Enhanced catabolism of glycine betaine and derivatives provides improved osmotic stress protection in Methylorubrum extorquens PA1

ABSTRACT

Integration of metabolites into the overall metabolic network of a cell requires careful coordination dependent upon the ultimate usage of the metabolite. Different stoichiometric needs, and thus pathway fluxes, must exist for compounds destined for diverse uses, such as carbon sources, nitrogen sources, or stress-protective agents. Herein, we expand upon our previous work that highlighted the nature of glycine betaine (GB) metabolism in Methylobacteria to examine the utilization of GB-derivative compounds dimethylglycine (DMG) and sarcosine into Methylorubrum extorquens in different metabolic capacities, including as sole nitrogen and/or carbon sources. We isolated gain-of-function mutations that allowed M. extorquens PA1 to utilize dimethylglycine as a carbon source and dimethylglycine and sarcosine as nitrogen source. Characterization of mutants demonstrated selection for variants of the AraC-like regulator Mext_3735 that confer constitutive expression of the GB metabolic gene cluster, allowing direct utilization of the downstream GB derivatives. Finally, among the distinct isolates examined, we found that catabolism of the osmoprotectant used for selection (GB or dimethylglycine) enhanced osmotic stress resistance provided in the presence of that particular osmolyte. Thus, access to the carbon and nitrogen and osmoprotective effects of GB and DMG are made readily accessible through adaptive mutations. In M. extorquens PA1, the limitations to exploiting this group of compounds appear to exist predominantly at the levels of gene regulation and functional activity, rather than being constrained by transport or toxicity.

IMPORTANCE

Osmotic stress is a common challenge for bacteria colonizing the phyllosphere, where glycine betaine (GB) can be found as a prevalent osmoprotectant. Though Methylorubrum extorquens PA1 cannot use GB or its demethylation products, dimethylglycine (DMG) and sarcosine, as a sole carbon source, utilization is highly selectable via single nucleotide changes for both GB and DMG growth. The innate inability to use these compounds is due to limited flux through steps in the pathway and regulatory constraints. Herein, the characterization of the transcriptional regulator, Mext_3735 (GbdR), expands our understanding of the various roles in which GB derivatives can be used in M. extorquens PA1. Interestingly, increased catabolism of GB and derivatives does not interfere with, but rather improves, the ability of cells to thrive under increased salt stress conditions, suggesting that metabolic flux improves stress tolerance rather than providing a distinct tension between uses.

INTRODUCTION

In nature, bacteria often experience shifting physiochemical environmental conditions and fluctuating nutritional needs; this is particularly true for bacterial epiphytes dwelling in the phyllosphere (1). These leaf surface-colonizing microbes face a plethora of challenges, including abiotic stressors (e.g., heat, desiccation, shifting osmolarity, and UV radiation), biotic stressors (predators and antagonistic microbes), and variable access to nutrients. Under osmotic stress, changes in surrounding solute concentrations impact water movement into and out of cells (2). When faced with osmotic stress, organisms uptake or manufacture compatible solutes, organic compounds that mitigate osmotic stress, categorized as osmoprotectants and osmolytes (3–6).

Osmoprotectants prevent damage by protecting membrane integrity and stabilizing proteins to achieve cellular osmotic balance (7–9). Prominent among these classes of molecules are amino acid derivatives including the quaternary amines, such as choline, carnitine, and glycine betaine (“GB,” i.e., trimethylglycine) (10–13). Often these osmoprotectants are host-produced or host-associated. Thus, the uptake of these osmoprotectants is potentially of great importance for bacteria colonizing host-associated sites that regularly experience osmotic stress. While bacteria can maintain measurable cellular pools of GB-related compounds, many bacterial species only accumulate these osmolytes transiently under conditions of heightened stress (14, 15). Adding to the complexity, numerous bacterial species can utilize these compounds as sources of carbon and nitrogen (16). For example, root-associated members of the Rhizobiales have been noted to utilize GB as an osmoprotectant or catabolize it directly as a carbon source (17–19). Whether it is beneficial to uptake and use a compound as an osmolyte (meaning in order to regulate cellular osmotic pressure), or catabolize it for carbon and energy generation, is situation specific and has been reported to be regulated directly via osmotic conditions (20). Importantly, the utilization of a host-produced compound may play a critical role in the establishment and maintenance of a healthy and balanced association between colonizing microbes and their hosts (16). Therefore, the use of host-produced osmoprotectants by the surrounding microbiome could hold great ecological significance for host-microbe interaction dynamics (21–25).

GB is among the most common and critical osmoprotectants produced in response to salinity stress (26, 27). While some bacteria may produce GB directly (28), GB and related compounds are commonly produced by eukaryotic hosts (both animals and plants) with which beneficial microbes associate (29). Many plant species produce GB to thrive under osmotic stress (30), and the GB precursors choline and carnitine are known to be produced by plants and accumulate in plant tissues and on plant surfaces (31–33). Under salt stress, concentrations of GB have been found to range as high as hundreds of millimolars in plant tissues (34). Other related compounds have also been measured at high environmental concentrations: in grassland soil, carnitine was found to be the third most abundant soluble nitrogen compound detected (35). Low concentrations of GB present near known producers are also suggestive that osmolytes released by the host are quickly imported and/or catabolized by the associated microbiome (6).

While many plants produce substantial amounts of GB, this is not the case for all plant species, including the model organism Arabidopsis thaliana. Instead, A. thaliana has been reported to produce large amounts of carnitine, especially in response to osmotic stress (36). Among the most prominent microbial phyllosphere residents that might benefit from plant-produced osmolytes are members of the Methylobacterium and Methylorubrum genera. These well-studied epiphytes can utilize host-produced one-carbon compounds as a source of carbon and energy while providing the plants they colonize with potential benefits (37–40). A bacterial isolate from A. thaliana, Methylorubrum extorquens PA1, cannot natively use GB as a carbon source (41). However, the capacity to grow on GB can be easily evolved with appropriate selective pressure, as experimentally demonstrated in our previous work (42).

Under aerobic conditions, GB is sequentially demethylated by distinct enzymes, yielding glycine that can then be incorporated into central metabolism [Fig. 1 (43)]. Compounds derived from the demethylation of GB, including dimethylglycine (DMG) and methylglycine (sarcosine), also have suggested roles as potential, but often less effective, osmolytes when compared to GB (44–49). Interestingly, DMG can be used to synthesize GB for osmoprotection by halotolerant and halophilic organisms (50, 51) or catabolized further, either by fermentation or aerobically, to glycine and fed into central metabolism (43, 52–55). This provides GB and derivative compounds with dualistic, and potentially interfering, roles in the physiology of cells, due to the potential depletion of internal pools caused by catabolism. GB, DMG, and sarcosine also provide intriguing metabolic connections in the methylotrophic metabolism of M. extorquens, as each demethylation step has the potential to generate highly reactive free formaldehyde, which methylotrophs are uniquely evolved to tolerate and metabolize (56).

Fig 1.

Glycine betaine catabolism. Diagram of GB utilization pathways in M. extorquens PA1. The specific enzymes catalyzing each demethylation reaction are listed above the arrows.

Despite the observed utilization of GB as a source of carbon and energy in the closely related strain M. extorquens AM1 (41), the specific mechanistic steps and genes involved had not been identified or characterized until our recent work (42). In our work with M. extorquens PA1, we leveraged its inability to use GB as a carbon source and performed selection experiments that demonstrated that cells encoded the metabolic potential for GB carbon source utilization (42). By characterizing the evolved pathway, we provided the first annotation and characterization of the GB catabolic gene cluster in M. extorquens, which includes all genes required for the uptake and catabolic breakdown of GB through serial demethylation reactions to glycine. Furthermore, the gene cluster identified in M. extorquens PA1 was conserved among the genus and bore some similarities, but also distinct differences, compared to previously characterized GB-catabolizing systems, such as those of Pseudomonas aeruginosa and Sinorhizobium meliloti (33, 56).

