Time-course Transcriptome of Parageobacillus thermoglucosidasius DSM 6285 Grown in the Presence of Carbon Monoxide and Air

Parageobacillus thermoglucosidasius is a metabolically versatile, facultatively anaerobic thermophile belonging to the family Bacillaceae. Previous studies have shown that this bacterium harbours co-localised genes coding for a carbon monoxide (CO) dehydrogenase (CODH) and Ni-Fe hydrogenase (Phc) complex and oxidises CO and produces hydrogen (H2) gas via the water-gas shift (WGS) reaction. To elucidate the genetic events culminating in the WGS reaction, P. thermoglucosidasius DSM 6285 was cultivated under an initial gas atmosphere of 50% CO and 50% air and total RNA was extracted at ~8 (aerobic phase), 20 (anaerobic phase), 27 and 44 (early and late hydrogenogenic phases) hours post inoculation. The rRNA-depleted fraction was sequenced using Illumina NextSeq, v2.5, 1x75bp chemistry. Differential expression revealed that at 8 vs.. 20, 20 vs.. 27 and 27 vs.. 44 h post inoculation, 2190, 2118 and 231 transcripts were differentially (FDR < 0.05) expressed. Cluster analysis revealed 26 distinct gene expression trajectories across the four time points. Of these, two similar clusters, showing overexpression at 20 relative to 8 h and depletion at 27 and 44 h, harboured the CODH and Phc transcripts, suggesting possible regulation by O2. The transition between aerobic respiration and anaerobic growth was marked by initial metabolic deterioration, as reflected by up-regulation of transcripts linked to sporulation and down-regulation of transcripts linked to flagellar assembly and metabolism. However, the transcriptome and growth profiles revealed the reversal of this trend during the hydrogenogenic phase.


Introduction
Hydrogen gas (H 2 ) has the highest energy per unit mass of all fuels (heating value of 141.9 MJ/kg) [1]. In contrast to traditional fuels, H 2 is a zero-emission fuel when undergoing complete combustion in the presence of oxygen (2 H 2 + O 2 → 2 H 2 O) [2]. As such H 2 has become prominent as a clean and sustainable alternative energy source [3]. Unlike industrial strategies for H 2 production, such as coal gasification, steam reformation of natural gas and partial oxidation of oil, which require high energy inputs and rely on the use of non-renewable substrates [4], biohydrogen production is less energy intensive and the process results in almost zero waste products [4,5]. Biohydrogen is principally generated via biophotolysis of water by algae, and bacterial photo-fermentation and dark fermentation of organic substances [6]. However, these methods are constrained by both economic and technological P. thermoglucosidasius DSM 6285 was cultivated with an initial gas atmosphere of 50% air and 50% CO. Total RNA was extracted from cells harvested at four time points: 8, 20, 27, and 44 h post-inoculation. These timepoints approximately coincide with the middle of the aerobic growth phase, the early anaerobic phase when O 2 is exhausted, when H 2 is first detected and in the middle of the WGS phase, respectively (Figure 1). A total of~106 million reads (7.9 Gb) were generated using Illumina NextSeq, v2.5, 1x75bp chemistry. The average number of reads ranged from~1.46 × 10 7 (mid-aerobic phase) to~7.4 × 10 6 (mid-WGS). Between 93% and 98% of these reads were trimmed and aligned to 3888 protein coding regions in the draft genome of P. thermoglucosidasius DSM 6285 (Table S1). A multidimensional scaling (MDS) plot, used to assess pattern of gene expression across the four sampling points revealed a clear separation of transcripts between the samples and commonality of expression between biological replicates within the sample groups ( Figure S1). These results suggest that the transcriptome datasets are sufficiently robust to validly assess transcript dynamics across the studied conditions.
