Selection of tobacco etch virus protease variants with enhanced oxidative stability for tag-removal in refolding of two disulde-rich proteins

Tobacco etch virus protease (TEVp) is a powerful enzymatic reagent for removing fusion tag. In this work, we constructed nine TEVp variants with introducing one to three mutations of C19S, C110S and C130S into the soluble TEVp variant, TEVp 5M . Using the C-terminal green uorescent protein (GFP) variant reporter, all constructs showed different solubility levels among four E. coli strains. The TEVp 5M containing the C110S and/or C130S mutations in the hyperoxic strain showed the enhanced the cleavage activity. Addition of dithiothreitol to the cultural medium increased the activity of certain constructs produced in the BL21(DE3), contrary to the added hydrogen peroxide, due to cytoplasmic redox change measured by the redox sensitive GFP construct. The more cysteine residues in the puried TEVp 5M were modied specically than those in the other variants. All puried constructs showed similar specic activities in the presence of 5 mM dithiothreitol. In the buffer containing the compounds to aid disulde bond formation of the refolded protein, the double mutant TEVp 5M C110S/C130S exhibited the highest cleavage eciency. This variant was ecient for removing the fusion tag after refolding of cellulose-binding module tagged disulde-rich proteins including bovine enteropeptidase and maize peroxidase absorbed on the regenerated amorphous cellulose.


Introduction
Owing to high speci city, the mutated TEVp variant with the decreased auto-inactivation via selfcleavage, and the improved protein solubility is a commonly-used tool in various biotechnological applications [1]. Other mutated amino acid resides are contributed to enhance the thermostability [2], inhibit the translational modi cation in eukaryote cells [3], change of the thiol protease to serine protease [4], alter substrate speci city [5], and increase the cleavage activity [6].
Because of its easier culture, short life cycle, well-known genetics, and easy genetic manipulation, Escherichia coli is the preferred choice for production of recombinant protein [7]. However, many proteins are formed into insoluble aggregates, also called as inclusion bodies (IBs). Since IBs are produced at higher level, easily prepared and puri ed, target protein in IBs is refolded for studying protein structure and function [8]. Several fusion tags improve the refolding e ciency, and the speci c protease is used for removing the fused tag [9]. On-column refolding dependent on the fused a nity tag saves the refolding period and increases yields of refolded protein [10]. The cellulose-binding module (CBM) tag fused with the single-chain antibodies is e cient for renaturation of the fusion protein bound to the cellulose [11].
For renaturing the disul de-rich proteins, the reagents for enabling disul de bond formation were required, such as the reduced and oxidized glutathione (GSH and GSSG), cysteine and cystine [9]. Little information of these reagents on affecting the TEVp activity is available.
Previously, we combined missense mutations in the TEVp to generate the TEVp 5M variant with inhibiting the auto-cleavage, and improving protein folding and solubility [12]. The TEVp activity analysis based on the coupled assay is developed [13]. Once the designed fusion protein is cleaved by the sequence-speci c protease, the released tag-free E. coli diaminopropionate ammonia-lyase (eDAL) from the glutathione S-transferase (GST) tagged eDAL increases activity signi cantly. In this study, we mutated and combined three cysteine residues in the TEVp 5M and identi ed the change of protein solubility and cleavage activity in different E. coli strains. Among nine variants, only C110S and/or C130S mutations showed little impact on protein production, and afforded the TEVp 5M with the improved oxidative stability in vivo and in vitro.
The double mutant was e cient for removing the CBM tag as the fusion partner for oriented immobilization and matrix-assisted refolding of the disul de-bonded bovine enterokinase (also called enteropeptidase) light chain (bEK) and maize peroxidase (mPex) as the isoenzyme C homolog of horseradish peroxidase (HRP). Both disul de-rich enzymes are widely applied in biotechnology.

