YgfY Contributes to Stress Tolerance in Shewanella oneidensis Neither as an Antitoxin Nor as a Flavinylation Factor of Succinate Dehydrogenase

YgfY(SdhE/CptB) is highly conserved while has controversial functions in bacteria. It works as an antitoxin and composes a type IV toxin–antitoxin system with YgfX(CptA) typically in Escherichia coli, while functions as an flavinylation factor of succinate dehydrogenase and fumarate reductase typically in Serratia sp. In this study, we report the contribution of YgfY in Shewanella oneidensis MR-1 to tolerance of low temperature and nitrite. YgfY deficiency causes several growth defects of S. oneidensis MR-1 at low temperature, while YgfX do not cause a growth defect or morphological change of S. oneidensis MR1-1 and E. coli. YgfY do not interact with FtsZ and MreB nor with YgfX examined by bacterial two-hybrid assay. YgfY effect on growth under low temperature is not attributed to succinate dehydrogenase (SDH) because a mutant without SDH grows comparably with the wild-type strain in the presence of succinate. The ygfY mutant shows impaired tolerance to nitrite. Transcription of nitrite reductase and most ribosome proteins is significantly decreased in the ygfY mutant, which is consistent with the phenotypes detected above. Effects of YgfY on growth and nitrite tolerance are closely related to the RGXXE motif in YgfY. In summary, this study demonstrates pleiotropic impacts of YgfY in S. oneidensis MR-1, and sheds a light on the physiological versatility of YgfY in bacteria.


Introduction
Shewanella species are emerging environmental bacteria with unique and versatile respiration, therefore playing a role in the biogeochemical cycle of metals [1,2]. In their favored niches, Shewanella species might confront stresses, such as heavy metals, toxic chemicals, high pressure and cold temperature [3]. However, the current knowledge regarding stress tolerance in the Shewanella species is still limited.
Physiological functions of bacterial toxin-antitoxin (TA) systems are commonly investigated in a number of laboratory and clinical isolates. In contrast, their physiological significance in environmental bacteria is relatively obscure. Recent studies have revealed a novel TA system in S. oneidensis MR-1 that plays a role in cell motility [4].
TA systems include a toxin protein that hinders cell growth, and an antitoxin as a protein or an RNA that protects cells from toxins via various mechanisms. Up to date, six types of TA system have been reported [5]. The type IV TA system is different from all the other TA systems in which antitoxins directly target the toxin. Instead, the antitoxin protein in type IV system protects cells from the toxin through interacting with and stabilizing the targets of the toxin. All type IV TA systems reported so far are involved in cell division through interfering functions of cytoskeleton proteins FtsZ and MreB. CptB(YgfY)/CptA(YgfX) and CbtB/CbtA are the first two identified type IV TA systems. Both toxins of CptA and CbtA inhibit the cell division through interacting with cytoskeleton proteins, FtsZ and MreB [6,7]. CbtB antagonizes the CbtA toxicity by enhancing the bundling of cytoskeletal polymers of MreB and FtsZ in Escherichia coli [6]. YkfI and YpjF, CbtA homologous in E. coli also inhibit cell growth and lead to morphological changes of cells by interacting with FtsZ and MreB [8].
This study aims to reveal the physiological function of YgfY/YgfX and explores the underlying mechanism in S. oneidensis MR-1. Our results demonstrate that YgfY has pleiotropic impacts in S. oneidensis MR-1 including adaptive growth at low temperature and tolerance to nitrite. However, these impacts of YgfY were not attributed to functions as an antitoxin nor as a flavinylation factor of SDH. Lines of evidence in this study suggest a novel mechanism of YgfY function in stress tolerance of Shewanella species.

Bacteria and Growth Conditions
Bacterial strains were cultured in Lysogeny Broth (LB) medium or mineral medium as reported previously [12]. Strains of E. coli were cultured at 37 • C, and those of S. oneidensis MR-1 were cultured at 30 • C or 16 • C as indicated. Chemicals were added when needed at the following concentrations: 100 µg/mL diaminopimelic acid, 50 µg/mL kanamycin, 100 µg/mL ampicillin, and 20 µg/mL gentamycin for strains of E. coli; 10 µg/mL gentamycin for strains of S. oneidensis MR-1.

Construction of Strains and Plasmids
The construction of transposon library and the location of transposon insertion were conducted as described previously [13], except that S. oneidensis MR-1 was used as the parent strain of the library.
Mutants with an in-frame deletion of desired genes were constructed as described previously [14]. Briefly, flanking regions of a target gene were amplified and ligated with digested suicide vector pRE112 [15] using ClonExpress MultiS One Step Cloning Kit (Vazyme, Nanjing, China). The ligation products were transformed into E. coli SM10 and verified by PCR. Verified plasmids were extracted and transformed into E. coli WM3064 [16] and then introduced into strains of S. oneidensis MR-1 by conjugation. After two rounds of selection, a mutant was obtained with the deletion of a target gene and confirmed by sequencing.
Site-directed mutation was conducted to construct pBBP1_ygfY G16R/E19A as described previously [13]. Briefly, primers with desired substitution at the 5 ends of joint primers sites were designed and used for overlapping PCR. PCR products were ligated into digested pBBP1 using ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China). All generated plasmids were verified by sequencing before phenotype tests. All strains and plasmids used in this study were listed in Table S1 and all primers were listed in Table S2.