Comparisons of the expression profiles at the different time points revealed substantial differences between different phases. Comparison of the transcriptome profiles of P. thermoglucosidasius DSM 6285 at 8 and 20 h post-inoculation showed statistically significant (FDR < 0.05) differential expression of 2190 transcripts, of which~67% and 33% were up-and down-regulated at 20 h, respectively (Figure 2a  Comparisons of the expression profiles at the different time points revealed substantial differences between different phases. Comparison of the transcriptome profiles of P. thermoglucosidasius DSM 6285 at 8 and 20 hours post-inoculation showed statistically significant (FDR < 0.05) differential expression of 2190 transcripts, of which ~67% and 33% were up-and downregulated at 20 hours, respectively (Figure 2a     Comparisons of the expression profiles at the different time points revealed substantial differences between different phases. Comparison of the transcriptome profiles of P. thermoglucosidasius DSM 6285 at 8 and 20 hours post-inoculation showed statistically significant (FDR < 0.05) differential expression of 2190 transcripts, of which ~67% and 33% were up-and downregulated at 20 hours, respectively (Figure 2a

Metabolic and Functional Shifts in Response to Changing Gas Atmosphere
To determine the functional implication of the observed differences in expression profiles during the different growth phases, gene ontology (GO) and KEGG orthologues of the proteome of P. thermoglucosidasius DSM 6285 were predicted based on annotation obtained from UniProt [28] and using BlastKOALA [29]. Approximately 54.8% and 57.6% of the 3888 P. thermoglucosidasius DSM proteins were assigned to GO terms or KEGG categories, respectively.
Evaluation of the up-regulated transcripts at 20 h (relative to 8 h) post-inoculation, revealed significant (FDR < 0.05) enrichment of genes of annotated with 14 GO terms including four biological processes (BP), seven cellular components (CC) and three molecular functions (MF) ( Table S2). By contrast, the down-regulated transcripts are significantly (FDR < 0.05) predominated by GOs linked to 91 BP, 3 CC and 1 MF. Evaluation of the enriched functions showed that among the BPs, GOs of anion transport (GO:0006820), developmental processes (GO:0032502) and sporulation (GO:0043934) led the up-regulated terms while drug metabolic process (GO:0017144) and locomotion (GO:0040011) were among the top down-regulated functions. Additionally, the GOs nucleotide and ATP biosynthesis were also down-regulated at 20 h relative to 8 h. The differentially expressed transcripts were also evaluated for KEGG pathway enrichment. At 20 h relative to 8 h, differentially expressed transcripts were significantly associated with between 14 (q < 0.05) and 18 (p < 0.05) KEGG pathways (Table 1). Genes linked to amino acid metabolism (valine, leucine and isoleucine degradation), carbohydrate metabolism (citrate cycle, glycolysis / gluconeogenesis, pyruvate, propanoate and butanoate metabolism), cell motility (flagellar assembly), energy metabolism (oxidative phosphorylation), biosynthesis of antibiotics and secondary metabolites, metabolism of cofactors and vitamins and translation (ribosomes) were down-regulated while only one category, membrane transport (ABC transporters), was significantly (q < 0.05) up-regulated. In addition to various roles in wide range of cellular processes, including transport of molecules [30], ABC transporters overexpression has been observed in Mycobacterium smegmatis grown under oxygen starvation [31].
Considering the differentially expressed genes at 27 h relative to 20 h, the GO enrichment analysis showed 10 GOs including 3 BPs, 5 CCs and 2 MFs were significantly (FDR < 0.05) up-regulated, with GOs of transport (GO:0006810), spore germination (GO:0009847) and developmental process (GO:0032502) being the top up-regulated processes (Table S2). Conversely, 44 GOs were significantly (FDR < 0.05) depleted, including 28 BPs and 8 CCs and MFs each. Differentially expressed transcripts at 27 h (relative to 20 h) enrich between nine (q < 0.05) and 18 (p < 0.05) KEGG pathways (Table 1). These include 13 of the pathways that have been down-regulated at 20 h, including the TCA cycle, ABC transport, biosynthesis of amino acids, antibiotics, secondary metabolites; porphyrin and chlorophyll metabolism and ribosome synthesis. Additionally, transcripts associated with fatty acid metabolism, carbohydrate (glyoxylate and dicarboxylate metabolism) and energy (methane metabolism) metabolism, folding, sorting and degradation (RNA degradation) as well as translation (aminoacyl-tRNA biosynthesis) were also down-regulated. Based on the KEGG pathway enrichment analysis, however, no specific function was up-regulated at this point. At the final stages of the experiment (44 vs. 27 h), with cells undergoing the WGS reaction, 34 and one GOs were significantly (FDR < 0.05) enriched among the up-and down-regulated transcripts, respectively (Table S2). The most up-regulated function appeared to be translation (GO:0006412), while the alcohol catabolic process (GO:0046164) was down-regulated. Only one KEGG pathway, translation (ribosomes), was significantly (q < 0.05) up-regulated (Table 1).
