Core and auxiliary functions of one-carbon metabolism in Pseudomonas putida exposed by a systems-level analysis of transcriptional and physiological responses

ABSTRACT The soil bacterium Pseudomonas putida is a robust biomanufacturing host that assimilates a broad range of substrates while efficiently coping with adverse environmental conditions. P. putida is equipped with functions related to one-carbon (C1) compounds (e.g. methanol, formaldehyde, and formate) oxidation—yet pathways to assimilate these carbon sources are largely absent. In this work, we adopted a systems-level approach to study the genetic and molecular basis of C1 metabolism in P. putida. RNA sequencing identified two oxidoreductases, encoded by PP_0256 and PP_4596, transcriptionally active in the presence of formate. Quantitative physiology of deletion mutants revealed growth defects at high formate concentrations, pointing to an important role of these oxidoreductases in C1 tolerance. Moreover, we describe a concerted detoxification process for methanol and formaldehyde, the C1 intermediates upstream formate. Alcohol oxidation to highly-reactive formaldehyde by PedEH and other broad-substrate-range dehydrogenases underpinned the (apparent) suboptimal methanol tolerance of P. putida. Formaldehyde was mostly processed by a glutathione-dependent mechanism encoded in the frmAC operon, and thiol-independent FdhAB and AldB-II overtook detoxification at high aldehyde concentrations. Deletion strains were constructed and characterized towards unveiling these biochemical mechanisms, underscoring the worth of P. putida for emergent biotechnological applications—e.g. engineering synthetic formatotrophy and methylotrophy. IMPORTANCE C1 substrates continue to attract interest in biotechnology, as their use is both cost-effective and ultimately expected to mitigate the impact of greenhouse gas emissions. However, our current understanding of bacterial C1 metabolism remains relatively limited in species that cannot grow on (i.e., assimilate) these substrates. Pseudomonas putida, a model Gram-negative environmental bacterium, constitutes a prime example of this sort. The biochemical pathways active in response to methanol, formaldehyde, and formate have been largely overlooked—although the ability of P. putida to process C1 molecules has been previously alluded to in the literature. By using a systems-level strategy, this study bridges such knowledge gap through the identification and characterization of mechanisms underlying methanol, formaldehyde, and formate detoxification—including hitherto unknown enzymes that act on these substrates. The results reported herein both expand our understanding of microbial metabolism and lay a solid foundation for engineering efforts toward valorizing C1 feedstocks.


Bacterial strains, medium composition and culture conditions
All bacterial strains and plasmids are listed in Table 1 and Table 2, respectively. E. coli DH5α λpir was used as a cloning host, while the reduced-genome P. putida strain EM42 (21) was selected for quantitative physiology and engineering purposes. Lysogeny broth (LB) complex medium (containing 10 g L −1 tryptone, 5 g L −1 yeast extract, and 10 g L −1 NaCl) and de Bont minimal (DBM) medium were used for all cultivations (34). DBM medium contained 3.88 g L −1 K 2 HPO 4 , 1.63 g L −1 NaH 2 PO 4 , 2 g L −1 (NH 4  Overnight cultures in LB medium were diluted 1/100 to inoculate a 5 mL preculture of DBM medium with 20 mM glucose in a 50mL culture tube and incubated at 30°C and 250 rpm for ca. 18 h. This overnight culture was washed with DBM medium without any carbon source prior to the inoculation of the main culture in either 96-well microtiter plates or baffled Erlenmeyer flasks (shaken-flask cultures) with the appropriate carbon source(s) as described in the text. For cultivations in 96-well microtiter plates, 200 µL of a cell suspension at an OD 600 of 0.05 were incubated in an ELx808 microtiter plate reader (BioTek Instruments Inc., Winooski, VT, USA). Shaken-flask cultivations were performed in 50 mL of medium in 250 mL Erlenmeyer flasks at 30°C and 200 rpm. The specific growth rate (μ) and the extension of the lag phase (λ) were calculated by performing a smooth spline fit on the growth data (40).

Construction of mutant P. putida strains
The corresponding suicide pGNW2-derivative plasmid was delivered into the cells by electroporating 500 ng of plasmid DNA into 50 µL of freshly prepared electrocompetent P. putida EM42 cells, previously washed three times with 300 mM sucrose. Electroporation was performed with a Gene Pulser XCell (Bio-Rad) set to 2.5 kV, 25 µF capacitance, and 200 Ω resistance in a 2 mm gap cuvette. Positive co-integration events were further transformed with pQURE6·H (Table 2), a conditionally replicative plasmid bearing the meganuclease gene I-SceI. I-SceI cuts pGNW2 co-integrants within the chromosome, thus forcing a second homologous recombination event. Cells were recovered in 1 mL of LB medium supplemented with 2 mM of 3-methylbenzoate (3-mBz) for at least 3 h at 30°C and plated onto LB medium agar containing the corresponding antibiotic(s) and 1 mM 3-mBz to induce both plasmid replication and I-SceI expression. Positive clones were identified by colony PCR, verified by DNA sequencing, and cured from the resolving plasmid by serial dilution under nonselective conditions.

Analytical procedures for quantitative physiology experiments
The concentration of glucose, formate, and methanol was measured in culture supernatants in a high-performance liquid chromatography (HPLC) instrument (UltiMate 3000 Basic Automated System ThermoFisher Scientific Co) equipped with an Aminex HPX-87P column (Bio-Rad) and a refractive index detector Shodex RI-101 (Showa Denko America Inc., NY, USA). Specific glucose (q G ), formate (q F ), and methanol (q M ) consumption rates were normalized to the cell dry weight (CDW) with the following equation where q S corresponds to the biomass-specific substrate consumption rate (mmol g CDW −1 h −1 ), X is the average biomass concentration between two sampling time points (g CDW L −1 ), ∆S is the difference in substrate concentration between two sampling time points (mM), and ∆t refers to the time between two sampling points (h). The q G, q F, and q M values correspond to an average of the values individually determined in three biological replicates.

