Bacterial growth, morphology, and cell component changes in Herbaspirillum sp. WT00C exposed to high concentration of selenate

Selenium (Se) is a nonmetallic element of the chalcogens. It is primarily available in natural environments as selenate and selenite oxoanions. Although selenate/selenite reduction in many microbes is widely studied at low concentrations (<50 mM), the effects of high selenate stress on bacterial growth, morphology, and cell components have not yet been studied. In this study, the response of Herbaspirillum sp. WT00C to selenate stress at high concentration is investigated by microbiological and scanning electron microscopy (SEM) techniques as well as proteomic analysis. Bacterial growth was seriously inhibited under high selenate concentrations and its growth‐inhibitory phase was prolonged with the increase of selenate concentrations. More interestingly, this bacterium was able to recover its growth even if the selenate concentration was up to 400 mM. Its growth inhibition period shortened to 6 h when the bacterium growing in 200 mM selenate for 28 h was reinoculated to the Luria‐Bertani medium containing 200 mM selenate. The high concentration of selenate also induces marked changes in the cell dimension and surface roughness, as revealed by SEM, along with compositional changes in the cell wall shown by proteomic analysis. The bacterial growth inhibition results from the marked downregulation of the α‐subunit of DNA polymerase III and RNA helicase, whereas its growth recovery is related to its high antioxidative activities. More NADPH synthesis and the upregulation of thioredoxin reductase and GPx are beneficial for Herbaspirillum sp. WT00C to establish and maintain a balance between oxidant and antioxidant intracellular systems for defending selenate toxicity. This study is an important contribution to understanding why Herbaspirillum sp. WT00C survives in a high concentration of selenate and how the bacterial cells respond physiologically to selenate stress at high concentration.


| INTRODUCTION
Selenium (Se) is a member of the chalcogens. As a nonmetallic element, it was recognized to be an essential trace element for both plants and mammals sixty-two years ago [1]. At a low dosage, Se stimulates the growth of the plant, whereas at high dosages it causes plant damage [2][3][4]. The deficiency of Se is thought to be associated with over 40 human diseases [5,6] but the excessive intake of Se seriously damages human health [7]. Medical research also shows that Se is effective against cancer [8,9]. In nature, Se usually occurs in organic and inorganic forms. Its organic form (e.g., selenocysteine, selenomethionine, or selenoprotein) mainly presents in living organisms where organic selenocompounds can be metabolized [8,9]. However, its inorganic form (e.g., selenate and selenite oxoanions) primarily exists in natural environments. Selenate (SeO 4 −2 ) and selenite (SeO 3

−2
) are water-soluble so that they have potential mobility and bioavailability in the environment [10]. Soluble Se 6+ and Se 4+ are able to be reduced to insoluble nontoxic elemental selenium (Se 0 ) by reducing agents or microbes. The reduction of selenate and selenite by microbes to elemental selenium is an effective way to remove them from contaminated soil, water, and drainage [11]. Se-nanoparticles (SeNPs) are also frequently achieved via reduction of selenate or selenite by using biogenic synthesis [12], and the synthetic SeNPs have wide applications in semiconductor technology, electronic engineering, glass, and rubber industries, as well as biomedicine.
Many microbes were found to have the capability of reducing selenate/selenite to form elemental selenium, and the mechanisms of redox reaction, SeNP synthesis, as well as potential application, have been wildly studied [13][14][15][16][17][18][19][20][21][22]. Nevertheless, most of these studies were performed under low selenate/selenite concentrations (<50 mM). Due to the low tolerance of most microbia, the effects of high concentration of selenate on bacterial growth and metabolism have not drawn public attention. Herbaspirillum sp. WT00C, isolated from the tea plant (Camellia sinensis L), was a novel member of the genus Herbaspirillum and successfully incubated in Luria-Bertani (LB) medium under laboratory conditions [23]. Its genome was also sequenced and deposited in the GenBank database (Acc#: KV880769.1) [24]. As shown in Figure 1, this bacterium has an intact selenate reduction pathway in its genome, and it is indeed able to reduce selenate (Se 6+ ) to elemental selenium (Se 0 ). Electron microscopy and energy-dispersive X-ray spectroscopy confirmed that this bacterium not only reduced selenate to form zerovalent SeNPs but also secreted SeNPs outside bacterial cells, where SeNPs grew large to form Se-nanospheres and then crystallized to form selenoflowers [25]. More interestingly, bacterial growth was seriously inhibited in the early stage but its growth subsequently recovered and finally approached the level of bacterial cells untreated with selenate when the concentration of selenate was larger than 50 mM. This bacterium was able to recover its growth and reduce selenate to form elemental selenium even if the selenate concentration was up to 200 mM [25]. In this study, we investigated the response of Herbaspirillum sp. WT00C to selenate stress under high concentrations, and found steady prolongation of the bacterial growth inhibition period with increasing selenate concentrations and changes in bacterial morphology, extracellular structure, and cell function as well as the soluble proteome. Possible mechanisms for changing bacterial morphology, cell structure, and function under high selenate concentrations were also discussed.