Overall, the differential expression analysis provides some insight into the physiological adaptation of P. thermoglucosidasius DSM 6285 in response to changing gas atmospheres and shifts in metabolism. The observed up-regulation of sporulation transcripts (Table S2) and down-regulation of flagellar assembly (Table 1) at 20 h suggests a degree of metabolic shut-down in response to energy limitation and build-up of acid in the medium. At the 8 h time point, DSM 6285 attained the maximum OD600 of 0.37 while the medium pH dropped from 6.69 to 6.43 (Figure 1a), indicating the accumulation of acetic acid under aerobic conditions. This phenomenon is characteristic of the 'overflow mechanism' observed when bacteria are grown with excess carbon source as a strategy to fast track acetyl-coA metabolism and maximise ATP synthesis when the key enzymes of the TCA cycle are inhibited [32,33]. Aerobic acetate production has been reported in P. thermoglucosidasius M10EXG [34]. Although only 0.75 g L −1 of glucose was used in the current experiment, the presence of carbon monoxide (CO), which is known to inhibit aerobic respiration via the inactivation of enzymes of the TCA cycle and the respiratory chain [35,36], may have resulted in a backlog of glucose which is channelled to acetate.
Inhibition of the respiratory chains will also result in the excess of reducing equivalents [35,36]. For instance, Mycobacterium tuberculosis H37Rv was demonstrated to maintain the balance of redox intermediaries by utilizing oxygen and, in its absence, by either decreasing the reduction of cofactors like FAD, NAD and NADP or via using pathways that convert cofactors to their oxidized forms [37].
As observed from the growth curve and gas analysis (Figure 1a,b), P. thermoglucosidasius DSM 6285 renewed the growth between 27 and 44 h under WGS. This is supported by the observed induction of cellular protein metabolic process, ribosome, and translational activities (Table S2). Interestingly, there was gradual restoration of flagellar function at 27 h (Figure 3), as indicated by the overexpression of fliO, fliQ, fliP and fliR associated with the type III secretory system. Furthermore, transcripts of RNA polymerase sigma factor FliA and the switch mediator protein FliK were overexpressed at 44 h. FliA and its inhibitor, FlgM have been shown to play a central role in regulating the assembly of flagellum in bacteria [38].

