbicide.The plates were incubated for 24 h at 37 C. (E) Growth of the parental strain and three isolated GSresistant mutants in C-Glc medium supplemented with increasing amounts of GS in a microplate reader at 37 C.

F
I G U R E 4 Relative quantification of the cellular PEP concentrations in Burkholderia anthina and Escherichia coli.(A) Relative quantification of the cellular PEP concentrations in the GS-resistant B. anthina mutants.The asterisk indicates that the PEP concentration was normalized to the parental strain S2.1.(B) Relative quantification of the cellular PEP concentrations E. coli strain W3110 expressing the native PpsA enzyme from plasmid pBP1224 (+PpsA).The PEP concentration was normalized to the E. coli strain W3110 carrying the empty plasmid pGP380.PEP concentrations in the suppressor mutants are shown as log 2 -fold changes compared to the wild type concentration.Mean value and standard deviation of three replicates are shown.
U R E 5 Effect of ppsA overexpression on GS resistance in Escherichia coli.Growth of the E. coli strain W3110 carrying the plasmids pGP380 (ÀPpsA) and pBP1224 (+PpsA) C-Glc minimal medium supplemented with the indicated amounts of GS and glutamate in a microplate reader at 37 C.

F
I G U R E 6 Characterization of the Burkholderia anthina mutant with increased GS resistance.(A) Evaluation of growth of the parental strain and isolated GS-resistant mutant S2.1 on C-Glc plates in the absence and in the presence of 35 mM GS.The plates were incubated for 24 h at 37 C. (B) Evaluation of growth of the parental strain and isolated GS-resistant mutant S2.1 in C-Glc liquid supplemented in the absence and in the presence of 35 mM GS. (C) The pyruvate kinase catalyses the conversion of ADP/Pi and PEP to ATP and pyruvate in the glycolytic pathway.The PEP synthetase converts pyruvate and ATP to AMP/Pi and PEP that is required for gluconeogenesis.(D) Growth of the B. anthina mutant S2.1 in C medium supplemented with 0.5% (w/v) glucose, 0.6% (w/v) succinate and 0.8% (w/v) glutamate in a microplate reader at 37 C.