Comparative transcriptome analysis by RNA-Seq
Differential transcriptome analysis was used to explore gene expression in shaken-flask cultures of P. putida grown in the presence of formate or methanol. At the time(s) indicated in the text, an equivalent of 1 mL of culture at an OD 600 of 1 was pelleted at 10,000×g for 10 min at 4°C, washed with phosphate-buffered saline and immediately frozen in liquid nitrogen. Pellets were stored at −80°C until further processed. RNA extraction, library preparation, and RNA-Seq were performed by BGI (Shenzhen, China) as indicated previously (43). Sequencing data were filtered by removing reads mapped to rRNA as well as sequencing reads containing low-quality, adaptor-polluted, and high content of unknown base reads. Next, clean reads were mapped to the reference bacterial genome (24) using HISAT2 (Hierarchical Indexing for Spliced Alignment of Transcripts) (44). The clean reads were mapped to reference transcripts using Bowtie2 (45), and the gene expression level for each sample was calculated with RSEM (46). RNA-Seq data were analyzed using VisomX version 0.0.0.9. Initially, missing values were filtered to remove genes that were not detected in every replicate of at least one condition. Furthermore, only genes with a count of ≥ 5 in at least three samples were retained. Differential expression analysis of log 2 -transformed read counts was performed with the function DESeq (with sfType = "ratio") of the DESeq2 package version 1.36.0 (47). P-values were adjusted by independent hypothesis weighting using the IHW package (48); log 2 (fold change) (FC) shrinkage was performed through the "apeglm" method of the apeglm package (49). Genes showing absolute log 2 (FC) values of ≥ 2 and adjusted P-values (q-values) of ≤ 0.01 were identified as differentially expressed.

Statistical analysis
Data analysis was performed using MS Excel, Prism 9.0.2 (GraphPad Software Inc., San Diego, CA, USA) and R studio (Integrated Development for R; RStudio, PBC, Boston, MA, USA) unless differently specified. Reported values are indicated as averages ± standard deviation of replicates as specified in the legend of the corresponding figures. The level of statistical significance of differences when comparing results across experiments and conditions was evaluated by ANOVA (Dunnett's test, Prism 9.0.2).

Native mechanisms of formate oxidation and tolerance in P. putida and other Gram-negative bacteria
Formic acid is a weak acid (pK a = 3.74), and its toxicity is associated with the diffusion across the cell membrane at low pH, thus acidifying the cytoplasm and impairing the proton motive force (50). Formate can also lead to oxidative stress and inhibit the activity of bacterial cytochrome c oxidases (51,52). To overcome the toxicity brought about by this C1 acid, microorganisms have evolved different dissimilation mechanisms. The growth of microbes with weak FDH activity, e.g., E. coli, is impaired at formate concentrations below 100 mM (53). Bacteria bearing highly active FDHs, in contrast, can tolerate several hundred mM formate (54). In a previous study, we showed that P. putida could tolerate formate at least at 240 mM in LB medium-suggesting high levels of endogenous FDH activity (36). P. putida possesses two bona fide FDHs (31, 32), i.e., FdoGHI-FdhE (PP_0489-PP_0492) and FmdEFGH (PP_2183-PP_2186). The fdoGHI-fdhE cluster encodes a membrane-bound FDH, possibly using quinol as a cofactor, while FmdEFGH is annotated to be a soluble, NAD + -dependent FDH. Both FDHs are predicted to contain iron-sulfur clusters and molybdenum-binding domains. Metal-containing FDHs exhibit enhanced kinetic properties toward formate oxidation than similar dehydrogenases, and they can also catalyze the reduction of CO 2 to formate (55, 56). In vitro enzymatic assays highlighted that P. putida KT2440 metabolizes formate mostly through the soluble, NAD + -dependent FDH (31). In order to identify the role and contribution of the different FDHs of P. putida EM42 toward formate oxidation, we constructed several mutants where fdoGHI-fdhE (strain ∆FDH1) and fmdEFGH (strain ∆FDH2) were deleted, either independently or jointly (strain ∆∆FDH; Table 1). The growth profile of the resulting mutants was analyzed in DBM medium in the presence of 20 mM glucose as the main carbon source and increasing formate concentrations (Fig. 1a). In the absence of formate, P. putida EM42 grew with a specific growth rate (μ) of 0.75 h -1 . A protracted lag phase and reduced μ were observed with rising formate concentrations-but the final OD 600 was not severely affected. The μ of wild-type strain EM42 decreased to 0.35 h -1 in cultures with 360 mM formate (Fig.  1b). A similar behavior was observed for the single and double FDH mutants, displaying a μ of ca. 0.37 h -1 with 360 mM formate (Fig. 1b). Since a reduced formate tolerance was expected upon deleting dissimilatory FDHs, we wondered whether FdoGHI-FdhE and FmdEFG are active and involved in formate oxidation under the conditions tested. E. coli cultures were included as a control in these experiments. We previously reported on the growth defects when E. coli was grown in rich LB medium supplemented with >120 mM formate (36). The same qualitative pattern was observed here when E. coli MG1655 was incubated in M9 minimal medium supplemented with 20 mM glucose and formate (Fig.  1). Bacterial growth was already impacted by formate at 120 mM, decreasing the final OD 600 by ca. two-thirds. Likewise, no growth was observed in the presence of formate concentrations >180 mM (Fig. 1a).
Next, P. putida EM42 and its double ∆∆FDH mutant derivative were cultivated in DBM medium with 20 mM glucose in shaken-flask cultures, with or without 240 mM formate, and changes in the substrates concentration were measured over time (Fig. 2c). Glucose consumption mirrored bacterial growth, with a specific glucose uptake rate (q G ) of 5.2 ± 0.1 mmol g of cell dry weight (CDW) -1 h -1 in the absence of formate and ca. 2.6 ± 0.1 mmol g CDW -1 h -1 in cultures added with formate ( Fig. 2c). Bacterial growth slowed down upon formate supplementation (at 6 h of cultivation, corresponding to an OD 600 of ca. 0.6-0.7) and resumed only after 2 h (Fig. 2c). This observation indicates that a sudden formate challenge results in an acute toxicity effect that stalls bacterial growth while detoxification mechanisms are being deployed. Feeding formate to cultures of P. putida KT2440 as an additional energy source has been shown to increase biomass yields (Y X/S ) (31). A similar behavior was observed in this study, with an increase in Y X/S of ca. 8.5% (on a glucose basis) upon supplementing formate to cultures of P. putida EM42. Glucose and formate were co-consumed (Fig. 2c). Surprisingly, the pattern and rate of formate uptake (q F ) were similar in all cultures (with q F = 37.7 ± 2.7 and 39.1 ± 2.4 mmol g CDW -1 h -1 for the wild-type EM42 strain and the ∆∆FDH mutant, respectively). With these results at hand and considering that no differences in formate tolerance or oxidation were observed upon deleting fdoGHI-fdhE and/or fmdEFGH, we explored the possibility Growth and specific consumption rates were calculated during exponential growth; a non-inoculated culture (labeled as "Medium") was included as a control. Average values ± standard deviation for the specific consumption rate of glucose (q G ) and formate (q F ), as well as specific growth rates, are indicated in the figure. CDW, cell dry weight; n.a., not applicable. that additional dehydrogenases might be present in P. putida. Hence, we set to study its genome-wide transcriptome to identify putative FDH candidates as explained below.