Temporal Dynamics of P. thermoglucosidasius DSM 6285 Transcriptome between Aerobic Metabolism to Water Gas SHIFT reaction
Clustering of the 3055 normalised transcript counts showing differential expression in at least two consecutive condition sets revealed that the transcripts were grouped into 26 clusters with varying trajectories over the four time points ( Figure S2). Of these, 11 clusters showing two distinct patterns of expression comprised over 75% of the analysed transcripts. Transcripts in first set of clusters (8, 11, 13, 18 and 19), comprising 1319 transcripts, were generally down-regulated at 8 hours and up-regulated at 44 hours, while the second set of 1031 transcripts (1, 3, 4, 9, 16 and 22) showed overexpression at 8 hours and down-regulation at 44 hours. GO enrichment analysis revealed that the first cluster set was dominated by transcripts associated with amino acid transport, cellular developmental and sporulation, while the second set was significantly enriched with transcripts of

Temporal Dynamics of P. thermoglucosidasius DSM 6285 Transcriptome between Aerobic Metabolism to Water Gas SHIFT reaction
Clustering of the 3055 normalised transcript counts showing differential expression in at least two consecutive condition sets revealed that the transcripts were grouped into 26 clusters with varying trajectories over the four time points ( Figure S2). Of these, 11 clusters showing two distinct patterns of expression comprised over 75% of the analysed transcripts. Transcripts in first set of clusters (8, 11, 13, 18 and 19), comprising 1319 transcripts, were generally down-regulated at 8 h and up-regulated at 44 h, while the second set of 1031 transcripts (1, 3, 4, 9, 16 and 22) showed overexpression at 8 h and down-regulation at 44 h. GO enrichment analysis revealed that the first cluster set was dominated by transcripts associated with amino acid transport, cellular developmental and sporulation, while the second set was significantly enriched with transcripts of metabolic processes, including oxidation reduction, the tricarboxylic acid cycle and tetrapyrrole metabolism (Figure 4). KEGG enrichment analysis for these sets of clusters also revealed that only the ABC transporters were significantly (q < 0.05) enriched for the first set (Table 2). By contrast, significant (q < 0.05) enrichment of eleven pathways, biosynthesis of antibiotics and secondary metabolites, carbon metabolism and oxidative phosphorylation was observed for the second set of clusters. metabolic processes, including oxidation reduction, the tricarboxylic acid cycle and tetrapyrrole metabolism (Figure 4). KEGG enrichment analysis for these sets of clusters also revealed that only the ABC transporters were significantly (q < 0.05) enriched for the first set (Table 2). By contrast, significant (q < 0.05) enrichment of eleven pathways, biosynthesis of antibiotics and secondary metabolites, carbon metabolism and oxidative phosphorylation was observed for the second set of clusters.    While the clusters highlight the overall trend in terms of a shift in metabolism from aerobic processes to the anaerobic WGS-driven phase, it masks some of the nuances. For example, the significant down-regulation of some the pathways coincides with the up-regulation of specific transcripts for pathways that may be of interest for the metabolism of the organism during the WGS reaction. A specific example is the oxidative phosphorylation pathways, where transcripts for genes across the respiratory chain are generally down-regulated at 44 h post-inoculation relative to the aerobic phase (8 h). These include transcripts of genes involved in the synthesis of the NADH-quinone oxidoreductase (nuoB, nuoC, nuoD, nuoJ, nuoK, nuoN and ndhF), cytochrome bc1 complex (cytb, cyt1 and isp), cytochrome c oxidase caa3 (ctaA and ctaB) and the terminal cytochrome bd oxidase (cydA). Together, these changes suggest a shift from aerobic respiration to anaerobic fermentation ( Figure 5).
synthetase are significantly up-regulated at the 44h timepoint relative to the aerobic (8h) timepoint. A combined transcriptome and mutagenesis study of Listeria monocytogenes also showed upregulation of these genes when this organism was grown under anaerobic conditions [39]. It has been postulated that, under anaerobic conditions, F1F0-ATPase functions in reverse to export protons at the expense of ATP. Proton export via the F1F0-ATPase may regulate the intracellular pH [39,40] in response to pH effects resulting from aerobic acetate accumulation. The DSM 6285 transcriptome profile also showed an up-regulation of acetate kinase (AckA) transcripts at time point 20 relative to 8 hours and those of acetyl-CoA synthetase (ACS) at 27 relative to 20 hours, suggesting a switch to acetate assimilation in response to acid accumulation. In E. coli, phosphotransacetylase Pta and the acetate kinase AckA have been shown to reversibly convert acetate to acetyl CoA [32,41]. This is also consistent with the observed up-regulation of the transcripts qoxA, qoxC and qoxA, which are upregulated in L. monocytogenes in response to acid stress [42].  However, two transcripts, atpD and atpC, coding for the β and ε subunits of the F 1 F 0 -type ATP synthetase are significantly up-regulated at the 44h timepoint relative to the aerobic (8h) timepoint. A combined transcriptome and mutagenesis study of Listeria monocytogenes also showed up-regulation of these genes when this organism was grown under anaerobic conditions [39]. It has been postulated that, under anaerobic conditions, F 1 F 0 -ATPase functions in reverse to export protons at the expense of ATP. Proton export via the F 1 F 0 -ATPase may regulate the intracellular pH [39,40] in response to pH effects resulting from aerobic acetate accumulation. The DSM 6285 transcriptome profile also showed an up-regulation of acetate kinase (AckA) transcripts at time point 20 relative to 8 h and those of acetyl-CoA synthetase (ACS) at 27 relative to 20 h, suggesting a switch to acetate assimilation in response to acid accumulation. In E. coli, phosphotransacetylase Pta and the acetate kinase AckA have been shown to reversibly convert acetate to acetyl CoA [32,41]. This is also consistent with the observed up-regulation of the transcripts qoxA, qoxC and qoxA, which are upregulated in L. monocytogenes in response to acid stress [42].

