Multiple exposures to high concentrations of selenate significantly improve selenate tolerability, red elemental selenium (Se0) and selenoprotein biosynthesis in Herbaspirillum camelliae WT00C

Herbaspirillum camelliae WT00C is a gram-negative endophyte isolated from the tea plant. It has an intact selenate metabolism pathway but poor selenate tolerability. In this study, microbiological properties of the strain WT00C were examined and compared with other three strains CT00C, NCT00C and NT00C, which were obtained respectively from four, six and eight rounds of 24-h exposures to 200 mM selenate. The selenate tolerability and the ability to generate red elemental selenium (Se0) and selenoproteins in H. camelliae WT00C has significantly improved by the forced evolution via 4–6 rounds of multiple exposures a high concentration of selenate. The original strain WT00C grew in 200 mM selenate with the lag phase of 12 h and 400 mM selenate with the lag phase of 60 h, whereas the strains CT00C and NCT00C grew in 800 mM selenate and showed a relatively short lag phase when they grew in 50–400 mM selenate. Besides selenate tolerance, the strains CT00C and NCT00C significantly improved the biosynthesis of red elemental selenium (Se0) and selenoproteins. Two strains exhibited more than 30% selenium conversion efficiency and 40% selenoprotein biosynthesis, compared to the original strain WT00C. These characteristics of the strains CT00C and NCT00C make them applicable in pharmaceuticals and feed industries. The strain NT00C obtained from eight rounds of 24-h exposures to 200 mM selenate was unable to grow in ≥ 400 mM selenate. Its selenium conversion efficiency and selenoprotein biosynthesis were similar to the strain WT00C, indicating that too many exposures may cause gene inactivation of some critical enzymes involving selenate metabolism and antioxidative stress. In addition, bacterial cells underwent obviously physiological and morphological changes, including gene activity, cell enlargement and surface-roughness alterations during the process of multiple exposures to high concentrations of selenate.


