CO2-Induced Transcriptional Reorganization: Molecular Basis of Capnophillic Lactic Fermentation in Thermotoga neapolitana

Capnophilic lactic fermentation (CLF) is a novel anaplerotic pathway able to convert sugars to lactic acid (LA) and hydrogen using CO2 as carbon enhancer in the hyperthermophilic bacterium Thermotoga neapolitana. In order to give further insights into CLF metabolic networks, we investigated the transcriptional modification induced by CO2 using a RNA-seq approach. Transcriptomic analysis revealed 1601 differentially expressed genes (DEGs) in an enriched CO2 atmosphere over a total of 1938 genes of the T. neapolitana genome. Transcription of PFOR and LDH genes belonging to the CLF pathway was up-regulated by CO2 together with 6-phosphogluconolactonase (6PGL) and 6-phosphogluconate dehydratase (EDD) of the Entner–Doudoroff (ED) pathway. The transcriptomic study also revealed up-regulation of genes coding for the flavin-based enzymes NADH-dependent reduced ferredoxin:NADP oxidoreductase (NFN) and NAD-ferredoxin oxidoreductase (RNF) that control supply of reduced ferredoxin and NADH and allow energy conservation-based sodium translocation through the cell membrane. These results support the hypothesis that CO2 induces rearrangement of the central carbon metabolism together with activation of mechanisms that increase availability of the reducing equivalents that are necessary to sustain CLF. In this view, this study reports a first rationale of the molecular basis of CLF in T. neapolitana and provides a list of target genes for the biotechnological implementation of this process.


INTRODUCTION
Thermotoga neapolitana is a hyperthermophilic anaerobic bacterium of the order Thermotogales (Belkin et al., 1986). The taxonomic group shares a rod shape and complex outer envelope called toga that surrounds the bacterial cell and forms a periplasmatic space around the poles (Angel et al., 1993). T. neapolitana and other sister species are good candidates for the sustainable and efficient conversion of food and agriculture residues to hydrogen (H 2 ) by Dark Fermentation (Conners et al., 2006;Manish and Banerjee, 2008;Hallenbeck and Ghosh, 2009;Guo et al., 2010;d'Ippolito et al., 2010;Elleuche et al., 2014;Pradhan et al., 2015Pradhan et al., , 2016a. In the last years, we reported that T. neapolitana also operates a novel, anaplerotic process named capnophilic lactic fermentation (CLF) for the synthesis of almost enantiopure L-lactic acid (LA) without affecting H 2 production d'Ippolito et al., 2014;Pradhan et al., 2016bPradhan et al., , 2019Nuzzo et al., 2019). The metabolic process is activated by CO 2 (capnophilic means "requiring CO 2 ") and, nominally, is dependent on a Janus pathway including a catabolic branch leading to acetyl-CoA (AcCoA) from sugars by glycolysis, and an anabolic branch that combines AcCoA and CO 2 to give LA through reduction of newly synthesized pyruvate (PYR) by a NADH-dependent lactic dehydrogenase (LDH) (Figure 1).
This second part of the pathway requires an additional burden of reducing equivalents and determines an unconceivable deviation from Dark Fermentation model for carbon and hydrogen balance . We showed that CLF is not equally active in all members of the order Thermotogales (Dipasquale et al., 2018) but the almost complete absence of molecular and biochemical studies on metabolism of this group of bacteria has been an insurmountable barrier to explore this process that presumably involves crosstalking of several pathways. The aim of the present work was to investigate the role of CO 2 as biochemical trigger in T. neapolitana and to correlate metabolic change with or without CO 2 to hydrogen and LA production. We based our analysis on a differential RNA-sequencing of the strain T. neapolitana subsp. capnolactica, a mutant that shows an incremented operation of CLF (Pradhan et al., 2017). Transcriptome studies were associated to the experimental FIGURE 1 | Capnophilic lactic fermentation (CLF) pathway in T. neapolitana with details of the enzymatic reactions in the catabolic (gray) and anabolic (yellow) branch in relation to production and use of CO 2. PFOR, pyruvate:ferredoxin oxidoreductase; LDH, lactate dehydrogenase; HYD, hydrogenase.
response of the bacterium to CO 2 . Under our experimental conditions, CO 2 sparging generates a complex equilibrium between carbon dioxide as a gas or dissolved in the aqueous phase, and its hydrated derivatives H 2 CO 3 , HCO − 3 , CO = 3 . For simplicity we refer to all these chemical forms as CO 2 throughout the manuscript.

