Glycine betaine metabolism is enabled in Methylorubrum extorquens PA1 by alterations to dimethylglycine dehydrogenase

ABSTRACT

Low nutrient availability is a key characteristic of the phyllosphere (the aerial surface of plants). Phyllospheric bacteria utilize a wide array of carbon sources generated by plant hosts. Glycine betaine (GB) is a plant-derived compound that can be metabolized by certain members of the phyllosphere microbiota. Metabolism of glycine betaine generates formaldehyde, an intermediate of methylotrophic metabolism, leading us to investigate how the ubiquitous plant colonizing bacterium Methylorubrum extorquens PA1 might metabolize GB encountered in its native environment. M. extorquens PA1 cannot utilize GB as a sole carbon source. Through suppressor mutation analysis, we show that M. extorquens PA1 encodes a conserved GB utilization pathway that can be activated by single point mutations conferring GB utilization as a carbon source. We identified the gene cluster encoding the GB catabolic enzymes and found that gene expression was induced in the presence of GB. We show that utilization of GB is conserved among representative Methylobacterium species and generates the one-carbon metabolism intermediate formaldehyde, which M. extorquens utilizes as a source of energy. Our results support a model where suppressor mutations in Mext_3745 or ftsH (Mext_4840) prevent the degradation of the dimethylglycine dehydrogenase subunit DgcB by the membrane integral protease FtsH, conferring the ability to utilize GB by either (i) restoring stable membrane topology of DgcB or (ii) decreasing FtsH protease activity, respectively. Both mutations alleviate the bottleneck at the second step of GB degradation catalyzed by DgcAB.

IMPORTANCE

Overcoming low nutrient availability is a challenge many bacteria encounter in the environment. Facultative methylotrophs are able to utilize one-carbon and multi-carbon compounds as carbon and energy sources. The utilization of plant-derived glycine betaine (GB) represents a possible source of multi-carbon and one-carbon substrates. The metabolism of glycine betaine produces formaldehyde and glycine, which may be used simultaneously by facultative methylotrophs. However, the genes required for the utilization of GB in the ubiquitous plant-associated bacterium Methylorubrum extorquens have yet to be identified or described. Our work identifies and validates the genes required for glycine betaine metabolism in M. extorquens and shows that it directly intersects with methylotrophic metabolism through the production of formaldehyde.

INTRODUCTION

The phyllosphere (aerial surface of plants) is a harsh and rapidly changing habitat, characterized by low nutrient availability, diurnal cycling of light, ultraviolet radiation exposure, variable water availability, and large temperature shifts (1, 2). The global phyllosphere comprises an area twice the size of the surface of the earth, making it one of the largest habitats on our planet (1). In phylogenetically and geospatially distinct host plants, phyllosphere microbial populations are dominated by alpha-proteobacteria from the genera Methylobacterium, Methylorubrum, and Sphingomonas (1, 3–5). The phyllosphere microbiome has an essential role in plant growth, stress tolerance, and protection against pathogens (2, 6, 7). However, comparatively little is known about how commensal and beneficial bacteria survive in the phyllosphere.

One major challenge to life in the phyllosphere is low nutrient availability; therefore, metabolic versatility is a common trait among phyllospheric bacteria (1, 2). In addition to being able to utilize diverse carbon sources, methylotrophic bacteria from the genera Methylobacterium and Methylorubrum can use one-carbon growth substrates such as methanol, produced by plants, unlike many other phyllospheric bacteria (1). Utilization of methanol as a source of carbon and energy requires significant flux through formaldehyde-generating pathways. Glycine betaine (GB) is an N-trimethylated plant-derived compound present in the phyllosphere, which is predicted to generate formaldehyde as a byproduct of its degradation to glycine via sequential demethylation of GB (8) (Fig. 1). Given methylotrophs’ ability to assimilate formaldehyde-derived carbon into biomass and tolerate formaldehyde during methylotrophic growth, it may be advantageous for Methylobacterium or Methylorubrum species to utilize plant-derived GB in the phyllosphere during periods of low methanol availability.

Fig 1.

Suppressor mutants that utilize GB as a sole source of carbon and energy can be isolated. (A) Diagram of GB utilization pathways in aerobic bacteria. Dimethylglycine, DMG; sarcosine, sar; glycine, Gly; serine, Ser; pyruvate, pyr; formaldehyde, CH₂O; methylene tetrahydrofolate, met-H₄F; glycine betaine monooxygenase, GBmo; dimethylglycine dehydrogenase, DMGdh; sarcosine oxidase, SOX; serine hydroxymethyltransferase, GlyA; and serine dehydratase, Sda. (B) Growth in liquid Methylobacterium piparazine-N,N′-bis(2-ethanesulfonic acid) medium supplemented with 8 mM GB was quantified for Methylorubrum extorquens PA1 [“wild type (WT),” black, filled circles], M. extorquens AM1 (black, open circles), and representative PA1 GB + suppressor isolates that were sequenced from Class I: JB258 (orange, closed circles) and JB263 (orange, open circles), Class II: JB267 (sky blue, closed circles) and JB283 (sky blue, open circles), and Class III: JB274 (pink closed circles). Error bars represent the 95% confidence interval of three biological replicates.

Methylorubrum extorquens AM1 can utilize GB as a sole source of carbon and energy; however, the pathway responsible has not been identified (9). There are two known aerobic GB degradation pathways in bacteria (Fig. 1A). In Pseudomonas species and Chromohalobacter salexigens, GB is converted to dimethylglycine (DMG) by GB monooxygenase, while in Sinorhizobium meliloti DMG is produced by a methyltransferase reaction with homocysteine (8, 10–12). Previous in vitro work suggests that the demethylation of GB and DMG produces formaldehyde, a potent toxin, while the demethylation of sarcosine may produce either formaldehyde or methylene-tetrahydrofolate (methylene-THF), depending on the enzyme carrying out the reaction (8, 13). In most bacteria, formaldehyde is oxidized to CO₂ by one of the numerous defined detoxification pathways (14), while the resulting glycine enters central metabolism via conversion to serine, which is then deaminated to pyruvate and ultimately oxidized in the tricarboxylic acid (TCA) cycle (15). In marine bacteria, formaldehyde detoxification during growth on GB proceeds via the THF-linked C1 oxidation pathway commonly found in environmental heterotrophic bacteria (16, 17).