Overexpression of YgfY and YgfX in E. coli
E. coli BL21(DE3) bearing pBAD_ygfX and pET_ygfY was cultured in LB overnight and then diluted in fresh LB to 0.01 of OD 600 for sub-cultivation. Inducers (isopropyl β-D-thiogalactoside (IPTG) and L-arabinose) or inhibitor (glucose) were added when subcultures grew to 0.2 of OD 600 . IPTG and/or L-arabinose were added to 1 mM of final concentration to induce expression of YgfY and YgfX, alone or simultaneously. Glucose was added to 1 mM of final concentration to suppress leaking expression both of YgfY and YgfX. Aliquots of subcultures were withdrawn to measure OD 600 and observe cell morphology using a microscope Axio Scope A1 (Carl Zeiss Microscopy, LLC., New York, NY, USA).

Bacterial Two-Hybrid Assay
The bacterial two-hybrid system based on adenylate cyclase reconstitution in E. coli was adopted to detect the interaction of YgfX with other proteins in vivo (Euromedex, Souffelweyersheim, France). ygfX was amplified and cloned into pKNT25 and pKT25 to express a chimeric protein that fused T25 fragment of adenylate cyclase to the N-terminal and C-terminal of YgfY, respectively. ftsZ, mreB, and ygfX were cloned into pUT18 and pUT18C. Detection of protein-protein interaction in vivo was conducted according to manufacturer's instruction (Euromedex, Souffelweyersheim, France). Overnight cultures of E. coli BTH101 bearing two plasmids were used to determine β-galactosidase activity in cells.

RNA Isolation and Sequencing
Overnight cultures in LB were diluted in fresh LB to 0.1 of OD 600 and then cultivated at 16 • C for 12 h. Four biological replicates of subcultures were mixed and collected by centrifugation (12,000× g, 4 • C, 1 min). Collected cells were subjected to extraction of total RNA using the Trizol Reagent (Takara Biotechnology, Beijing, China) according to the manufacturer's instruction. Total RNA was further purified using RNA clean kit (BioTeke, Beijing, China). The concentration, quality, and integrity of total RNA were determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and an Agilent 2100 bioanalyzer (Agilent Technologies, Waltham, CA, USA). Qualified total RNA was used to construct a sequencing library that was subsequently sequenced on a Hiseq platform (Illumina, New York, NY, USA), which was performed by Shanghai Personal Biotechnology Co., Ltd.
The data of RNA sequencing (RNA-Seq) were deposited to GOE. The accession numbers have not yet been obtained but will be provided before acceptance of this manuscript.

qRT-PCR Assay
cDNA was synthesized using 500 ng of qualified total RNA and PrimeScript RT reagent kit with gDNA eraser (Takara Biotechnology, Beijing, China). The qRT-PCR was conducted using the SYBR Premix Ex Taq kit (Takara Biotechnology, Beijing, China) and Lightcycler 96 (Roche, Mannheim, Germany). The gyrB was used as the reference gene. The relative expression value of target genes was obtained from three determinations with normalization against the reference gene using the method of 2 −∆∆Ct .

Assay of Transcription Capability Using Click Chemistry
To detect the transcription capability, a method of click chemistry using 5-ethynyluridine and an azide-modified fluorophore was adopted [10]. Overnight cultures of WT and ∆ygfY were diluted into fresh LB to 0.01 of OD 600 , and then cultured at 30 • C for 3 h. Then, cultures were collected, concentrated for 25 folds, and added with EU (Molecular Probes, Inc., Eugene, OR, USA) to a final concentration of 0.5 mM. Cultures were allowed to grow for another 2 h at 30 • C. After that, cells were collected, fixed with 4% formaldehyde, and neutralized with 2 mg/mL glycine. After washed with phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , pH 7.2), cells were 4 of 13 incubated in 0.5% Triton X-100/PBS for 10 min. After washed with PBS, cells were stained by 2 µM azide-modified Alexa Fluor 488 (Molecular Probes, Inc., Eugene, OR, USA) in Click-iT cell reaction buffer (Molecular Probes, Inc., Eugene, OR, USA) for 30 min at room temperature. Stained cells were washed with PBS, and then observed using a microscope Axio Scope A1 (Zeiss, Jena, Germany).