Introduction
As a nonmetallic element of the chalcogens, selenium occurs commonly in materials of the earth's crust and is an essential trace element for both plants and mammals. In nature, selenium is usually available in inorganic and organic forms. Its organic form includes selenocysteine, selenomethionine, selenodiglutathione, and selenoproteins, which are metabolized in the biological system (Ip et al. 2000;Miller et al. 2001). The inorganic form often appears as four different valence states: selenate (SeO 4 2− , Se 6+ ), selenite (SeO 3 2− , Se 4+ ), selenium (Se 0 ), and selenide (Se 2− ). Selenide is a volatile compound often using as an ingredient to synthesize selenoproteins in living organisms. In contrast, selenate and selenite are watersoluble compounds possessing potential mobility and bioaccessibility in natural environments (Chasteen and Bentley 2003). In an oxidative environment, selenate and selenite are numerically dominant, but both are toxic to plants and animals at high dosages. In the reductive environment, selenium tends to have thermodynamic stability, but it is poorly water-soluble (Strawn et al. 2002). Under highly reducing conditions, elemental selenium (Se 0 ) can be further reduced to selenide, which in turn binds with metals and organics to form metal selenides or organoselenides, respectively (Nancharalah and Lens 2015). Owing to its high photoconductivity and strong adsorption capacity, selenium has been widely applied in photoelectricity, semiconductors, and biosensors (Liu et al. 2004a, b;Wadhwani et al. 2016). In addition, elemental selenium also possesses low toxicity (26-fold less toxic than SeO 2 ). Its antioxidation and biological reactivity have displayed a potential application in immunomodulation, anti-aging, and anticancer anti-microbial infection (Wadhwani et al. 2016), etc. At present, selenium is mainly produced by physical and chemical methods, such as hydrothermal/solvothermal techniques, pulse laser ablation, chemical vapor deposition, acid decomposition, and catalytic reduction using ascorbic acid, glucose, sulfur dioxide, sodium dodecyl sulfate, etc. (Xi et al. 2006;Liu et al. 2004a, b;Quintana et al. 2002;Gatea et al. 2002;Iranifam et al. 2013;Dwivedi et al. 2011). However, these physical/chemical methods require high temperature, acidic pH and harsh chemicals (Dwivedi et al. 2011), which lead to high energy consumption and nonenvironmental friendliness. Furthermore, the insecurity of selenium products limits its application in the biopharmaceutical area. To overcome the disadvantage of physical/ chemical methods, biomass synthesis has been developed for producing selenium (Wadhwani et al. 2016;Husen and Siddiqi 2004). It is well known that the selenium nanoparticles prepared from biological material are less toxic than the bulk selenium nanoparticles prepared from chemicals. The biomolecules in the extracts of bacteria, algae, dry fruits, and plants have been used as reducing agents and stabilizers for selenium nanoparticles, and green synthesis of selenium nanoparticles has also been achieved by using the extracts of raisin (Vitis vinifera) and cayenne pepper (Capsicum annuum) (Sharma et al. 2014;Li et al, 2007). It has also been found that many different species of eubacteria and archaeobacteria effectively reduce selenate or selenite to elemental selenium (Se 0 ) (Li et al. 2007(Li et al. , 2014Gurunathan et al. 2009;Smith and Mondaca 2012;Torres et al. 2016;Oremland et al. 2004Oremland et al. , 1994Avendano et al. 2016;Lortie et al. 1992;Lampis et al. 2014). Thus, biogenic methods have been developed to synthesize red elemental selenium in order to circumvent the effect of toxic chemicals in the fabrication of selenium nanoparticles. The production of red elemental selenium by microbes should be one of the best ways to produce selenium because it can satisfy the requirements of industrial mass production. Nevertheless, the selenate/ selenite tolerance of most microbes is low, so that selenium production by microbes is inevitably limited. Developing new microbial methods to prepare nano-selenium has gained more attention, and it is also an effective way to realize environmentally friendly, low-cost, and large-scale production.
Herbaspirillum camelliae WT00C, as a novel member of the genus Herbaspirillum, was isolated from the tea plant (Camellia sinensis L) and incubated successfully in Luria-Bertani (LB) medium under laboratory conditions Liu et al. 2020). Its genome was also sequenced and deposited in the GenBank database (Acc#: KV880769.1) (Cheng et al. 2017) As shown in Fig. 1, this bacterium has an intact selenate reduction pathway in its genome, in which selenate (Se 6+ ) is reduced to selenite and then to selenium (Se 0 ) or selenoproteins. Electron microscopy and energy-dispersive X-ray spectroscopy confirmed that this bacterium not only reduced selenate to form zerovalent selenium-nanoparticles (SeNPs) but also secreted SeNPs outside bacterial cells, where SeNPs grew to form Se-nanospheres or seleno-flowers (Xu et al. 2020). The red elemental selenium (Se 0 ) purified from the culture of H. camelliae WT00C markedly improved fish immunoactivities (Tian et al. 2021). In this study, we aim to develop a bacterial strain displaying strong selenate tolerability with a short lag phase and holding strong capability of red elemental selenium and selenoprotein biosynthesis. After 4-6 rounds of 24-h exposures to 200 mM selenate, the strains CT00C and NTC00C were successfully obtained. Two strains able to grow in 800 mM selenate had a short lag phase when they grew in ≤ 400 mM selenate and exhibited more than 30% selenium conversion efficiency and a 40% increase in selenium integration into proteins.

Bacterial strain, growth medium, and chemicals
Herbaspirillum camelliae WT00C was isolated from the tea plant in Wuhan city, China Liu et al. 2020) and deposited in China Center for Type Culture Collection (CCTCC AB 2018017 T ). The strain WT00C was usually incubated at 37 °C in Luria-Bertani (LB) medium (5 g yeast extract, 10 g peptone, 1.5 g NaCl in 1 L dH 2 O, pH 7.0). The strains CT00C, NCT00C, and NT00C obtained in this study were routinely incubated in the LB medium containing 200 mM selenate. Selenate, selenite, inorganic and organic chemicals were purchased from Xiya Reagent (Shandong, China) and Zhong Ke (Shanghai, China), respectively.

Forced evolution by multiple exposures to high concentrations of selenate
Forced evolution was performed by multiple exposures to 200 mM selenate to obtain the bacterial strain growing in high concentrations of selenate. H. camelliae WT00C was firstly activated by inoculating into 5 mL LB medium and growing at 37 °C overnight. The activated bacterial cells were inoculated with the ratio of 1: 100 (v/v) into 25 mL of the LB medium containing 200 mM selenate and incubated at 37 °C, 200 rpm for 24 h. 0.25 mL of the 24 h-incubated culture was transferred to 25 mL of fresh LB medium containing 200 mM selenate and incubated at 37 °C, 200 rpm for another 24 h. This process was successively repeated eight times in total. Finally, the bacterial culture obtained from each exposure was respectively collected and stored in a − 80 °C freezer for subsequent study.