M. extorquens has been extensively studied as a model methylotroph, where formaldehyde is an essential intermediate in dedicated carbon assimilation pathways (18, 19). This opens the possibility that M. extorquens uses a combination of C₁ and multi-carbon metabolism to catabolize GB, whereby both end-products, formaldehyde and glycine, serve as carbon and energy sources. The genes required for the catabolism of GB in M. extorquens have not been identified or described, leading us to investigate how this model plant-colonizing methylotroph metabolizes GB encountered in association with host plants.

M. extorquens strain PA1, isolated from the leaf surface of Arabidopsis thaliana, is an emerging model organism for the study of phyllosphere colonization and plant/microbe interactions and intriguingly does not utilize GB as a sole source of carbon and energy (4–6, 20). In this work, we show through suppressor analysis that M. extorquens PA1 encodes a GB utilization pathway that can be activated via single point mutations. Using isolated suppressors, we identify and characterize the gene cluster encoding the GB catalytic pathway and show that suppressor mutations identified in this study prevent targeting of essential GB-catabolizing enzymes from degradation by FtsH, an essential protease. Collectively, our data support a model in which GB catabolism is bifurcated into energy generation and carbon assimilation, where M. extorquens utilizes the methyl group carbons (released as formaldehyde) for energy generation to support the assimilation of glycine carbons into biomass via the TCA cycle. Finally, we demonstrate that the GB pathway is present in a variety of Methylorubrum species and that genome pathway completeness is well-correlated with their ability to use GB, suggesting it is conserved among the genus.

RESULTS

Isolation of GB utilizing mutants of M. extorquens PA1

M. extorquens PA1 is unable to utilize GB as a source of carbon and energy, unlike other sequenced strains including M. extorquens AM1 (Fig. 1B) (9). A cursory genomic analysis did not reveal known GB catabolic genes in AM1 or PA1 strains. Additionally, we considered that methylotrophs may catabolize GB through previously undescribed means, as each demethylation step is expected to generate formaldehyde, which M. extorquens is highly adapted to and may incorporate into biomass as a methylotroph. To determine whether PA1 encodes the metabolic potential to catabolize GB, we sought to isolate mutants that suppress the initial growth defect of the wild-type (WT) strain and allow growth on GB (“GB+” phenotype). Ten independent stationary-phase cultures of WT grown in Methylobacterium piparazine-N,N′-bis(2-ethanesulfonic acid) (MP) medium supplemented with 3.5 mM succinate as a sole carbon and energy source were plated on solid medium supplemented with 8 mM GB as the sole carbon and energy source and incubated at 30°C for 14 days. In total, 30 colonies (three from each independent culture) were selected for characterization. Individual isolates were assayed for growth in a liquid medium supplemented with 8 mM GB and fell into three broad classes (Fig. S1). Class I contained four isolates that had average generation times (gt) of 5.97 ± 0.23 hours. This phenotype was similar to the growth phenotype of M. extorquens AM1 on GB (gt = 5.89 ± 0.19 hours). The 17 Class II isolates had an average generation time of 7.48 ± 1.16 hours, while the nine Class III isolates had significantly longer generation times (gt = 11.14 ± 2.50 hours) and often lower final cell densities (OD₆₀₀ < 0.2). The ready isolation of GB+ mutants suggested that PA1 encodes a GB utilization pathway, which can be activated by mutation.

Mutations in Mext_3745 and Mext_4840 confer GB utilization

Whole-genome sequencing analysis of several suppressor mutants identified putative causal GB+ mutations (Fig. 1B; Table 1). Each of the GB+ isolates had nonsynonymous mutations in one of two genes: Mext_3745 or ftsH (Mext_4840). Two Class I isolates (JB258 and JB263) contained only a single point mutation, a C→T transition at position 89 of Mext_3745, annotated as encoding an iron-sulfur-binding protein that contains a domain of unknown function (DUF3483) and a GlpC-like Fe-S oxidoreductase domain. Mext_3745 mutations, which conferred the most robust GB growth, resulted in a Mext_3745^P30L variant (Table 1). To confirm the causative nature of these mutations, we reconstructed the Mext_3745^P30L allele in the WT background and replaced the Mext_3745^P30L allele in JB258 with the WT Mext_3745 allele by allelic exchange to generate JB767 and JB768, respectively. In both instances, the introduction of the Mext_3745 alleles reversed the GB phenotype of the parental strain, demonstrating the Mext_3745 mutation was causative(Table 2). From these data, we concluded that gain-of-function mutations in Mext_3745 conferred the GB+ phenotype in Class I mutants.

TABLE 1.

Mutations identified by whole-genome sequencing of selected GB+ isolates of M. extorquens PA1

Strain	Isolate (culture.plate)	Class	Locus: annotation	Mutation (position)	Variant (codon change)
JB258	2.B	I	Mext_3745: (Fe-S)-binding protein	C→T (89)	P30L (CCC→CTC)
JB263	4.A	I	Mext_3745: (Fe-S)-binding protein	C→T (89)	P30L (CCC→CTC)
JB267	5.A	II	Mext_4840: ATP-dependent zinc metalloprotease FtsH	T→C (1,598)	V533A (GTG→GCG)
JB274	7.B	III	Mext_4840: ATP-dependent zinc metalloprotease FtsH	T→C (649)	F217L (TTC→CTC)
JB283	10.B	II	Mext_4840: ATP-dependent zinc metalloprotease FtsH	C→A (337)	P113T (CCG→ACG)

TABLE 2.