High-resolution RNA sequencing reveals novel routes of formate oxidation
Shaken-flask cultures of P. putida EM42 and P. putida ∆∆FDH (in DBM medium with 20 mM glucose and with or without 240 mM formate) were harvested in mid-exponential phase (corresponding to 3 h for the glucose-only condition and 4 h for cultures added with formate) and processed for RNA-Seq analysis. A concentration of 240 mM formate was selected for these experiments as E. coli was not able to grow under this condition, while P. putida had a noticeable growth defect (Fig. 1a). Thus, we hypothesized that the effects exerted by formate toxicity on gene regulation would be evidenced in this culture setup. Changes in transcriptional levels could be reproducibly quantified for 5,158 genes of P. putida ( Fig. 2a-c). Principal component analysis (PCA) of the total transcriptome data underscored high reproducibility across biological triplicate samples and illustrated different clustering patterns depending on medium composition (Fig. S1). When comparing the global transcriptome of P. putida EM42 grown on glucose versus glucose cultures supplemented with formate, 142 DEGs could be identified-while 143 DEGs were recognized when the transcriptome of P. putida ∆∆FDH growing in the presence of formate and glucose was compared to the control, glucose-only experiment (Fig. 2c). Cultivating either the wild-type or the ∆∆FDH mutant in the presence of formate triggered transcriptional changes in 133 genes that were shared by both P. putida strains. Then, we compared the fold change (FC) of these DEGs to the transcriptional levels observed in the wild type growing on glucose and grouped them into genes (i) upregulated or (ii) downregulated in the presence of formate ( Fig. 2a; Table S1). As expected, a major transcriptional difference between P. putida EM42 and its ∆∆FDH derivative grown in the presence of formate lies in the expression of fdoGHI-fdhE and fmdEFGH ( Fig. 2b)-as these genes were deleted in the double mutant strain ( Table 1). The majority of the other open reading frames (ORFs) were differentially expressed to a similar extent as compared to P. putida EM42 cultivated with glucose ( Fig. 2a-c; Table S2). In fact, and as indicated above, ca. 93% of the identified DEGs overlapped between the two strains-highlighting a strong similarity in the transcriptional response of P. putida to formate challenges, regardless of the presence of the two (supposedly major) FDHs ( Fig. 2c; Fig. S1). Quite surprisingly, neither fdoGHI-fdhE nor fmdEFGH was significantly upregulated when P. putida EM42 was cultivated in the presence of 240 mM formate (Fig. 2b). This observation could be indicative of either (i) high levels of constitutive expression for the fdo and fmd genes, (ii) post-transcriptional regulation of the cognate proteins, or (iii) a relatively minor role of these FDHs in formate processing. This result prompts the question of whether other dehydrogenases are involved in formate dissimilation.
PP_4596 and PP_0256, annotated to encode two soluble oxidoreductases that belong to the broad FdhF/YdeP family, were significantly overexpressed in both P. putida EM42 and strain ∆∆FDH upon exposure to the C1 acid (Fig. 2a). FdhF/YdeP oxidoreductases usually contain a molybdopterin cofactor and are associated with other dehydrogenase subunits featuring iron-sulfur clusters (57). The polypeptides encoded by PP_4596 and PP_0256 share little similarity to known FDHs of P. putida, Methylobacterium extorquens, E. coli, or Cupriavidus necator [20 to 35% identity, according to BLAST analysis (58)]. However, PP_4596 and PP_0256 display a significant degree of amino acid identity with Fdh4A from M. extorquens (59%; Fig. S2) and the YdeP protein of E. coli (47% and 49% for PP_4596 and PP_0256, respectively; Fig. S3). Fdh4A was demonstrated to be involved in formate oxidation (59), whereas YdeP was shown to be connected to acid stress responses (60). Interestingly, PP_4596 and PP_0256 also exhibit 1% amino acid identity to each other (Fig. S4), hinting at similar functionality of the polypeptides. The gene encoding FdhD (PP_0257), a FDH accessory sulfurtransferase, was also highly overexpressed in the presence of formate (33). The sequence coverage plot of these loci at the single-nucleotide resolution suggests that PP_4596 constitutes a separate, single transcriptional unit, whereas fdhD and PP_0256 are part of the same mRNA transcript (Fig. 2d). FdhD is a carrier protein that transfers sulfur to the molybdenum cofactor prior to its insertion into formate dehydrogenases. FdhD activity is induced upon cultivation of P. putida in the presence of sublethal formaldehyde concentrations (61). Taken together, the results described thus far suggest that FdoGHI-FdhE and FmdEFGH are not the primary FDHs; the experiments also indicate that the PP_4596 and PP_0256 oxidoreductases are actively involved in formate oxidation in P. putida.