Determination of growth curves
Bacterial strains from forced evolution were respectively inoculated with the ratio of 1: 100 (v/v) into 250 mL of the LB medium containing 0-1000 mM selenate or 0-100 mM selenite and incubated at 37 °C, 200 rpm for different times. 3 mL sample was taken out every two hours, and then the OD 600 values were measured at the wavelength of 600 nm on a UV-visible spectrophotometer (Shimadzu UV-2550 Japan). Each experiment was repeated three times independently, and the mean for each point was calculated. The growth curve was obtained by mapping between the bacterial density (OD 600 ) and incubation time.

Stability assays
The stability of physiological characteristics of two strains CT00C and NCT00C was also assayed in absence of selective stress. Two strains were respectively inoculated into LB broth and grown at 37 °C, 200 rpm for 12 h and then the bacterial cukture was inoculated with the ratio of 1:100 into LB broth and incubated at 37 °C, 200 rpm for 12 h. This operation was repeated 10 times under the same condition. In each round, bacterial cells were again inoculated with the ratio of 1:100 into the LB broth plus 200 mM selenate and incubated at 37 °C, 200 rpm for another 12 h. bacterial density or number was finally determined by measuring OD 600 . Meanwhile, selenium productivities by 1 × 10 8 cells were also examined according to the method described in the Sect. Determination of red elemental selenium and selenoprotein productivities. The strains NCT00C and CT00C growing directly in the LB plus 200 mM selenate at 37 °C, 200 rpm for 12 h were used as the control. The relative percentage (%) was calculated according to the Eq. 1.

Observation of red elemental selenium formation
The bacterial culture was inoculated with the ratio of 1: 100 (v/v) into 250 mL of the LB medium containing 0-1000 mM selenate or 0-100 mM selenite and incubated at 37 °C, 200 rpm for different times. Red elemental selenium (Se 0 ) formation in the culture was observed directly by eyes based on the color change of bacterial culture. The redder the color of bacterial culture became, the more the output of red elemental selenium was. The results were finally recorded by photographing.

Determination of red elemental selenium and selenoprotein productivities
Productivities of red elemental selenium (Se 0 ) in four different strains were examined according to the method reported by Biswas et al. 2011. Briefly, bacterial cells of the strains WT00C, CT00C, NCT00C, and NT00C were respectively inoculated to the LB broth containing 100 mM selenate and incubated at 37 °C, 200 rpm to logarithmic phase (OD 600 = 0.8). Bacterial cells and elemental selenium as a pellet were collected by centrifuging at 4 °C, 8000 rpm for 20 min. Pellets were washed twice with 1 M NaCl. The red colloidal selenium in the pellets was dissolved in 10 mL of 1 M Na 2 S. Then, the samples were again centrifuged at 4 °C, 1000 rpm for 15 min twice to remove all bacterial cells. The absorption of the red-brown solution was finally measured at the wavelength of 500 nm on a UV-visible spectrophotometer (Shimadzu UV-2550 Japan). The content of elemental selenium was calculated by reference to the standard curve. Selenium conversion efficiency (Se µg/10 8 cells) was defined as the content of red elemental selenium was produced by 1 × 10 8 bacterial cells in a specific time.
Productivities of selenoproteins in four different strains were also estimated via measuring selenium content in bacterial proteins according to the method reported by Xu et al. 2020. Briefly, the strains WT00C, CT00C, Relative percentage (%) = (The experimental group/The control) × 100 NCT00C, and NT00C were respectively cultured in LB broth containing 50 mM selenate at 37 °C until the OD 600 of 1.0. Bacterial cells were harvested by centrifugation at 8000 × g, 4 °C for 15 min. The cells were then resuspended in PBS (pH 7.2) and washed and centrifuged at 4 °C three times with the same buffer to remove LB medium. After the bacterial cells in PBS (pH 7.2) were sonicated, the supernatant and the pellet were separated by centrifuging at 10,000 × g, 4 °C for 20 min. The supernatant containing cytoplasmic proteins was collected, and protein concentrations were determined using the Bradford method. The selenium content was finally measured using the atomic fluorescence spectrometer (AFS200) according to the National Standards of the People's Republic of China (GB/T 21729). Data were collected and analyzed using SPSS24.0, and a P value of less than 0.05 was considered significant.