GB utilization phenotypes^a

Strain	Genotype	Locus	Growth rate	Doubling time (hours)
CM2720	AM1 WT (Δbcs)		0.103	6.71
CM2730	PA1 WT (Δbcs)		NG
JB258	dgcBP30L	Mext_3745	0.105	6.62
JB263	dgcBP30L	Mext_3745	0.101	6.88
JB267	ftsHV533A	Mext_4840	0.084	8.24
JB274	ftsHF217L	Mext_4840	0.062	11.25
JB283	ftsHP113T	Mext_4840	0.096	7.24
JB767	dgcBP30Lsynth	Mext_3745	0.104	6.69
JB768	dgcBL30P	Mext_3745	NG
JB578	dgcBP30LΔproVUX	Mext_3745, Mext_3731–3733	NG
JB562	dgcBP30LΔgbcBA	Mext_3745, Mext_3747–3748	NG
JB669	ΔdgbBA	Mext_3745–3746	NG
JB668	dgcBP30LΔMext_3743–3744	Mext_3745, Mext_3743–3744	NG
JB575	dgcBP30LΔsoxBGAD	Mext_3745, Mext_3736–3739	NG
JB652	dgcBP30LΔfae	Mext_3745, Mext_1834	NG
JB649	dgcBP30LΔftfL	Mext_3745, Mext_0414	0.115	6.02
JB650	dgcBP30LΔsga	Mext_3745, Mext_1795	0.105	6.63
JB680	dgcBP30LΔsdaB	Mext_3745, Mext_3740	NG
JB897	dgcBP30LΔftfLΔgcvPHT	Mext_3745, Mext_0414, Mext_0853–0855	0.112	6.14
JB989	dgcA-His dgcB-FLAG	Mext_3746, Mext_3745	NG
JB990	dgcA-His dgcBP30L-FLAG	Mext_3746, Mext_3745	0.104	6.64
JB991	ftsHV533A dgcA-His dgcB-FLAG	Mext_3746, Mext_3745	0.079	8.84
JB992	ftsHF217L dgcA-His dgcB-FLAG	Mext_3746, Mext_3745	0.060	11.47
JB993	ftsHP113T dgcA-His dgcB-FLAG	Mext_3746, Mext_3745	0.093	7.42

^a NG, no growth.

A variety of single, nonsynonymous point mutations in ftsH were identified in Class II and Class III mutants resulting in FtsH^V533A (JB267), FtsH^F217L (JB274), and FtsH^P113T (JB283) variants. Despite several attempts, we were unable to generate a markerless deletion of ftsH and suspect that ftsH is essential in PA1 as it is in many bacteria (21). We complemented the ftsH mutations with a WT copy of ftsH under the control of a cumate-inducible promoter on plasmid pDM001 (Fig. S2). All complemented strains (JB631, JB633, and JB635) reverted to GB− phenotypes, suggesting the ftsH mutations were causative and recessive (Fig. S2C and F). Furthermore, we observed an increased lag time in ftsH mutants when grown on alternative carbon sources (Fig. S2A and B), which was also alleviated by complementation (Fig. S2D and E). These results strongly suggest the observed mutations in ftsH decreased essential proteolytic activity in vivo. Consistent with its essentiality, these decreased-function mutations conferred a GB+ phenotype but simultaneously led to a general growth defect on non-GB substrates.

Identification of putative GB catabolism gene cluster

Since Mext_3745 was a common mutational target in our initial selection and was predicted to encode an oxidoreductase, we suspected that it encoded an enzyme directly involved in GB catabolism and investigated its genomic context further. A detailed analysis of this cluster downstream of Mext_3745 revealed an ABC-type transporter predicted to transport GB or proline (proVUX), an AraC-type regulator (Mext_3735), a heterotetrameric sarcosine oxidase (soxBDAG), and a putative serine dehydratase (sdaB) (Fig. 2). Sarcosine oxidases (SOX) catalyze the N-demethylation of sarcosine (i.e., N-methylglycine) producing glycine and either formaldehyde or methylene-THF, while serine dehydratase catalyzes the dehydration of serine resulting in spontaneous deamination producing pyruvate (Fig. 1A). Based on the genetic context, we expected that Mext_3745 was involved in an earlier step of the pathway, the N-demethylation of either GB or DMG.

Fig 2.

Comparison of the genomic context of GB catabolic enzyme encoding genes in M. extorquens PA1 and Pseudomonas aeruginosa PAO1. Locus tags are listed above the arrows. Gene annotations are listed below arrows. Predicted functions are listed below brackets, where brackets indicate genes expected to form a protein complex.

To identify whether Mext_3745 encoded a novel or conserved GB-catabolizing enzyme and which step of GB catabolism Mext_3745 is involved in, we compared Mext_3745 to genes encoding characterized DMG dehydrogenases (DMGdh) and GB monooxygenases (GBmo) in Pseudomonas aeruginosa PAO1 and C. salexigens DSM 3043 and to GB-homocysteine methyltransferase from S. meliloti. DMGdh consisted of an alpha subunit and a beta subunit transcribed immediately upstream from a heterodimeric electron transfer flavoprotein (8, 11). Mext_3745 is found in an analogous genetic arrangement to the beta subunit of DMGdh in other organisms with Mext_3746 upstream and Mext_3744 and Mext_3743 downstream (Fig. 2). The beta subunit of DMGdh was predicted to contain a clostridial type ferredoxin domain and an N-terminal transmembrane domain consisting of five transmembrane helices (11). Therefore, we generated a structural model of Mext_3745 using the Phyre2 server and predicted transmembrane domains using TMpred (22, 23). Both applications predicted five transmembrane helices near the N-terminus of Mext_3745, with the proline translated at position 30 near the cytoplasmic end of the first transmembrane helix. The alpha subunit of DMGdh was shown to be an NAD(P)H-dependent flavin oxidoreductase that localizes to the cytoplasm (11). This description is consistent with the annotated description of Mext_3746. Taken together, these data suggested that Mext_3745 encoded the beta subunit of DMGdh (dgcB), Mext_3746 encoded the alpha subunit of DMGdh (dgcA), and Mext_3744/Mext_3743 encoded the electron transport flavoprotein predicted to function with DMGdh (Fig. 2; Fig. S3) (11).

Finally, we identified Mext_3747 and Mext_3748 as encoding putative GBmo subunits gbcB, and gbcA, respectively, by comparison to characterized GBmo subunits in P. aeruginosa and C. salexigens using BlastP (e values: 2 × 10⁻¹⁰⁶ and 5 × 10⁻¹¹⁵) (Fig. S3) (8, 10). In total, the putative GB catabolism cluster spanned a 23-kb region and contained a GB-specific transporter and putative enzymes required for the successive demethylation of GB to glycine.