| Bacterial strain and chemical reagents
Herbaspirillum sp. WT00C was isolated from tea plants in Wuhan city, China [23] and deposited in the China Center for Type Culture Collection (CCTCC AB 2018017 T ). This strain was routinely cultured in NB or LB medium according to the method reported previously [23,25]. Inorganic and organic chemical reagents were purchased from Zhong Ke (Shanghai, China). Culture media came from Oxoid and Amresco, and selenate/ selenite was purchased from Xiya Reagent.

| Bacterial growth curves at different concentrations of sodium selenate
Herbaspirillum sp. WT00C was inoculated into 5 ml LB medium and incubated at 37°C, 200 rpm overnight. Then, The selenate reduction pathways of Herbaspirillum sp. WT00C. Arabic numbers in each red frame represent the EC-number of specific enzymes. EC, enzyme commission 0.1 ml of the culture was transferred into 10 ml fresh LB broth and incubated at 37°C, 200 rpm until the OD 600 value approached 0.8. The activated bacterial culture was inoculated with a ratio of 1:100 into 200 ml LB broth containing 0-1,000 mM Na 2 SeO 4 and incubated at 37°C, 200 rpm for 88 h. Here, the LB broth without the addition of Na 2 SeO 4 (0 mM) was used as control. During the incubation period, 1 ml of the bacterial culture was respectively taken out at specific intervals throughout the whole process of cultivation and the optical density at a wavelength of 600 nm was measured on a Shimadzu UV/ visible spectrophotometer (UV-2550). Bacterial growth curves at different selenate concentrations were obtained through the mapping between OD 600 values on the vertical axis and incubation times on the horizontal axis.
To know the growth traits of bacterial cells growing in the LB medium containing 200 mM Na 2 SeO 4 , 0.5 ml of the bacterial culture incubated in LB medium containing 200 mM Na 2 SeO 4 at 37°C for 28 h was again inoculated into 50 ml LB medium containing 0 and 200 mM Na 2 SeO 4 , and cultured at 37°C, 200 rpm for 34 h. Meanwhile, Herbaspirillum sp. WT00C incubated in LB medium under the same conditions was used as control. The optical density at 600 nm was measured as described above.

| The half-maximal inhibitory concentration (IC 50 ) measurement of selenate and selenite
To test the toxicities of selenate and selenite toward Herbaspirillum sp. WT00C, the IC 50 [26] for selenate and selenite was determined. Herbaspirillum sp. WT00C was activated in 5 ml LB medium at 37°C until the OD 600 value approached 0.8. Then, 0.1 ml of the bacterial culture was inoculated into 10 ml LB medium, respectively, containing 0-200 mM Na 2 SeO 4 or Na 2 SeO 3 and incubated at 37°C, 200 rpm for 12 h. Finally, the optical density at a wavelength of 600 nm was measured on a Shimadzu UV/visible spectrophotometer (UV-2550). The mean calculated from three independent tests was used to reckon the IC 50 value via mapping between the relative survival rate (%) and selenate/selenite concentrations based on the definition of the IC 50 .