Assay of Nitrite Tolerance
To determine the minimal inhibition concentration (MIC) of nitrite, S. oneidensis MR-1 was cultured at 30 • C for 16 h and then diluted in fresh LB to 0.01 of OD 600 . Nitrite was added into those diluted cultures to final concentrations of 0 to 1000 mM. Diluted cultures were cultivated at 30 • C for 24 h.
The nitrite tolerance of ∆ygfY and wild-type (WT) was determined by plating nitritetreated cells on LB agar plates and counting colony-forming units (CFU). In detail, strains were cultured in LB at 30 • C for 16 h and then diluted in fresh LB to 0.01 of OD 600 . Diluted cultures were subcultivated to 0.2-0.3 of OD 600 before nitrite of 470 mM (15 × MIC) was added. At indicated time points, aliquots of nitrite-treated subcultures were withdrawn, washed with 0.9% NaCl two times to remove remained nitrite. Washed cultures were serially diluted in 0.9% NaCl and spotted on LB agar plates. Cells on plates were cultivated at 30 • C for 24 h before counting CFU.

Assay of Tolerance to Heat Shock
Strains were cultured in LB at 30 • C for 16 h and then diluted in fresh LB to 0.01 of OD 600 . Diluted cultures were subcultivated to 0.2-0.3 of OD 600 at 30 • C, and then incubated at 42 • C. At indicated time points, aliquots of heat-shocked subcultures were withdrawn, diluted with 0.9% NaCl, and spotted on LB agar plates. If needed, 100 U/µL catalase solution of 5 µL was overlayed on culture spots. Cells on plates were cultivated at 30 • C for 24 h before counting CFU.

Quantification of Succinate
Strains were cultured in LB overnight. Cells were collected by centrifugation (3000× g, 10 min) for 1 min and washed three times with a mineral medium (MM) [12]. After that, cells were resuspended in MM containing 10 mM succinate to final density of 2.5 of OD 600 . These cultures were incubated at 30 • C with vigorous shaking, and aliquots were withdrawn at indicated time points. Aliquots of samples were centrifugated at 12,000× g for 10 min and treated with 0.22 µm filters. Succinate remained in MM was quantified using Agilent 1260 liquid chromatography (Agilent Technologies Inc., Santa Clara, CA, USA) equipped with an Aminex HPX-87H column (Bio-Rad Laboratories, Inc., Hercules, CA, USA). H 2 SO 4 of 5 mM was used as a running buffer at a flow rate of 0.5 mL/min. The standard calibration curve of succinate was obtained using MM containing succinate over range of 1-50 mM.

YgfY Is Required for Normal Growth of S. oneidensis MR-1
A transposon library of S. oneidensis MR-1 was constructed to screen genes that contribute to tolerance to various stresses. We accidently found a colony showing smaller size than WT (data not shown). The transposon in this mutant inserted into a putative two-gene operon containing ygfY (SO_1339) and ygfX (SO_1340). The ygfY and ygfX are predicted as an operon coding a type IV TA system, in which ygfY encodes an antitoxin and ygfX encodes a toxin.
To confirm the effect of YgfY and YgfX on cell growth, mutants with in-frame deletion of ygfX and ygfYX were constructed. Intriguingly, we also successfully constructed a ygfY mutant that is theoretically inaccessible if YgfY was a cognate antitoxin of YgfX. ∆ygfY and ∆ygfYX rather than ∆ygfX showed defect in growth at 30 • C ( Figure S1). ∆ygfY and ∆ygfYX showed more severe defect of growth when cultivated at 16 • C ( Figure 1A). Growth was Microorganisms 2021, 9, 2316 5 of 13 quantified by the optimal density of cultures at 600 nm that could be affected by the cell density as well as the cell size. The cell density of cultures was further determined by counting colony forming units (CFU). After cultivated at 16 • C for 48 h, cultures of ∆ygfY reached to (2.02 ± 0.30) × 10 9 CFU/mL and those of WT reached to (3.62 ± 0.15) × 10 9 CFU/mL. The CFU of ∆ygfY cultures was about 56% of those of WT, which was consistent with the result of optimal density measurement. Meanwhile, cells in these cultures were observed by microscopy, and no morphological difference was observed among ∆ygfY, ∆ygfX, ∆ygfYX and WT ( Figure S2). Complementation of ygfY completely restored the growth of ∆ygfY and ∆ygfYX to a similar level of WT ( Figure 1B). Expression of ygfY from its native promoter also restored the growth of ∆ygfY and ∆ygfYX to a similar level of WT ( Figure S3).
To confirm the effect of YgfY and YgfX on cell growth, mutants with in-frame deletion of ygfX and ygfYX were constructed. Intriguingly, we also successfully constructed a ygfY mutant that is theoretically inaccessible if YgfY was a cognate antitoxin of YgfX. ΔygfY and ΔygfYX rather than ΔygfX showed defect in growth at 30 °C ( Figure S1). ΔygfY and ΔygfYX showed more severe defect of growth when cultivated at 16 °C ( Figure 1A). Growth was quantified by the optimal density of cultures at 600 nm that could be affected by the cell density as well as the cell size. The cell density of cultures was further determined by counting colony forming units (CFU). After cultivated at 16 °C for 48 h, cultures of ΔygfY reached to (2.02 ± 0.30) × 10 9 CFU/mL and those of WT reached to (3.62 ± 0.15) × 10 9 CFU/mL. The CFU of ΔygfY cultures was about 56% of those of WT, which was consistent with the result of optimal density measurement. Meanwhile, cells in these cultures were observed by microscopy, and no morphological difference was observed among ΔygfY, ΔygfX, ΔygfYX and WT ( Figure S2). Complementation of ygfY completely restored the growth of ΔygfY and ΔygfYX to a similar level of WT ( Figure 1B). Expression of ygfY from its native promoter also restored the growth of ΔygfY and ΔygfYX to a similar level of WT ( Figure S3). These results indicated that YgfY rather than YgfX positively affected the growth and did not support YgfY/YgfX in S. oneidensis MR-1 as a TA system, because ∆ygfYX lacking both YgfY and YgfX should show growth recovery compared with ∆ygfY if YgfX functions as a toxin and YgfY as a cognate antitoxin.