Bacterial Growth
Bacterial precultures (30 mL) were incubated overnight at 80 • C without shaking and used to inoculate (6% v/v) cultures in 120 ml serum bottles with a final culture volume of 30 mL. Oxygen was removed by heating until solution was colorless. Cultures were sparged with CO 2 gas (CLF condition, three bottles) or N 2 gas (control, three bottles) for 5 min at 30 mL/min. pH was monitored and adjusted to approximately 7.5 by 1 M NaOH. Sparging followed by pH adjustment was repeated every 24 h. Inoculated bottles were maintained in a heater (Binder ED720) at 80 • C. Cell growth was determined by optical density (OD) at 540 nm (UV/Vis Spectrophotometer DU 730, Beckman Coulter). Samples (2 ml of medium) were collected from each bottle after 0, 24, and 48 h. After centrifugation at 16,000 × g for 15 min (Hermle Z3236K), residues and supernatants were kept at −20 • C until analysis. Cell morphology was monitored by microscope observation (Axio VertA1, Carl Zeiss, magnification of 100×).

Gas Analysis
Gas (H 2 and CO 2 ) measurements were performed by gas chromatography (GC) on an instrument (Focus GC, Thermo Scientific) equipped with a thermoconductivity detector (TCD) and fitted with a 3 m molecular sieve column (Hayesep Q). N 2 was used as carrier gas. Gas sampling was carried out at 24 and 48 h.

Chemical Analysis
Glucose concentration was determined by the dinitrosalicylic acid method calibrated on a standard solution of 2 g/L glucose (Bernfeld, 1995). Organic acids were measured by ERETIC 1 H NMR as described by Nuzzo et al. (2019). All experiments were performed on a Bruker DRX 600 spectrometer equipped with an inverse TCI CryoProbe. Peak integration, ERETIC

RNA-Seq Bioinformatic Analysis
Quality check of the sequenced reads was performed using the FAST QC software 1 . The obtained high-quality reads were used for mapping by the Bowtie software (Langmead and Salzberg, 2012). Genome of T. neapolitana DSM_4359 at GenBank database (assembly accession: GCA_000018945.1) was used as 1 http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ reference. The quality statistics of the mapping process is showed in the Supplementary Table 1. The counting of the mapped reads was performed by the software HTScount. In order to define the set of expressed genes, raw read counts were normalized using the TMM method (Trimmed mean). Differentially expressed genes (DEGs) were obtained by the DESeq2 package of R language at a false discovery rate (FDR) ≤ 0.05 (Love et al., 2014). Expression values were reported as "Fold Change" (ratio of the normalized expression value in sample over control). Values <1 are shown as (−1/FC) to display negative regulation. Cluster analysis was performed using the MeV software based on the respective normalized reads count values (Howe et al., 2011). The number of clusters was determined by Figure of Merit (FOM) analysis. Clusters were generated by employing k-means clustering with Euclidian distances. In gene ontology enrichment analysis (GOEA), categories with a number of entries lower than 4 were excluded in the final output.

Real-Time PCR
RNAseq results were validated by real-time PCR. Triplicate quantitative assays were performed using a Platinum SYBR Green qPCR SuperMix (Life Technologies, Carlsbad, CA, United States). Cells from control cultures (N 2 sparging) were used as calibrators and RNA 16S served as endogenous reference gene (Okonkwo et al., 2017). Calculation of gene expression was carried out using the 2 − Ct method as in Livak and Schmittgen (2001). For each sample, mRNA amount was calculated relatively to the calibrator sample for the corresponding genes. Primers used for genes expression analysis are listed in Supplementary Table 2.

Western Blotting
Bacterial cells were grown under CLF condition and control as described above. Samples were collected after 12, 15, 36, 39, 42, and 45 h. Proteins extraction was performed on 1.5 ml of culture. After centrifugation at 10,000 × g for 20 min at 4 • C, the pellets were suspended in cracking buffer (100 mM Tris-HCl pH 7.5, 30% glycerol, 2% SDS, 4 mM EDTA, 28 mM β-mercaptoethanol) and incubated at 100 • C for 10 min. After centrifugation at 10,000 × g at 4 • C for 1 h, the supernatants were recovered and protein content was measured by Bradford reagent (Bio-Rad, Hercules, CA, United States Western blot filters were visualized by Clarity TM Western ECL substrate (Bio-Rad, Hercules, CA, United States) and subjected to analysis using Image Lab 6.0 Software (Bio-Rad, Hercules, CA, United States). Analysis included the determination of intensity of total Stain-Free fluorescence and intensity of PFOR-A and β-HYD blots for each lane by the "Lane and Bands" tool. Results were expressed as relative percentage of antibody blot intensity respect to intensity of total proteins.