| Scanning electron microscopy
(SEM) observation of bacterial morphology SEM was used to observe bacterial cells in different periods, as reported previously [20,25]. Briefly, Herbaspirillum sp. WT00C was inoculated into 5 ml LB broth and incubated at 37°C overnight. The activated bacterial culture was inoculated with a ratio of 1:100 into a 25 ml LB broth containing 0 or 200 mM selenate and incubated at 37°C. Here, Herbaspirillum sp. WT00C growing in LB medium without the addition of selenate was used as control. In the control, each sample (5 ml) was, respectively, collected at 0, 10, and 25 h. To keep bacterial cells growing at the same phase as the control, 5 ml of the sample from the bacterial cells growing in the LB medium with 200 mM selenate were collected at 0, 12, 22, and 28 h. Similarly, the samples (5 ml for each) for those bacterial cells, coming from Herbaspirillum sp. WT00C growing in LB medium containing 200 mM selenate for 28 h and growing again in the LB medium containing 0 and 200 mM selenate, were, respectively, collected at 0, 6, 10, 25 h and 0, 6, 17, 25 h. Bacterial cultures for each sample were centrifuged at 6,000g for 15 min and the pellets were collected. Bacterial cells of each sample were fixed in 2.5% glutaraldehyde for 30 min, rinsed three times in 100 mM phosphate buffer (pH 7.2), and dehydrated in an ethanol series (20%, 50%, 70%, 90%, and 100% ethanol). The ethanol was then displaced by isoamyl acetate. Each sample was mounted onto microscope slides and dried using a BAL-TEC CPD030 critical point drying apparatus. Finally, all samples were sputter-coated with gold to a thickness of approximately 20 nm and observed under a JSM7100F scanning electron microscope (JEOF, Tokyo, Japan). Meanwhile, cell sizes were measured at a magnification of ×50,000 and the values were represented as the mean from five different cells.

| Analysis of bacterial proteome
Herbaspirillum sp. WT00C was activated in LB medium at 37°C overnight and then inoculated into 5 ml fresh LB medium with a ratio of 1:100 and incubated at 37°C until the OD 600 value approached 0.8. A total of 2.5 ml of the bacterial culture was inoculated into 250 ml LB medium containing 0 or 200 mM sodium selenate and incubated at 37°C, and then bacterial cells were, respectively, collected at 0, 12, and 26 h for 0 mM selenate incubation and 0, 12, 23, and 30 h for 200 mM selenate incubation. The bacterial culture growing in 200 mM selenate for 28 h was again inoculated with a ratio of 1:100 into 250 ml LB medium containing 200 mM Na 2 SeO 4 and incubated at 37°C for 34 h. Bacterial cells growing at different periods (6,18, and 24 h) were collected. All cells were harvested by centrifuging at 8,000 rpm, 4°C for 15 min and washed three times with PBS (pH 7.2) to remove all residual medium. Finally, bacterial cells were suspended in 50 mM Tris-HCl (pH 8.0) for protein preparation.
Soluble proteins were prepared using a Qproteome Bacterial Protein Prep Kit (Qiagen) according to the manufacturer's instructions, and protein concentration was determined using a Pierce™ BCA Protein Assay Kit (Pierce) as per the manufacturer's instruction. After each protein sample was prepared according to the universal sample preparation method [27], the shotgun proteomic analysis via PatternLab tool [28] was completed by Shanghai Applied Protein Technology (Shanghai, China). The proteins with significant (p < .05) upregulation (>1.2 fold) or downregulation (<0.83-fold) were selected via Fisher's exact test [29,30], and then protein identification and quantitative analysis [31,32] were executed by searching the database P18011-Herbaspirillum-NCBI-174124-20180423.fasta with the software Mascot 2.3 (https://www.matrixscience.com). The numbers of bacterial proteins identified by this method were in the range of 1,000-2,000. Functional annotations of those proteins with significant differences between the control and different treatments were also performed by using Blast2GO software (https://www.blast2go.com) [29,33] and the Kyoto Encyclopedia of Genes and Genomes (KEGG; http://www.kegg.jp) [30,34].