Mutations in the GB pathway abolish GB utilization

To confirm the role of genes in the predicted GB cluster of M. extorquens PA1, we generated markerless deletions of the predicted transporter (ΔproVUX) and enzymes that catalyze each demethylation step (ΔMext_3747–3748, ΔMext_3745–3746, ΔMext_3743–3744, and ΔsoxBDAG) (Fig. 2) in the GB+ genetic background (JB258, dgcB^P30L). Each resulting strain was assayed for growth in a liquid medium with GB as the sole carbon and energy source and for the presence of GB pathway intermediates, post-GB exposure. As expected, none of the deletion mutants could grow on GB (Table 2). We extracted metabolites from strains incubated in the medium with GB as the sole carbon source and analyzed for GB, DMG, sarcosine, and glycine. Using thin layer chromatography, we found the ΔMext_3747–3748 mutant (JB562) accumulated cellular GB but no downstream intermediates (Fig. 3). In ΔMext_3745–3746 and ΔMext_3743–3744, we observed accumulation of DMG. Finally, in the ΔsoxBDAG mutant, we observed an accumulation of sarcosine. These data confirmed our predicted roles for each of the operons in the GB pathway and allowed us to annotate Mext_3745–3746 as dgcB and dgcA, and Mext_3747–3748 as gbcB and gbcA. Additionally, because Mext_3743–3744 (putative electron transport flavoproteins) are required for DMGdh function and are within the dgcB operon, we annotated them as dgcD and dgcC. We observed the accumulation of GB and DMG in ΔproVUX, suggesting that alternative GB transporters are encoded in PA1. Notably, we found that DMG accumulates in WT but not in the dgcB^P30L mutant (Fig. 3). Taken together, our data suggest that in WT, GB utilization is hindered at the DMGdh step and that the P30L nonsynonymous substitution in the dgcB^P30L mutant restores DMGdh activity.

Fig 3.

DMG accumulates in WT PA1 and GB pathway intermediates accumulate in GB pathway deletion mutants. Thin layer chromatographic analysis of metabolites isolated from M. extorquens PA1 strains incubated with GB. Lanes were spotted at the origin, and the migration positions of authentic standards are indicated. Mext_3745^P30L = dgcB^P30L, ΔMext_3747–3748 = ΔgbcBA, ΔMext_3745–3746 = ΔdgcBA, and ΔMext_3743–3744 = ΔdgcDC.

The GB catabolism gene cluster is found in other GB-utilizing Methylobacteriaceae species

We assessed whether the presence of a GB utilization cluster correlates with growth on GB for other Methylorubrum and Methylobacterium strains. We assayed seven additional well-studied strains for growth in liquid medium supplemented with 8 mM GB as a sole source of carbon and energy. M. extorquens AM1, M extorquens CM4, M. extorquens DM4, Methylobacterium nodulans, and Methylobacterium radiotolerans utilized GB, while Methylobacterium populi and Methylobacterium sp. 4-46 did not (Fig. S4A). Based on genomic context, we identified complete GB utilization clusters in each of the five GB+ species (Table S1; Fig. S4B). Methylobacterium spp. 4-46 was the lone exception that lacked a GB gene cluster altogether, while M. populi BJ001 contained most of the GB cluster but lacked proV and proU genes encoding two components of the ABC transporter with specificity for GB. M. nodulans contained two nearly complete GB clusters. Cluster 1 lacked the proV component of the ABC transporter, while cluster 2 lacked the entire proVUX transporter. From these data, we concluded that the conserved GB gene clusters encoded the GB pathway enzymes required for GB utilization and that the ABC transporter encoded by proVUX may be required in some species such as M. populi but may be compensated for by another transport mechanism in others such as M. nodulans.

Expression of GB cluster genes is increased during the utilization of GB

We expected that genes in the GB cluster would be expressed specifically when GB was a growth substrate and sought to determine whether the GB− phenotype was due to transcriptional inactivity of GB utilization genes. We assayed the gene expression of a representative gene from each operon in the GB cluster (proV, gbcB, dgcB, soxA, and sdaB) in the dgcB^P30L mutant when utilizing GB or pyruvate. Expression of proV, gbcB, dgcB, soxA, and sdaB increased substantially in cells utilizing GB compared to cells utilizing pyruvate, confirming GB-mediated induction of gene expression (Fig. S4A). To rule out the possibility that the GB pathway is inactive in WT PA1 due to a lack of transcription, we assayed the expression of genes in the GB neighborhood using methanol-grown cells treated with and without 1 mM GB. Expression of each representative gene was increased by the presence of GB (Fig. S4B). From this, we concluded that the GB cluster of genes is activated in response to GB and that the bottleneck in the DMGdh-catalyzed step of the pathway observed in WT PA1 is not due to a lack of transcription.

GB catabolism produces formaldehyde, which is utilized as an energy source

It is unknown whether the demethylation of GB produces free formaldehyde or an activated intermediate such as methylene-THF. Free formaldehyde is a toxin that can be detoxified by non-methylotrophs, but for methylotrophs, it also serves as a potential source of carbon and energy. M. extorquens strains lacking fae, which encodes formaldehyde-activating enzyme, are sensitive to the presence of both exogenous and endogenous formaldehyde (24). We generated markerless deletions of fae in the dgcB^P30L mutant background and assayed for growth in liquid GB or pyruvate medium. The dgcB^P30LΔfae strain failed to grow with GB, but its growth was indistinguishable from the dgcB^P30L parent when grown with pyruvate, suggesting that catabolism of GB produces free formaldehyde (Fig. S5; Table 2). Furthermore, we measured the formaldehyde secreted into the supernatant as a proxy for intracellular formaldehyde production and saw a significant increase in the formaldehyde concentration in GB-grown cells compared to pyruvate-grown cells (Fig. 4A). As a positive control, formaldehyde concentrations were also elevated in methanol-grown cells as previously seen (25, 26) (Fig. 4A). Based on this experiment, we concluded that utilization of GB produces free formaldehyde, which is detoxified by methylotrophy pathways, thus linking GB metabolism and C₁ metabolism.

Fig 4.

Catabolism of GB produces free formaldehyde. (A) Mid-log phase formaldehyde concentrations in the growth medium of the dgcB^P30L mutant were measured by colorimetric assay from cells grown in MP with 8 mM pyruvate, 15 mM methanol (MeOH), or 8 mM GB. (B) Expression of genes involved in C₁ metabolism (fae, mtdB, ftfL, and mtdA), serine cycle (hprA and mtkA), and TCA cycle (aceE and sucA) of cells utilizing GB compared to cells utilizing pyruvate or (C) methanol. Individual values shown are averages of three technical replicates. The mean of three independent biological replicates is indicated. Error bars represent the 95% confidence interval of three biological replicates.