Data Availability
The raw sequencing data from this study are stored in the NCBI SRA database 2 and are retrievable under the accession code PRJNA574556.

Statistics
Each experiment was performed at least in triplicate. Values were expressed as mean ± standard deviation (SD). The statistical significance of OD 540 , qRT-PCR, sugar consumption, organic acid, and H 2 yields was evaluated through Student's t-test (p ≤ 0.05). The differentially expressed data from the RNA-seq approach were filtered using an FDR ≤ 0.05.

Massive Molecular Rearrangement
Induced by CO 2 in T. neapolitana subsp. capnolactica In order to clarify the molecular effects of CO 2 , an RNAsequencing approach was performed by using N 2 -treated cells as control. The transcriptomic analysis gave from 19,390,095 to 31,800,340 reads with a percentage of alignment rate >98.59% (Supplementary Table 1). We identified 1601 DEGs between experiments under CO 2 (CLF samples) and N 2 (control) (see Supplementary Information). In particular, 612 DEGs showed a significant fold change up to ≥1.5 whereas 593 DEGs were down-regulated (fold change below −1.5) (Figure 2). Considering the whole genome of T. neapolitana, CO 2 -dependent DEGs accounted for 83% of the bacterial genes. GOEA of DEGs highlighted change of 45 metabolic networks under CLF conditions, with 26 related to up-regulation and 19 to down-regulation (Supplementary Figure 1). The analysis indicated increased expression of genes related to RNA translation (GO:0006412), ribosome organization (GO:0015934, GO:0015935), transcription factor (TF) (GO:0003700), and amino acids biosynthesis (GO:0009089, GO:0009088, GO:0006526, GO:0009097, GO:0009098, GO:0009073, GO:0006541). Expression and regulation of proteins related to active transport, including major facilitator proteins, and various membrane channels and antiporters, put forward a general down-regulation of sugar transport and vice versa an increased mobilization of phosphate, polyamine, and amino acids. The GOEA also suggested a significant impact of CO 2 on organization of the cell-wall (GO:0071555) with a marked down-regulation of the degradation of polysaccharides of the outer membrane (xylan catabolic process, GO:0045493; endoglucanase activity, GO:0004519). Interestingly, two of the three genes encoding for the outer membrane proteins that form the structure of toga (Petrus et al., 2012), namely CTN_RS01400 and CTN_RS01395, showed a similar decrease of expression. A number of TFs were also regulated by CO 2 treatment, including a number of TFs involved in the regulation of sugar catabolism (AraC and LacI). As expected by the simultaneous production of LA and H 2 , we found the enrichment of metabolic categories related to redox reactions (Cell redox homeostasis -GO:0045454; Electron carrier activity -GO:0009055; and Oxidoreductase activity -GO:0016491) and carbon metabolism.

Central Carbon Metabolism Under CLF Conditions
Sparging by CO 2 induced up-regulation of glucokinase (GCK -CTN_RS05065) and glucose 6-phosphate isomerase (PGI -CTN_RS01845) that catalyze the starting steps of Embden-Meyerhof-Parnas glycolysis (EMP) ( Table 2). On the other hand, we found that other enzymes of EMP were significantly downregulated. Inspection of other genes related to central carbon metabolism revealed a significant transcription of glucose-6-phosphate dehydrogenase (G6PDH -CTN_RS07110) and 6-phosphogluconolactonase (6PGL -CTN_RS07115). These enzymes preside over the synthesis of 6-phospho gluconate that is the first metabolite of the main alternative pathways of glucose catabolism, namely ED and oxidative pentose phosphate (OPP) pathway. On the whole, these data suggest a diversion of carbon flux from EMP to ED or OPP pathways (Figure 3). This evidence was further validated by RT-PCR that showed up-regulation of 6-phosphogluconate dehydratase (EDD -CTN_RS00550) that control the key step of ED and down-regulation of 6-phosphofructokinase (PFK -CTN_RS02345) that is the regulator enzyme of EMP (Supplementary Figure 2). Analysis of the genes of Krebs cycle reveals that T. neapolitana apparently lacks the whole pathway. However, we found a significant upregulation of the expression of citrate synthase (CTN_RS01940) and isocitrate dehydrogenase (CTN_RS07145) ( Table 2).