Exposure of P. putida to formate triggers network-wide redox balancing mechanisms
In addition to genes encoding putative FDHs, moaA and moaB-II, which encode the cyclic pyranopterin monophosphate synthase (PP_4597) and the molybdenum cofactor biosynthesis protein B (PP_4600), were highly expressed in the presence of formate [log 2 (FC) = 2.4 and 4.3 for moaA and moaB-II, respectively; Fig. 2e]. Overexpression of genes within the molybdopterin biosynthesis pathway positively correlates with the activation of molybdenum-dependent FDHs. This transcriptional pattern was shared with several genes of the ped cluster that encodes PQQ-dependent alcohol dehydrogenases (Fig. 2e), known to be active when P. putida is cultivated in the presence of alcohols (25). PedE and PedH act on an extensive range of alcohols and aldehydes with a strong dependence on PedF, a cytochrome c oxidase that regenerates the PQQ cofactor (27). All of these genes were overexpressed in strain EM42 cultivated in the presence of formate [log 2 (FC) = 2.6 for pedE, 5.1 for pedH, and 5.7 for pedF; Fig. 2e]. Genes encoding respiratory components were also differentially regulated in the presence of formate. P. putida harbors five terminal oxidases, i.e., Cio, Cyo (a bo3-type oxidase), and the cytochromes aa3, cbb3-1, and cbb3-2 oxidases (24). These terminal oxidases display differential redox potentials, regulatory patterns, O 2 affinity, and H + pumping efficiency (62). The cioA and cioB genes, encoding the cyanide-insensitive ubiquinol oxidase subunits I and II (PP_4651 and PP_4650), and members of the cytochrome bd family were among the most significantly differentially expressed ( Fig.  2a and e). Cio receives electrons from the ubiquinol pool in the periplasmic space and is involved in oxidative and nitrosative stress defense (63). In E. coli, cytochrome bd expression and content increase at low O 2 availability (64), medium alkalization (65), high temperature (66), and in the presence of cyanide (67). Hence, the pH increase triggered by formate addition could correlate with cioAB overexpression in P. putida. Conversely, the expression of genes encoding NuoABCEFGHIJKLMN, the NADH-quinone oxidoreductase component of the electron transport chain (complex I; PP_4119-PP_4131), was significantly downregulated in the presence of formate [log 2 (FC) < -2; Fig. 2e]. Fine-tuning adjustments of the respiratory chain activity in response to the C1 acid remain to be fully elucidated-but these observations point to a differential pattern of redox balancing. Changes in the expression of genes involved in intracellular redox homeostasis support this notion. Indeed, the membrane-bound, proton-pumping pyridine nucleotide transhydrogenase genes (pntAA, PP_0156, pntAB, PP_5747 and pntB, PP_0155) were overexpressed in formate-challenged P. putida cells (Fig. 2e). In E. coli, PntAB catalyzes the reduction of NADP + by NADH, regenerating NADPH while transferring H + inside the cell (68). While NADH is linked to ATP generation via oxidative phosphorylation, NADPH is mainly used as a cofactor for anabolism and as a reductant in oxidative stress responses (69). Our results indicate that formate addition activates the expression of NAD + -dependent FDHs in P. putida, thereby affecting redox homeostasis; pntAB overexpression hints at redox balancing mechanisms that ultimately yield NADPH via transhydrogenation (70).