In non-methylotrophic organisms, the catabolism of GB proceeds via a linear pathway where glycine produced by serial demethylations is converted to serine by serine hydroxymethyltransferase (GlyA), which is deaminated to pyruvate by serine dehydratase (SdaB). The formaldehyde generated by each demethylation is detoxified by oxidation to CO₂ (8). This process is more complex in serine cycle methylotrophs such as M. extorquens PA1 because the formaldehyde byproduct represents a potential source of carbon for biomass (Fig. 5). Furthermore, glycine and serine are intermediates of the serine cycle, and the GlyA reaction is an essential step in the assimilation of C₁ units during methylotrophic growth. Previous work has shown that when M. extorquens AM1 is grown in the presence of multi-carbon (succinate) and a C₁ (MeOH) carbon sources, the C₁ carbon is dissimilated as a source of energy, while the multi-carbon compound is assimilated into biomass (27). To test whether GB utilization would follow an analogous pattern, we disrupted the first step of the assimilatory branch of the C₁ pathway by deleting ftfL in the dgcB^P30L mutant background (Fig. 5). The dgcB^P30LΔftfL strain maintained its GB+ phenotype, which was indistinguishable from the dgcB^P30L parent (Fig. S5B; Table 2). This demonstrated that under these conditions, the formaldehyde generated by the demethylation of GB and DMG is oxidized to CO₂ and preferentially used as a source of energy.

Fig 5.

Diagram of expected carbon flow in PA1 when utilizing GB as a sole source of carbon and energy. Proteins encoded by genes in the GB cluster are shown in orange, dark orange, blue, green, and yellow. Genes’ names for genes encoding enzymes are shown above or beside the corresponding reactions in bold italic text. Enzyme abbreviations are in the standard text above reactions. Heavyweight arrows represent expected carbon flow during growth on GB. Lightweight arrows represent non-essential pathways. Light blue arrows correspond to reactions where deletions have no effect on the metabolism of GB. Red arrows correspond to reactions where deletion resulted in the inability to grow on GB as a source of carbon. Dotted lines represent observed reactions in GB metabolic networks, which are not present in M. extorquens PA1. Asterisks denote gene sets where a representative gene was evaluated by real-time quantitative PCR. Enzymes: glycine betaine monooxygenase (GBmo), dimethylglycine dehydrogenase (DMGdh), sarcosine oxidase (Sox), serine dehydratase (Sda), and betaine homocysteine methyltransferase (BHMT). Metabolites: sarcosine (sar), glycine (Gly), serine (Ser), pyruvate (pyr), acetyl-CoA (ace-CoA), citrate (cit), 2-oxoglutarate (2-oxg), succinyl-CoA (suc-CoA), oxaloacetate (oxa), 2-phosphoglycerate (2P-glc), glycerate (glc), hydroxypyruvate (hpr), glyoxylate (glx), ammonia (NH₃), methyl group (CH₃), formaldehyde (CH₂O), formate (for), methylene tetrahydrofolate (met-H₄F), tetrahydrofolate (H₄F), tetrahydromethanopterin (H₄MPT), carbon dioxide (CO₂), homocysteine (Hcys), and methionine (Met).

Real-time quantitative PCR of C₁ genes further supported formaldehyde generation during GB utilization. Specifically, the expression of genes required for formaldehyde oxidation to formate (fae and mtdB) was increased in GB-utilizing cells when compared to pyruvate-utilizing cells that do not generate formaldehyde (Fig. 4B) and similar to methanol-utilizing cells that generate formaldehyde as an obligate intermediate of C₁ metabolism (Fig. 4C). Expression of ftfL was decreased in GB-utilizing cells when compared to methanol-utilizing cells but expressed at similar levels to cells grown on pyruvate (Fig. 4C), further supporting that formaldehyde from GB demethylation is not assimilated into biomass. Expression of a gene further downstream in the assimilatory C₁ pathway (mtdA) had an identical expression pattern, leading us to conclude that serial demethylation of GB produces formaldehyde, which PA1 utilizes for energy and reducing power via successive oxidation reactions to formate and carbon dioxide by the tetrahydromethanopterin (H₄MPT) pathway and formate dehydrogenase.

Although our data show that GB utilization generates free formaldehyde, it is unclear whether the sarcosine oxidase-catalyzed demethylation contributes to the free formaldehyde pool or generates methylene-THF (13). To test which scenario is most likely in PA1, we generated a markerless deletion of the glycine cleavage complex, a major cellular source of methylene-THF, in the dgcB^P30LΔftfL background. The resulting genotype (dgcB^P30LΔftfL ΔgcvPHT, JB897) causes a drastic growth defect on most carbon substrates, including pyruvate, as cells lack both primary sources of THF-linked C1 intermediates (Fig. S5A). However, during growth on GB as a sole source of carbon, the dgcB^P30LΔftfL ΔgcvPHT mutant did not have a demonstrable growth defect (Fig. S5B; Table 2), suggesting that the demethylation of sarcosine generated sufficient methylene-THF to support growth, drove the serine hydroxymethyltransferase reaction forward, and supported one-carbon metabolism. Based on these results, we expect that demethylation of GB and DMG produces free formaldehyde, which is oxidized to formate by the H₄MPT pathway and then to CO₂ by formate dehydrogenase to produce energy and reducing equivalents. Furthermore, our results support a model where demethylation of sarcosine produces methylene-THF and glycine, which are utilized by GlyA to produce serine (Fig. 5).

GB metabolism utilizes the TCA cycle

Whereas non-methylotrophs utilize GB via the TCA cycle, serine cycle methylotrophs must integrate C₁ pathways, the serine cycle, and the TCA cycle to efficiently utilize GB and thus have numerous potential entry points for GB-derived carbon (Fig. 5). We predicted that GB is most likely utilized via the TCA cycle as sdaB, which encodes serine dehydratase, is located within the GB cluster. We generated clean deletions at the entry point to the serine cycle and the linear GB pathway, sga and sdaB in the GB+ parent strain background (dgcB^P30L), and assayed each strain for growth in liquid GB medium. The dgcB^P30LΔsga strain maintained the ability to utilize GB, while the dgcB^P30LΔsdaB strain did not (Table 2). For additional confirmation, we assayed the expression of genes involved in C₁ assimilation (ftfL, Mext_0414, and mtdA, Mext_1797), serine cycle (hprA, Mext_1796, and mtkA, Mext_1800), and TCA cycle (aceE, Mext_2786 and sucA, Mext_1646) genes. In all cases, gene expression levels of GB-utilizing cells were similar to pyruvate-grown cells, which use the TCA cycle (Fig. 4B). Additionally, expression levels of C₁ assimilation and serine cycle genes were lower in GB-utilizing cells than in methanol-utilizing cells, while TCA cycle genes were expressed at high levels (Fig. 4C). These results strongly suggest that Sda deaminates the serine generated from GB, and the resulting pyruvate is decarboxylated to acetyl-CoA before entering the TCA cycle (Fig. 5).