YgfY and YgfX do not Fulfil the Criteria of Type IV TA System
Both ygfY and ygfX are supposed to compose a two-gene operon in which the start codon of ygfY is 37 bp away from the stop codon of ygfX ( Figure 2A). The RT-PCR detected the transcription of the intergenic region between ygfY and ygfX, indicating that they belong to the same operon ( Figure 2B). The similarity of amino acid sequences between YgfY homologs in S. oneidensis MR-1 and E. coli MG1655 was 62.5%, and that between YgfX homologs in two strains was 34.8%. EMBOSS Needle was used to analyze the similarity of amino acid sequences. These results indicated that YgfY rather than YgfX positively affected the growth and did not support YgfY/YgfX in S. oneidensis MR-1 as a TA system, because ∆ygfYX lacking both YgfY and YgfX should show growth recovery compared with ∆ygfY if YgfX functions as a toxin and YgfY as a cognate antitoxin.

YgfY and YgfX Do Not Fulfil the Criteria of Type IV TA System
Both ygfY and ygfX are supposed to compose a two-gene operon in which the start codon of ygfY is 37 bp away from the stop codon of ygfX ( Figure 2A). The RT-PCR detected the transcription of the intergenic region between ygfY and ygfX, indicating that they belong to the same operon ( Figure 2B). The similarity of amino acid sequences between YgfY homologs in S. oneidensis MR-1 and E. coli MG1655 was 62.5%, and that between YgfX homologs in two strains was 34.8%. EMBOSS Needle was used to analyze the similarity of amino acid sequences.
Hetero-expression of toxins in E. coli commonly causes growth defect. To further examine physiological function of YgfY and YgfX, they were overexpressed in E. coli BL21(DE3). Expression of YgfY was driven by an IPTG-induced promoter in pET-28a(+) and that of YgfX was driven by an arabinose-induced promoter in pBAD24. As the blank controls, cultures of E. coli BL21(DE3) in the early exponential phase was added with 1 mM glucose as an inhibitor to minimize leaking expression of YgfY and YgfX. After addition of inducers or glucose for 4 h, cultures expressing either YgfX or YgfY showed growth defect, and those expressing both YgfX and YgfX showed a more severe growth defect compared with the glucose-added cultures ( Figure 3A). This result did not support YgfX and YgfY functioning as a toxin and cognate antitoxin, respectively. Otherwise, expression of YgfY should rescue the growth defect caused by YgfX expression. Many proteins become toxic when highly overexpressed as published previously [18], which probably explains the growth defect observed upon overexpression of YgfX or YgfY. Overexpression of toxin YgfX (CptA) from E. coli in its native host causes change of cell morphology to the lemon shape [7]. Therefore, the cell morphology of E. coli BL21(DE3) after expressing YgfY and/or YgfX was observed. No morphological change was observed for cells expressing YgfX or YgfY or both ( Figure 3B). No phenotype was observed in corresponding deletion mutants and overexpression strains, showing that YgfY and YgfX did not follow one of criteria of the type IV TA system. Hetero-expression of toxins in E. coli commonly causes growth defect. To further examine physiological function of YgfY and YgfX, they were overexpressed in E. coli BL21(DE3). Expression of YgfY was driven by an IPTG-induced promoter in pET-28a(+) and that of YgfX was driven by an arabinose-induced promoter in pBAD24. As the blank controls, cultures of E. coli BL21(DE3) in the early exponential phase was added with 1 mM glucose as an inhibitor to minimize leaking expression of YgfY and YgfX. After addition of inducers or glucose for 4 h, cultures expressing either YgfX or YgfY showed growth defect, and those expressing both YgfX and YgfX showed a more severe growth defect compared with the glucose-added cultures ( Figure 3A). This result did not support YgfX and YgfY functioning as a toxin and cognate antitoxin, respectively. Otherwise, expression of YgfY should rescue the growth defect caused by YgfX expression. Many proteins become toxic when highly overexpressed as published previously [18], which probably explains the growth defect observed upon overexpression of YgfX or YgfY. Overexpression of toxin YgfX (CptA) from E. coli in its native host causes change of cell morphology to the lemon shape [7]. Therefore, the cell morphology of E. coli BL21(DE3) after expressing YgfY and/or YgfX was observed. No morphological change was observed for cells expressing YgfX or YgfY or both ( Figure 3B). No phenotype was observed in corresponding deletion mutants and overexpression strains, showing that YgfY and YgfX did not follow one of criteria of the type IV TA system.  YgfX (CptA) in E. coli hinders cell division through interacting with cytoskeleton proteins, FtsZ and MreB [7]. CbtA, the toxin in another type VI TA system, also interacts with FtsZ and MreB to hinder division of E. coli cells [6]. The interaction of YgfX with FtsZ and MreB was examined using bacterial two hybridization system. Compared to the negative and positive controls, no obvious interaction was detected either between YgfX and FtsZ nor between YgfX and MreB ( Figure 4A,B). In addition, no interaction between YgfY and YgfX was detected ( Figure 4C). YgfX (CptA) in E. coli hinders cell division through interacting with cytoskeleton proteins, FtsZ and MreB [7]. CbtA, the toxin in another type VI TA system, also interacts with FtsZ and MreB to hinder division of E. coli cells [6]. The interaction of YgfX with FtsZ and MreB was examined using bacterial two hybridization system. Compared to the negative and positive controls, no obvious interaction was detected either between YgfX and FtsZ nor between YgfX and MreB ( Figure 4A,B). In addition, no interaction between YgfY and YgfX was detected ( Figure 4C). YgfX (CptA) in E. coli hinders cell division through interacting with cytoskeleto teins, FtsZ and MreB [7]. CbtA, the toxin in another type VI TA system, also interact FtsZ and MreB to hinder division of E. coli cells [6]. The interaction of YgfX with Fts MreB was examined using bacterial two hybridization system. Compared to the ne and positive controls, no obvious interaction was detected either between YgfX an nor between YgfX and MreB ( Figure 4A,B). In addition, no interaction between Yg YgfX was detected ( Figure 4C).