In this study, we expand upon our previous work to investigate the specificity and versatility of the substrates of the activated GB catabolic pathway and determine the impact of their catabolism on safeguarding cells against osmotic stress. Specifically, we demonstrate that the native GB pathway in M. extorquens PA1 is not wholly inactive but rather allows for GB to serve as a nitrogen source. Furthermore, we demonstrate the ability to select for the utilization of DMG, the demethylation product of GB, as a carbon source. Mutational identification in selected DMG+ variants led to the characterization of Mext_3735, which encodes a functional homolog of the regulatory protein, GbdR. GbdR is an AraC-like protein that has formerly been described in other bacteria such as P. aeruginosa; however, the overall scheme of regulation appears to differ in M. extorquens PA1 (56, 57). Here, we find that the selected modifications to Mext_3735 confer constitutive expression of the GB gene cluster, the ability to utilize DMG as a carbon source, and the use of DMG and sarcosine as nitrogen sources, while deletion of Mext_3735 abolishes the ability to use GB and its derivatives. Molecular modeling supports the binding of GB by wild-type Mext_3735 (GbdR) and a “locked-on” conformation for the variants that provide the constitutive expression phenotype. Finally, and most significantly, selection for DMG utilization leads to a shift in the predominant osmoprotectant that provides benefits under salt stress. Specifically, selection for GB utilization enhances the osmoprotection by GB (and DMG to a lesser extent) and selection for DMG utilization enhances osmoprotection by DMG (and GB to a lesser extent).

RESULTS

The native GB catabolic pathway is functional and allows GB to serve as a nitrogen source for M. extorquens PA1

The catabolism of glycine betaine (N,N,N-trimethylglycine) has been previously reported in Methylorubrum extorquens PA1, where it was observed that utilization of GB as a sole carbon and energy source (which we denote phenotypically as “GB+”) was possible following selection for single gain-of-function mutations (dgcB^P30L) that alleviated a bottleneck in the native pathway (42). In other bacteria such as Pseudomonas aeruginosa, the GB catabolic pathway allows the demethylation products N,N-dimethylglycine (DMG) and N-methylglycine (sarcosine) to serve as sole carbon sources (56). To comparably assess the ability of M. extorquens strains to utilize DMG and sarcosine, we assayed the M. extorquens PA1 wild-type and dgcB^P30L suppressor strains, as well as the closely related M. extorquens AM1 strain, for growth in MP media supplemented with 8 mM DMG or 8 mM sarcosine. In contrast to P. aeruginosa, neither wild-type M. extorquens PA1 nor the GB+ dgcB^P30L suppressor mutant can grow with DMG or sarcosine as a sole carbon source (Fig. 2). Interestingly, M. extorquens AM1 did not possess the same constraints and was able to utilize DMG, but not sarcosine, as a carbon source.

Fig 2.

M. extorquens PA1 is unable to utilize dimethylglycine or sarcosine as a sole source of carbon. Growth of M. extorquens PA1 strains [wild type (black) and dgcB^P30L (red)], as well as the closely related M. extorquens AM1 (gray) strain in liquid MP media supplemented with (A) 8 mM DMG or (B) 8 mM sarcosine as a carbon source. Error bars represent 95% confidence intervals based on three biological replicates.

Additionally, as GB pathway compounds also contain nitrogen, we investigated whether GB, pathway intermediates (DMG, sarcosine), or the pathway end product (glycine) could serve as nitrogen sources. We observed that both focal PA1 strains (wild type and dgcB^P30L) grew with GB provided but failed to grow when downstream intermediates DMG and sarcosine were provided as a sole nitrogen source (Fig. 3; Fig. S1; Table 1). These data suggested that in M. extorquens PA1, the native GB pathway is partially functional and provides sufficient flux to allow GB utilization to satisfy the cells’ nitrogen requirements. These data also demonstrated that analogous to the carbon source results, M. extorquens PA1 could only use GB or glycine but neither intermediate (DMG and sarcosine), while M. extorquens AM1 could additionally use DMG for nitrogen.

Fig 3.

M. extorquens PA1 is unable to utilize dimethylglycine or sarcosine as a sole source of nitrogen. Growth of M. extorquens PA1 strains (wild type and dgcB^P30L), as well as the closely related M. extorquens AM1 strain in liquid MP media supplemented with 15 mM methanol as a carbon source and the following sole nitrogen sources: (A) glycine betaine, (B) dimethylglycine, and (C) sarcosine. Experimental controls including ammonium, glycine, and no additional nitrogen are shown in Fig. S1. Error bars represent the 95% confidence intervals based on three biological replicates.

TABLE 1.

Growth rates of strains utilizing GB and intermediates as sole sources of nitrogen^a

Strain	Genotype	Glycine betaine	Dimethylglycine	Sarcosine	Glycine	NH4
CM2720	AM1 WT (Δbcs)	4.4 ± 0.3	4.3 ± 0.2	N.G.	9.5 ± 0.7	3.4 ± 0.1
CM2730	PA1 WT (Δbcs)	9.1 ± 0.1	N.G.	N.G.	15 ± 1	3.47 ± 0.02
JB258	dgcBP30L	4.6 ± 0.7	N.G.	N.G.	15 ± 1	3.4 ± 0.6

^a N.G., no growth.

Given this combined phenotypic information, three major possibilities related to the inability to use DMG or sarcosine emerged: (i) the GB/proline ABC transporter ProVUX previously described in M. extorquens PA1 (42) only has high specificity for GB, (ii) DMG and sarcosine exert toxic effects upon M. extorquens PA1 when accumulated at sufficiently high concentrations, or (iii) the regulatory network responsible for the activation of the GB cluster responds specifically to GB and not to DMG or sarcosine.

Compatible solutes GB and its demethylation derivatives do not display toxicity in M. extorquens

Given the inability of M. extorquens PA1 to use DMG or sarcosine as nitrogen sources, it was possible that these compounds may be toxic. Therefore, we evaluated minimal inhibitory concentrations (MICs) imposed by GB, DMG, and sarcosine to evaluate their potential positive or negative impacts upon M. extorquens PA1 (Fig. 4). Neither GB, DMG, nor sarcosine provided any substantial growth inhibition until very high concentrations were reached, at or beyond 200 mM concentrations for each. Thus, we concluded that the lack of DMG and sarcosine utilization by M. extorquens PA1 is not due to any inherent toxicity.

Fig 4.

Minimum inhibitory concentrations of glycine betaine, dimethylglycine, and sarcosine. The wild-type M. extorquens PA1 strain was exposed to a range of concentrations of glycine betaine (G), dimethylglycine (D), and sarcosine (S). Growth was measured after 48 hours of incubation in liquid MP media supplemented with 15 mM methanol and the indicated GB-related compound and concentration. Error bars represent 95% confidence intervals based on three biological replicates.

Isolation of DMG-utilizing gain-of-function mutant derivatives of M. extorquens PA1 dgcB^P30L

Although the M. extorquens PA1 dgcB^P30L mutant is GB+, it is unable to utilize any downstream GB pathway intermediates as a source of carbon or nitrogen. To understand this constraint, we sought to isolate mutants that gained the ability to use GB pathway intermediates DMG and/or sarcosine (which we denote phenotypically as “DMG+” and/or “Sarc+”). Five independent stationary-phase cultures of wild type and the dgcB^P30L mutant were grown in MP medium supplemented with 15 mM succinate or 8 mM GB, respectively, and then plated on solid MP medium supplemented with 8 mM DMG or 8 mM sarcosine as the sole carbon and energy source, incubated at 30°C and observed periodically. No growth was observed on solid MP medium supplemented with sarcosine for either strain. Although colonies of wild type were observable on DMG plates after 90 days, none of the colonies obtained from those plates could be recultured. However, DMG+ mutants were successfully isolated from DMG plates spread with dgcB^P30L cells after 28 days (~30 per plate).

In total, 10 DMG+ isolates (seven large, three small, and two from each independent replicate culture) from the dgcB^P30L populations were selected for further characterization. Individual isolates were assayed for growth in liquid MP supplemented with 8 mM DMG and they fell into two classes that aligned with their observed colony sizes (Fig. S2). The first class (herein called “Class I”) contained seven isolates that had average generation times (gt) of 5.92 ± 0.22 hours. This phenotype was nearly identical to the generation time of the GB+ dgcB^P30L mutant when utilizing GB as a sole source of carbon and energy (42). Consistent with the smaller colony size, the three isolates from the second class (“Class II”) had average generation times that were significantly longer than Class I, at 11.32 ± 0.44 hours.