Regulation of CLF Pathway
Genes encoding key enzymes of the anabolic branch of CLF pathway were differential expressed by CO 2 ( Table 3). Pyruvate:ferredoxin oxidoreductase (PFOR, E.C. 1.2.7.1) is a heterotetramer enzyme that operates the reversible coupling of AcCoA and CO 2 to PYR under CLF conditions (d 'Ippolito et al., 2014). Each subunit of this enzyme (CTN_RS03380, CTN_RS03385, CTN_RS03390, and CTN_RS03395) showed a significant increase of expression. In consideration of the double function of PFOR as catabolic (from PYR to AcCoA) and anabolic (from AcCoA and CO 2 to PYR) enzyme, the up-regulation is in good agreement with both the general acceleration of metabolism induced by CO 2 and the enhanced demand of PYR to feed the downfield synthesis of LA. According to this view, LDH (E.C. 1.1.1.27 -CTN_RS03950) was also up-regulated under the experimental conditions. A specific biochemical character of CLF is the increase of the rate of H 2 synthesis during the fermentation process . The molecular analysis revealed a significant upregulation of the expression of the α-subunit of the bacterial hydrogenase (CTN_RS05285) together with two of the three proteins (CTN_RS06525, HYD E; CTN_RS06535, HYD G) that are necessary to the complex process of maturation of this enzyme. The α-subunit (HYDα) contains the catalytic H-cluster (Schut and Adams, 2009) that is committed to reduction of H + to hydrogen gas. In their model of the hydrogenase of Thermotoga maritima, Schut and Adams (2009) have discussed a trimeric bifurcating enzyme that catalyzes multiple electron transfer events simultaneously. HYDα is the final collector of this transfer of electrons from NADH and reduced ferredoxin; thus, it is plausible that the experimentally observed boost of the H 2 production rate requires an accelerated turnover of this protein under CLF conditions. The CO 2 -induced upregulation of PFOR and HYD was further confirmed by western blotting analysis carried out on PFORα and the β-subunit of hydrogenase (HYDβ) (Figure 4).

Redox Re-organization Under CLF Conditions
Enhancement of cell reductants, e.g., NAD(P)H or ferredoxin (Fd), represents a major requirement to sustain the simultaneous production of H 2 and LA during CLF d'Ippolito et al., 2014). RNA-seq approach revealed up-regulation of several genes involved in redox reactions and electron transport. Intriguingly, among up-regulated DEGs we found a significant expression of genes coding for enzymes related to the flavin-based oxidoreductase enzymes NADH-dependent reduced ferredoxin:NADP oxidoreductase (NFN) (CTN_RS04025 and CTN_RS04020) and NAD-Ferredoxin oxidoreductase (RNF) (CTN_RS02165, CTN_RS02160, and CTN_RS02155). Both these proteins are energy conservation systems. NFN controls supply of reduced ferredoxin and NADH by oxidation of NADPH while RNF couples reversible consumption of NADH with synthesis of reduced Fd and trans-membrane transport of Na + or H + (Hess et al., 2013). K-means cluster analysis of DEGs identified five different groups of genes with similar expression profile (not shown). One of the upregulated cluster (Supplementary Figure 3) included genes coding for PFOR (CTN_RS03385, CTN_RS03390, and CTN_RS03395), flavin-based protein complexes (RNF and NFN -CTN_RS02160, CTN_RS02165, CTN_RS04020, and CTN_RS04025), thioredoxin (CTN_RS05860 and CTN_RS08515), ferredoxin (CTN_RS08225), Na/P cotransporter family protein (CTN_RS07190), and G6PDH (CTN_RS07110). The clustering was coherent with the fermentation results and suggested a CO 2 -stimulated rearrangement of the bacterial metabolism that links NAD(P)H from ED/OPP, PYR synthesis by PFOR, and production of NADH by flavin-based energy conservation systems. According to the experimental results, up-regulation of these pathways could be consistent with increase of the availability of redox potential for the synthesis of LA and H 2 during CLF.