| Reverse transcriptase polymerase chain reaction (RT-PCR)
Semiquantitative RT-PCR [35] was used to assess the relative expression quantity of seven genes of Herbaspirillum sp. WT00C, in which glutathione synthase (gs), glutathione S-transferase (gst), glutathione reductase (gr), glutathione peroxidase (gpx), thioredoxin reductase (TrxB) were involved in glutathione metabolism or antioxidant and redox regulation, whereas cystathionine β-lyase (cbl) and sulfate adenylyltransferase subunit (cysD) were randomly selected as controls. Meanwhile, the 16S ribosomal RNA (rRNA) gene was used as an internal reference. Herbaspirillum sp. WT00C was incubated in the LB medium containing 0, 20, 50, and 100 mM Na 2 SeO 4 at 37°C for 12 h, and then bacterial cells were harvested by centrifugation at 8,000 rpm, 4°C for 15 min. Total RNA was prepared using an RNAiso™ Kit, and complementary DNA (cDNA) was synthesized using a Prime Script™ RT Reagent Kit (TaKaRa Bio Inc.) with random hexamer primers, as per manufacturer's instruction. Next, 3 μg/μl cDNA was used as a template, and the transcripts of genes were amplified with the specific primers listed in the Table 1 at 94°C for 5 min, followed by 30 cycles of 30 s at 94°C, 30 s at 48°C, 30 s at 72°C, and a final incubation at 72°C for 5 min. Meanwhile, DNA amplification with the primers 16S rRNA-F and 16S rRNA-R specific for the 16S rRNA gene of Herbaspirillum sp. WT00C was also executed under the same conditions. The DNA products amplified via PCR were electrophoretically separated on 1.2% agarose gel and observed under ultraviolet light after staining with ethidium bromide. The fluorescence intensity for each DNA band was measured using the software Gel-Pro Analyzer 4.0. The optical density ratio of DNA bands between each sample and the internal reference was calculated as the value of relative expression quantity for each desired gene. Finally, the data obtained from semiquantitative RT-PCR with three replicates were analyzed by using SPSS 17.0 software (IBM SPSS Inc., Chicago, IL), and a p < .05 was considered significant.
its growth (see Figure 2a). When the selenate concentrations were ≥600 mM, bacterial growth was not detectable in 88 h. As expected, bacterial growth was completely inhibited in 12 h, and then bacterial cells gradually recovered their growth after 12 h incubation and finally approached the level of the cells growing in LB medium without selenate at 30 h when 200 mM Na 2 SeO 4 was added to LB broth [9]. Herbaspirillum sp. WT00C had a 12-h growth-inhibitory period and a growth recovery period including exponential and stationary phases when it grew in 200 mM Na 2 SeO 4 as shown in Figure 2a. In its growth-inhibitory period, bacterial growth was seriously inhibited by a high concentration of selenate. More interestingly, when the bacterial cells growing in 200 mM Na 2 SeO 4 for 28 h were transferred into the LB medium plus 200 mM Na 2 SeO 4 , the growth-inhibitory period was shortened to 6 h, which was only a half of that for the original cells (see Figure 2b). This result suggested that continuous cultivation under the high concentration of selenate could improve bacterial tolerance to selenate stress. In addition, the growth curve was the same as that for the original cells when the same bacterial cells were returned back to the LB medium without selenate (see Figure 2b). Herbaspirillum sp. WT00C, as shown in Figure 1, can reduce selenate (Se 6+ ) to selenite (Se 4+ ), and then the latter can further reduce to form the elemental selenium (Se 0 ) or enter into the pathway of selenoprotein synthesis. In addition, both selenate and selenite, as selenium oxyanions, were thought to be quite toxic [29]. Was bacterial growth inhibited by selenate or selenite or both together in this study? To test the toxicities of selenate and selenite to Herbaspirillum sp. WT00C, an IC 50 was determined. Bacterial toxicity tests gave the IC 50 value of 37 mM for selenate and 35 mM for selenite. The minimal inhibitory concentration (MIC) values, estimated roughly  Figure 3, were 200 mM for selenate and 150 mM for selenite. Two IC 50 values indicated that both selenate and selenite were similarly toxic to Herbaspirillum sp. WT00C. Selenate, together with selenite, inhibited the growth of Herbaspirillum sp. WT00C in its growth-inhibitory period. Here, this raised the question of why bacterial cells were able to recover their growth after a long growth inhibition period. One possibility was that the selenate/selenite concentration in the medium might decrease to less than the MIC value after the growth inhibition period. Another possibility was that bacterial cells might develop certain physiological or genetic mechanisms against the toxicity of selenate/selenite under the high concentration of selenate. To test the first possibility, selenate and selenite concentrations in the culture of the bacterium growing in 200 mM selenate were examined according to the previous method [36]. The results showed that 198 ± 0.64, 194 ± 0.68, and 189 ± 0.52 mM selenate and 0.16 ± 0.04, 0.22 ± 0.06, and 0.31 ± 0.08 mM selenite existed, respectively, in the bacterial culture at the 5-, 12-, and 20-h incubations. At the 12-h incubation, 194 mM selenate remained in the culture, which was close to its MIC value. Obviously, such a small change of selenate concentration might not be the reason for bacterial growth recovery after 12 h.