Fermentative production of red elemental selenium
The strain NCT00C was inoculated into 50 mL of the LB medium containing 200 mM selenate and incubated at 37 °C, 200 rpm, until the OD 600 value approached 0.8. Then, 30 mL of the bacterial culture was added to 3 L of the LB broth containing 400 mM selenate in a 5 L fermentor (BIOTECH, Baoxing, Shanghai, China). Fermentation was performed at 37 °C, 200 rpm, pH 7.0, and maximum voluntary ventilation of 1 v/v . min. After fermentation was completed, the culture was used for the isolation and purification of red elemental selenium.
Isolation and purification of red elemental selenium were performed according to the method reported in a previous study (Tian et al. 2021). The bacterial culture was collected by centrifugation at 4 °C 1000 rpm for 10 min twice to remove bacterial cells. The red supernatant was collected and centrifuged at 4 °C, 8000 rpm for 30 min. Pellets were collected and washed three times with physiological saline by centrifuging at 4 °C, 12,000 rpm for 10 min. Se-nanospheres were collected and resuspended in physiological saline and then treated by ultrasonication at 100 W for 10 min. After sonication, red precipitates were collected by centrifugation at 4 °C, 12,000 rpm for 30 min, and washed twice with the solution (1.5 M Tris-HCl and 1% SDS (pH 8.3)) and dH 2 O by centrifuging at 4 °C, 12,000 rpm for 10 min. After the red precipitate was resuspended in dH 2 O, n-octyl alcohol was added according to the ratio of 2:1 (the suspension: n-octyl alcohol, v/v). The mixture was placed at 4 °C for 24 h and then centrifuged again at 4 °C, 12,000 rpm for 10 min to remove all liquid. The precipitate was then washed once respectively with chloroform, ethanol, and dH 2 O to remove all proteins. Finally, the purified elemental selenium was dried using a vacuum freeze drier (LGJ-10, Songyuan Huaxing, Beijing, China) and stored in sealed, cool, and dry conditions.

SEM and TEM observation
Scanning electron microscopy (SEM) was used to observe bacterial cells, as reported by Xu et al. 2020. Briefly, the strain WT00C in LB medium and the strains CT00C, NCT00C and NT00C in the LB broth containing 200 mM selenate were incubated at 37 °C, 200 rpm for different times. The bacterial cultures were centrifuged at 6,000 × g for 15 min, and the pellets 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 (50, 70, 90, and 100% ethanol). Then, the ethanol was displaced by isoamyl acetate, and the samples were mounted onto microscope slides and dried using a BAL-TEC CPD030 critical point drying apparatus. Finally, the samples were sputtercoated with gold to a thickness of approximately 20 nm and observed under a JSM7100F scanning electron microscope (JEOF, Tokyo, Japan).
Transmission electron microscopy (TEM) was also used to observe selenium nanospheres, as described by Xu et al. 2020. The bacterial cultures from fermentation were taken out and diluted 20-fold with phosphate buffer (100 mM, pH 7.2). The diluted culture was aliquoted onto a formvar carbon support film and then incubated at room temperature for several hours to allow water evaporation. Finally, the samples were observed under a Tecnai G 2 20 transmission electron microscope (FEI, USA) at 200 kV.

qPCR assay
HiScript III qRT super Mix, SYBR Green Supermix, and Trizol reagent were respectively purchased from Vazyme (Nanjing, China), Bio-Red, and Ambion (USA). PCR Master Mix was purchased from TransGen Biotech (Beijing, China). Four different strains were respectively activated by growing in LB broth at 37 °C to the OD 600 value of 0.8. The bacterial culture for each strain was then inoculated with the ratio of 1:100 into the LB medium containing 100 mM selenate, and incubated at 37 °C, 200 rpm, until the OD 600 value approached 0.8. After bacterial cells were collected by centrifugation at 4 °C, 6000 rpm for 15 min, total RNA was prepared using Trizol reagent. Quantitative real-time PCR (qPCR) was performed according to the manufacturer's instructions. The primers for qPCR were listed in supplementary data 1. The qPCR program was: 1 cycle of 95 °C for 10 min, 40 cycles of 95 °C for 15 s and 60 °C for 1 min, followed by dissociation curve analysis (60-95 °C) to verify the amplification of a single product. The cycle threshold (Ct) value was determined using the manual setting on the 7500 Real-Time PCR System (ABI, USA) and exported into a Microsoft Excel Sheet for further data analyses. The expression of the 16S rRNA gene was used as an internal reference. The relative expression ratios of each gene were calculated by the 2 −△△CT method (Livak and Schmittgen 2001).