YgfY does not Attribute to Succinate Catabolism
YgfY in S. oneidensis MR-1 shares 60.2% similarity with its homolog SdhE in S sp. SdhE in Serratia sp. functions as a flavinylation factor of SDH and fumarate red [11,19]. We examined whether growth defect of ΔygfY was attributed to succinate m olism. S. oneidensis MR-1 encodes SDH and can consume succinate. However, it c use succinate as a sole carbon source to support growth [20]. Therefore, we first com the growth of ΔygfY, Δsdh and WT in the mineral medium supplemented with lacta succinate. Lactate was added as a carbon source. There was no obvious differe growth between Δsdh and WT, while ΔygfY showed a growth defect compared wi

YgfY Does Not Attribute to Succinate Catabolism
YgfY in S. oneidensis MR-1 shares 60.2% similarity with its homolog SdhE in Serratia sp. SdhE in Serratia sp. functions as a flavinylation factor of SDH and fumarate reductase [11,19]. We examined whether growth defect of ∆ygfY was attributed to succinate metabolism. S. oneidensis MR-1 encodes SDH and can consume succinate. However, it cannot use succinate as a sole carbon source to support growth [20]. Therefore, we first compared the growth of ∆ygfY, ∆sdh and WT in the mineral medium supplemented with lactate and succinate. Lactate was added as a carbon source. There was no obvious difference in growth between ∆sdh and WT, while ∆ygfY showed a growth defect compared with WT in the mineral medium ( Figure 5A). Then, we examined succinate consumption by resting cells in the mineral medium. After incubation for 12 h, ∆ygfY showed a comparable ability to consume succinate with WT, while ∆sdhE showed an obvious defect in succinate consumption ( Figure 5B). in the mineral medium ( Figure 5A). Then, we examined succinate consumption by resting cells in the mineral medium. After incubation for 12 h, ΔygfY showed a comparable ability to consume succinate with WT, while ΔsdhE showed an obvious defect in succinate consumption ( Figure 5B). SdhE in Serratia sp. possesses an RGXXE motif that is essential for its activity. SdhE variants with substitution mutation of G16R or E19A in RGXXE motif are nonfunctional [19]. YgfY in S. oneidensis MR-1 also possess this motif ( Figure 6A). Surprisingly, the RGXXE motif is highly conserved in SDH flavinylation factor in diverse organisms such as Saccharomyces cerevisiae, Arabidopsis thaliana and Homo sapiens. Mutation of G in the RGXXE motif of this factor is closely related to hereditary paraganglioma [21]. Therefore, we examined whether the RGXXE motif is also essential for the activity of YgfY in S. oneidensis MR-1. A YgfY mutant (YgfY G16R/E19A ) was constructed with substitution of glycine SdhE in Serratia sp. possesses an RGXXE motif that is essential for its activity. SdhE variants with substitution mutation of G16R or E19A in RGXXE motif are nonfunctional [19]. YgfY in S. oneidensis MR-1 also possess this motif ( Figure 6A). Surprisingly, the RGXXE motif is highly conserved in SDH flavinylation factor in diverse organisms such as Saccharomyces cerevisiae, Arabidopsis thaliana and Homo sapiens. Mutation of G in the RGXXE motif of this factor is closely related to hereditary paraganglioma [21]. Therefore, we examined whether the RGXXE motif is also essential for the activity of YgfY in S. oneidensis MR-1. A YgfY mutant (YgfY G16R/E19A ) was constructed with substitution of glycine and glutamic acid in RGXXE motif by arginine and alanine, respectively. YgfY G16R/E19A was readily expressed in ∆ygfY ( Figure S3), while did not rescue the growth defect of ∆ygfY ( Figure 6B). These results confirmed that RGXXE motif was also essential for the activity of YgfY in S. oneidensis MR-1. These results indicated that the growth effect of YgfY should attribute to other activity rather than SDH flavinylation. Figure 5. YgfY did not involve in succinate metabolism. (A) Growth in the mineral medium with 50 mM lactate and 10 mM succinate. (B) Succinate consumption by resting cells for 12 h in the mineral medium containing 10 mM succinate. The data are the mean ± SD (n = 3). Error bars for some data points were too small to be shown. SdhE in Serratia sp. possesses an RGXXE motif that is essential for its activity. SdhE variants with substitution mutation of G16R or E19A in RGXXE motif are nonfunctional [19]. YgfY in S. oneidensis MR-1 also possess this motif ( Figure 6A). Surprisingly, the RGXXE motif is highly conserved in SDH flavinylation factor in diverse organisms such as Saccharomyces cerevisiae, Arabidopsis thaliana and Homo sapiens. Mutation of G in the RGXXE motif of this factor is closely related to hereditary paraganglioma [21]. Therefore, we examined whether the RGXXE motif is also essential for the activity of YgfY in S. oneidensis MR-1. A YgfY mutant (YgfY G16R/E19A ) was constructed with substitution of glycine and glutamic acid in RGXXE motif by arginine and alanine, respectively. YgfY G16R/E19A was readily expressed in ΔygfY ( Figure S3), while did not rescue the growth defect of ΔygfY ( Figure 6B). These results confirmed that RGXXE motif was also essential for the activity of YgfY in S. oneidensis MR-1. These results indicated that the growth effect of YgfY should attribute to other activity rather than SDH flavinylation.