Transcriptome Pattern of P. thermoglucosidasius DSM 6285 CODH and Hydrogenases
Evaluation of the gene expression profiles of the genes coding for the carbon monoxide dehydrogenase (CODH) and hydrogen evolving hydrogenase (Phc) [25] across the different points revealed that transcripts associated with CODH and Phc were more abundant at 8 and 20 h (aerobic phase) compared to 27 and 44 h (anaerobic and WGS phase; Figure 6a). Furthermore, the transcripts are generally up-regulated at time point 20 relative to 8 h and down-regulated at 27 relative to 20 h and at 44 relative to 27 h (Figure 6d). However, differential expression analysis revealed that only transcripts of cooF (coding for the CODH iron-sulphur protein) and phcABCE (coding for the Ni-Fe hydrogenase 4 B, C, E and D subunits, respectively) were significantly (FDR < 0.05) up-regulated at 20 h. By contrast, all transcripts except phcIKL (coding for the hydrogenase maturation protein, maturation nickel metallochaperone and accessory protein, respectively) were significantly (FDR < 0.05) down-regulated at 27 h. Cluster analysis showed that the CODH and Phc transcripts followed similar trajectories as depicted in clusters 9 and 7 ( Figure S3). The trajectories of cooC and phcDFGHJ suggest that these genes were up-regulated at both timepoint 8 and 20 h, and down-regulated at 44 h. By contrast, cooSF and phcABCE show rapid up-regulation at timepoint 20 relative to 8 h and subsequent down-regulation at 44 h. The initial overexpression of the latter group of transcripts, however, coincides with the depletion of O 2 at 20 h, suggesting that these genes may constitute an O 2 sensing mechanism; ie., anaerobic CODH and Ni-Fe group 4a hydrogenase are O 2 -sensitive [43][44][45]. It may be significant that transcripts of two putative CO-sensing transcriptional factors (Crp/Fnr; WP_090948317.1 and WP_064552491.1) were up-regulated (cluster 7; Figure S3

Transcriptome Pattern of P. thermoglucosidasius DSM 6285 CODH and Hydrogenases
Evaluation of the gene expression profiles of the genes coding for the carbon monoxide dehydrogenase (CODH) and hydrogen evolving hydrogenase (Phc) [25] across the different points revealed that transcripts associated with CODH and Phc were more abundant at 8 and 20 hours (aerobic phase) compared to 27 and 44 hours (anaerobic and WGS phase; Figure 6a). Furthermore, the transcripts are generally up-regulated at time point 20 relative to 8 hours and down-regulated at 27 relative to 20 hours and at 44 relative to 27 hours (Figure 6d). However, differential expression analysis revealed that only transcripts of cooF (coding for the CODH iron-sulphur protein) and phcABCE (coding for the Ni-Fe hydrogenase 4 B, C, E and D subunits, respectively) were significantly (FDR < 0.05) up-regulated at 20 hours. By contrast, all transcripts except phcIKL (coding for the hydrogenase maturation protein, maturation nickel metallochaperone and accessory protein, respectively) were significantly (FDR < 0.05) down-regulated at 27 hours. Cluster analysis showed that the CODH and Phc transcripts followed similar trajectories as depicted in clusters 9 and 7 ( Figure  S3). The trajectories of cooC and phcDFGHJ suggest that these genes were up-regulated at both timepoint 8 and 20 hours, and down-regulated at 44 hours. By contrast, cooSF and phcABCE show rapid up-regulation at timepoint 20 relative to 8 hours and subsequent down-regulation at 44 hours. The initial overexpression of the latter group of transcripts, however, coincides with the depletion of O2 at 20 hours, suggesting that these genes may constitute an O2 sensing mechanism; ie., anaerobic CODH and Ni-Fe group 4a hydrogenase are O2-sensitive [43][44][45]. It may be significant that transcripts of two putative CO-sensing transcriptional factors (Crp/Fnr; WP_090948317.1 and WP_064552491.1) were up-regulated (cluster 