Functional validation of PP_04596 and PP_0256 as major FDHs of P. putida
The overexpression of genes encoding PP_4596, PP_0256, and FdhD in the presence of formate (Fig. 2a) suggested a role for these oxidoreductases in processing the C1 acid. To further investigate this possibility, the PP_4596 and PP_0256-fdhD genes were independently deleted in P. putida EM42 to yield strains ∆PP_4596 and ∆PP_0256-0257, respectively (Table 1). Next, the growth profile of these mutant strains was analyzed in DBM medium cultures supplemented with 20 mM glucose and increasing formate concentrations (Fig.  3A). The deletion of PP_0256-0257 drastically affected formate tolerance, with growth deficiency starting at 120 mM and no growth observed above 240 mM formate (Fig. 3a). The deletion of PP_4596, on the other hand, had a milder effect on formate tolerance by P. putida-an increase in the lag phase was evident starting from 240 mM, and no growth was observed at >300 mM formate (Fig. 3a). Indeed, when cultivated with 240 mM formate, both mutant strains had a lag phase longer than 5.5 h (whereas the wild-type strain had a ca. 1.8 h lag time under the same conditions; Fig. 3b). The next step was combining all four deletions in a single mutant strain in order to explore the synergies of the FDHs in the formate tolerance phenotype of P. putida. To this end, strain ∆fmdEFGH ∆fdoGHI-fdhE ∆PP_0256-0257 ∆PP_4596 (termed EM42 ∆∆∆∆FDH for the sake of simplicity) was constructed (Table 1), and its growth phenotype was tested in DBM medium with different formate concentrations (Fig. 3c). As expected, eliminating all four FDHs of P. putida resulted in a severe effect in formate tolerance, rather similar to that observed for the single PP_0256-0257 deletion (Fig. 3a). Indeed, the quadruple mutant could barely grow in the presence of 120 mM formate, and no growth was observed at >180 mM (unlike the parental strain EM42)-as reflected in the specific growth rates calculated under these conditions (Fig. 3d).
Since a similar growth defect was observed for P. putida ∆PP_0256-0257 and ∆∆∆∆FDH when exposed to the C1 acid, complementation assays were set by cloning PP_0256 and its neighboring PP_0257 (fdhD) gene as a single transcriptional unit in plasmid pS621·PP_0256-57 (Fig. 3e). In this plasmid, both genes are placed under the control of the constitutive P trc promoter ( Table 2). Either the empty vector (p621c, used as a control) or plasmid pS621·PP_0256-57 was introduced in different FDH-depleted strains, and the growth profiles of the resulting recombinants were analyzed in DBM medium supplemented with 20 mM glucose and increasing formate concentrations (Fig. 3f). Whereas P. putida ∆PP_0256-57 carrying the empty p621c vector had a formate sensibility similar to that observed for the plasmid-less strain (Fig. 3c), plasmid pS621·PP_0256-57 reverted the growth deficiency of the mutant virtually to the same levels observed for the wild-type strain (Fig. 3f). The μ values for strain EM42 and its ∆PP_0256-57 derivative carrying plasmid pS621·PP_0256-57 were practically the same in cultures containing ≤ 240 mM formate (Fig. 3g). Furthermore, the complementation effect was verified in strain ∆∆∆∆FDH (Fig. 3f), where expression of PP_0256-57 restored growth in the presence of the C1 acid up to 240 mM. Overall, these results underscore the native ability of P. putida to withstand high formate concentrations and expose a hitherto unknown role for PP_4596 and PP_0256 as FDHs. We were also interested in exploring native mechanisms for processing the metabolites upstream formate, i.e., methanol and formaldehyde, as explained in the next section.
Native methanol tolerance mechanisms in P. putida are linked to efficient routes for alcohol oxidation Methanol, a sustainable, easy-to-produce and naturally abundant one-carbon compound, has been studied as a substrate for biotechnology (71). Yet, methanol is a bacterial cytotoxic (72), and this effect is tied to either (i) alcohol intercalation between fatty acids, altering the fluidity and functioning of cell membranes (73,74) or (ii) methanol oxidation to formaldehyde, which is highly reactive toward proteins and nucleic acids (75). Thus, microbial methanol tolerance comes from either limited alcohol oxidation or high dissimilatory formaldehyde activity to formate (and ultimately CO 2 ). While P. putida KT2440 has been reported to tolerate up to 1.5 mM formaldehyde (61), no data on methanol tolerance is available. Hence, we determined the intrinsic methanol tolerance levels of P. putida EM42 in microtiter plate cultivations, using E. coli MG1655 as a control (Fig. 4). E. coli is known to grow in the presence of up to 10% (v/v) [ca. 2.5 M] methanol (76). Surprisingly, P. putida EM42 featured an apparently lower tolerance to methanol in our experiments when compared to E. coli. Strain EM42 tolerated methanol up to 1.35 M (although the final OD 600 under these conditions was only one-fifth of that in control cultures), whereas E. coli was only affected by >3 M methanol (Fig. 4a). The E. coli growth profiles were noisier than those of P. putida-compounded by the high In order to better quantify physiological responses to the alcohol challenge, shakenflask experiments with P. putida EM42 were carried out in DBM medium with 20 mM glucose, with or without methanol addition at 1 M-and the concentration of substrates and potential oxidation products (e.g., formate) was measured over time (Fig. 4b).
Glucose consumption increased by 20% in the presence of methanol (q G = 6.2 ± 0.1 mmol g CDW -1 h -1 , Fig. 4b) accounted for the extra energy obtained from methanol oxidation in addition to the standard, PQQ-dependent oxidation of glucose to gluconate in the periplasm (2). Consistent with our previous observations (Fig. 4a), the growth rate of strain EM42 was diminished by ca. 40% in methanol cultivations (Fig. 4b). We could not observe any significant reduction in methanol concentration over time, yet formate was secreted at q F = 0.57 ± 0.03 mmol g CDW -1 h -1 ( Fig. 4b; Fig. S5)-with an increase in the acid concentration particularly noticeable by the end of the cultivation (8 h onward). This experimental evidence supports the hypothesis that formaldehyde is produced from methanol, and the aldehyde is further detoxified via oxidation to formate. To gain an insight into the regulatory and metabolic mechanisms underlying this a non-inoculated culture (labeled as "Medium") was included in the analysis as control. Average values ± standard deviation for the specific consumption rate of glucose (q G ) and secretion rate of formate (q F ), as well as specific growth rates, were calculated during exponential growth and are indicated in the figure. CDW, cell dry weight; n.a., not applicable.

Research Article mSystems
methanol-sensitive phenotype, we resorted to genome-wide transcriptional fingerprinting as detailed in the next section.
High-resolution RNA sequencing in P. putida exposed to methanol In order to identify key players involved in methanol and formaldehyde oxidation in P. putida EM42, cells were cultivated in DBM medium containing 20 mM glucose, with or Research Article mSystems without 1 M methanol, and harvested during mid-exponential growth (5 h for cultures added with methanol) for RNA-Seq analysis (Fig. 5). Significantly DEGs between the glucose and methanol versus glucose-only conditions accounted for 116 ORFs [q-value (false discovery rate, FDR) ≤ 0.01 and |log 2 (FC)| > 2; Fig. 5a and b]. Among these DEGs, 42 genes (ca. 36%) were also present in the glucose and formate versus glucose comparison (Fig. 2c). Similarities in the transcriptional landscapes in formate and methanol cultivations could also be observed in a PCA of the total data. Here, the methanol and formate conditions clustered together in PC2 (with 30.1% variance) compared to the glucose-only control (Fig. S1). Indeed, the similarity between these two conditions hints at a mechanistic overlap among the detoxification strategies deployed for either C1 compound-possibly because these are intermediates within the same oxidative pathway(s). Transcriptional levels of DEGs in cells grown with methanol were compared to those in control conditions (Fig. 5a-c), testing the patterns both for their biological significance and association of the cognate functions with methanol and/or formaldehyde oxidation (Fig. 5d).
As P. putida can endure relatively high formaldehyde concentrations, the direct oxidative product of methanol, aldehyde detoxification systems are expected to be active upon cultivation of the cells in the presence of the C1 alcohol. Formaldehyde detoxification involves the concerted action of formaldehyde dehydrogenases and FDHs that eventually yield CO 2 (33,79). P. putida KT2440 harbors two NAD + -binding, glutathione-independent formaldehyde dehydrogenases (32), encoded by fdhA (PP_0328) and fdhB (PP_3970, also known as ybdR). A glutathione-dependent system, encoded by frmAC (PP_1616-PP_1617), is likewise present in this bacterium (61). The frmAB operon encodes the main formaldehyde detoxification system of E. coli (80)(81)(82), yet the role of the corresponding genes in P. putida remains to be studied. The FrmA and FrmC polypeptides of P. putida KT2440 are 72% and 53% identical to the E. coli K12 homologues, respectively. We initially assessed whether the detoxification systems mentioned above were differentially expressed in cells exposed to methanol compared to formate (with both conditions initially contrasted to glucose-only cultures). None of the formaldehyde dehydrogenase genes was differentially expressed in any of the C1 conditions as compared to glucose cultures (Fig. 5b and c). However, two aldehyde dehydrogenase genes, aldB-II [or pedI, PP_2680; log 2 (FC) = 9.6] and PP_2589 [log 2 (FC) = 2.9], were significantly overexpressed upon methanol treatment. The nonspecific, NAD + -dependent dehydrogenase encoded by aldB-I (PP_0545), upregulated in response to ethylene glycol and butanol treatment (83), was only increased under formate conditions and slightly downregulated with methanol (Fig. 5c). These results suggest that oxidative (detoxification) activities on formaldehyde may be associated to constitutive expression of the cognate gene(s) (81)-also indicating that some of the dehydrogenase activities of P. putida can act on different C1 substrates.