YgfY Contributes to Capability of Transcription and Translation in S. oneidensis MR-1
To explore the underlying mechanism of YgfY effect on growth in S. oneidensis MR-1, the transcriptomes of ∆ygfY and WT were compared. ∆ygfY and WT were cultured to the early exponential phase and subjected to analysis of RNA sequencing. Some housekeeping genes involved in transcription, translation, cell wall synthesis, and cell division were down-regulated (Table S3) in ∆ygfY. Among them, genes encoding ribosome proteins were noticeable. Fifty out of fifty-four genes for ribosome proteins were down-regulated for more than two folds (Table S3). The result of qRT-PCR confirmed the decrease in transcription of representative genes (Figure 7). The rate of ribosome formation is an integral part of the regulation of cell growth [22]. The decreased transcription of ribosome proteins in ∆ygfY is consistent with growth defect of this mutant.
The transcription capability of ∆ygfY was compared with WT using EU-based click chemistry [10]. EU is an uracil analog, and is incorporated into nascent RNA after addition. A weaker signal was detected in ∆ygfY than in WT after addition of EU for 2 h (Figure 8), indicating that the transcription capability was impaired in ∆ygfY.
Then, the translation capacity was compared by examining the tolerance to lethal stresses that cause global protein misfolding. Previous research reports that bactericidal aminoglycosides cause mistranslation, misfolding and aggregation of nascent proteins, which correlates to reactive oxygen species (ROS)A production and consequent cell death [23,24]. Inhibitors of protein synthesis, such as chloramphenicol, block ROS-related death of bacterial cells stressed by bactericidal antibiotics [25]. Therefore, we reasoned that if YgfY deficiency causes the lowered capacity of translation, ∆ygfY should show an increase in the tolerance to these stresses. The heat shock causes protein misfolding and aggregation [26], and there is a numerical correlation between the protein denature and the death of bacterial cells after the heat shock [27]. Consistent with our presumption, ∆ygfY showed an increased tolerance to heat killing than WT ( Figure 9A). Moreover, catalase supplementation to quench ROS greatly improved the survival rate of WT, while not so much for that of ∆ygfY ( Figure 9B). the early exponential phase and subjected to analysis of RNA sequencing. Some housekeeping genes involved in transcription, translation, cell wall synthesis, and cell division were down-regulated (Table S3) in ΔygfY. Among them, genes encoding ribosome proteins were noticeable. Fifty out of fifty-four genes for ribosome proteins were down-regulated for more than two folds (Table S3). The result of qRT-PCR confirmed the decrease in transcription of representative genes (Figure 7). The rate of ribosome formation is an integral part of the regulation of cell growth [22]. The decreased transcription of ribosome proteins in ΔygfY is consistent with growth defect of this mutant. Figure 7. Transcription of representative genes that were down-regulated in ΔygfY. ΔygfY and WT were cultured at 16 °C for 12 h for transcriptome analysis (RAN-Seq). Transcription (log2(fold change)) of genes was examined using qRT-PCR and compared with RNA-seq data. Error bars indicate standard deviations of results from three biological replicates.