DISCUSSION
Thermophilic bacteria, such as Thermotogales, have been acquiring an emerging attention in biotechnology for hydrogen production by digestion of organic residues (Conners et al., 2006;Elleuche et al., 2014;Pradhan et al., 2015). Among thermophilic bacteria, T. neapolitana shows an interesting potential for the simultaneous production of hydrogen and LA by CLF d'Ippolito et al., 2014;Pradhan et al., 2017). In the present study, a selected strain of this bacterium, namely T. neapolitana subsp. capnolactica (Pradhan et al., 2017) was characterized by comparative analysis of gene expression profiling under CO 2 (CLF conditions) and N 2 (control). The study showed transcriptional changes that well correlate with the promoting effect of CO 2 on both the synthesis of LA and the rate of hydrogen production and glucose consumption. The appearance of this bacterial phenotype was accompanied by variation in the expression of 1601 genes related to central biochemical pathways including glycolysis, TCA, cell wall construction, and protein synthesis. Under light microscopy, the cells showed a typical rod shape with a more homogeneous length than the control bacteria under N 2 that vice versa were round-shaped (Supplementary  Figure 4). In agreement with the acceleration of the fermentation FIGURE 4 | Expression of pyruvate:ferredoxin oxidoreductase (PFOR) and Fe-Fe hydrogenase (HYD) in T. neapolitana subsp. capnolactica. Bacterial cultures were grown under N 2 or CO 2 , and collected at 12, 15, 18, 36, 39, 42, and 45 h. PFOR levels were detected by using a polyclonal antibody raised in rabbits against peptide 149-359 of subunit α. HYD levels were detected by using a polyclonal antibody raised in rabbits against peptide 1-192 of subunit β. Western blot filters were subjected to analysis using Image Lab 6.0 Software (Bio-Rad, Hercules, CA, United States). Analysis included the intensity determination of total Stain-Free fluorescence and intensity of PFOR-A and β-HYD blots for each lane. Results were expressed as relative percentage of antibody blot intensity respect intensity of total proteins. rate, CO 2 also triggered modification of the expression of 105 genes related to control of extracellular transport. Latif et al. (2015) have recently underlined that adaptation of Thermotoga maritima to achieve improved growth fitness requires plasticity of the ABC transporters. We found that expression of several ABC transporters of T. neapolitana was differently regulated (Supplementary Information), which is consistent with the modulation of these proteins in response to the different environmental conditions. These molecular changes were driven by a number of key TFs that were differentially expressed under CO 2 . In particular, we observed different regulation of genes of the LacI family, such as LacI, AraC, and XylR, that are tightly linked to distinct regulons of carbohydrate catabolism (Ravcheev et al., 2014). These proteins have been also identified as regulators of sugar metabolism in Thermotogales, including T. neapolitana and T. maritima (Rodionov et al., 2013). Thus, control of their expression seems to corroborate the observation that the metabolic acceleration triggered by CO 2 is sustained by a change of sugar catabolism in association with a more effective import/export of substrates.
According to this view, under CLF conditions, the bacterial cells showed an interesting shift of glucose utilization through down-regulation of EMP and activation of the alternative ED and/or OPP (Figure 3). EMP is energetically more effective than other glycolytic pathways because of the higher ATP yields per glucose unit. However, the energetic cost for the synthesis of the proteins carrying out the 10 reactions of this pathway may represent a growth-limiting factor (Molenaar et al., 2009;Flamholz et al., 2013). Recently, Singh et al. (2018) have discussed the role of OPP in providing additional reductants for H 2 production in T. maritima. ED and OPP show a number of energetic advantages per glucose unit [reduced protein cost or increased amount of NAD(P)H, respectively] in comparison to EMP, thus switch from EMP to ED or OPP can be functional to the increased demand of reducing equivalents during CLF.
enzymes NFN and RNF that are both up-regulated by CO 2 . In T. maritima, NFN is a bifurcating enzyme that couples the oxidation of NADPH to the exergonic reduction of NAD + and the endergonic reduction of Fd (Demmer et al., 2015). Electron bifurcation is one the mechanisms of energy conservation that biological organisms use to minimize freeenergy loss in redox reactions. Bifurcating enzymes catalyze endoergonic and exoergonic electron transfer reactions to circumvent thermodynamic barriers and to carry out reactions that have unfavorable free energy variations (Buckel and Thauer, 2018). Deletion of the NFN complex (Tsac_2085 NfnA and Tsac_2086 NfnB) genes led to the total loss of NADPH-dependent activity and reduces ethanol fermentation in the thermophilic anaerobe Thermoanaerobacterium saccharolyticum (Lo et al., 2015). On the other hand, the RNF complex couples the reversible flow of electrons from reduced ferredoxin to NAD + . The process is associated with the transport of Na + across the cell membrane and the direction of the electron flow is dependent on the sodium gradient (Biegel et al., 2011;Buckel and Thauer, 2013;Hess et al., 2013). The high production of hydrogen in T. neapolitana is associated to consumption of hydrogen ions (H + ). RNF activity could be related to Na + or H + transport to balance the loss of H + and to maintain the ionic homeostasis in the cell. As suggested in Figure 5, the combination of RNF with NFN may generate a cyclic process leading to production of NADH. In this view, synthesis of LA could be another mechanism to get rid of the excess of reducing equivalents. A similar cycle has been postulated between RNF and ferredoxin-dependent NADH oxidoreductase (FNOR) in T. saccharolyticum in order to operate a net transfer of protons across the cell membrane and eliminate the proton gradient that could be lethal for the cells (Tian et al., 2016).
As already reported by Krebs (1941), CO 2 affects carboxylation reactions also in heterotrophs. In T. neapolitana, in addition to the indirect effect on the bacterial metabolism by gene regulation, CO 2 plays a direct role as chemical reagent in the reaction to give PYR from acetate by PFOR. Thus, increase of the concentration of CO 2 is very likely the driven force to reverse the oxidative decarboxylation of PYR and to yield carbon fixation. The occurrence of this reaction could explain the experimentally observed presence of PFOR even after the end of sugar consumption by glycolysis (d 'Ippolito et al., 2014). At the moment we do not know whether this occurs for inversion of the functionality of PFOR or it is due to a newly synthesized pool of the enzyme (Figure 1). Synthetic PFOR is the key enzyme of the autotrophic fixation by the Wood-Ljungdahl pathway in acetogens (Furdui and Ragsdale, 2000). After our publication of the synthetic activity of PFOR in T. neapolitana (d 'Ippolito et al., 2014), Xiong et al. (2016) have reported a similar mechanism in Clostridium thermocellum. In this work, the authors report assimilation of CO 2 during cellobiose metabolism with fixation of over 40% of labeled carbon in PYR that derives from PFOR-dependent synthesis. This process is very similar to that of T. neapolitana thus suggesting that the activation of the anaplerotic reductive C1 pathway dependent on PFOR can be more common than it is generally believed. It is obvious that the energy requirements of this process need to be considered within a comprehensive organization of the cellular redox pathways.

