C1-Pathways in Methyloversatilis universalis FAM5: Genome Wide Gene Expression and Mutagenesis Studies

Methyloversatilis universalis FAM5 utilizes single carbon compounds such as methanol or methylamine as a sole source of carbon and energy. Expression profiling reveals distinct sets of genes altered during growth on methylamine vs methanol. As expected, all genes for the N-methylglutamate pathway were induced during growth on methylamine. Among other functions responding to the aminated source of C1-carbon, are a heme-containing amine dehydrogenase (Qhp), a distant homologue of formaldehyde activating enzyme (Fae3), molybdenum-containing formate dehydrogenase, ferredoxin reductase, a set of homologues to urea/ammonium transporters and amino-acid permeases. Mutants lacking one of the functional subunits of the amine dehydrogenase (ΔqhpA) or Δfae3 showed no growth defect on C1-compounds. M. universalis FAM5 strains with a lesion in the H4-folate pathway were not able to use any C1-compound, methanol or methylamine. Genes essential for C1-assimilation (the serine cycle and glyoxylate shunt) and H4MTP-pathway for formaldehyde oxidation showed similar levels of expression on both C1-carbon sources. M. universalis FAM5 possesses three homologs of the formaldehyde activating enzyme, a key enzyme of the H4MTP-pathway. Strains lacking the canonical Fae (fae1) lost the ability to grow on both C1-compounds. However, upon incubation on methylamine the fae1-mutant produced revertants (Δfae1R), which regained the ability to grow on methylamine. Double and triple mutants (Δfae1RΔfae3, or Δfae1RΔfae2 or Δfae1RΔfae2Δfae3) constructed in the revertant strain background showed growth similar to the Δfae1R phenotype. The metabolic pathways for utilization of methanol and methylamine in Methyloversatilis universalis FAM5 are reconstructed based on these gene expression and phenotypic data.

lesion in the H4-folate pathway were not able to use any C1-compound, methanol or methylamine. Genes essential for C1-assimilation (the serine cycle and glyoxylate shunt) and H4MTP-pathway for formaldehyde oxidation showed similar levels of expression on both C1-carbon sources. M. universalis FAM5 possesses three homologs of the formaldehyde activating enzyme, a key enzyme of the H4MTP-pathway. Strains lacking the canonical Fae (fae1) lost the ability to grow on both C1-compounds. However, upon incubation on methylamine the fae1-mutant produced revertants (Δfae1 R ), which regained the ability to grow on methylamine. Double and triple mutants (Δfae1 R Δfae3, or Δfae1 R Δfae2 or Δfae1 R Δfae2Δfae3) constructed in the revertant strain background showed growth similar to the Δfae1 R phenotype. The metabolic pathways for utilization of methanol and methylamine in Methyloversatilis universalis FAM5 are reconstructed based on these gene expression and phenotypic data.
In addition to methanol, all tested members of the genera Methyloversatilis are capable of utilizing methylated amines [6,10]. M. universalis FAM5 oxidizes methylamine via the N-methylglutamate pathway [12]. The pathway includes three enzymes: the N-methylglutamate synthase (NMGS, encoded by mgsABC gene cluster), which can produce N-methylglutamate from glutamate and methylamine; the N-methylglutamate dehydrogenase (NMGDH, encoded by mgdABCD gene cluster), which is predicted to produce methylene-tetrahydrofolate (H4F) and regenerated glutamate from N-methylglutamate; and the gamma-glutamylmethylamine synthetase (GMAS, gmas), a reversible enzyme which produces gamma-glutamylmethylamine (GMA) from glutamate and methylamine using ATP. The gamma-glutamylmethylamine synthetase is predicted to serve as the first enzyme of the pathway, however the exact functional role of the enzyme and its product, NT, not tested; "++++" correspond to doubling time of 6 h; "+++" correspond to doubling time of 9 h; "++" correspond to 12 h doubling time; and "-" indicates no growth. Methanol (5 mM) was added to test if a mutant strain is sensitive to formaldehyde (2) or unable to produce methyl-H 4 folate for purine synthesis (1) .