The transcription capability of ΔygfY was compared with WT using EU-based click chemistry [10]. EU is an uracil analog, and is incorporated into nascent RNA after addition. A weaker signal was detected in ΔygfY than in WT after addition of EU for 2 h (Figure 8), indicating that the transcription capability was impaired in ΔygfY. Then, the translation capacity was compared by examining the tolerance to lethal stresses that cause global protein misfolding. Previous research reports that bactericidal aminoglycosides cause mistranslation, misfolding and aggregation of nascent proteins, which correlates to reactive oxygen species (ROS)A production and consequent cell death [23,24]. Inhibitors of protein synthesis, such as chloramphenicol, block ROS-related death of bacterial cells stressed by bactericidal antibiotics [25]. Therefore, we reasoned that if YgfY deficiency causes the lowered capacity of translation, ΔygfY should show an increase in the tolerance to these stresses. The heat shock causes protein misfolding and aggregation [26], and there is a numerical correlation between the protein denature and the death in the tolerance to these stresses. The heat shock causes protein misfolding and aggregation [26], and there is a numerical correlation between the protein denature and the death of bacterial cells after the heat shock [27]. Consistent with our presumption, ΔygfY showed an increased tolerance to heat killing than WT ( Figure 9A). Moreover, catalase supplementation to quench ROS greatly improved the survival rate of WT, while not so much for that of ΔygfY ( Figure 9B). In addition to house-keeping genes, transcription of some functional genes was also changed, such as two genes (nrfA and nsrR) involved in nitrite reduction. Transcription of nrfA and nsrR were down-regulated in ∆ygfY (Table S3), which was also confirmed by qRT-PCR ( Figure 6). The nitrite tolerance of ∆ygfY was examined next. When exposed to nitrite at a concentration of 15× MIC, ∆ygfY showed significantly more impaired tolerance to nitrite than WT, and the impairment was readily reversed by expression of YgfY but not YgfY G16R/E19A (Figure 10). In addition to house-keeping genes, transcription of some functional genes was also changed, such as two genes (nrfA and nsrR) involved in nitrite reduction. Transcription of nrfA and nsrR were down-regulated in ΔygfY (Table S3), which was also confirmed by qRT-PCR ( Figure 6). The nitrite tolerance of ΔygfY was examined next. When exposed to nitrite at a concentration of 15× MIC, ΔygfY showed significantly more impaired tolerance to nitrite than WT, and the impairment was readily reversed by expression of YgfY but not YgfY G16R/E19A (Figure 10).

Discussion
YgfXY homologs in E. coli and Acinetobacter baumannii were previously found to form a type IV TA system pair, which could not be confirmed later in Serratia sp. by different groups [28,29]. Similarly, we are unable to reproduce the toxic activity of YgfXY homologs in S. oneidensis MR-1. On the other hand, YgfY homologs in E. coli, Serratia sp. and Acetobacter pasteurianus show the activity as assembly factor for flavinylation of SDH [30,31]. In this study, we reveal that YgfY has pleiotropic impacts in physiology of S. oneidensis MR-1, but neither as an antitoxin nor as a flavinylation factor of SDH.
Genome annotation proposes YgfY and YgfX in S. oneidensis MR-1 as a TA system presumably for two reasons: (i) YgfY shares high similarity with the antitoxin CptB in E. coli, and (ii) ygfY locates adjacent to and in front of ygfX in the genome, which might compose a two-gene operon. Antitoxin CptB and toxin CptA in E. coli compose a type IV TA system in which CptA hinders cell division and causes the morphological change of cells through interacting with FtsZ and MreB, while CptB antagonizes CptA through stabilizing FtsZ and MreB [7]. Interaction of CptB with FtsZ and MreB causes morphological