Multiple alcohol dehydrogenases and related functions are transcriptionally activated upon methanol treatment
We observed a considerable upregulation of the ped cluster (comprising PP_2662 to PP_2683), the pqq biosynthetic operon (encoded by the PP_0375-PP_0379 genes), as well as PP_2736-PP_2738, encoding a putative short-chain dehydrogenase, in cells cultivated in the presence of methanol ( Fig. 5d; Table S2). The ped cluster encodes PedE and PedH (27), two PQQ-dependent alcohol dehydrogenases with a range (83). PedH has been identified as a lanthanide-dependent enzyme (84). YiaY, a third NAD + -dependent dehydrogenase recently proposed to act as a transcriptional regulator (26), is encoded within this cluster-together with the previously mentioned formaldehyde dehydrogenase gene aldB-II (or pedI). The high overexpression of pedE and pedH in the methanol condition [log 2 (FC) = 6.3 and 7.4, respectively] and the broad substrate specificity of the cognate dehydrogenases suggest a relevant role in processing the C1 alcohol. Since PedE and PedI are PQQ-dependent enzymes, it does not come as a surprise that the pqqFABCDEG operon was significantly upregulated in the methanol condition (Fig. 5d), with log 2 (FC) > 4. The PQQ biosynthetic pathway is required by multiple PQQ-dependent oxidases, e.g., Gcd (PP_1444), involved in the initial steps of glucose processing (85). Moreover, we analyzed different putative or characterized alcohol dehydrogenase genes, the products which could potentially act on methanol. PP_3839 (AdhP), for instance, has been previously characterized to act on short-chain alcohols (86). In our experiments, the expression of adhP was slightly downregulated upon methanol treatmentperhaps as a mechanism to prevent excessive formaldehyde accumulation. The putative short-chain alcohol dehydrogenases encoded by PP_2736 and PP_2737, in contrast, were significantly overexpressed on methanol (Fig. 5d)-unlike yqhD and adhB, which were largely unaffected under the same conditions. Taken together, these results showcase a relatively small subset of alcohol dehydrogenase genes induced in P. putida by methanol treatment. Yet, other more general detoxification mechanisms were activated in response to the C1 alcohol, as detailed below.

General detoxification mechanisms may help P. putida to cope with high methanol concentrations
We observed that formaldehyde toxicity could be partially counteracted through its oxidation to secretion (Fig. 4b), yet active aldehyde export may also contribute to detoxification. We detected overexpression of mdtABC/ompB (PP_3582-PP_3585), encoding a putative RND (resistance-nodulation-division superfamily) efflux transporter (87,88). In contrast, the mexEF/oprN operon was not differentially regulated in our dataset, but a putative oxalate/formate antiporter gene (PP_0288) was significantly overexpressed in the methanol condition (Table S2). Also, the cbb3-type cytochrome oxidase encoded by the cco-I operon (PP_4255-PP_4258) was overexpressed upon methanol treatment. Such oxidases exhibit a high O 2 affinity that allows for growth under O 2 -deprivation (89), and they are often involved in general defense mechanisms in several bacterial species (90). Paired with these highly transcribed DEGs, 47 ORFs were significantly downregulated in the methanol condition. Among them, some genes potentially involved in C1 metabolism [e.g., sahR, metE, ahcY, and metF; with log 2 (FC) < -2.9] and several uncharacterized genes, including the dapF diaminopimelate epimerase gene and the aspC aminotransferase gene [log 2 (FC) < -9.4, Table S2]. While the involvement of the cognate functions in the tolerance of P. putida to C1 compounds may not seem immediately evident, the transcriptional landscape discussed herein indicates a large rearrangement of gene expression patterns that involves both specific and general mechanisms. Hence, the genetic dissection of the main enzymatic activities that could process methanol and formaldehyde, as hinted in the RNA-Seq analysis, was the next step in our study.