Selenate tolerance of four different strains
Herbaspirillum camelliae WT00C was an endophytic bacterium isolated from the tea plant , and it was able to grow in 400 mM selenate, as shown in Fig. 2a  (a). Clearly, the lag phase of this bacterium was prolonged with the increase of selenate concentrations. For instance, its lag phases in 50, 100, 200, and 400 mM selenate were 3, 6, 12, and 60 h, respectively. When selenate concentration was ≥ 500 mM, this bacterium did not grow. To improve the selenate tolerance, H camelliae WT00C was exposed to 200 mM selenate for multiple rounds, and each round was 24 h. After four rounds of exposure to 200 mM selenate, the lag phases of the strain named CT00C reduced to 2 and 3 h in 200 and 400 mM selenate (see Fig. 2b (a)). Figure 2b (a) also showed that the strain CT00C was able to grow in 600 and 800 mM selenate although its growth required a relatively long time. After six rounds of exposure, the bacterium named NCT00C further shortened its lag phase. Different from the original strain WT00C, the growth of the strain NCT00C in the range of 0-400 mM selenite did not show a significant difference (see Fig. 2c(a)). Moreover, the lag phases of the strain NCT00C were only 2 and 8 h when it grew in 600 and 800 mM selenate. Shortening lag phase and growing in > 500 mM selenate illustrated that the selenate tolerance of bacterial cells was greatly improved via 4-6 rounds of exposures. Figure 2c (a) also showed that the strain NCT00C grew much better than the strain CT00C under the conditions of 600 and 800 mM selenate. The results from the growth-curve comparison indicated that the selenate tolerance of the strain NCT00C was the strongest among the four strains. We also performed eight rounds of exposures and found that the strain NT00C was able to grow in 200 mM selenate, but it did not grow when selenate concentration was ≥ 400 mM (see Fig. 2d (a)). Its lag phases in 50, 100, and 200 mM selenate were 1, 2, and 4 h. Although its lag phases were shorter than those of the strain WT00C in ≤ 200 mM selenate, the strain NT00C was unable to grow when selenate concentrations were ≥ 400 mM. Obviously, too many exposures to high concentrations of selenate were detrimental to the selenate tolerance of bacterial cells.
The growth status of four strains in different concentrations of selenite was also examined. The results (see supplementary data 2) showed that H. camelliae WT00C only grew in ≤ 10 mM selenite, and its lag phases in 0.5, 1, 5, and 10 mM selenite were 10, 20, 30, and 60 h, respectively. Compared to the strain WT00C, the strains NCT00C and CT00C not only markedly shortened lag phases but also grew in 40 mM selenite. After eight rounds of exposures, the strain NT00C displayed good growth status in 0-10 mM selenite, but it did not grow when selenite concentration was > 20 mM. Clearly, the strains NCT00C and CT00C also displayed strong selenite tolerance, which agreed with those results obtained from bacterial growth status in selenate.

Red elemental selenium (Se 0 ) formation of four different strains
When bacterial cells grew in the medium containing selenate or selenite, red color formation in the culture indicated the generation of red elemental selenium (Se 0 ) (Xu et al. 2020;Avendaño et al. 2016;Chen et al. 2020). As shown in Fig. 2a (b), the cultures of H camelliae WT00C became red in 50-400 mM selenate. When 600 and 800 mM selenate were used, the bacterial cultures did not show any color change. Different from H camelliae WT00C, the strains CT00C and NCT00C changed the color red in 50-800 mM selenate (see Fig. 2b (b) and Fig. 2c (b)). Unlike the strain CT00C and NCT00C, the strain NT00C only changed the color of cultures containing 50-200 mM selenate. When selenate concentration was over 200 mM, the culture color did not change anymore (see Fig. 2d (b)). Table 1 also showed the time of the red-color appearance in the cultures containing different selenate concentrations. The strain NCT00C turned the cultured red more rapidly than the strain CT00C, NT00C and WT00C. The red color appears in the culture containing 200 mM selenate was 16 h for WT00C and 5 h for the strain CT00C, 4 h for the strain NCT00C and 10 h for the strain NT00C. In 400 mM selenate, red color appearance was 66 h for the strain WT00C, 7 h for CT00C and 4 h for NCT00C, respectively. These results indicated that the strain NCT00C growing in high concentrations of selenate was able to effectively catalyze selenate reduction to form red elemental selenium (Se 0 ) in a relatively short time.
When bacterial cells grew in different concentrations of selenite, H camelliae WT00C only turned the culture red when selenite concentration was ≤ 10 mM. Different from the strain WT00C, the strains CT00C and NCT00C were able to change the color of the culture containing 40 mM selenite. However, the strain NT00C only changed the culture color when selenite concentration was ≤ 20 mM (see supplementary data 2). Among four strains, the strains CT00C and NCT00C were the best candidates for effectively catalyzing selenite reduction to form red elemental selenium (Se 0 ).