Mutations to Mext_3735 confer DMG utilization and constitutive expression of the GB metabolic gene cluster

To identify mutations that confer DMG utilization, we performed whole genome sequencing analysis on three Class I isolates and two Class II isolates. Each of the Class I DMG+ isolates had nonsynonymous mutations in the Mext_3735 gene (Table 2; Fig. 5). Mext_3735 encodes a putative AraC-type regulatory protein with similarity to gbdR from P. aeruginosa PAO1 (Fig. S3) and is contained within the larger GB cluster in the M. extorquens PA1 genome (42, 56). The observed Mext_3735 mutations resulted in three distinct Mext_3735 variants: Mext_3735^D167N, Mext_3735^G165S, and Mext_3735^E170V. Sequencing of Class II isolate variants yielded the following: both JB535 and JB537 had gene amplifications that included the entirety of the GB metabolic gene cluster (Table 2; Fig. S4), further highlighting the importance of these genes for the growth phenotypes being examined.

TABLE 2.

Causative mutations identified by whole genome sequencing of selected DMG+ isolates of M. extorquens PA1^a

Strain	Class	Locus: annotated description	Mutation (position)	Variant (codon)
JB528	1	Mext_3735: helix-turn-helix-domain-containing protein, AraC-type/GlxA family transcriptional regulator	G→A (499)	D167N (GAC→AAC)
JB529	1	Mext_3735: helix-turn-helix-domain-containing protein, AraC-type/GlxA family transcriptional regulator	G→A (493)	G165S (GGC→AGC)
JB533	1	Mext_3735: helix-turn-helix-domain-containing protein, AraC-type/GlxA family transcriptional regulator	A→T (509)	E170V (GAG→GTG)
JB535	2	Mext_0926A56G: hypothetical gene, genomic amplification of sequence spanning Mext_3711– Mext_3785	C→G (167)	A56G (GCG→GGG)
JB537	2	Genomic amplification of sequence spanning Mext_3685–Mext_3790	N.A.	N.A.

^a N.A., not applicable.

Fig 5.

Selection for mutations in M. extorquens dgcB^P30L allows dimethylglycine utilization as a sole source of carbon and energy. Growth of M. extorquens GB+ background strain dgcB^P30L (red circles), DMG+ evolved isolates dgcB^P30L Mext_3735^D167N (JB528, yellow closed circles), dgcB^P30L Mext_3735^G165S (JB529, olive squares), and dgcB^P30L Mext_3735^E170V (JB533, green open circles) was quantified in liquid MP medium supplemented with 8 mM DMG. Error bars represent 95% confidence intervals based on three biological replicates.

In our prior work, we demonstrated that exogenous GB activates the GB gene cluster (42). Here, we examined the specific role of Mext_3735 upon GB gene cluster expression by performing reverse transcriptase quantitative PCR (RT-qPCR). Specifically, we measured the expression of a representative gene from each operon in the GB cluster (proV, gbcB, dgcB, soxA, and sdaB) in the GB+ mutant (dgcB^P30L), the GB+ mutant lacking Mext_3735 (dgcB^P30L ΔMext_3735), and a representative DMG+ mutant (dgcB^P30L Mext_3735^D167N) compared to wild type (Fig. 6). For both the wild-type and GB+ strains, DMG failed to activate the expression of all the operons in the gene cluster (Fig. 6A). This suggested that the regulation of GB genes is dependent upon GB and that, unlike GB, DMG is not an activating ligand in M. extorquens PA1. However, the DMG+ strain displayed high levels of expression for all GB genes assayed, independent of the presence of GB (Fig. 6B). It is also notable that despite the constitutive expression phenotype, expression was still elevated in the presence of GB. Conversely, if Mext_3735 was deleted, activation of GB genes by GB was lost and cells became unable to utilize GB, DMG, or sarcosine as a carbon or nitrogen source (Fig. 6C and 7; Fig. S5). Due to the predicted annotation, synteny of the gene locus, and associated mutant phenotypes, we will from here on refer to Mext_3735 as gbdR and characterize our variants as constitutively expressing Mext_3735, or “gbdR^*.” From these data, we concluded that gain-of-function mutations were observed in Mext_3735/gbdR following selection, resulting in constitutive gene expression of the GB cluster and conferring the DMG+ phenotype.

Fig 6.

Mutations in gbdR lead to constitutive activation of GB cluster genes, regardless of ligand presence. The relative expression of representative genes from each operon of the M. extorquens GB gene cluster was examined, including proV, soxA, gbcB, sdaB, and dgcB. Growth was carried out in MP supplemented with 15 mM MeOH with or without 1 mM GB or DMG. (A) Expression of genes in the GB cluster in PA1 dgcB^P30L mutant, comparing conditions with versus without DMG, shows that DMG cannot induce the expression of GB cluster genes, in contrast to GB itself (42). (B) Expression of genes in the GB cluster in the dgcB^P30L Mext_3735^D167N mutant in comparison to PA1 dgcB^P30L grown in MP supplemented with only 15 mM MeOH. (C) Expression of genes in the GB cluster in the dgcB^P30L ΔMext_3735 mutant in comparison to PA1 wild type grown in MP supplemented with only 15 mM MeOH. Individual values shown are averages of three technical replicates. Error bars represent the standard deviation of three biological replicates.

Fig 7.

Constitutive expression of GB cluster allows utilization of DMG and sarcosine as sole nitrogen sources. Final OD₆₀₀ readings of M. extorquens strains grown in MP media lacking nitrogen supplemented with 15 mM methanol and 5 mM of the indicated nitrogen sources. Bars represent the mean value of three independent biological replicates (shown as circles). Error bars represent 95% confidence intervals based on three biological replicates.

Molecular simulations characterize the role of GbdR and its variants and the binding of substrates to GbdR

The GbdR variants we isolated had sequence changes that occurred within a five amino acid stretch and resulted in an altered charge or polarity of the originally encoded residue. While GbdR homologs have been characterized phenotypically in other organisms, no physical structure for it yet exists. To investigate the potential impact of the residue changes, we first generated a 3D structural model of Mext_3735 using AlphaFold2 [(58) Fig. 8A]. The predicted structure revealed a disordered N-terminal region (residues 1–24) followed by two domains linked via a loop (residues 208–232). Predicted Local Distance Difference Test (pLDDT) values per residue that highlight confidence in the predicted structure for that specific residue indicated a high confidence (>80) in the overall predicted structure except for the residues 1–24 (Fig. S6). There is no full-length structure for GbdR from any organism; however, there is a structure for Pseudomonas putida SouR (3GRA in PDB), which has a structure expected to be very similar to GbdR, containing a glutamine amidotransferase-1-like domain and only lacking the DNA-binding domain. We thus used this structure to evaluate our AlphaFold prediction and found good correspondence. The superimposition of both structures revealed a significant overlap of the majority of secondary structure elements, with a root mean square deviation (RMSD) of 1.10 Å between them (Fig. S7). This observation underscores the accuracy of the generated GbdR model, serving to validate our computational approach. Modeling of electrostatic surface charges in our GbdR model shows that the residues G165, D167, and E170 reside in a negatively charged environment (Fig. 8B), and each variant results in an overall reduction of local negative charge on the surface (characterized DMG+ variants were G165S, D167N, and E170V).

Fig 8.

Predicted protein structure of M. extorquens PA1 GbdR. (A) AlphaFold 2 predicted structure of wild-type GbdR (gray) with variant sites mutated in the selected DMG+ evolved strains highlighted in distinct colors. (B) Wild-type GbdR structure shown with the electrostatic potential surface. Each color indicates the following: red, negatively charged surface; blue, positively charged surface; and white, hydrophobic. Black circle highlights the location of three variant sites on the electrostatic potential surface. The model file can be found at https://github.com/BazurtoLab/Glycine-betaine-osmotic-stress/blob/main/IFD_glycine_betaine_top.pdb.

To better understand the specific impacts of the gbdR mutations on the constitutive GB phenotype, we first carried out molecular dynamic simulations to compare the conformational changes of the GbdR variants to wild-type GbdR (total four simulations) in the absence of GB. We were especially interested in revealing any putative shifts in structural conformation conferred by the selected mutations. Simulations revealed a large conformational change for the wild-type form of the protein, whereas variant forms appear to be “locked” as indicated by their low RMSD along the course of the simulation (Fig. 9). This suggested that the GbdR* mutants act as constitutive regulators by being permanently fixed in the ligand-bound form, independent of the presence of its preferred ligand. When investigating the ability of the wild-type GbdR to bind GB and the related substrates DMG and sarcosine, we observed that there was an enhanced binding affinity for GB (docking score: −5.2 kcal/mol, lower score indicates strong binding) compared to DMG (docking score: −4.0 kcal/mol) and sarcosine (docking score: −3.9 kcal/mol). Interestingly, the top docking pose of GB, which is physically near, but on the opposing face of the location of the observed GbdR variant residues, shows binding via four hydrogen bonds with T182 and R214 and forms hydrophobic contacts with L168, A160, and A179 side chains (see Fig. 10). To validate our ligand-binding prediction, we investigated the binding capability of a R214A GbdR variant to GB and noted a discernible alteration in the binding pocket preference (Fig. S8). Specifically, GB no longer engages the binding pocket seen on the wild-type GbdR. Instead, it occupies an alternative pocket characterized by a lower binding affinity, with a score of −4.8 kcal/mol. These observations underscore the key role of R214 in dictating binding site specificity within the GbdR protein structure.