| Bacterial morphology at high concentration of selenate
The cell morphology of Herbaspirillum sp. WT00C growing in the LB broth with 200 mM selenate or without selenate was also observed by SEM. As shown in Figure 4, the bacterial cells growing in LB medium appeared as a conjoined multicellular morphology at 6-h incubation, suggesting that the bacterial cells entered the proliferative state. Although cell sizes in the logarithmic growth and stationary phases were slightly different, the size change was quite small. The bacterial cells in the stationary phase (26-h incubation) appeared as a lanky stick with the size of around 1.17 ± 0.04 μm × 327 ± 6 nm. However, obvious morphological changes occurred when the cells grew in the LB broth containing 200 mM selenate. The bacterial cells at 12-h incubation did not proliferate and the cell surface was quite rough. At 20-h incubation, the bacterial cells with a lanky stick shape had entered the proliferative state. It was notable that bacterial cells in the midlogarithmic phase appeared as a thick-rod shape, and this morphology remained until the stationary phase. The cell size at 30-h incubation was 1.19 ± 0.08 μm × 580 ± 17 nm, in which the cell width was nearly twofold of the cells growing in LB medium. As compared with the cells growing in LB medium to stationary phase, the surface of bacterial cells growing in 200 mM selenate was relatively smooth, to which only Senanospheres adhered. More interestingly, when the bacterial cells growing in the LB medium containing 200 mM selenate for 28 h were inoculated back to the same medium containing 200 mM selenate, the bacterial cells had maintained the thick-rod shape over the whole growth period. The cell size measured at 26-h incubation was 1.30 ± 0.12 μm × 685 ± 28 nm, and the cell surface was less smooth as compared with the bacterial cells growing for the first time in 200 mM selenate. When the bacterial cells growing in the LB medium plus 200 mM selenate for 28 h were returned to LB medium and incubated at 37°C for 24 h, the size of bacterial cells in the stationary phase was 1.32 ± 0.16 μm × 491 ± 12 nm and the cell surface returned back to roughness again, as  Figure 2b). Gene Ontology annotation and functional enrichment analysis (see Figure 5a (B1/A0)) showed the proteins displaying significant difference were mainly involved in organic, carboxylic and α-amino acid catabolism, carbohydrate transport, secondary metabolic and toxin metabolic processes, short-chain fatty acid, cellular amino acid, and small molecule catabolic processes, oxidoreductase, and RNA helicase activities, and so forth, the major upregulated/downregulated proteins identified by KEGG function annotation are shown in  (227-fold upregulation) catalyzing the formation of NADPH could help bacterial cells to establish and enhance antioxidant capacity. In the downregulated proteins, the more important protein was the α-subunit of DNA polymerase III [EC 2.7.7.7], 74.3% less α-subunit of DNA polymerase III could severely affect DNA polymerase III activity, which might result in hindering bacterial DNA replication. A total of 57.5% less DnaA (chromosomal replication initiator protein) might also affect bacterial DNA replication. In addition, 99.9% less RNA helicase should affect bacterial transcription.
Proteomic comparison between the bacterial cells growing in 200 mM selenate for 23 h and 0 mM selenate for 12 h (at the midlogarithmic phase) was also executed. Ninety-two proteins including the upregulation of 45 and the downregulation of 47 displayed a statistically significant difference (p < .05) in abundance. As shown in Figure 5b (B2/A1), those proteins with significant differences were associated with the iron coordination entity, iron chelate, siderophore transport, iron ion, and transition metal ion binding, receptor, and molecular transducer activities, alcohol dehydrogenase, and oxidoreductase activities, as well as the external encapsulating structure, outer membrane, and cell envelope, and so forth. Table 2B also showed that the upregulated proteins involving NADPH formation were malonate-semialdehyde/methylmalonate-semialdehyde  This enzyme catalyzed the synthesis of lipopolysaccharide (LPS) which was the major component of the outer membrane of Gram-negative bacteria. A total of 99.94% less lipid-A-disaccharide synthase could result in the shortage of LPS on the outer membrane of bacterial cells, which would affect the structure of the bacterial cell wall. Meanwhile, the marked downregulation (99.9% less) of the branched-chain amino acid transport system permease, iron (III) transport system substrate-binding protein, D-methionine transport system substrate-binding protein, cell envelope biogenesis protein TolA, and chemotaxis proteins CheZ and CheV indicated that the function of the bacterial cell wall and membrane were abnormal. After the bacterial cells entered the growth recovery period, the difference of proteins associated with DNA replication and transcription, such as the α-subunit of DNA polymerase III, DnaA, and RNA helicase, was not observed. This suggested that the bacterium had recovered its capacity for DNA replication. At this stage, the bacterium generated enough energy for cell growth via enhancing glycolysis and gluconeogenesis. Nevertheless, its external encapsulating structure and outer membrane were not in normal status due to the insufficiency of LPS biosynthesis under a high concentration of selenate. When bacterial cells grew in 200 mM selenate for 30 h or in 0 mM selenate for 25 h, both cells approached the saturation phase. Proteomic comparison (see Figure 5c (B3/A2) and  were significantly different (p < .05), in which 11 proteins were upregulated but eight proteins were downregulated. These proteins with different abundance were related to gas, oxygen, pentose, and D-ribose transport, "de novo" L-methionine biosynthesis and the outer membrane, and so forth. Thus, the bacterial cells growing to the saturation phase in 200 mM selenate appeared to restore partially physiological status although the cell outer membrane components were still different from that of those cells growing in the medium without selenate.
To understand why the bacterial cells growing in 200 mM selenate for 28 h showed stronger survivability under high oxidative stress, proteomic comparison between the bacterial cells growing and regrowing in 200 mM selenate for 12 and 8 h (late growth inhibition, C1/B1) was performed. Twenty-four proteins showed significant differences (p < .05) in abundance, which were mainly associated with cellular response to an extracellular stimulus, liposaccharide metabolic and carbohydrate biosynthetic processes as well as enzymatic activities related to the RNA transcription process. The markedly upregulated proteins were lipid- showed that those different proteins in abundance were mainly related to response to extracellular stimulus, short-chain fatty acid biosynthetic and branched-chain amino acid metabolic processes, rRNA binding, and ATPase and oxidoreductase activities, in which lipid-A-disaccharide synthase was markedly upregulated (4,039-fold). The continuous upregulation of lipid-A-disaccharide synthase in different growth phases indicated that the bacterium regrowing in 200 mM selenate at least restored the partial structure of cell envelope.