7; Figure S3) only at 44 hours relative to the three other time points. The two Crp/Fnr proteins share similar protein sizes (230-236 amino acids in length) and domain architecture (PR018490, IPR000595 and IPR012318) as the CooA protein (223 aa) of C. amalonaticus Y19 [46]. These genes occur ~0.3 Mb upstream the CODH-H2-evolving hydrogenase locus in the DSM 6285 genome. In addition to the CODH-Phc complex, two loci coding for predicted uptake hydrogenases, Pha and Phb, have been previously predicted on the DSM 6285 genome [25]. These share orthology to the unidirectional uptake Ni-Fe group 1d hydrogenases and Ni-Fe group 2a hydrogenases, respectively In addition to the CODH-Phc complex, two loci coding for predicted uptake hydrogenases, Pha and Phb, have been previously predicted on the DSM 6285 genome [25]. These share orthology to the unidirectional uptake Ni-Fe group 1d hydrogenases and Ni-Fe group 2a hydrogenases, respectively [24,25]. Both uptake hydrogenases are found in aerobic and facultative anaerobic bacterial taxa, are O 2 tolerant and play a role in recycling of H 2 for use in aerobic and hydrogenotrophic respiration [43,44,47].
Transcripts of the Phb genes showed similar expression patterns as observed for the Phc genes, with general up-regulation at time point 8 h and down-regulation at time point 44 h (Figure 6b). However, differential expression analysis (Figure 6e Pha transcripts showed the most diverse expression pattern (Figure 6c). While phaA showed the highest expression at 27 h, phaBCDEFGH had maximum expression at 8 h. By contrast, the maximum expression of phaIJK was observed at 27 and 44 h. Comparison of transcript profiles at the different time points revealed, however, that phaBCDFGH expression was significantly (FDR < 0.05) down-regulated at 20 relative to 8 h while only phaK was significantly (FDR < 0.05) up-regulated at this time (Figure 6f). Comparing time points 27 and 20 h, only transcripts of two genes phaJK showed significant (FDR < 0.05) up-regulation. No significant differentially gene expression was observed at time point 44 relative to 27 h. The overall expression profiles for Pha revealed that these transcripts follow four distinct trajectories ( Figure S3).
The general trend suggests that the uptake hydrogenases Pha and Phb are active during aerobic phases of P. thermoglucosidasius growth as previously predicted [24]. However, the phaJK genes, which code for hydrogenase expression/formation proteins (HypD/HypE) are the only uptake hydrogenase transcripts with increased transcriptional activity under anaerobic conditions. This is perhaps surprising since HypD and HypE, which are involved in Ni incorporation and purine derivative binding, respectively, are part of the machinery that generates the matured hydrogenase [48].

Discussion
RNA-seq based transcriptome profiling of the facultative anaerobe P. thermoglucosidasius DSM 6285 growing under an initial gas atmosphere of 50% air and 50% CO revealed a metabolic flexibility reflective of its ability to switch between aerobic growth, anaerobic survival, and anaerobic growth to recovery via the WGS reaction. Annotation of differentially expressed transcripts using KEGG and GOs revealed the suppression of aerobic metabolism during O 2 depletion [49], as indicated by downregulation of cytochrome oxidases and ATPase transcripts. Strikingly, transcripts of cytochrome aa3-600 menaquinol oxidases remained induced throughout the period of cultivation. Menaquinol oxidases serve as proton-pumping O 2 reductases that oxidize membrane-bound quinols directly without the need for electrons from cytochrome c [50]. Although part of aerobic respiratory chains, their evolution has been predicted to predate atmospheric O 2 [51] and it is therefore perhaps not surprising that they are induced under low O 2 or anaerobic conditions in P. thermoglucosidasius.