Fig 9.

Root mean square deviation of wild-type GbdR and three variants during molecular dynamics simulations. The RMSD plot shows the root mean square deviation of three protein variants (G165S, D167N, and E170V) compared to the wild type (“WT”) over time. The x-axis shows the time in nanoseconds (ns), and the y-axis shows the root mean square deviation in Angstroms. The wild-type GbdR exhibits a higher RMSD after 70 ns compared to the three variants, indicating that the wild-type form undergoes a large conformational change, which is not seen for the three variants.

Fig 10.

Docking of GB to GbdR. (A) This panel shows the top docked pose of GB to GbdR in electrostatic surface representation. The molecule is colored according to its electrostatic potential, with red representing negative charges and blue representing positive charges. The circle around the docked GB represents the binding pocket of GbdR. (B) This panel shows the amino acid residues in the predicted binding pocket of GbdR, with dashed lines indicating hydrogen bonds between the residues and GB (gray). The residues (green) are labeled with their one-letter amino acid codes and position numbers.

Efficient GB catabolism increases fitness under high levels of osmotic stress

We sought to determine how the altered catabolic capacity for GB and its derivatives would impact their ability to serve as osmoprotectants. To achieve this, we measured the growth and competitive fitness of strains with varying capacities to uptake and catabolize GB and its derivatives under a range of osmotic stress. This included wild type (uses GB for nitrogen), dgcB^P30L (uses GB for carbon and nitrogen), dgcB^P30L gbdR^D167N (uses both GB and DMG for carbon and nitrogen), dgcB^P30L ΔgbdR (cannot activate GB gene expression), dgcB^P30L ΔproVUX (cannot uptake GB), dgcB^P30L ΔgbcBA (cannot enzymatically convert GB into DMG), and ΔdgcBA (cannot enzymatically convert DMG to sarcosine).

To identify conditions of osmotic stress, we grew all strains in the presence of increasing 100 mM increments of NaCl up to 500 mM (Fig. 11A). We found that, in the absence of GB, all strains grew comparably well and were growth inhibited by 300 mM NaCl. Additionally, we established that GB could act as an osmoprotectant to wild type, GB+, and DMG+ mutants under these conditions (Fig. 11B). Importantly, strains lacking the GbdR regulator, the ProVUX transporter, or GbcAB, encoding GB monooxygenase that catalyzes the first step of the GB catabolic pathway, were not protected by GB. Finally, these data suggested that the ability to catabolize GB as a carbon source conferred greater osmotic protection.

Fig 11.

Minimum inhibitory concentrations of NaCl. Different genotypes were exposed to a range of concentrations of up to 500 mM NaCl and (A) without or (B) with supplementation of 1 mM GB. Growth was measured after 48 hours of incubation in liquid MP media supplemented with 15 mM methanol and the indicated NaCl concentration. Error bars represent 95% confidence intervals derived from three biological replicates.

To measure the competitive fitness of mutants when experiencing osmotic stress, each of the aforementioned strains was independently competed against a fluorescently tagged wild-type strain (CM3839) in liquid MP supplemented with 15 mM methanol and either with or without 1 mM GB and/or 200 mM NaCl (Fig. 12). Fitness scores were calculated relative to wild type in each condition. No significant differences in strain fitness were observed without the addition of both GB and NaCl, again indicating that GB can act as an osmolyte for M. extorquens (Fig. 12). By contrast, when GB and NaCl were both present, the GB+ dgcB^P30L strain was dramatically more fit than wild type (w = 1.743, P < 0.0001). The DMG+ dgcB^P30L gbdR^D167N strain was also more fit than wild type but experienced a dramatic fitness decrease compared to the dgcB^P30L single mutant (w = 1.246, P < 0.0001). The dgcB^P30L ΔgbdR and dgcB^P30L ΔproVUX mutants faired far worse than the other strains, confirming the need for expression of the GB genes and dedicated GB transport for successful GB mediation of salt stress (dgcB^P30L ΔgbdR: w = 0.735, dgcB^P30L ΔproVUX: w = 0.692, P < 0.0001). Notably, the dgcB^P30L ΔgbcBA mutant, whose GB catabolism has been completely abolished, had lower fitness scores than all other strains (w = 0.289), confirming that the ability to efficiently catabolize GB is advantageous under osmotic stress.

Fig 12.

Regulated GB utilization improves competitive fitness under osmotic stress conditions. Relative Malthusian fitness scores of M. extorquens PA1 wild type (blue), dgcB^P30L (red), dgcB^P30L gbdR^D167N (green), dgcB^P30L ΔgbdR (yellow), dgcB^P30L ΔgbcBA (cyan), and dgcB^P30L ΔproVUX (magenta) following a 24-hour competition period against a fluorescently tagged wild-type PA1 strain under growth conditions in liquid MP media supplemented with 15 mM methanol and addition of (A) neither NaCl nor GB, (B) 1 mM GB, (C) 200 mM NaCl, and (D) 200 mM NaCl and 1 mM GB. Error bars represent the 95% confidence interval of three independent biological replicates. ns, no significant difference. ****P < 0.0001.

A closer examination of the growth dynamics of each strain found that they all displayed decreasing growth rates with increasing concentrations of NaCl before growth was entirely inhibited at 300 mM NaCl (Fig. S9 and S10). Inclusion of 1 mM GB had subtle impacts at 100 mM NaCl where the dgcB^P30L ΔgbcBA mutants showed a minor but statistically significant decrease in growth rate (doubling times: wild type = 4.00 ± 0.10 hours, dgcB^P30L ΔgbcBA mutant = 4.49 ± 0.14 hours, P < 0.0001; Table 3; Fig. S9 and S10). All strains grew with 200 mM NaCl but demonstrated a clear salt-induced growth defect. At this stress level, the GB+ and DMG + mutants grew identically well to one another and outperformed wild type, while wild type outperformed strains lacking the transporter and the GbdR regulator in the presence of GB. The dgcB^P30L ΔgbcBA strain was growth impaired by the addition of GB where its doubling time greatly increased from 5.21 ± 0.22 hours to 26.37 ± 1.31 hours (Fig. S8 and S10C and G; Table 3). Taken together, these data suggest that in M. extorquens PA1, catabolizing GB increases osmotic stress resistance and that strains with an increased catabolic capacity for GB (conferred by the dgcB^P30L mutation present in the GB+ and DMG+ strains) demonstrate increased resistance. The data also suggest that the ability to uptake GB but not catabolize it to DMG significantly decreases osmotic stress tolerance.

TABLE 3.