DgcB is targeted for degradation by FtsH

Given that FtsH is a membrane integral protease responsible for protein quality control of membrane-bound proteins (28–31) and DgcB has a membrane-bound domain, we hypothesized that FtsH was targeting DgcB for degradation and that the P30L substitution of DgcB in the GB+ mutant alleviated FtsH-mediated degradation. To test this hypothesis, we performed Western blot analysis of genomically tagged DMGdh subunits (DgcA-His and DgcB-FLAG) after first confirming that both dgcA and dgcB are expressed in response to GB in all strains (Fig. 6A). Tagging the DMGdh subunits had no measurable effect on growth rate (Table 2). Our data showed that, as expected, DgcA is present in all strains at similar levels when GB is present (Fig. 6B). By contrast, DgcB was undetectable in WT, but was present at varying levels in the dgcB^P30L, ftsHV^533A, ftsH^F217L, and ftsH^P113T mutants (Fig. 6B). The level of DgcB present in each ftsH mutant correlated positively with the growth rate of each strain on GB, confirming that DMGdh is rate limiting in the ftsH mutants and that strains carrying ftsH mutations have decreased FtsH-mediated protein degradation. Based on these results, we concluded that FtsH degrades WT DgcB, preventing WT PA1 from catabolizing DMG and resulting in an enzymatic block in the GB catabolic pathway (Fig. S6). This is consistent with the results described above showing the accumulation of DMG in WT PA1. In the dgcB^P30L GB+ mutant, FtsH no longer degrades DgcB, allowing for the efficient conversion of DMG to sarcosine and the utilization of GB as a source of carbon and energy.

Fig 6.

Proline 30 targets DgcB for degradation by FtsH. (A) Expression of dgcA and dgcB in M. extorquens PA1 cells in the presence of GB compared to succinate. Individual values shown are averages of three technical replicates. The mean of three independent biological replicates is indicated. Error bars represent the 95% confidence interval of three biological replicates. (B) Western blot analysis of whole cell extracts of M. extorquens PA1 strains targeting 6×His-tagged DgcA or FLAG-tagged DgcB of cells grown in the presence (left) or absence (right) of GB in the growth medium.

DISCUSSION

Given that M. extorquens PA1 was isolated from the surface of A. thaliana leaves and has been utilized as a model system for methylotrophy/plant interactions (4, 6, 7), we sought to understand GB utilization in conditions reflective of its native environment. GB is a prevalent metabolite in the phyllosphere, where it is produced by many plant species as an osmolyte (32, 33). Nutrient limitation is a ubiquitous feature of the phyllosphere; however, phyllosphere microbiota utilization of plant-produced compounds such as GB is poorly understood. Using suppressor mutation analysis, we isolated and characterized GB+ strains of PA1, ultimately unveiling a common pathway for utilizing GB as a sole carbon and energy source in Methylobacterium. A gain of function mutation in a putative DMGdh subunit-encoding gene (dgcB) resulted in a DgcB^P30L variant protein that enabled cells to efficiently convert DMG to sarcosine, alleviating an apparent bottleneck in the native GB utilization pathway of M. extorquens PA1. The dgcB gene was found within a 23-kb cluster that also contained genes encoding a putative transporter and enzymes for all three demethylation steps of GB catabolism to glycine. Herein, our analyses are limited to the use of GB as a source of carbon; however, we address other uses of GB, as a nitrogen source and osmoprotectant, in our companion manuscript (34).

The ecological reason for the inability of M. extorquens PA1 to metabolize GB remains unknown. A. thaliana, the host from which M. extorquens PA1 was isolated, does not naturally produce GB, suggesting there is little selective pressure to maintain a functional GB-catabolizing pathway (33, 35). Another potential explanation lies in the metabolic conundrum posed by the simultaneous utilization of multi-C and C₁ carbon sources. In PA1, the C₁ utilization assimilates carbon via the tetrahydrofolate pathway and the serine cycle, whereas utilization of C₂-C₄ carbon sources proceeds via the TCA cycle. Both the serine cycle and the TCA cycle require malate dehydrogenase; however, each pathway requires the enzyme to run in opposing directions making the concurrent utilization of the serine and TCA cycles conflictual. Indeed, when PA1 utilizes C₁ (MeOH) and multi-C (succinate) carbon sources simultaneously, MeOH is completely oxidized as an electron source, while nearly 100% of the carbon assimilated into biomass is derived from succinate (27). The fate of each C₁ unit derived from GB was unknown in PA1. Our results suggest that PA1 uses a bifurcated metabolism, where C₁ units generated by demethylation are oxidized completely while carbon from the core glycine molecule is incorporated into biomass. This pattern of assimilation is consistent with how non-methylotrophs metabolize GB and a similar scheme involving oxidation of C1 units by the THF-linked C1 pathway to produce ATP and reducing equivalents to support assimilation of the core glycine molecule as a source of carbon (15, 17).

We confirmed each putative step in the successive demethylation of GB to glycine in PA1. Demethylation of GB to DMG can be catalyzed by either GBmo, as in P. aeruginosa, or GB:homocysteine methyltransferase, as in S. meliloti (8, 12). Despite the evolutionary distance between M. extorquens PA1 and P. aeruginosa being greater than with S. meliloti, we identified a GBmo in PA1. Furthermore, the PA1 genome does not encode annotated putative GB:homocysteine methyltransferases. This suggests the GB metabolic cluster may have been acquired by horizontal gene transfer, possibly from a phyllospheric Pseudomonas species. However, the genomic organization of the GB cluster in PA1 differs from that of P. aeruginosa (Fig. 2). Intriguingly, the cluster (and the PA1 genome at large) lacks homologs of PA5396 and PA5397, which have been shown in P. aeruginosa to be required for DMGdh functionality (8).