Discussion
YgfXY homologs in E. coli and Acinetobacter baumannii were previously found to form a type IV TA system pair, which could not be confirmed later in Serratia sp. by different groups [28,29]. Similarly, we are unable to reproduce the toxic activity of YgfXY homologs in S. oneidensis MR-1. On the other hand, YgfY homologs in E. coli, Serratia sp. and Acetobacter pasteurianus show the activity as assembly factor for flavinylation of SDH [30,31]. In this study, we reveal that YgfY has pleiotropic impacts in physiology of S. oneidensis MR-1, but neither as an antitoxin nor as a flavinylation factor of SDH.
Genome annotation proposes YgfY and YgfX in S. oneidensis MR-1 as a TA system presumably for two reasons: (i) YgfY shares high similarity with the antitoxin CptB in E. coli, and (ii) ygfY locates adjacent to and in front of ygfX in the genome, which might compose a two-gene operon. Antitoxin CptB and toxin CptA in E. coli compose a type IV TA system in which CptA hinders cell division and causes the morphological change of cells through interacting with FtsZ and MreB, while CptB antagonizes CptA through stabilizing FtsZ and MreB [7]. Interaction of CptB with FtsZ and MreB causes morphological change of E. coli cells. Although YgfY and YgfX are co-transcribed and deficiency of YgfY causes growth defect of S. oneidensis MR-1 (Figure 1), other lines of evidence did not support that YgfY and YgfX function as a type IV TA system. Firstly, ∆ygfYX, deficient of both YgfX and YgfY, shows the same growth defect as ∆ygfY (Figure 1). If YgfX functioned as a toxin and was antagonized by YgfY, deletion of ygfX in ∆ygfY should rescue the growth defect of ∆ygfY. Secondly, overexpression of YgfX in E. coli does not cause morphological change of cells ( Figure 3). Thirdly, YgfX does not interact with FtsZ or MtrB ( Figure 4). Overall, YgfY and YgfX do not fulfill the criteria of a type IV TA system.
The severe growth defect of ∆ygfY also cannot be attributed to SDH because ∆sdh shows an unimpaired growth ( Figure 5), although interaction between YgfY and SDH cannot be excluded. ∆ygfY shows a global change in transcriptome compared with WT ( Figure 6, Table S3). It is very unlikely that YgfY directly regulates transcription for the absence of any known DNA-binding motifs. Hence, transcriptome change for YgfY deficiency is more likely a feedback regulation of cells through other signals and regulators. Interestingly, SdhE in Serratia sp. also demonstrates pleiotropic impacts. Except for inability to use succinate, a mutant of Serratia sp. without sdhE shows a decrease in the transcription of an operon responding to synthesize an antibiotic prodigiosin [30], which is also hardly to be explained by SdhE functioning as a flavinylation factor of SDH.
Results from site mutation of YgfY provides some hints about the mechanism underlying pleiotropic impacts of YgfY. YgfY possesses a conserved RGXXE motif in flavinylation factor of SDH that is surprisingly conserved in all three kingdoms ( Figure 5). Mutation of G16 in RGXXE motif abolishes the in vivo activity of YgfY in S. oneidensis MR-1 ( Figures 5 and 7). Structure investigations show that Gly in the RGXXE motif of the flavinylation factor of SDH plays the exact same role in all investigated organisms crossing three kingdoms. Gly in RGXXE motif forms a critical hydrogen with SdhA in SDH in E. coli and humans [32,33]. Moreover, Gly in RGXXE of SdhE(CptB) can also form a hydrogen bond with FrdA in the fumarate reductase complex and is indispensable for FrdA flavinylation in E. coli [34]. Based on those investigations, we propose that YgfY likely functions as a flavinylation factor as do its homologs in diverse organisms, while it has unidentified targets that eventually impose effects on growth and nitrite tolerance. Identification of those targets are on the way to deepen our view of this family of flavinylation factors.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/microorganisms9112316/s1, Figure S1: Growth of ∆ygfY and WT at 30 • C, Figure S2: Cell morphology of ∆ygfY, ∆ygfX, ∆ygfYX and WT. Stains were cultured in LB at 16 • C for 48 h and subjected to observation under a microscope, Figure S3: YgfY expressed from its native promoter revered the growth defect of ∆ygfY. Region containing ygfY coding sequence and 285 bp upstream was cloned into pBBP1MCS5 to express ygfY from its native promoter. Resulted plasmid pBBR1_ygfY was then transformed into strains of Shewanella. Strains transformed with the empty vector pBBP1MCS5 were set as controls. The data are the mean ± SD (n = 3), Table S1: Strains and plasmids used in this study, Table S2: Primers used in this study, Table S3: Genes differentially expressed and discussed in this study.  Data Availability Statement: The data supporting the findings of this study are available within the article.

Conflicts of Interest:
The authors declare no conflict of interest.