CONCLUSION
In conclusion, the present work showed a correlation between the gene expression, the metabolic adaptation and the biochemical phenotype of the bacterium T. neapolitana under conditions that promote CLF. In according with acceleration of H 2 synthesis, the CO 2 -induced effects clearly indicated an increased energetic metabolism of T. neapolitana that is sustained by a concerted reorganization of the cell metabolism including at least variation of central carbon metabolism, activation of anaplerotic oxidoreductase processes and increase of the exchange across the membrane. Similar effects on enhancement of the energetic metabolism or cell growth upon exposure to CO 2 have been reported in other anaerobic bacteria, such as Streptococcus thermophilus (Arioli et al., 2009) and C. thermocellum (Xiong et al., 2016). The most significant trait of the molecular adaptation induced by CO 2 in T. neapolitana is related to additional production of reductants by the EMP/ED-OPP switch and the bifurcating mechanisms based on the overexpression of the flavin-based complexes, such as NFN and RFN.

DATA AVAILABILITY STATEMENT
The raw sequencing data from this study are stored in the NCBI SRA database (http://www.ncbi.nlm.nih.gov) and are retrievable under the accession code PRJNA574556.

AUTHOR CONTRIBUTIONS
Gd'I conceived the whole research project, and wrote and amended the manuscript. SL conceived the idea of writing the manuscript, bioinformatic analysis, and wrote and amended the manuscript. NE performed the microbiology and western blot analysis. ML and MV performed the western blot. LD and NP performed the microbiology and sequencing analysis. AF planned and coordinated the whole research project and the experiments along with Gd'I, and wrote and amended the manuscript.