Redundant formaldehyde oxidation activities in P. putida
Formaldehyde detoxification in P. putida is assumed to occur in a concerted manner, given the large number of aldehyde oxidases encoded in the genome-yet these features remained thus far understudied. The enzymes encoded by fdhA and its fdhB homologue are involved in aldehyde detoxification (32), but glutathione-independent dehydrogenases have not been identified in P. putida. Thus, we delved into the action of FrmAC and AldB-II, considering that the cognate genes were highly overexpressed in cells grown in the presence of methanol (Fig. 5) and the interaction of these activities with other aldehyde oxidases. Single-deletion mutants for frmAC, aldB-II, fdhA, and fdhB were constructed in the P. putida EM42 background, together with a quadruple mutant harboring all these knockouts (i.e., EM42 ∆frmAC ∆aldB-II ∆∆fdhAB, Table 1). Next, the growth profile of the mutants was compared to that of the wild-type EM42 strain in DBM medium supplemented with 20 mM glucose and increasing methanol concentrations (Fig. 6). P. putida ∆frmAC showed the starkest growth effect of all single-deletion mutants, already evident at 250 mM methanol, the lowest concentration tested (Fig. 6a). The growth rate of this strain was diminished by ca. 20% at 500 mM methanol (Fig.  6b). At higher alcohol concentrations (e.g., 1 M), all single-deletion mutants were more affected than the wild-type strain. Furthermore, the most severe growth impairment was observed in cultures of the ∆frmAC ∆aldB-II ∆∆fdhAB mutant, with a ca. 40% reduction in μ at 250 mM methanol when compared to P. putida EM42 (Fig. 6b). These results showcase synergistic effect of individual formaldehyde dehydrogenases of P. putida in aldehyde detoxification, with P. putida ∆frmAC as the most affected strain among all single mutants even at a low methanol concentration.
Broad-substrate-range alcohol dehydrogenases mediate methanol oxidation in P. putida P. putida KT2440 is able to oxidize a range of short-chain alcohols (91), e.g., 1-butanol (92), 1,4-butanediol (93), ethanol (26), and ethylene glycol (83,94,95), through the action of broad-substrate-range dehydrogenases. These activities underline the native P. putida preparedness to withstand toxic alcohols and aldehydes, regularly present in the natural environments where this bacterium thrives. The transcriptome of P. putida EM42 exposed to methanol indicated that the dehydrogenases encoded by pedE, pedH, and adhP (as well as the putative transcriptional regulator encoded by yiaY) are involved in alcohol oxidation. Hence, we constructed single-mutant strains of the above mentioned genes in strain EM42, as well as a quadruple mutant harboring the combined ∆yiaY ∆∆pedEH ∆adhP deletions (Table 1). Again, the growth profile of the deletion mutants was compared to that of the wild-type EM42 in DBM medium supplemented with 20 mM glucose and 1.5 M methanol (Fig. 6c), the highest alcohol concentration tolerated by all of these strains. Interestingly, all single mutants had a shorter lag phase compared to that of P. putida EM42, especially the ∆pedE strain (Fig. 6d). The quadruple ∆yiaY ∆∆pedEH Research Article mSystems ∆adhP knockout strain had a lag phase shorter than that of any of the single mutants (Fig. 6d), a feature that was repeated across a range of methanol concentrations (Fig. 6e). The shorter λ times were paired with increased μ values compared to P. putida EM42 (ca. 38% higher in the presence of 2 M methanol; Fig. 6f). Hence, the mutant strain depleted of all alcohol dehydrogenase activities tolerated a methanol challenge much better than the parental strain-confirming the hypothesis that synergistic oxidation activities of P. putida yield toxic formaldehyde.

DISCUSSION
In this work, a genome-wide transcriptome analysis provided mechanistic insights into the metabolism of methanol, formaldehyde, and formate in P. putida. Long thought to be non-native substrates (8), these C1 molecules are processed and interconverted by a number of specific and peripheral activities in P. putida of formate, the most oxidized form among these soluble C1 substrates, has been assumed to occur via the two annotated FDHs encoded by fdoGHI-fdhE and fmdEFGH (31,32). Our results show that, in reality, Fdo and Fmd are not the primary FDHs of P. putida. Upon deleting each FDH gene, either independently or jointly, similarly high q F values were observed for all mutants in cultures containing 240 mM formate-suggesting that alternative dehydrogenases are involved in formate detoxification. Indeed, omic-guided transcriptional screening revealed that fdoGHI-fdhE and fmdEFGH were not significantly expressed under these conditions, whereas two oxidoreductase genes and fdhD were transcriptionally active in both the wild-type strain and the ΔfdoGHI-fdhE ΔfmdEFGH strain. Deletion of PP_4596 alone or in combination with fdhD negatively affected growth in the presence of formate, providing evidence that these activities are involved in processing the C1 acid. Interestingly, PP_4596 and PP_0256 share low identity with other α-subunit components of known FDHs-yet they display high amino acid identity with Fdh4A of M. extorquens. Fdh4A is involved in formate oxidation (59), and overexpression of the cognate gene was observed when the cells were cultivated on methanol in the presence of 2 µM La 3+ (96). In M. extorquens, fdhB is located right next to fdhA, and it was reported to be functionally coupled to Fdh4A. PP_0257, in contrast, does not share any amino acid identity with Fdh4B, and no other subunit or accessory protein was identified in relation to PP_4596-suggesting a divergent evolutionary trajectory for the P. putida enzymes.
Mutants of E. coli and Salmonella enterica, lacking either fdhD or fdhE, display limited FDH activity (97,98). However, almost identical 14 CO 2 evolution levels have been reported for P. putida ΔfdhD compared to the parental strain when fed with 10 mM 14 C-formate or 14  Although the in vivo cofactor dependence of these soluble dehydrogenases remains to be explored, increased biomass yields and pntAB overexpression triggered by formate addition suggest that NAD + -dependent oxidation mechanisms prevail in P. putida, an NADH excess that can either be used for energy conservation or channeled into anabolism via transhydrogenation-a general strategy for redox balancing (70). Formate is also the terminal C1 molecule in an oxidation sequence that starts with methanol and its direct product, formaldehyde. Hence, we also delved into the tolerance, regulation, and metabolism of P. putida exposed to these compounds. The genome of P. putida KT2440 (24) encodes 16 alcohol dehydrogenases (over half of them annotated as putative activities) as well as more than 43 aldehyde dehydrogenases (out of which three are properly characterized formaldehyde dehydrogenases). Though the in vivo ability of these oxidases to act on methanol or formaldehyde remains unknown, their presence underlies the native robustness of P. putida toward a range of toxic compoundsharnessing some of them as potential energy or carbon sources (91, 100). Our findings revealed a concerted action of both alcohol and formaldehyde dehydrogenases. Alcohol dehydrogenases mediate formaldehyde formation, decreasing methanol tolerance of P. putida to the C1 alcohol as compared to other organisms-e.g., E. coli or Bacillus cereus. P. putida KT2440 harbors numerous broad-substrate-range, short-chain alcohol dehydrogenases, e.g., AdhP and the PQQ-dependent enzymes encoded by pedE and pedH. PedE and PedH are highly active toward ethanol and long-chain-length alcohols, but the specific activity on methanol is relatively low at < 0.80 U mg −1 (27). YiaY has been assumed to encode a NAD + -dependent methanol dehydrogenase, and overexpressing the cognate gene resulted in increased methanol consumption (28). However, YiaY was recently hypothesized to function as the regulatory partner of a two-component sensor system, given its sequence similarity to the well-characterized ercA gene of P. aeruginosa PAO1 (26). Hence, YiaY would sense methanol through its oxidation, generating (IM), to be oxidized in the cytoplasm by other broad-substrate-range dehydrogenases, e.g., the NAD(P) + -dependent AdhP or YiaY proteins. The latter may also induce a signal transduction cascade to transcriptionally activate the ped cluster. Formaldehyde, in turn, can be oxidized to formate through different pathways, e.g., the glutathione-dependent (FrmAC) and the thiol-independent (FdhA, FdhB, and AldB-II) mechanisms or the tetrahydrofolate methylation loop. Finally, formate may be fully oxidized to CO 2 via the NAD(P) + -dependent PP_0256-0257 and PP_4596 activities reported in this work or, to a lesser extent, via FdoGHI-FdhE and FmdEFGH. The transport mechanisms through bacterial membranes for one-carbon molecules are yet unknown (indicated with a "?" symbol). Electrons harvested by quinones (Q) or PQQ may directly be channeled into the electron transport chain (ETC) for energy conservation. sp.,