Doubling times of PA1 utilizing GB under osmotic stress conditions^a

		Doubling time (hours)
	[NaCl] (mM)	0			100			200			300
Strain	Genotype	NA	+GB	+DMG	NA	+GB	+DMG	NA	+GB	+DMG	NA	+GB	+DMG
CM2730	Wild type	3.60 ± 0.14	3.28 ± 0.04	3.74 ± 0.08	3.77 ± 0.17	4.00 ± 0.10	3.89 ± 0.03	5.57 ± 0.03	7.53 ± 0.16	6.63 ± 0.20	NG	15.19 ± 0.35	NG
JB258	dgcBP30L	3.66 ± 0.08	3.51 ± 0.09	3.72 ± 0.17	3.84 ± 0.02	4.07 ± 0.01	3.95 ± 0.12	5.32 ± 0.07	6.17 ± 0.07	7.19 ± 0.90	NG	13.57 ± 0.57	NG
JB528	dgcBP30L gbdRD167N	3.57 ± 0.09	3.55 ± 0.06	3.84 ± 0.45	3.52 ± 0.05	3.82 ± 0.02	4.02 ± 0.31	5.11 ± 0.04	7.23 ± 0.22	5.87 ± 0.24	NG	NG	21.32 ± 0.86
JB577	dgcBP30L ΔgbdR	3.62 ± 0.07	3.36 ± 0.13	3.72 ± 0.32	3.56 ± 0.04	3.89 ± 0.13	4.32 ± 0.49	4.92 ± 0.14	8.52 ± 0.35	6.64 ± 0.23	NG	NG	NG
JB578	dgcBP30L ΔproVUX	3.89 ± 0.07	3.33 ± 0.16	4.13 ± 0.68	3.95 ± 0.07	3.96 ± 0.11	3.76 ± 0.09	5.04 ± 0.01	7.37 ± 0.25	6.56 ± 0.27	NG	NG	NG
JB562	dgcBP30L ΔgbcBA	3.90 ± 0.13	3.74 ± 0.24	NT	3.93 ± 0.06	4.49 ± 0.14	NT	5.21 ± 0.15	26.37 ± 1.41	NT	NG	NG	NT

^a NG, no growth; NT, not tested; and NA, no addition of osmolyte. GB and DMG were both added at 1 mM final concentrations.

At 300 mM NaCl, only wild type and the dgcB^P30L mutant could grow, suggesting that GB must be catabolized for cells to cope with elevated levels of osmotic stress (Fig. S10H). However, while dgcB^P30L gbdR^D167N grew as well as dgcB^P30L in the 200 mM NaCl + GB condition, it failed to grow at all at 300 mM NaCl + GB. This suggests a potential optimal level of GB gene expression and resulting GB catabolism under the presence of heightened salt stress, as the DMG+ mutant strain can catabolize GB and DMG at higher levels than wild type and dgcB^P30L (Fig. S11). From these results, we concluded that wild-type PA1 must catabolize GB during high osmotic stress (300 mM) and that catabolizing GB in this condition improves strain fitness. By contrast, in less stressful conditions (200 mM), GB uptake without GB catabolism inhibits growth compared to wild type, and GB generally induces minor decreases in growth rates (Table 3). Namely, both transport and catabolism of GB as well as GB responsive regulation are required for maximum fitness of M. extorquens PA1 under increasing salt stress, and this is mediated by the regulation imposed by GbdR (Fig. 12).

Evaluating the role of DMG as an osmolyte in M. extorquens PA1

Considering the higher growth and fitness measures that GB could provide to the GB+ and DMG+ strains possessing enhanced flux through the GB pathway (and resulting catabolism), we analogously assessed the ability of DMG to provide osmoprotection under salt stress. Growth measurements revealed that only the DMG+ strain experiences improved growth when supplemented with DMG under salt stress (Fig. S9 and S10). DMG+ grew better than any other strain tested with DMG under 200 and 300 mM NaCl conditions (Fig. S10K and L). Similarly, while the addition of DMG further delayed growth for other strains, it did not delay the growth of DMG+ at 200 mM NaCl and improved the growth of DMG+ at 300 mM NaCl, analogous to the effect GB has upon the wild-type, GB+, and DMG+ strains (Fig. S9). The limited number of strains impacted by DMG is consistent with the observation that DMG fails to increase gene expression for the GB+ strain, and the DMG+ strain experiences high expression with or without DMG (Fig. 6). The ability of the DMG+ strain to utilize DMG effectively could be due to heightened transport and/or catabolism.

We similarly assessed competition outcomes of these strains under different salt and/or DMG combinations (Fig. 13). Here, the competition assays demonstrated that heightened fitness was only observed under conditions containing both salt (200 mM NaCl) and DMG (1 mM), and again only to strains with heightened pathway flux, the GB+ and DMG+ mutant strains. Interestingly, in this case, the ordering was reversed, with DMG+ experiencing higher fitness (GB+: w = 1.128, P < 0.001; DMG+: w = 1.248, P < 0.0001, GB+ versus DMG+: P = 0.0005), and at similar levels to how it performed in the presence of GB under salt stress (Fig. 12). These fitness data contrast with the results in isolation, where improved growth rate is only observed in DMG+ (Fig. S10). Although all deletion mutant strains performed worse than the evolved mutants, none performed worse than wild type, suggesting that these mutations brought them back to the same baseline fitness that wild-type PA1 experiences in the presence of DMG. These results further support an overall model in which flux through the GB pathway (i.e., catabolism of the pathway intermediates, as in Fig. 1) confers the biggest growth and fitness gains under imposed osmotic stress.

Fig 13.

DMG utilization improves competitive fitness under osmotic stress conditions for strains with increased GB pathway flux. Relative Malthusian fitness scores of M. extorquens PA1 wild type (blue), dgcB^P30L (red), dgcB^P30L gbdR^D167N (yellow), dgcB^P30L ΔgbdR (green), dgcB^P30L ΔdgcBA (blue), and dgcB^P30L ΔproVUX (magenta) following a 24-hour competition period against a fluorescently tagged wild-type PA1 strain under growth conditions in liquid MP media supplemented with 15 mM methanol and addition of (A) neither NaCl nor DMG, (B) 1 mM DMG, (C) 200 mM NaCl, and (D) 200 mM NaCl and 1 mM DMG. Error bars represent the 95% confidence interval of three independent biological replicates. ns, no significant difference. ***P < 0.001 and ****P < 0.0001.

DISCUSSION

A major challenge to bacteria living in the phyllosphere is low water availability, due to intense sunlight exposure and fluctuations in access to moisture, which induces osmotic stress (1, 59). To cope, many bacteria accumulate GB and related small molecules that serve as osmoprotectants. These small molecules can often serve as carbon and/or nitrogen sources, potentially creating a physiological conflict. M. extorquens PA1 is a competitive phyllosphere colonizer that has been extensively studied as a model methylotroph; however, it does not grow on GB and related compounds as a carbon source (41, 60–62). This led us to investigate whether this model plant-colonizing methylotroph might utilize GB and related osmolytes encountered in nature depending on environmental conditions and nutritional needs (42). By selecting for mutations that conferred catabolism of GB (42), and herein its downstream intermediate DMG, we identified genes encoding major catabolic functions and alleles that provide increased ability to metabolize these compounds. Notably, we found that osmotic stress resistance conferred by either osmolyte was enhanced when catabolism of that osmolyte was increased.

We recently uncovered an adaptive GB utilization pathway in M. extorquens PA1 (Fig. 1). Altered function of dgcB or ftsH genes alleviated a bottleneck at the DMG dehydrogenase catalytic step and activated a seemingly nonfunctional GB catabolic pathway (42). Our analyses of the primary GB+ mutant (dgcB^P30L) revealed a 23 kb GB gene cluster that contained genes for the uptake and catabolism of GB and was well conserved among other closely related strains belonging to the Methylorubrum and Methylobacterium genera. The GB cluster contained genes encoding enzymatic functions for the breakdown of GB to glycine, hypothetically allowing it to be catabolized and fed into central metabolism. However, wild-type PA1 does not grow with GB as a carbon source (41) and does not utilize downstream intermediates of GB in any previously noted metabolic capacity (Fig. 1 and 6). In the present study, we leveraged the activated pathway identified in reference (42) to investigate its ability to catabolize DMG and sarcosine for carbon or nitrogen and the effects that differing degrees of catabolism may have on osmotic stress resistance.

We found that even without adaptive mutations, the GB pathway flux in the wild-type strain is sufficient to allow GB use as a nitrogen source. These data demonstrated that the native pathway encoded by the GB gene cluster is functional and provides sufficient flux for GB to satisfy the cell’s nitrogen requirements. Curiously, neither of the downstream intermediates of the pathway, DMG nor sarcosine, could be utilized as a nitrogen source in wild type. The GB+ mutant (dgcB^P30L) characterized in our previous work has a higher flux version of the GB pathway that alleviates the bottleneck at the conversion of DMG to sarcosine. However, like wild type, it still fails to use DMG and sarcosine exclusively for carbon or nitrogen, suggesting that the inability to use downstream intermediates cannot be solely due to pathway flux. Here, we were able to select for and isolate derivatives of the GB+ mutant that can also use DMG, but not sarcosine, as a sole carbon source. Characterization of these DMG+ isolates ultimately led us to identify and define Mext_3735 as a central player in the utilization of GB-derived compounds.