Productivities of red elemental selenium and selenoproteins
Productivities of red elemental selenium (Se 0 ) in four different strains were examined by growing bacterial cells in LB broth containing 100 mM selenate to logarithmic phase. Figure 3 showed that selenium contents (µg/mL) of the strains CT00C and NCT00C were 36% more than that of the strain WT00C (see Fig. 3a), and selenium conversion efficiencies (µg/10 8 cells) were also 30% higher than that of the strain WT00C (see Fig. 3b). High selenium conversion efficiencies indicated that the capability of the strains CT00C and NCT00C to generate red elemental selenium via selenate reduction was more effective than the strain WT00C. We also used the strains NCT00C and WT00C-se (Chen et al. 2020) (two rounds of forced evolution) to produce red elemental selenium via fermentation in LB broth containing 200 mM selenate. The yields of pure elemental selenium gave 1028 mg/L for the strain NCT00C and 560 mg/L for the strainWT00C-se. A high yield of red elemental selenium demonstrated that the strains NCT00C indeed was a better candidate for effectively catalyzing selenate reduction to produce red elemental selenium (Se 0 ). Figure 4 showed fermentation in 5L fermentor, Se-nanospheres observed by TEM in bacterial culture and the red elemental selenium (Se 0 ) purified from the fermentation culture.
Productivities of selenoproteins synthesized in four different strains were also examined by growing bacterial cells in LB broth containing 50 mM selenate to the stationary phase. Selenoproteins were estimated by measuring selenium contents in bacterial proteins (µg/mg). As shown in Fig. 3c, selenium contents in bacterial proteins of the strains CT00C and NCT00C were 40% more than that of the strain WT00C. Thus, the strains CT00C and NCT00C integrated selenium into proteins to form selenoproteins more effectively as compared to the strain WT00C.

Transcriptional levels of the genes involving in selenate metabolism and tolerance
Transcriptional levels of 19 genes involving elemental selenium (Se 0 ) and selenoprotein synthesis, as well as selenate tolerance in four different strains, were investigated by using qPCR. Six genes for elemental selenium synthesis were cysD (sulfate adenylyltransferase (SAT), EC 2.7.7.4), gsdA (glucose-6-phosphate 1-dehydrogenase (G6PD), EC 1.1.1.49), icd (isocitrate dehydrogenase (IDH), EC 1.1.1.42), gshA (glutamylcysteine synthetase (GCL), EC 6.3.2.2), gshB (glutathione synthase (GSS), EC 6.3.2.3) and gsr (glutathione reductase (GR), EC 1.8.1.7), in which GCL and GSS are responsible for glutathione (GSH) formation. 4 genes for selenoprotein synthesis were trxR (thioredoxin reductase (TRX), EC 1.8.1.9), selD (selenide, water dikinase, EC 2.7.9.3), selA (L-seryl-tRNA(Ser) seleniumtransferase, EC 2.9.1.1) and metG (methionyl-tRNA synthetase (MARS), EC 6.1.1.10). Those genes were involved in selenate metabolism as showed in Fig. 1. 8 genes (mdh, ligK, pntA, iolA, ghrA, gluC, mmsB, argC) thought to be related with selenate tolerance [34] were also examined. In addition, the expression of lpxB gene encoding lipid-A-disaccharide synthase [EC 2.4.1.182] was incidentally tested. To keep the consistency of analytical conditions, all four strains were incubated at 37 °C, 200 rpm in LB medium containing 100 mM selenate. As shown in Fig. 5, the transcriptional levels of 6 genes for elemental selenium synthesis in CT00C, NCT00C and NT00C strains displayed an increase of ≥ tenfold as compared to the wild-type strain WT00C (see Fig. 5a). Although the expression of icd and gshA genes in the NT00C strain increased significantly, the expression of cysD gene responsible for the reduction of selenate to selenite was much less that those of the strains CT00C and NCT00C. In the pathway of selenoprotein synthesis, the transcriptional levels of trxB, selD, selA and metG genes in CT00C and NCT00C Fig. 4 The fermentation of the strain NCT00C and zerovalent selenium product. a Bacterial fermentation in a 5L fermentor; b Se-nanospheres in the fermentation culture; c Red elemental selenium (Se 0 ) purified from the fermentation culture 5 Page 10 of 13 strains were about tenfold more than those for the strain WT00C. The strain NT00C displayed the increase of trxB and selD gene expressions but its transcription of selA and metG genes did not show a significant increase as compared to the strain WT00C (see Fig. 5b). Those results suggested that two strains CT00C and NCT00C held a strong capacity of synthesizing elemental selenium and selenoproteins. Figure 5c showed the transcriptional levels of mdh, ligK, pntA, iolA, ghrA, gluC, mmsB and argC genes in four different strains. Compared to the strain WT00C, the strains CT00C and NCT00C significantly increased transcriptional levels of eight genes (≥ fivefold), whereas the strain NT00C only increased the transcription of five genes (mdh, pntA, iolA, ghrA, gluC). Although the expression of mdh, pntA and gluC genes was markedly increased in the strain NT00C, its mmsB and argC genes were poorly transcribed. The transcriptional levels of the mmsB and argC genes in the strain NT00C was even less than that of the strain WT00C. Clearly, the strains CT00C and NCT00C indeed held strong selenate tolerability. In addition, Fig. 5d also showed the expression of lpxB gene involving in lipopolysaccharide (LPS) biosynthesis. The strains CT00C and NCT00C displayed tenfold less expression, whereas the strain NT00C increased seven folds as compared to the strain WT00C. Less transcription of lpxB gene suggested that the activity of lpxB gene was seriously inhibited by high concentration of selenate, which was in accord with the results observed in the previous study [34] . Nevertheless, the expression of lpxB gene was gradually increased along with the increase of exposure times in selenate as shown in Fig. 5d. After eight rounds of exposure to high concentration of selenate, the lpxB gene in the strain NT00C was highly activated.