DNA Manipulations
DNA was isolated using the QIAamp DNA Mini kit (Qiagen, Venlo, Netherlands). Plasmid DNA was purified using the Qiagen Miniprep Kit (Qiagen). E. coli transformation, restriction enzyme digestion and ligation reactions were carried out as described by Sambrook et al. [29]. Polymerase chain reaction (PCR) amplifications were performed using hot-start Taq polymerase (Qiagen) in accordance with the manufacturer's instructions. Primers used for PCR amplification of upstream or downstream regions (approximately 450-600 bp) of each targeted gene are listed in Table S1. Amplified fragments were cloned into pCR2.1, sequenced and then sub-cloned into pCM184 using appropriate restriction sites. After verification of the nucleotide sequence, the plasmids were transformed into E. coli S17-1, and the resulting donor strains were mated with wild-type M. universalis FAM5 via biparental mating as previously described [8]. The identity of the double-crossover mutants was verified by diagnostic PCR with primers specific to the insertion sites. Mutant phenotypes were assessed on solid media and in liquid culture with succinate, pyruvate, methanol or methylamine as carbon sources.

RNA-seq Experiments
Trascriptomic experiments were carried out with M. universalis strain FAM5. Cells pre-grown on methanol (10 mM) were used for inoculation of methanol (25 mM) or methylamine (35 mM) cultures at 5:100 ratios (inoculum:fresh media). Cultures were grown to mid-exponential phase (OD600 0.4 ± 0.05). The cellular activities were terminated by addition of "stop solution" as described [30]. The cells were collected by centrifugation at 4300× g at 4 °C for 10 min. mRNA samples were isolated and enriched as previously described [30] and submitted to the University of Washington's High-Throughput Sequencing Solutions Center on dry ice for single-read Illumina ® sequencing (Department of Genome Sciences, University of Washington; http://www.htseq.org/). The RNA-Seq sequence data sets ranged from 12.19 to 25.46 million reads (36 bp) per sample. The M. universalis FAM5 genome to be used as the alignment scaffold was downloaded from MaGE (https://www.genoscope.cns.fr/agc/microscope/mage/) on 31 March 2014. The raw reads were aligned to the scaffold using BWA version 0.7.4-r385 with default parameters [31]. The alignments were post-processed and sorted into BAM files with SAMTools version 0.1.19-44428cd [32]. Reads were attributed to open reading frames (ORFs) using the htseq-count tool from the "HTSeq" framework version 0.5.4p5 in the "intersection-nonempty" mode [33]. The ratios of reads mapped per ORF in the two conditions were calculated along with p-values using DESeq2 [33]. The p-values in Table 2 were corrected for multiple testing using the q-value method of Storey [34]. Proportions of reads mapped to rRNA were 78%-90% for RNA-libraries prepared from cells grown on methanol, and 91%-92% for RNA-libraries prepared from cells grown on methylamine. For display and presentation purposes, the Reads per Kilobase per Million reads sequenced (RPKM) were calculated [35].

Enzyme Assays
All cell extracts (50 mL) for activity were prepared from methylamine-grown cultures. Cells were harvested by centrifugation at 4500× g using a Sorvall RC-5B centrifuge at 4 °C for 10 min. The supernatant was removed and the cell pellet was stored at −80 C. Cell pellets were thawed and resuspended in 1 mL of 100 mM potassium phosphate buffer, pH 7.6 or 7.2 and broken using a French press (three times, 1000 psi). The extracts were centrifuged at 28,000× g for 5 min to remove cell debris. Qhp activity was detected by measuring amine dehydrogenase activity using a spectrophotometric assay measuring the reduction rate of potassium ferricyanide (500 μM, at 420 nm) in the presence of cell free extract (100 μg), and methylamine (25 mM) as substrate at room temperature. The extinction coefficient of potassium ferricyanide is 1.02 mM −1 cm −1 at pH = 7.6. The total volume of the reaction was 200 μL. Activity was measured using a Spectramax 190 plate reader (Molecular Devices, Sunnyvale, CA, USA); a minimum of two biological replicates was assayed.