Given that the identified mutated locus, Mext_3735, was annotated as encoding an AraC-like protein with a helix-turn-helix domain and impacted GB gene cluster expression, we predicted that it had a role as the primary regulator of those genes. A comparison to GB regulators in other GB-catabolizing bacteria detected strong sequence similarity (66%) to the GbdR protein of Pseudomonas aeruginosa PAO1 (Fig. S3), which positively regulates GB catabolism genes in response to GB, DMG, and ethyl glycine betaine (54, 58, 63). In M. extorquens PA1, we found that both the dgcB^P30L mutant and wild-type strains upregulated each operon in the GB cluster in response to GB (42), but not in response to DMG (Fig. 6A). By contrast, the Mext_3735 variants we isolated allow for GB gene cluster expression regardless of ligand presence/absence (Fig. 6B), suggestive of a “locked-on” configuration. Our supposition was further supported by molecular modeling, which demonstrated the ability of GbdR to bind to GB and provided evidence that, independent of ligand, the structure of each of the Mext_3735 variants appears consistent with conformational changes that wild-type GbdR experiences upon ligand binding (Fig. 9 and 10). Thus, it appears that in its native form, GB catabolism in M. extorquens is activated by the presence of GB and that direct responsiveness to and utilization of DMG and sarcosine requires altered regulation. Yet, even for the dgcB^P30L gbdR* (DMG+) mutants, only DMG can serve as a carbon source, while both DMG and sarcosine serve as nitrogen sources (Fig. 4 and 6). Together, our data provide clear evidence that the M. extorquens PA1 genome has the potential to utilize downstream intermediates of GB, despite the wild-type strain not doing so. This is also suggestive that wild-type PA1 may use GB and derivative compounds predominantly as osmolytes in its native environment. Furthermore, the restriction in the wild type to utilize DMG is impaired at the level of regulation, rather than an issue of transport or catabolic capacity.

Intriguingly, the closely related M. extorquens AM1 strain is capable of growth on DMG (Fig. 1) and possesses a GbdR homolog that only differs from GbdR^PA1 at residues 83 (AM1 = G, PA1 = A) and 215 (AM1 = G, PA1 = E) (Fig. S3, cyan boxes). The residue change at position 215 was suspected to be of greatest significance, as it was located directly next to the arginine 214 predicted to be involved in H-bond formation with GB in our ligand docking studies (Fig. 10). We tested the possibility that substituting wild-type gbdR with the homologous gene Mex_1p4107 from M. extorquens AM1 could provide the ability to grow on DMG directly via complementation studies and found that indeed it could (Fig. S12). This confirmed the fact that GbdR is a major determinant of pathway substrate specificity, though this does not rule out the possibility of added layers of regulation. We additionally tested an R214A mutant derivative of PA1 GbdR and found that this variant no longer supported growth on either GB or DMG, confirming the essentiality of this residue for normal GbdR function, likely via ligand binding.

This study employed molecular simulations to shed light on the role of GbdR in relation to GB gene expression, and its selected variants in governing the constitutive GB phenotype (Fig. 9 and 10). The simulation results suggest that specific mutations alter GbdR’s structure and binding affinity, potentially explaining their functional impact. The examined GbdR variants (G165S, D167N, and E170V), all of which exhibited altered charge and polarity, displayed a consistent “locked-on” conformational structure. This rigidity, contrasting with the dynamic wild-type GbdR protein, suggests a potential mechanism for their constitutive GB phenotype. The reduced negative charge on the surface of the variant proteins might further contribute to this altered function, but further molecular modeling and empirical investigations are needed to understand the precise functional consequences. The observed higher binding affinity of wild-type GbdR for GB compared to DMG and sarcosine underscores its specific role in GB metabolism and is consistent with experimental RT-qPCR and phenotypic results. The identified docked pose characterized by hydrogen bonds and hydrophobic interactions offers valuable clues about the binding site (Fig. 10). However, future longer molecular simulations are crucial to better understand the stability of the observed docked pose and experimental structures. The predictions generated from these simulations additionally require biochemical confirmation. Particularly with regard to ligand binding, no experimental evidence demonstrating GATR family regulator (including GbdR) binding to ligand currently exists. This could be for a variety of reasons, including a requirement for another co-ligand, though our simulation data suggest it is due to binding and corresponding protein conformational changes being transient in nature (Fig. 9).

The GbdR homolog of P. aeruginosa is directly responsive to both GB and DMG, which is distinctly different from M. extorquens PA1 GbdR, and likely explains its innate capacity to catabolize these compounds. Additionally, P. aeruginosa can utilize sarcosine as a sole carbon source and possesses an additional regulator, SouR, which responds directly to sarcosine (53). A BLAST alignment of SouR (encoded by PA4184) against the M. extorquens PA1 genome also yielded Mext_3735 as the sequence hit with the highest score, with a 52% similarity and 36% identity. While SouR impacts gene expression in response to the presence of sarcosine, neither the GbdR of Pseudomonas nor M. extorquens PA1 appear to be responsive to sarcosine. Thus, it appears that wild-type PA1 lacks an authentic SouR homolog and provides an explanation as to why (i) wild-type PA1 does not successfully utilize sarcosine as a carbon source (Fig. 1) and (ii) the constitutive GbdR variant allows for the utilization of sarcosine as a nitrogen source (Fig. 7). Overall, these data suggest that M. extorquens PA1 possesses a different regulatory scheme than P. aeruginosa concerning the GB derivatives DMG and sarcosine, consistent with the inability of wild type to utilize these compounds. Finally, the lack of a SouR homolog may also explain the inability to readily evolve sarcosine utilization.

The use of plant-produced and endogenous microbe-produced osmoprotectants at the plant-microbe interface by beneficial phyllosphere bacteria is poorly understood. Even organismal preferences in terms of osmolyte production and utilization are often unknown. GB has been studied as a primary osmoprotectant for many biological organisms, but it has been demonstrated that DMG can also serve a role as an osmoprotectant in some organisms (48, 53). The evidence for sarcosine providing osmoprotection is sparser, and there are specific instances in which it has been noted as comparatively less effective (48), although it has proved to be effective in some systems (49). Interestingly, not all organisms use GB and derivates as osmolytes, and it has even been found to interfere with other endogenous osmolytes in rare cases (43). Here, we observed that both GB and DMG serve as potential osmoprotectants for M. extorquens PA1, as supplementation conferred salt stress resistance in PA1 strains and allowed growth in otherwise non-permissive conditions (Fig. S9 and S10).

How bacteria utilize compounds such as GB falls into a few discrete categories, with some only utilizing it as an osmoprotectant, while others catabolize it, and still others do both (43). For bacteria capable of both uses, it is of interest to determine how use is coordinated and whether these uses work in opposition to each other. Previous studies have suggested that for dual users, environmental conditions can guide usage; for example, in S. meliloti, catabolism is partially inhibited by the osmolarity of the growth medium, allowing GB to sufficiently accumulate to function as an osmoprotectant (18). Surprisingly, we found that the ability to catabolize the potential osmolyte of interest, whether GB or DMG, was advantageous under osmotic stress conditions, based on both observed strain growth and competition results (Fig. 12 and 13; Fig. S9 and S10). This suggests that catabolism of GB-related compounds is environmentally relevant for M. extorquens PA1 and may also be the case for other bacteria that may encounter the intersection of variable osmotic stress and host-produced osmolytes. Though seemingly antithetical, a positive correlation between catabolism and osmotic stress tolerance has been previously observed, with respect to GB specifically (64). In other organisms, it has been observed that GB and related compounds can interfere with the production of endogenous osmoprotectants such as ectoine and trehalose (64, 65), suggesting that while GB and derivatives can aid in osmoprotection for M. extorquens in the short term, they may be suboptimal for longer-term protection. As a result, it may also be the case that specific osmoprotectant pools maintained and used by M. extorquens shift over time, as has been observed in other bacteria (66–69). In this study, we notably find that both the GB+ (dgcB^P30L) and DMG+ (dgcB^P30L gbdR^D167N) mutant strains outperform the wild-type PA1 strain under direct competitive conditions in the presence of intermediate salt stress (200 mM) and supplemented GB (Fig. 12). Analogously, under the same salt stress with supplemented DMG, we see that only the GB+ and DMG+ strains, which experience heightened pathway flux and DMG catabolism, experience increased growth and/or fitness (Fig. 13; Fig. S10K and L). In the case of both GB and DMG supplementation, strains that cannot upregulate gene expression, transport the specific compatible solute, or catabolize the specific osmoprotectant being supplemented have a demonstrable growth defect and competitive disadvantage. Of note, the extent of this impact is more fixed for the constitutive DMG+ strain, but quite variable for the GB+ that is specifically able to regulatorily respond to GB and not DMG (Fig. 12 and 13) Altogether, we see that for different intermediates of the GB pathway, GB and DMG, heightened pathway flux and catabolism (Fig. S11) correspond to better outcomes in the face of increased osmolarity. Here, we did not directly quantify the accumulation of pathway intermediates or other endogenously produced osmolytes in M. extorquens under salt stress; however, we have previously shown the accumulation of pathway intermediates upon the introduction of genetic blocks that disrupt each catalytic step (42). Quantifying endogenously produced osmolytes is an interesting future research direction to explore.