In addition to the DMGdh subunits, dgcB is encoded in an operon with three other genes. Mext_3743–3744 encodes a putative electron transport flavoprotein that is the expected electron acceptor of DMGdh, but this function has not been definitively shown despite these genes’ presence in other organisms at the same location. Our results showed DMG accumulating in a ΔMext_3743–3744 mutant lacking these putative electron transport flavoproteins, confirming both their role and specificity in the demethylation of DMG to sarcosine. Accordingly, we recommend annotating Mext_3743 and Mext_3744 as dgcD and dgcC, respectively. Though Mext_3742 is encoded in an operon with dgcBCD, its function remains unclear. Mext_3742 is annotated as encoding a YbaK-type prolyl-tRNA synthetase; these transferases serve an editing function whereby prolyl-tRNA that have been incorrectly charged with cysteine are deacylated to prevent misincorporation of cysteine (36). If charged cysteinyl-tRNA reacts with formaldehyde, a thioprolyl-tRNA would be generated and incorporation of thioproline into nascent polypeptides may result in proteotoxicity (37). As Mext_3742 is encoded with a formaldehyde-producing enzyme, we surmise that Mext_3742 may function to deacylate formaldehyde-damaged charged cysteinyl-tRNA, preventing the problematic incorporation of thioproline into nascent polypeptides.

We hypothesized that GB catabolism results in the production of formaldehyde. Using biochemical and genetic methods, we confirmed that GB demethylation produces formaldehyde and thus GB catabolism represents a potential compounding source of formaldehyde when PA1 is living on leaf surfaces and consuming other formaldehyde-generating carbon sources such as MeOH. However, unlike formaldehyde generated in the periplasm during methylotrophic growth, GB-derived formaldehyde is cytoplasmic and is more proximal to formaldehyde-sensitive macromolecules such as DNA, RNA, and protein. Thus, GB catabolism represents a novel endogenous source of cytoplasmic formaldehyde stress. Formaldehyde-responsive proteins that coordinate the central cellular response to endogenous formaldehyde stress have recently been identified in PA1 (25, 38, 39). Therefore, GB utilization, alone or in combination with other formaldehyde-generating growth substrates could be leveraged to understand how PA1 responds to formaldehyde stress originating from different cellular compartments and from mixed metabolism in general.

In addition to mutations in dgcB, we also observed mutations in ftsH resulting in a GB+ phenotype. FtsH is a membrane integral protease responsible for diverse functions, including the regulation of lipopolysaccharide synthesis, stress tolerance, and protein quality control of membrane-bound proteins (28–31). Consistent with the wide-ranging role of FtsH, GB+ strains with ftsH mutations had altered phenotypes in all conditions tested. We linked the phenotype of gain-of-function dgcB mutants and ftsH mutants by showing that proline 30 (Pro30) of DgcB targets DgcB for degradation by FtsH in M. extorquens PA1. Pro30 is predicted to be at or near the cytoplasmic face of the first transmembrane helix of DgcB. Proline is known to disrupt alpha helices and interrupt transmembrane helices (40, 41). Other Methylobacterium and Methylorubrum strains with the ability to utilize GB as a carbon source assayed herein encoded either alanine or serine at position 30 of dgcB (Fig. S7). These residues are structurally and chemically similar to the leucine of the DgcB^P30L variant we observed in PA1, which suggests the possibility that PA1 acquired a loss-of-function mutation in dgcB prior to isolation. In Staphylococcus aureus, the substitution of leucine with proline in a transmembrane helix in SaeS results in SaeS degradation by FtsH (42). This suggests that Pro30 of DMGdh results in an altered, unstable membrane topology in DgcB that causes it to be targeted for degradation by FtsH, consistent with the role of FtsH in membrane protein quality control.

The link between mutations observed in dgcB and ftsH leads to questions about the potential benefit of the accumulation of dimethylglycine and the origins of the GB− phenotype of PA1. It is possible that PA1 accumulates DMG as a supplementary osmoprotectant while preventing the accumulation of GB, which has been shown to inhibit the production of intracellular osmolytes during salt stress in Pseudomonas syringae (43). It has been predicted that DMGdh in S. meliloti is also able to demethylate the plant-derived osmolyte stachydrine (N,N dimethylproline) (44). This leaves open the possibility that mutations in PA1 DMGdh allow for the accumulation of stachydrine as an osmoprotectant.

Together, this work confirms the rapid transition from GB− to GB+ phenotype by M. extorquens PA1, enabled by a single mutational change to dgcB. We further describe the generation of formaldehyde through the subsequent demethylation steps of GB and suggest a bifurcated usage of carbons derived from GB, with C₁ carbons passing through the H₄MPT-dependent methylotrophic pathway in PA1, while glycine-derived carbons are incorporating via central metabolism (TCA cycle). Finally, our study suggests an integral role of FtsH in the function of DgcB in the cell.

MATERIALS AND METHODS

Bacterial strains, media, and chemicals

Bacterial strains used in this study (Table S2) are derived from Methylorubrum extorquens PA1 (4) with cellulose synthase genes deleted (ΔbcsSABZCN, previously reported as Δcel) to prevent aggregation and optimize growth measurements in liquid culture (45). Bacterial strains were cultivated using Methylobacterium piparazine-N,N′-bis(2-ethanesulfonic acid) medium (45) with 3.5 mM succinate, 15 mM methanol, 8 mM glycine betaine, or 8 mM pyruvate as a carbon source. When grown on a solid MP medium, carbon source concentrations were increased for succinate (15 mM) and methanol (125 mM). When present in the medium, tetracycline was used at a final concentration of 12.5 µg/mL, and Bacto agar was used at 15 g/L to solid media. Chemicals and reagents were purchased from Sigma-Aldrich.

Isolation of GB+ suppressors

Ten independent colonies of M. extorquens PA1 were inoculated into 5 mL of MP medium supplemented with 15 mM succinate. Cultures were incubated at 30°C on a rotary drum apparatus until the early stationary phase (~30 hours). A volume of 100 µL of each culture was plated onto three independent plates (30 in total) of solid MP media supplemented with 8 mM GB. The plates were incubated at 30°C for 14 days before colonies arose. One isolate from each plate was chosen for characterization. After initial characterization, genomic DNA was isolated from five GB+ suppressor strains and sequenced by short-read Illumina sequencing. Suppressor mutations were identified by comparison to the M. extorquens PA1 reference genome using Breseq (46).

Genetic approaches

Markerless deletions were generated by allelic exchange as previously described with modifications (47). Bi-parental conjugations were performed by mixing E. coli S17-1 cells carrying the pCM433-based or pCH07-based donor plasmid with M. extorquens PA1 (Table S3). The mixture was grown overnight on nutrient agar plates at 30°C, resuspended in MP medium lacking carbon and nitrogen, serially diluted, and plated on a selective medium supplemented with 15 mM succinate, 5 mM methylamine (as the sole source of nitrogen for counterselection against E. coli), and 12.5 µg/mL tetracycline. Sucrose selection for pCM433-mediated allelic exchange was accomplished by streaking isolated colonies from selection plates on an MP medium supplemented with 15 mM succinate and 5% sucrose. Donor plasmids and primers were designed using SnapGene software. Plasmids were assembled using NEB HiFi assembly kits.