Previous work in B. subtilis revealed that at least one quinol oxidase is essential for the formation of spores [52,53]. Indeed, the up-regulation of aa3-600 coincides with the up-regulation of predicted sporulation genes as P. thermoglucosidasius DSM 6285 transitions from aerobic growth to the anaerobic phase where initiation of spore formation may serve as a survival strategy under energy limitation. This is further supported by the observed down-regulation of motility. The change from motile to non-motile forms is typically observed in the transition between exponential and stationary phase and often the two phases coexist [54,55].
The up-regulation of transcripts of the anaerobic CODH and H 2 -evolving Phc seems to coincide with O 2 depletion and the locus does not appear to be under the control of the putative CO-sensing transcriptional factor (Crp/Fnr) as reported for other carboxydotrophs [56]. Sensitivity of the anaerobic CODH to oxygen has been observed at both protein (enzymic activity) and genetic levels [57,58]. However, the precise regulation of the CODH genes in P. thermoglucosidasius DSM 6285 remains unclear as the current fermentation strategy, which was optimised for early hydrogen production, also incorporated CO from the inception. Further investigations should focus on elucidating the regulation of CODH loci in P. thermoglucosidasius DSM 6285.
The commencement of the biological WGS reaction after~27 h, as indicated by the consumption of CO and evolution of H 2 , coincided with the resumption of several metabolic activities. Both GO and KEGG enrichment analyses showed that ribosome biosynthesis was more active at 44 compared to 27 h. Up-regulation of ribosome biosynthesis has been reported during recovery of Nitrosomonas europaea impaired by CeO 2 nanoparticles [59]. Overexpression of ribosome biosynthesis has also been suggested as a means by which microorganisms compensate for low temperature-induced reduction in the rate of protein biosynthesis [60][61][62]. Although the current experiment did not directly quantify the amount of energy obtained by the organism via the WGS reaction, the up-regulation of genes coding for key central metabolic enzymes-e.g., ATP-dependent 6-phosphofructokinase and pyruvate, as well as the observable increase in biomass (Figure 1)-suggest that the WGS provides sufficient energy to support growth.

Microorganism and Media
P. thermoglucosidasius DSM 6285 was obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ, Braunschweig, Germany) and stored at −80 • C in glycerol.
For cultivation of the pre-cultures, mLB (modified Luria-Bertani) medium was used as previously described [24]. The experiment was performed in a modified ammonium sulphate medium (ASM) [

Experimental Set up and Analytical Methods
A two-step pre-culture system was used in the study. All pre-culture cultivation was carried out in 20 mL mLB medium. The first pre-culture was inoculated with 20 µL of a glycerol stock and cultured aerobically for 24 h at 55 • C, 120 rpm (Infors Thermotron, Infors AG, Bottmingen, Switzerland). Thereafter, a second pre-culture was inoculated to an OD 600 of 0.1 and cultivated aerobically for 12 h (60 • C, 120 rpm). The experimental set-up (49 mL ASM medium) was inoculated with 1 mL of the second pre-cultures in 250-mL stoppered serum bottles with an initial gas atmosphere of 50% air and 50% CO. Samples for RNA-Seq analysis were taken in the aerobic phase (8 h), the anaerobic lag phase (20 h), the anaerobic phase prior to H 2 production (27 h) and the H 2 -production phase (44 h). The experiment was conducted with two biological replicates, each comprised of six identical bottles. To obtain adequate amounts of RNA for sequencing, 4 mL of culture was collected from each bottle and pooled. Pressure was measured before and after every sample step using a manometer (GDH 14 AN, Greisinger electronic, Regenstauf, Germany). Absorbance (OD 600 ) was measured as an indication of bacterial growth using an Ultrospec 1100 pro spectrophotometer (Amersham Biosciences, Picataway, NJ, USA) at each sample point. pH was monitored with a pH meter (Profilab pH 597, Xylem Analytics Germany Sales GmbH and Co. KG, WTW, Oberbayern, Germany). To determine the gas composition in the bottle, 4 mL gas samples were taken and injected to a 300 Micro GC gas analyzer (Inficon, Bad Ragaz, Switzerland). The calculation of the gas composition was carried out as previously described [24].