Despite conferring critical gains in growth and fitness under salt stress, the optimality of osmolyte catabolism versus osmoprotection may be restricted to a narrow range of osmotic conditions. As an example, it is interesting to note that the DMG+ mutant grows better than wild type at 200 mM NaCl with 1 mM GB but then fails to grow completely in the presence of 300 mM NaCl with 1 mM GB, where wild type is still capable of growth. This could be due to a tradeoff related to the costly nature of constitutive GB gene cluster expression (70, 71) or due to enhanced pathway flux that results in insufficient levels of GB to confer osmoprotection. Similarly, when DMG is supplemented under increasing salt stress, the DMG+ strain grows better than in the absence of DMG; however, DMG failed to rescue wild-type growth under any heightened osmotic stress conditions evaluated here (Fig. S10). Moreover, the extent to which DMG rescues the DMG+ strain under the 300 mM NaCl condition is much more restricted than the rescue of the GB+ strain by GB, both in terms of growth rate, growth yield, and competitive fitness, possibly due to incomplete activation of pathway flux in the absence of GB (Fig. 6B).

Altogether, we interpret this body of data to indicate that GB is a preferred osmolyte for wild-type M. extorquens PA1, in comparison to either DMG or sarcosine. The wild-type strain also maintains the metabolic capacity to catabolize GB as a sole nitrogen source, suggesting that the inherent capacity for some flux through this pathway provides some osmotic stress protection (as compared to the ΔgbcBA catabolic mutant), which may be advantageous in its native phyllosphere environment. Low-level pathway flux may be important for proper turnover of osmolytes to maintain homeostatic levels within the cell. Increasing the catabolic potential of the pathway (GB+ strain) or generally upregulating all aspects of the pathway (DMG+ strain) does not appear to be disadvantageous to PA1. However, these experiments were carried out in conditions that may lack major aspects of the leaf environment, where limitations or constraints of these mutant strains may become clearer. Importantly, the capacity for increased GB pathway flux also does not inherently provide heightened resistance to salt stress, as the fitness benefits we observed for both the GB+ and DMG+ strains were dependent on the presence of the respective osmoprotectants. Finally, our results shed light on the conditions under which quaternary amines and related compounds may provide benefits to bacteria and in which capacity they may utilize these compounds and demonstrate the intertwined nature of cellular metabolism and stress response for M. extorquens PA1, as well as its ability to evolve new metabolic capacities with respect to the osmolytes it may encounter in natural environments. Our findings suggest that the regulation of GB catabolism critically impacts the specific osmoprotectant utilization profile, which likely has strong ecological ramifications and may be vital during M. extorquens plant-associated growth.

MATERIALS AND METHODS

Bacterial strains, media, and chemicals

Chemicals and reagents used in this study were purchased from Sigma-Aldrich. All strains and plasmids in this study are listed in Table 4. All strains used to produce data in the present study were derived from Methylorubrum extorquens PA1 ΔbcsABZC (Mext_1367–Mext_1370) for minimal cell aggregation in liquid culture growth (62, 72). Strains were cultivated using Methylobacterium piparazine-N,N′-bis(2-ethanesulfonic acid) (“MP”) media supplemented with 3.5 mM succinate, 15 mM methanol, 8 mM of GB-related compounds [glycine betaine (N,N,N-trimethylglycine), N,N-dimethylglycine, or 8 mM sarcosine (N-methylglycine)] as carbon sources. When tested for their potential as alternative nitrogen sources, 5 mM GB, DMG, sarcosine, or glycine were used to supplement nitrogen-free (lacking ammonium sulfate) MP medium. Solid MP media contained Bacto agar at 15 g/L and carbon sources succinate at 15 mM and methanol at 125 mM when used. When required for selection in the media, tetracycline was added to a final concentration of 12.5 µg/mL.

TABLE 4.

Bacterial strains used in this study

Strain or plasmid	Genotype	Description
Escherichia coli strains
DH5α		Strain used for general cloning. Sourced from NEBiosciences.
S17-1		Strain used for allelic exchange.
JB554	S17-1 pZH030 (pCM433_ΔgbdR)	Contains conditional replication vector to delete gbdR (Mext_3735) in M. extorquens PA1.
M. extorquens strains
CM2730	ΔbcsABZC	Wild-type (“WT”) parental strain of strains used in this study.
CM3839	hpt::mCherry	WT strain with constitutively expressed mCherry inserted at the hpt locus.
JB258	dgcBP30L	Spontaneous GB+ mutant. Parental GB+ strain used in this study.
JB528	dgcBP30L gbdRD167N	Spontaneous DMG+ mutant isolated from JB258 parent strain.
JB529	dgcBP30L gbdRG165S	Spontaneous DMG+ mutant isolated from JB258 parent strain.
JB533	dgcBP30L gbdRE170V	Spontaneous DMG+ mutant isolated from JB258 parent strain.
JB562	dgcBP30L ΔgbcBA	Deletion of gbcBA (Mext_3747–3748) in JB258 background.
JB577	dgcBP30L ΔgbdR	Deletion of gbdR (Mext_3735) in JB258 background.
JB578	dgcBP30L ΔproVUX	Deletion of proVUX (Mext_3731–3733) in JB258 background.
JB669	ΔdgcBA	Deletion of dgcBA (Mext_3745–3746) in JB258 background.
JB1151	dgcBP30L ΔgbdR pCH07	Strain testing complementation with empty expression vector.
JB1152	dgcBP30L ΔgbdR pEK17	Strain testing complementation with the wild-type gbdR allele from M. extorquens PA1.
JB1180	dgcBP30L ΔgbdR pEK18	Strain testing complementation with the wild-type gbdR allele from M. extorquens AM1.
JB1181	dgcBP30L ΔgbdR pEK21	Strain testing complementation with a mutant allele of gbdR from M. extorquens PA1 encoding the R214A variant.
Plasmids
pZH030		Allelic exchange vector made in pCM433 background to delete gbdR (Mext_3735) in M. extorquens PA1.
pCH07		Tetracycline-resistant derivative of pLC290 containing RBS from fae to drive gene expression.
pEK17		Derivative of pCH07 expressing gbdR from M. extorquens PA1.
pEK18		Derivative of pCH07 expressing gbdR from M. extorquens AM1.
pEK21		Derivative of pCH07 expressing a mutant allele of gbdR from M. extorquens PA1 encoding the R214A variant.

Isolation of DMG+ mutants

Ten independent colonies of M. extorquens PA1 genotype dgcB^P30L 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 entering the early stationary phase (~30 hours). A total volume of 100 µL of each culture was plated onto three independent plates (producing 30 plates per media type in total) of solid MP media supplemented with either 8 mM DMG or 8 mM sarcosine as a sole carbon source. Plates were incubated at 30°C for 28 days before colonies arose. Plates that did not have growth at this point were further incubated for up to 90 days, but no colonies that arose after longer incubation were able to be regrown in liquid culture. Isolate colonies from different DMG plates (10 in total) were chosen for characterization. After initial characterization, genomic DNA was isolated from five DMG+ isolate strains and sequenced by short-read Illumina sequencing. Specific mutations were identified by comparison to the M. extorquens PA1 reference genome using Breseq (73).