| Effects of selenate on transcription
of five genes involving in GSH metabolism and redox regulation Glutathione (GSH) and thioredoxin reductase (TrxR) were thought to play an important role in determining the redox state of cells [37,38], and GSH was also involved in selenate/selenite reduction [12,39,40]. Thus, five genes gs, gst, gr, gpx, and trxB involving GSH metabolism, antioxidant, and redox regulation were selected to test if selenate affected the expression of these genes at the transcriptional level. The results obtained from semiquantitative RT-PCR are shown in Figure 6. The relative expression quantities of five genes among the seven genes were significantly changed (p < .05) when sodium selenate was added into bacterial media. trxB, gpx, and gst genes were markedly upregulated, whereas gr and gs genes were significantly downregulated. cysD and cbl genes, irrelevant with the redox state of cells, did not display significant change at the transcriptional level. It was noted that the upregulation of trxB and gpx genes would benefit antioxidant regulation of bacterial cells. Nevertheless, the upregulation of two genes and the downregulation of two genes among four F I G U R E 6 Relatively transcriptional quantities of seven genes under different selenate concentrations. All data were collected from three independent semiquantitative reverse transcription polymerase chain reaction, and a significant difference was marked with *p < .05 or **p < .01 genes related to GSH metabolism implied that GSH output might not closely associate with bacterial antioxidation in Herbaspirillum sp. WT00C.