RNA Isolation, Library Preparation, and RNA-Seq
Total RNA extraction, rRNA depletion and sequencing were conducted by Microsynth AG (Balgach, Switzerland). Stranded, ribodepleted RNA libraries were prepared using MICROBExpress (Thermo Fisher, Balgach, Switzerland) according to the manufacturer's instructions. Sequencing was performed using Illumina NextSeq, v2.5, 1x75bp chemistry. The reads were then demultiplexed and trimmed of Illumina adaptor residuals. All sequencing data obtained from this study have been deposited in the European Nucleotide Archive (ENA) under study accession number PRJEB36750.

RNA-Seq Data Analysis
The RNA-Seq reads were initially analysed using SPARTA [64] with default settings. Briefly, the reads were trimmed and analysed for quality using Trimmomatic [65] and FastQC [66], respectively.
High-quality reads were mapped against the draft genome sequence of P. thermoglucosidasius DSM 6285 using Bowtie [67]. Transcript feature abundance was subsequently determined with HTSeq [68]. To determine overexpressed transcripts between time points, differential gene expression analysis was performed using edgeR [69]. The normalised counts obtained from edgeR [69] were used to profile the time-dependent expression of transcripts using DP_GP_cluster [70]. The draft genome sequence of P. thermoglucosidasius DSM 6285 was annotated using gene ontology (GO) and KEGG orthologues obtained from UniProt [28] and using BlastKOALA [29]. GOEAST [71] and Pathview Web [72] were used to predict the functions of overexpressed transcripts. MDS plots and heat maps were generated using edgeR [69] and PAST [73], respectively.

Conclusions
We analysed the genetic response of P. thermoglucosidasius DSM 6285 transitioning from aerobic respiration to anaerobic growth and supported by CO metabolism and the WGS reaction. The transcriptome data revealed no obvious impact of 50% CO in the atmosphere on the initial aerobic metabolic phase in the presence of O 2 with P. thermoglucosidasius DSM 6285 only switching to CO metabolism after O 2 depletion, possibly due to the O 2 -dependent inactivation of the WGS enzymes (CODH and hydrogenase) by O 2 . Furthermore, a shift from a motile to non-motile physiology, a reduction in metabolic activity and the activation of stress response and sporulation pathways are all observed once O 2 is depleted, where P. thermoglucosidasius adapts to circumvent energy limitations through the WGS reaction. Overall, the data revealed a clear partitioning of transcripts which reflects a metabolic shift to cope with energy limitation and subsequent resumption of growth driven by the WGS reaction.
Supplementary Materials: The following are available online at http://www.mdpi.com/1422-0067/21/11/3870/s1, Figure S1: MDS plot showing pattern of transcripts expression in four RNA-seq samples of P. thermoglucosidasius DSM 6285 cultivated in two biological replicates (indicated by '_replicate number') in stoppered serum bottles with an initial gas atmosphere composition of 50% CO and 50% air over four time points. Figure S2: DP_GP_cluster analysis of differentially expressed transcripts of P. thermoglucosidasius DSM 6285 cultivated in stoppered serum bottles with an initial gas atmosphere composition of 50% CO and 50% air and samples over four time points. Figure  S 3. Trajectories of transcripts in the carbon monoxide dehydrogenase (CODH), uptake hydrogenase (Pha and Phb) and H 2 -evolving hydrogenase (Phc) loci. Table S1: Overview of RNA-seq read metrics of P. thermoglucosidasius DSM 6285. Culture were cultivated in two biological replicates in stoppered serum bottles with an initial gas atmosphere composition of 50% CO and 50% air over four time points. Table S2: Gene ontology (GO) enrichment of differentially Table S2