Growth quantification

Isolated colonies were inoculated into 2 mL MP medium supplemented with 3.5 mM succinate in biological triplicate. Cultures were incubated at 30°C on a rotary drum apparatus and then subcultured (1/64) into 5 mL of MP medium supplemented with a relevant carbon source for acclimation. After acclimation (~30 hours: succinate or pyruvate, ~40 hours: methanol, and ~72 hours: glycine betaine), stationary phase cultures were subcultured (1/64) into 640 µL of MP medium supplemented with appropriate carbon sources in 48-well polystyrene plates (Falcon, Ref. No. 351178). The plates were sealed with adhesive optical film (VWR, Cat. No. 60941-064) to prevent the evaporation of volatile metabolites. The cultures were incubated at 30°C with a 2°C gradient to prevent condensation in a BioTek Epoch 2 microplate spectrophotometer with single orbital shaking at 800 rpm. Cell density was determined by measuring absorbance at 600 nm every hour. Cell growth rate (μ) and generation time (gt) were determined by non-linear regression of data points determined to be within the exponential growth phase where μ = ln(X_T/X₀)/T and gt = ln2/μ, where X is OD₆₀₀ and T is elapsed time.

Analysis of GB pathway intermediates

GB, DMG, sarcosine, and glycine present in the samples were determined by thin-layer chromatography, as previously described (48, 49). Briefly, samples were obtained by extracting 5 mL of cells normalized to OD₆₀₀ of 0.3 after incubation for 6 hours in MP supplemented with 8 mM GB twice with 70% ethanol at 65°C before filtration and drying by vacuum evaporation. Dried extracts were resuspended in 50 µL sterile H₂O. A 10-µL aliquot of standard or experimental sample was spotted onto Silica Gel 60 plates (Sigma-Aldrich) that were pretreated with MeOH. The plates were run in 90% (vol/vol) phenol in H₂O, air dried, and stained with 0.04% (wt/vol) bromocresol green in 100% ethanol titrated to dark green and heated with a heat gun to develop. Developed plates were scanned with a Canon ImageCLASS digital scanner, and images were converted to monochrome.

Real-time quantitative PCR

M. extroquens PA1 cultures (5 mL) were harvested in the mid-log phase (OD₆₀₀ 0.30 ± 0.02) and centrifuged at 3,234 × g for 10 min. Cell pellets were frozen at −80°C before resuspension in RNAsnap and incubated at 95°C for 7 min. Cell debris was removed by centrifugation at 16,000 × g for 5 min (50). RNA was extracted from the RNAsnap solution using the RNA Clean and Concentrator kit (Zymo Research). RNA was quantified using a NanoDrop One, and quality was assessed by denaturing agarose gel electrophoresis. Genomic DNA was removed by incubation with DNase I (ThermoFisher) per the manufacturer’s instructions. cDNA was synthesized using a High-Capacity cDNA Reverse Transcription kit (Applied Biosystems) according to the manufacturer’s instructions. qPCR reactions were run using Maxima SYBR Green/ROX qPCR Master Mix (ThermoFisher) in Fast Optical 96-Well reaction plates (MicroAmp) in 10 µL total volume. Reactions were run on a StepOnePlus Real-Time PCR system (Applied Biosystems) using the following conditions: 95°C for 10 min, followed by 40 cycles of 95°C for 20 seconds, 60°C for 20 seconds, and 72°C for 60 seconds with a plate-reading step after each cycle. Melt-curve analysis was performed under the following conditions: 95°C for 20 seconds, and a 70°C – 95°C ramp increasing at 0.3°C/second. Each reaction was carried out in technical triplicate for three independent biological replicates. StepOne Software v2.3 was used to obtain and analyze the qPCR reaction data and calculate cycle threshold values (C_T) values for each reaction. Technical triplicate values were averaged and normalized to recA and compared using the 2^-ΔΔCt method (51).

Formaldehyde quantification

Formaldehyde concentrations in culture media were measured using the colorimetric Nash method as described previously (52). Briefly, 1,000-µL aliquots from the mid-log phase cultures were centrifuged at 16,000 × g to pellet cells. A volume of 180 µL of the resulting supernatant was combined with 20 µL of Nash reagent B (2 M ammonium acetate, 50 mM glacial acetic acid, and 20 mM acetylacetone) in 96-well polystyrene plates in technical triplicate. Reaction plates were incubated for 10 min at 60°C and cooled to room temperature before reading absorbance at 432 nm on a SpectraMax i3x spectrophotometer. Formaldehyde standards (100 nM–1 mM) were prepared from 1 M formaldehyde stock solutions and used to generate a standard curve alongside all unknown sample measurements.

Western blot analysis

A volume of 5 mL of cells at OD₆₀₀ = 0.6 was resuspended in 150 µL SDS-PAGE loading buffer and incubated at 95°C for 15 min. Proteins were separated by SDS-PAGE in Mini-PROTEAN TGX gels (Bio-Rad) before transfer to PVDF membrane using a Trans-Blot Turbo transfer system (Bio-Rad). DgcA-His was detected using rabbit polyclonal antibody to 6×-His tag (Abcam: AB9108) at a 1:1,000 dilution. DgcB-FLAG was detected using rabbit polyclonal antibody to DDDDK tag (Abcam: AB1162) at a 1:5,000 dilution. Goat anti-rabbit secondary antibody linked to horseradish peroxidase (Abcam: ab205718) was used at a 1:10,000 dilution with Novex ECL HRP Chemiluminescent Substrate Reagent Kit (Invitrogen) and detected by an Odyssey XF Imager using LI-COR Acquisition software.

DATA AVAILABILITY

The DNA sequence reads for the characterized isolates of M. extorquens PA1 listed in Table 1 are available in the NCBI Short Read Archive database under accession number PRJNA1084801, which includes BioSamples with accession numbers SAMN40284052 through SAMN40284056https://www.ncbi.nlm.nih.gov/biosample/40284056.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.02090-23.

Supplemental material. aem.02090-23-s0001.pdf.

Supplemental figures and tables.

Supplemental data set. aem.02090-23-s0002.xlsx.

All data used to generate figures.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.