spontaneous.
Research Article mSystems a signal that activates the putative histidine kinase PP_2683. Indeed, deleting yiaY in P. putida lowered pedH and aldB-II expression upon an ethanol challenge (26). In line with these interpretations, we have observed reduced methanol toxicity for P. putida ∆yiaY, likely due to low expression of the abovementioned genes and leading to limited formaldehyde formation. Furthermore, the combined deletion of ∆yiaY ∆∆pedEH and ∆adhP resulted in the highest methanol resistance and growth rates compared to the parental EM42 strain across a range of alcohol concentrations. Formaldehyde detoxification systems may be constitutive or inducible (101). In P. putida KT2440, fdhB (PP_3970) has been proposed to be constitutively expressed (61, 79), yet the frmAB operon of E. coli is induced by the FrmR regulator (80,102). We observed a similarly constitutive expression pattern for fdhAB in our study, whereas aldB-II represented the highest upregulated gene of the whole dataset. Interestingly, the expression of the frmAC operon was not induced significantly as in E. coli. The regulator to frmAC in P. putida (PP_1615) has no significant homology to FrmR of E. coli K12-but shares >80% amino acid homology with GfnR (PA_3630) of P. aeruginosa. The GfnR protein has been characterized to be activated upon sarcosine (N-methylglycine) supplementation, thereby upregulating the glutathione-dependent detoxification system encoded in the adhC/PA_3628 operon (102,103). Sarcosine is converted into glycine and formaldehyde via oxidative demethylation catalyzed by sarcosine oxidase (104). FrmA and FrmC have an 86% and 77% amino acid similarity to their P. aeruginosa orthologues, respectively. As we observed a severe growth defect for the ∆frmAC mutant of P. putida EM42 even at the lowest methanol concentration, we hypothesize that the FrmAC detoxification system, regulated by PP_1615, may be the most active to deal with low formaldehyde levels. In contrast, the concerted action of multiple enzymes seems to detoxify higher aldehyde concentrations. We also conclude that AldB-II (PedI) may be involved in formaldehyde oxidation at high concentrations, given the growth impairment of the corresponding deletion mutant and the broad substrate specificity of alcohol and aldehyde dehydrogenases encoded in the ped gene cluster. Finally, we note that a C1 transfer pathway, linked to tetrahydrofolate (THF) metabolism and widespread in Nature (105), exists in P. putida KT2440. This route involves the bifunctional FolD enzyme, possessing both methylene-THF dehydrogenase and methenyl-THF cyclohydrolase activities, and the non-reversible formyl-THF hydrolase (PurU) enzyme. Although we cannot rule out that FolD/PurU participate in formaldehyde processing by P. putida, the cognate genes could not be identified in the transcriptome datasets.
Taken together, our results enabled us to propose an updated model for the metabolism of C1 compounds in P. putida (Fig. 7). Besides expanding the current understanding of how microbes process methanol, formaldehyde, and formate, our study showcases the possibility of valorizing C1 feedstocks by engineering the versatile routes native to P. putida. While further research will help elucidate the mechanism of action for the novel activities reported herein, our work serve as a cornerstone toward establishing synthetic methylotrophy in P. putida. Meanwhile, we envisage this soil bacterium to be a favored microbial host for metabolic engineering of C1 assimilation given its natural ability to process methanol and its direct oxidation products, formaldehyde and formate.

ACKNOWLEDGMENTS
The financial support from The Novo Nordisk Foundation through grants NNF20CC0035580, LiFe (NNF18OC0034818), and TARGET (NNF21OC0067996), and the European Union's Horizon 2020 Research and Innovation Programme under grant agreement No. 814418 (SinFonia) to P.I.N. are gratefully acknowledged. J.T. is the recipient of a fellowship from The Novo Nordisk Foundation as part of the Copenhagen Bioscience Ph.D. Programme, supported through grant NNF18CC0033664.

DATA AVAILABILITY STATEMENT
All the raw RNA sequencing data generated in this study are publicly available at the NCBI database with accession number PRJNA955793.

ADDITIONAL FILES
The following material is available online.