Genetic approaches

The allelic exchange was utilized to produce clean markerless deletions of genes in the M. extorquens PA1 genome, as previously described (74). Specifically, E. coli S17-1 containing chromosomally expressed tra genes and a vector derived from pCM433 (tetracycline resistant) were combined with M. extorquens recipient strains in biparental matings. The cell mixture was plated on solid nutrient agar and incubated overnight at 30°C and then resuspended in MP media without a carbon or nitrogen source, serially diluted in this media in a 10-fold dilution series, and plated on solid MP media containing 15 mM succinate as a carbon source, 5 mM methylamine as a nitrogen source, and 12.5 µg/mL tetracycline, and incubated for several days at 30°C until visible colonies grew. The allelic exchange was completed by plasmid counterselection arising from the presence of sucrose: colonies were restreaked onto solid MP media containing 15 mM succinate and 5% sucrose and incubated at 30°C for several days. SnapGene software was used to design plasmids and primers, and following PCR amplification, plasmids were constructed using HiFi assembly kits (NEB). Gene edits were confirmed via PCR amplification of deleted loci.

Growth quantification

Isolated colonies from solid agar plates were picked and placed into liquid media. All growth experiments were conducted in biological triplicate, each individual colony representing a biological replicate. Initial cultures were grown in 2 mL liquid MP media supplemented with 3.5 mM succinate as a carbon source. These cultures underwent incubation at 30°C using a rotary drum apparatus and were subsequently subcultured (1/64) into 5 mL of MP medium containing the respective carbon source for acclimation. Following acclimation periods of approximately 30 hours for succinate, around 40 hours for methanol, and about 72 hours for GB, stationary phase cultures were further subcultured (1/64) into 640 µL of MP medium supplemented with the appropriate carbon sources. This step took place in 48-well polystyrene plates (Falcon, Ref. No. 351178), sealed with adhesive optical film (VWR, Cat No. 60941-064) to prevent volatile metabolite evaporation. For testing alternative nitrogen sources, nitrogen-limited cells were obtained by growing cells on MP medium with only 400 µM ammonium sulfate in starter and acclimation cultures. The cultures were then incubated at 30°C with a 2°C gradient to prevent condensation in a BioTek Epoch 2 microplate spectrophotometer, utilizing single orbital shaking at 800 rpm. Cell density was monitored by measuring absorbance at 600 nm every hour. The cell growth rate (μ) and generation time (gt) were calculated through non-linear regression of data points identified within the exponential growth phase, where μ = ln(X_T/X₀)/T, and gt = ln2/μ, with X representing OD₆₀₀ and T denoting elapsed time.

Real-time quantitative PCR

M. extorquens PA1 cultures in the mid-log phase (OD₆₀₀ 0.30 ± 0.02) were collected and subjected to centrifugation at 3,234 × g for 10 minutes. The resulting cell pellets were frozen at −80°C until further processing. To extract RNA, the frozen cell pellets were resuspended in RNAsnap and incubated at 95°C for 7 minutes. Cell debris was eliminated by centrifugation at 16,000 × g for 5 minutes, as described previously (75). The RNA was isolated from the RNAsnap solution using the RNA Clean and Concentrator kit from Zymo Research. The quantity of RNA was determined using a NanoDrop One spectrophotometer, and its quality was evaluated through denaturing agarose gel electrophoresis. Genomic DNA was removed by treating the RNA with DNase I following the manufacturer’s instructions (ThermoFisher). Subsequently, cDNA was synthesized using the High-Capacity cDNA Reverse Transcription kit from Applied Biosystems, following the manufacturer’s protocol. For qPCR analysis, Maxima SYBR Green/ROX qPCR Master Mix (ThermoFisher) was used in 10 µL reaction volumes on Fast Optical 96-well reaction plates (MicroAmp). The qPCR reactions were performed on a StepOnePlus Real-Time PCR system (Applied Biosystems) with the following cycling conditions: an initial denaturation step at 95°C for 10 minutes, followed by 40 cycles of amplification consisting of denaturation at 95°C for 20 s, annealing at 60°C for 20 s, and extension at 72°C for 60 s. A plate reading step was conducted after each cycle. Melt-curve analysis was carried out by subjecting the samples to a temperature ramp from 70°C to 95°C at a rate of 0.3°C/s, following a brief denaturation step at 95°C. Each qPCR reaction was performed in triplicate for three independent biological replicates. StepOne Software v2.3 was utilized to acquire and analyze the qPCR data, and cycle threshold (CT) values were calculated for each reaction. The technical triplicate values were averaged, normalized to recA, and analyzed using the 2-ΔΔCt method (76).

Competition assays

Methanol-acclimated cultures were combined in equal volumetric ratios with the hpt::mCherry strain CM3839, as described previously (77). Subsequently, cell suspensions were introduced into 5 mL of MP medium (methanol), with or without 200 mM NaCl and with or without 1 mM GB. The cultures were incubated at 30°C with shaking. Upon reaching the early stationary phase (OD₆₀₀ ~ 0.4), 1 mL of each culture was mixed with DMSO (final concentration of 8%) and stored at −80°C until analysis.

For analysis, the cultures were thawed, subjected to centrifugation, and then resuspended in 1 mL of PBS, followed by a 1:10 dilution before examination of population composition using flow cytometry. The frequency of competitors was assessed by passing mixed population samples from the experiment’s start (F₀) and end (F₁) through an LSRFortessa X-20 flow cytometer (Becton Dickinson Biosciences). Approximately 100,000 cells per sample were gated based on forward and side scatter. Competitors were distinguished by comparing fluorescence, excited at 561 nm and emitted through a 610/620 nm BP filter.

Malthusian fitness measurements (w) were determined using the methods previously outlined (77).

Statistical methods

Statistical analyses were executed using GraphPad Prism version 10.2.0 software. Comparisons between growth rates were performed using multiple t-tests with Bonferroni correction. Comparison of fitness measurements reported in Fig. 12 and 13 was conducted via Welch ANOVA with Dunnet’s T3 post hoc correction. RT-qPCR statistical analyses were performed using raw ΔCt values and a distribution-free Wilcoxen two group test with Dunnet’s T3 post hoc correction (78).

GbdR structure prediction and mutant structures

ColabFold2 v1.5.5: AlphaFold2 using MMseqs2 was employed to generate a 3D protein structure model based on the amino acid sequence of wild-type GbdR (79, 80). Default model building parameter values were used except for the following parameters specified here: template_mode = pdb100 and num_relax = 5. pLDDT values per residue led to the selection of the top wild-type GbdR model out of five generated (Fig. S6). Wild-type GbdR model was then used as an input in ChimeraX to generate three mutant (G165S, D167N, and E170V) structures (81).

Molecular dynamic simulations

Wild-type GbdR and three mutant structures were used as inputs to conduct a total of four molecular dynamics simulations using GROMACS 2020.4 software with the Charmm36 force field (82, 83). Prior to performing simulations, residues 1–24 forming disordered regions were removed to increase the computational efficiency and with the awareness that they are distant from the GbdR variant sites. Each structure was solvated in a dodecahedral box filled with TIP3P water molecules and neutralized with 0.15 M NaCl. The steepest descent minimization (10,000 steps) was followed by equilibration simulations: 1 ns with harmonic restraints on protein-heavy atoms and 1 ns without restraints. During restrained simulations, temperature was linearly increased to 310 K, while pressure was maintained at 1 atm using the Berendsen algorithm. Production simulations (100 ns) utilized the Parrinello-Rahman method for pressure control. All simulations employed the LINCS algorithm for bond constraints enabling a 2 fs timestep. Temperature control was achieved with the v-rescale option. Electrostatics were calculated using Particle mesh Ewald with a 1.2 nm real-space cutoff. Van der Waals interactions were cut off at 1.2 nm using the Potential-shift-Verlet method. Each system was simulated for 100 ns, and trajectories were analyzed using VMD and GROMACS analysis tools (84).

Docking of substrates to GbdR

Energy minimized structure of wild-type GbdR was used for docking using Glide (Schrödinger). Glide SP docking was performed following established protocols provided by the developers (85). GbdR structure was prepared using a protein preparation wizard to ensure proper atom typing and protonation states. Substrate structures of GB, DMG, and sarcosine were downloaded from Pubchem and prepared using the Epik module to assign correct protonation states. The docking grid was kept sufficiently large to allow exploration of the entire protein. Glide SP docking was then performed with default settings. The poses with the highest GlideScore were visually inspected and used for binding site characterization.

DATA AVAILABILITY

Sequence data for characterized evolved isolates of M. extorquens PA1 (listed in Table 2) are available on the NCBI Short Read Archive database under the accession number PRJNA1110189, including BioSamples with the accession numbers SRR28991164-SRR28991168.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.00310-24.

Supplemental material. aem.00310-24-s0001.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.