| DISCUSSION
Selenium, as a trace element, is essential for the growth and development of plants and mammals as well as human nutrition. As a growth factor, selenium is involved in thyroid hormone homeostasis, immunity, and fertility and has powerful antioxidant and anticancer properties [41]. In the natural environment, selenium is normally available as selenate (Se 6+ ) and selenite (Se 4+ ) oxyanions, and both SeO 4 −2 and SeO 3 −2 are more soluble but toxic [12]. Toxic effects were initially thought to be due to its nonspecific incorporation into proteins by replacing sulphur in Cys and Met residues [31]. At present, toxic effects are widely thought to result from its interaction with essential sulfhydryl-containing enzymes and structural proteins, which led to oxidative stress at high concentrations [37,38,42]. According to this theory, the oxidative stress of bacterial cells should be enhanced with the increase of selenate concentration. In our study, the inhibitory time of bacterial growth was indeed prolonged with the increase of selenate concentrations. During growth inhibition, the marked downregulation of α-subunit of DNA polymerase III and those proteins participating in carboxylic, amino acid, and fatty acid catabolism led to the disability of bacterial cells in DNA replication and physiological metabolism. More surprisingly, the bacterium recovered its growth after undergoing growth inhibition for more than 12 h when the concentration of Na 2 SeO 4 was ≥200 mM. Bacterial growth recovery implies that bacterial cells have not died yet under the high selenate concentration. Although high oxidative stress inhibits cell growth and proliferation, the bacterium in the growth-inhibitory period may also establish its own antioxidant mechanism to balance the oxidative stress in cells. If Herbaspirillum sp. WT00C did not establish an antioxidative mechanism, it could not recover its growth under high oxidative stress. Both NADPH and GSH concentrations are thought to be related to the redox state of cells because TrxR and GR transfer reducing equivalents from NADPH to thioredoxin (Trx) and glutathione disulfide, respectively, and form Trx(SH) 2 and GSH acting as effective intracellular antioxidants [43]. A balance between oxidant and antioxidant intracellular systems is vital for cell function, regulation, and adaptation to diverse growth conditions [37,42,44]. In plants and mammals, TrxR in conjunction with Trx, as a ubiquitous oxidoreductase system, plays an important role in antioxidant and redox regulation [38,45], and this oxidoreductase system has been positioned at the core of cellular thiol redox control and antioxidant defense based on the properties of TrxR and the functions of Trx [38]. During the bacterial growth recovery period, both TrxR and the enzymes catalyzing the formation of NADPH exhibited an enhanced expression. The induction of these antioxidant proteins in Herbaspirillum sp. WT00C confirms again that the toxic effect of selenate is the formation of reactive oxygen species as reported in Rhodobacter sphaeroides [13]. Although GSH does not show an obvious increase, more NADPH formation and the upregulation of TrxR and GPx may help Herbaspirillum sp. WT00C to establish and maintain a balance between oxidant and antioxidant intracellular systems for defending selenate toxicity. In Herbaspirillum sp. WT00C, the oxidative stress caused by high concentration of selenate not only affected bacterial physiological metabolism and cell function but also changed cell morphology. Many metal elements (e.g., chromium (Cr) and manganese (Mn)) were reported to change bacterial morphology [46,47]. For instance, high concentration of Mn 2+ and chromate caused fourfold to eightfold changes in cell size and surface roughness increase [46,47]. Herbaspirillum sp. WT00C mainly changed bacterial cell width or height and decreased cell surface roughness under high selenate concentrations. Cell enlargement and surface-roughness change may not be considered as specific responses to selenate stress, as these variances have been also observed upon exposure to other stress conditions (e.g., high salt and ultraviolet radiation) [48,49]. Nevertheless, it is certain that changes in cell size and surface roughness enable Herbaspirillum sp. WT00C to escape the killing effects of the oxidative stress caused by high concentration of selenate. In Herbaspirillum sp. WT00C, the variation of surface roughness is possibly associated with LPS biosynthesis. When the bacterial cells growing in 200 mM selenate for 28 h were again incubated in 200 mM selenate, cell surface roughness was partially restored due to the continuous upregulation of lipid-A-disaccharide synthase. Once the same bacterial cells were cultured back to LB medium, cell surface roughness was almost recovered to the level of the original cells. These results illustrate that the variation of cell surface roughness merely is bacterial physiological response to selenate stress.
The encapsulating structure and outer membrane in both bacterial cells growing and regrowing in 200 mM selenate are obviously different from those of the original cells. The variation of those cell components ultimately leads to functional changes in bacterial cells, especially in the bacterial membrane system. For instance, cellular response to extracellular stimulus, ion, and molecule transports, as well as receptor activity is markedly affected by high selenate stress. Obviously, the detailed molecular mechanism for cell functional change is still needed to be further studied. Nevertheless, the bacterial cell changing morphology may perhaps be used to replace the original cells for the production of elemental selenium (Se 0 ) at high selenate concentration due to its shorter growth-inhibitory period and high antioxidative activity.

ACKNOWLEDGMENTS
This study was supported by a grant (2016YFD0200905) from the Ministry of Science and Technology of the People's Republic of China and also assisted by an innovation-driven power program of the Hubei Association for Science and Technology, Hubei province, China.