Accession Numbers
The RNA-seq data were uploaded into NCBI/GEO under accession number GSE63822.
Most of the previously recognized metabolic modules essential for oxidation of C1-compounds fell into very high to moderate categories. Expression of central metabolic pathways, such as the citric acid cycle, amino acid biosynthesis, gluconeogenesis and energy generating pathway (respiratory chain and ATP-synthase) genes remained mostly unchanged (Table 3). Genes encoding the H4MPT synthesis and the H4MPT-linked C1-transfer enzymes and the serine cycle enzymes also remained unchanged, suggesting that these pathways contribute in a similar fashion to methanol and methylamine utilization (Table 3).
Cells growth on methanol showed a slightly higher (about 1.4 fold) abundance of transcripts for methanol dehydrogenase (mdh2) and the associated cytochrome gene (cyt) ( Table 3; Figure 1). The relative abundance of formate-tetrahydrofolate ligase/formyltetrahydrofolate synthetase (ftfL) transcripts, a tungsten-containing aldehyde ferredoxin oxidoreductase, a distant homolog of the PQQ-dependent methanol dehydrogenase (xoxF3G3J3), cytochrome bc1 reductase complex, a putative NADP(H)-dependent aldo/keto reductase and a number of hypothetical proteins were higher (1.6-3 folds) in methanol grown cells (Table 3; Figure 1). Genes encoding the N-methylglutamate pathway enzymes, co-clustered conserved proteins and a putative transcriptional regulator, were up-regulated (6−10-fold) during growth on methylamine (Table 3, Figure 1). Among other genes altered by the shift to methylamine were: a seven gene cluster (qhpRADCBFE) encoding a heme-containing amine dehydrogenase (Qhp, qhpADC), and associated proteins involved in the posttranslational modification and regulation; a ferredoxin; and a distant homolog of the formaldehyde activating enzyme (fae3). The fae3 gene was almost 10 fold higher in cells shifted to methylamine.

Mutagenesis Studies: Methylamine Oxidation
It has previously been shown that deletion of two subunits of NMGS (encoded by mgsA and mgsB) completely abolished the ability of the strain M. universalis FAM5 to utilize methylamine; however the mutants were still able to generate small amounts of N-methylglutamate [12]. It has been speculated that the third subunit of the protein (encoded by mgsC) can contribute to the residual production. Here we mutated mgsC and showed that the lack of the gene eliminates the ability of the strain to utilize methylamine (Table 1).
We also tested involvement of the gamma subunit of the NMGDH (mgdD) in methylamine utilization. The mgdD is located at the end of the four-gene cluster encoding the NMGDH. The fourth gene has been identified in all microbes possessing the N-methylglutamate pathway; however, in contrast to the catalytic subunits (mgdA and mgdC) and mgdB, which are typically quite conserved, the gamma subunit of the enzyme is surprisingly divergent [12]. Relatively low homology (30%-45% amino acid identity) has been observed between the mgdD subunit from the Methyloversatilis species and the corresponding protein from betaproteobacterial methylotrophs, such as Methylotenera versatilis 301 or Methylophilus methylotrophus DSM 46235. The gene from Methyloversatilis species shared 25% AA identity with the mgdD gene from Agrobacterium tumefaciens. No homology was observed among mgdD genes from betaproteobacteria and gamma subunit of the NMGDH from alphaprotebacterial methylotrophs, such as Methylocellla silvestris, Methylopila spp, and Methylobacterium extorquens spp. The exact function of the mgdD gene product is still not known. We mutated the mgdD gene in M.universalis FAM5, and the mutant strain lost the ability to utilize methylamine (Table 1). These two additional mutations additionally confirmed the importance of the N-methylglutamate pathway genes for methylamine utilization by M. universalis FAM5.
The transcriptomic study presented above revealed a number of additional functions indicating that the metabolic arrangement of the methylamine utilization in M. universalis FAM5 could be more intricate (Figure 1). Two of the most noticeable differences between cells grown on methanol vs. grown on methylamine are the overexpression of the heme-containing amine dehydrogenases gene cluster and fae3 in cells grown on methylamine. We generated a strain of M. universalis FAM5 lacking the alpha subunit of the Qhp gene (ΔqhpA). The activity of the Qhp enzyme was reduced to background level in the mutant strain (Table 4). However, the strain lacking Qhp gene (ΔqhpA) was able to grow on methylamine similarly to wild type ( Table 1). The phenotype of the fae3 mutant is described below.