Morphological characteristics of bacterial cells
Cell morphologies of the strain WT00C growing in the LB medium and the strains CT00C, NCT00C and NT00C exposing to 200 mM selenate for four, six and eight rounds were observed by scanning electron microscopy. As shown in Fig. 6, bacterial sizes were varied at different growth stages. The cells of the strains WT00C, CT00C and NCT00C began to grow and divide at 2 h incubation, whereas the cells of the strain NT00C were in the inhibitory state. At the stationary phase (24 h), the cell sizes were 1.17 ± 0.04 μm × 327 ± 6 nm for the strain WT00C growing in LB medium [34] , 1.81 ± 0.06 × 1.02 ± 0.05 μm for the strain CT00C. 1.91 ± 0.05 × 1.03 ± 0.04 μm for the strain Fig. 5 Relative expressions of 19 genes in four different strains incubated in LB broth containing 100 mM selenate. a Six genes related to the synthesis of red elemental selenium (Se 0 ); b Four genes involving in selenoprotein synthesis; c Eight genes related to selenate tolerance; d lpxB gene ▸ NCT00C and 2.10 ± 0.07 × 1.06 ± 0.08 μm for NT00C. Clearly, the size of bacterial cells became bigger as the number increase of exposures to high concentration of selenate. The size change mainly manifested in the increase of cell width, although cell length was also increased. Three strains CT00C, NCT00C and NT00C in cell width were almost threefold of the strain WT00C. The strain NT00C had doubled in cell length as compared to the original strain. Those enlarged cells were incubated in LB medium for 24 h, their sizes did not return to the original size of the strain WT00C. We also observed the changes of cell surface roughness under 50,000 magnifications. Bacterial cells displayed surface changes from the rough to the smooth during multiple forced evolutions. After eight rounds of forced evolution, the cell surface of the strain NT00C did not recover to the original roughness, although its lpxB gene was markedly upregulated. These results clearly demonstrated that bacterial cells underwent obviously physiological and morphological changes when H camelliae WT00C was exposed to high concentration of selenate.