Mutagenesis Studies: Metabolic Arrangement Downstream from the Methylamine Oxidation
During growth on methanol, formaldehyde, the end product of alcohol oxidation, is oxidized to formate via the H4MPT pathway. Formate is either oxidized to CO2 to generate energy, or converted to methylene-H4F via an FtfL-Fch-MtdA variant of the H4F pathway. It has been predicted that the end product of the N-methylglutamate pathway is methylene-H4F, which could be incorporated directly into the serine cycle. However it is not known how it is oxidized. The FtfL-Fch-MtdA variant of the H4F pathway is typically linked to assimilation rather that oxidation [36]. In order to investigate the fate of the methylene-H4F in M. universalis FAM5 upon growth on methylamine, two possible scenarios were investigated: (1) The H4F pathway contributes to methylene-H4F oxidation by operating in the reverse direction; (2) The H4F pathway in M. universalis FAM5 contributes only to formate assimilation, and thus it should not be essential for growth on methylamine. However, in order to generate energy and reducing equivalent from C1-oxidation, the M. universalis FAM5 cells should possess an additional system for converting methylene-H4F to formaldehyde, or transfer C1-units from H4F to H4MPT. One candidate for C1-unit decoupling/shuttling is fae3, a homolog of formaldehyde activating enzyme, which has a very strong expression during growth on methylamine, but not methanol.
To verify these scenarios, mutations in H4F (ΔmtdA) and H4MPT pathways (Δfae1) were made. In addition we mutated homologs of the formaldehyde dehydrogenase (Δfae2 and Δfae 3), and generated multiple mutants (Δfae1Δfae2, Δfae1Δfae3, Δfae2Δfae3 and Δfae1Δfae2Δfae3). In addition, a strain lacking fae1 and qhpA was constructed. The mutant phenotypes are shown in Table 1.
Strains lacking mtdA or fae1 genes were not able to grow on methanol or methylamine. Interestingly, revertants were observed for the fae1-mutant. The number of revertants arising on methylamine plates was 10 ± 3 per 10 9 cells plated. The revertant strains grew on methylamine, albeit with a growth defect compared to the wild type strain (the growth rate was about 30% of the wild type). The revertant strains did not restore the ability to utilize methanol for growth. Single Δfae 2 or Δfae3 mutations, or the double Δfae2Δfae3 mutation displayed a growth rate similar to the wild-type strain on both methanol and methylamine. Δfae1Δfae2 and Δfae1Δfae3 double mutants were not able to grow on C1-compounds, similarly to Δfae1. No revertants were observed for Δfae1Δfae3 mutant upon transfer to methylamine plates. However, the incorporation of the Δfae3 mutation into the revertant Δfae1 R strain did not abolish the ability of the strain to utilize methylamine. A similar phenotype was observed for a triple Δfae1 R Δfae2Δfae3 mutant. The double Δfae1ΔqhpA mutant phenotype was similar to the single Δfae1 mutant. Like the latter, there were revertants of the double mutant (Δfae1ΔqhpA R ) that regained the ability to utilize methylamine for growth. The frequency of reversion was similar to the single fae1-mutant. The triple Δfae3Δfae1ΔqhpA R mutant retained the ability to use methylamine. Table 4. Activity of the heme-containing amine dehydrogenase in wild type and ΔqhpA-mutant strains upon growth on methylamine.

Discussion
Methylotrophic capabilities of the members of families Rhodocyclaceae/Burholderiaceae have relatively recently been discovered [1,[3][4][5]7], and thus not much is known about the pathways for the single carbon utilization in this betaproteobacterial lineage. The M. universalis is the first representative of the family isolated in a pure culture and formally described [6]. The genome of the strain has been sequenced opening up new approaches for investigation of pathways for single carbon or methylated multicarbon compounds investigation [23]. Over the past few years the strain became the model system for understanding molecular mechanisms of C1-carbon utilization and denitrification [8,12,13].
The mutagenesis performed in this study suggests that the N-methylglutamate pathway is the main route for methylamine oxidation. Based on the available genomic and enzymatic data the end product of the N-methylglutamate pathway is methylene-H4F [12,16,20]. Thus it could be predicted that, first, the H4folate pathway contributes to C1-oxidation; and, second, the H4MPT-linked transfer should not be required for growth on methylamine. M. universalis FAM5 possesses the Mtd-Fch-FtfL variant of the H4MPT pathway, which is typically associated with formate assimilation, rather than formaldehyde/methylene-H4F oxidation [25,26]. It has been shown that this portion of the pathway is not essential for growth on methylamine in Methylobacterium spp. [16,20]. We found here that the mtdA mutant is not able to grow on both C1-compounds. The phenotypic data indicate that the enzyme of the pathway can play some additional role in C1-utilization in M. universalis.
We also found that the H4MPT pathway is essential for growth on methylamine. That observation indicates that M. universalis somehow produces methylene-H4MPT during growth on methylamine. Since the heme-dependent amine dehydrogenase (which is predicted to produce formaldehyde) is not essential for growth on methylamine, the C1-units entering H4MPT pathway should come from methylamine oxidation via the N-methylglutamate pathway. Based on the available evidences it could be proposed that the NMGDH releases both methylene-H4F and formaldehyde and that the ratio of the produced compounds depends on the intracellular pool of H4F. It has been shown that the pool of H4F is only a quarter of the pool of H4MPT in facultative methylotrophs [37]. In M.universalis, the relative expression of the dihydrofolate reductase (folA), the only enzyme essential for biosynthesis of H4F (folA) from a supplied source of folate, showed similar levels of expression in methanol or methylamine grown cells. That suggests that the overall capacity of the cells to produce the folate-cofactor did not change. If methylene-H4folate is the only product of methylamine oxidation, the cellular needs for H4F should be much higher upon growth on methylamine. Because formaldehyde is a toxic compound, it is reasonable to speculate that formaldehyde is produced only upon the depletion of the H4F pool. The Fae1 and/or Fae3 enzymes handle the cellular formaldehyde pool, and deliver C1-units into the H4MPT pathway for oxidation. However, while this scenario explains the growth deficiencies of the fae1 mutant, it fails to explain the revertants. It could be speculated that in vivo the N-methylglutamate dehydrogenase uses both H4F and H4MPT cofactors. The preferable compound is H4F, which could be substituted with H4MPT upon depletion of H4F and buildup of H4MPT (due to the lack of Fae enzymes). The latter scenario is also supported by the previous study of methylamine utilization via the NMG-pathway [16], which showed that the H4MPT-biosynthesis, rather than the Fae-driven condensation of formaldehyde, is essential for growth of Methylobacterium extorquens PA1 on methylamine.
The initial reconstruction of the methylotrophy in methylotrophic Rhodocyclaceae/Burholderiaceae has been made based on enzymatic and genetic investigations [13,38,39]. In this study we further evaluated pathways for aerobic utilization of methanol and methylamine via transcriptomics and mutagenesis. The updated reconstruction of the core metabolic pathways in M. universalis FAM5 is shown in Figure 1. It has been demonstrated that methanol oxidation via PQQ-dependent enzymes is operating as a redox arm and is coupled with ATP generation with 0.6-1 mol of ATP produced per 1 mol of methanol oxidized [40]. Thus, the growth on the C1-alcohol is predicted to be reducing-power limited. Interestingly, the transcriptomic studies show that cells of M. universalis FAM5 expressed complex III (cytochrome bc1). As the majority of the cellular energy needs are fulfilled by the first step of methane oxidation, it could be speculated that M.universalis FAM5 uses complex III to supply reducing power for carbon assimilation upon growth on methanol. Methylamine oxidation in M.universalis FAM5 might also be connected to NADPH production, via a ferredoxin-NADP reductase. An uphill electron transfer has been proposed for a number of chemolitotrophs [41][42][43], but it has never been investigated in non-phototrophic methylotrophs. Here we present the first evidence that methylotrophic bacteria might develop a mechanism for restoring reducing power upon growth on C1-compounds. The exact role of the complex III and FNR in cellular energetics will be evaluated in further investigations.

Conclusions
Overall, the metabolic arrangement of C1-utilization in M. universalis appears to be quite complex and to some degree redundant. Methylamine utilization is supported by two different enzymatic systems: the N-methylglutamate pathways and heme-dependent amine dehydrogenase. Genes encoding both enzymatic systems are chromosomally co-located and show a very high level of expression upon switch from growth on methanol to growth on methylamine. Like other heme-containing amine dehydrogenases, the enzyme from M. universalis FAM5 is capable of direct conversion of methylamine to formaldehyde. However, from the ΔqhpA phenotypic data it could be suggested that the enzyme is not essential, at least under the growth condition tested, for methylamine utilization.
The functional implication of the enzymatic redundancy of the C1-utilization is not apparent in the experiments described here. It could be speculated, that in the presence of heterogeneous substrates, the simultaneous activation of different enzymatic pathways might provide some benefit. However, it remains to be determined via additional simulation experiments or in-situ investigations.