Discussion
Although forced evolution has been used as a conventional technique for improving biological properties and functions of cells, microbes, or biomolecules (Pooggin et al. 1998;Gullck et al. 1995;Wang et al. 2010), it still is quite effective to improve the selenate tolerance of H. camelliae WT00C in our study. The lag phase of the strain NCT00C in 400 mM selenate was shorted to 2 h, which was 1/30 of the original strain. This bacterium was also able to grow in 800 mM selenate with a lag phase of 8 h. The strain CT00C, obtained from 4-rounds of 24 h-exposure to high concentration of selenate, also showed marked improvement in selenium tolerance. Clearly, the selenate tolerance of H. camelliae WT00C was greatly improved through 4-6 rounds of 24 h-exposure to high concentration of selenate. This conclusion was also supported by the results from qPCR, in which those genes involving in selenium metabolism and tolerance were markedly up-regulated. The strain NT00C was unable to grow in ≥ 400 mM selenate after eight rounds of forced evolution. Poor selenium tolerance of the strain NT00C suggested that 4-6 rounds of forced evolution was enough to create a bacterial strain with strong selenium tolerance, and too many exposures to high concentration of selenate might perhaps cause gene inactivation of key enzymes involving in selenate metabolism or antioxidative stress. The stability of physiological characteristics of two strains CT00C and NCT00C was also assayed in absence of selective stress. As shpwn in Fig. 2, the strain NCT00C was more stable than the strain CT00C in the aspect of selenate tolerance or selenium productivities. This result indicated that four rounds of forced evolution could offer the strain CT00C more physiological alteration, while six rounds of forced evolution might give the strain NCT00C more genetic changes.
The strains CT00C and NCT00C were able to grow in the LB medium containing 800 mM selenate due to strong selenate tolerance. Here it was raised a question if these bacterial strains kept the same activity of selenate reduction to produce red elemental selenium (Se 0 ) as the original strain. The strains CT00C and NCT00C turned the cultured red more rapidly as compared to the strain WT00C. For example, the red color appears in the culture containing 400 mM selenate was at 72 h for the strain WT00C, 7 h for the strain CT00C and 5 h for the strain NCT00C. The selenium conversion efficiencies of the strains CT00C and NCT00C were 30% higher than that of the strain WT00C at log phase. More than 40% of selenium was integrated into bacterial proteins indicated that the strains CT00C and NCT00C had a strong capability of synthesizing selenoproteins. More selenium conversion and selenoprotein synthesis were also supported by qPCR assay in which those genes involving in selenate reduction and selenoprotein synthesis were up-regulated markedly. The strains CT00C and NCT00C not only displayed strong selenate tolerance but also showed high activity of selenate reduction and selenoprotein synthesis. Thus, two strains CT00C and NCT00C obtained through 4-6 rounds of exposures to the high concentration of selenate might be used as candidates for industrial production of elemental selenium and selenoproteins. The strain NT00C was unable to grow in the LB medium containing 400 mM selenate. Markedly enlarged cells and less expression of cysD, mmsB and argC genes should be main reasons for its poor selenate tolerance. In the pathway of selenate metabolism, the cysD is a key gene responsible for the first step of selenate reduction. Less expression of the cysD gene certainly slows down the reduction of selenate to selenite. Two genes mmsB and argC are mainly involved in amino acid metabolism. Less expression of both genes affects the synthesis and degradation of some amino acids (e.g., Val, Leu, Ile, Lys, Arg and Pro). Thus, too many exposures to the high concentration of toxic compounds are not beneficial for ameliorating properties and functions of living organisms.
Bacterial sizes and surface roughness underwent obvious changes during forced evolution. Both length and width of bacterial cells obviously increased after 4-8 rounds of forced evolution in 200 mM selenate. In addition, surface roughness also changed from rough to smooth. The physiological responses to high selenate concentration enlarged the size of bacterial cells. Toxic effects of selenate/selenite were widely thought to be related to their interactions with essential sulfhydryl-containing enzymes and structural proteins, and such interactions finally led to oxidative stress at high concentrations inside bacterial cells (Aborode et al. 2016;Nordberg and Armer 2001;Carmel-Harel and Storz 2000). Bacterial cell enlargement was a physiological response of escaping lethal effects of the oxidative stress caused by high selenate concentration (Carmel-Harel and Storz 2000).
Furthermore, the variation of cell surface roughness was associated with LPS biosynthesis. The lpxB gene encoding lipid-A-disaccharide synthase was one of the genes responsible for lipopolysaccharide (LPS) biosynthesis (Raetz 1986;Osborn 1979). When the bacterial cells were initially incubated in 200 mM selenate, the lpxB gene was seriously inhibited at its lag phase, whereas the lpxB gene was continuously up-regulated when the bacterium was again incubated in 200 mM selenate (Avendaño et al. 2016). Although the lpxB gene was markedly up-regulated in the strain NT00C, LPS biosynthesis was not enough to satisfy the need of its enlarged cells.