Plasmids construction
The C19S, C110S and C130S mutations or their mixed mutations were introduced into the TEVp 5M gene in the plasmid pET28-TEVp 5M independently and successively using each of the primer pairs C19S1 and C19S2, C110S1 and C110S2, C130S1 and C130S2, and pET28-TEVp 5M as the template (Fig. S1). After ampli cation, the products were phosphorylated, ligated, and sequenced. Then, each of the sequence was excised with Nco I/Xho I and subcloned into Nco I/Sal I sites of pET28-GFP for expressing the constructed TEVp tagged emerald GFP (EmGFP) to quantitatively analyze protein solubility. Due to the Origami (DE3) strain bearing kanamycin resistance, each of the coding sequence through Xba I and Xho I excision was inserted into the Xba I/Xho I sites of the pET-22b vector. In this strain, the genes encoding thioredoxin reductase and glutathione reductase are disrupted for generating the oxidative cytoplasm [16].
Based on comparison of the mature HRP amino acid sequence (Fig. S1), the sequence encoding the mature mPex Q45-S350 with deletion of the hydrophobic N-terminal leader sequence responsible for directing the protein to the endoplasmic reticulum (ER) was ampli ed by RT-PCR using the total RNAs extracted from maize leaves as the template, and primers mPex1 and mPex2. The PCR amplicon was incubated with BamH I and Xho I, and placed between the BamH I-Xho I site of pCBM-tevS-GFP to replace the GFP sequence. The similar substitution using the BamH I-Xho I excised bEK coding sequence from the pEK plasmid was conducted for expressing the CBM tagged bEK. The linker between the CBM and the tevS was introduced by ampli cation of the CBM tag using the primers CBM1 and CBM2, digestion with Nde I and BamH I and insertion into Nde I/Bgl II sites of the pCBM-tevS-GFP plasmids. The exible linker between the tevS and the bEK was introduced as the similar procedure by using the primers bEK1 and bEK2 for ampli cation of the bEK coding sequence to prolong sequence upstream of the original BamH I site encoding the extra six-amino-acid, and excising with Bgl II and Xho I for introduction of the BamH I/Xho I sites of the former constructed plasmids. The BamH I-Xho I excised fragment encoding bEK was then substituted with that encoding the mPex.
For constructing the redox sensitive GFP reporter roGFP [17], the sequence encoding the GFP variant mGFP5 with improved stability in E. coli [18], was amplified using the plant expression vector pCAMBIA1302 as the template and primers mGFP5-1 and mGFP5. The puri ed PCR amplicon with the Nco I and Hind III was subcloned into the Nco I-Hind III site of the pET-28b plasmid. The mutations C48S, S147C and Q204C were introduced into the mGFP5 step by step by using the primers for PCR. All inserts were sequenced to identify correction. The primers used in this study were listed in the table S1.
Solubility determination of the overexpressed TEVp constructs in different E. coli strains Except where noted, induction and extraction of recombinant proteins in this study were conducted as follows. The constructed plasmids were transformed independently into the different E. coli strains. The recombinant cells were cultured overnight at 37 °C in 5 ml of Luria-Bertani (LB) broth, diluted to 50-fold and grown at 37°C. Induction of the target protein was performed by addition of isopropylthio-β-Dgalactoside (IPTG) at nal concentration 0.5 mM, when OD 600 value reached about 0.5. After cultured at 28 °C for 12 h in 10 ml liquid culture of a 50-ml shake ask, cells were collected by centrifugation, washed with buffer A (20 mM Tris-HCl, pH 8.0, 100 mM NaCl), sonicated and centrifuged, and analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). The protein amounts were measured by Bradford method, using bovine serum albumin as the reference. After electrophoresis, proteins on the gel were transferred to polyvinylidene fluoride membrane, immunoblotted with anti-His6 monoclonal antibodies, and treated with HRP conjugated anti-rabbit IgG. The band representing the target protein was visualized by addition of 4-chloro-1-naphthol solution dissolved in 20% methanol and 0.08% The C-terminally fused EmGFP reporter is used for quantitative analysis of the soluble TEVp amounts, based on the uorescence emitted from soluble fractions on the F-4500 uorescence spectrometer (Hitachi, Japan) with excitation and emission maximum 488 and 515 nm [13].
Puri cation of the constructed TEVp with C110S and/or C130S mutations The expression plasmids were transformed into the Rossetta (DE3) strain with supply rare tRNAs for augmenting soluble expression level of the TEVp construct. After induction at 28 °C for 12 h, cells were collected by centrifugation and washed with buffer B (50 mM sodium phosphate, 300 mM NaCl, and 10 mM imidazole, pH 8.0), and disrupted by sonication. The supernatants were loaded on a Ni-NTA spin, centrifuged twice with buffer B, washed twice with 40 mM imidazole in buffer B (pH 8.0), and eluted with 250 mM imidazole in buffer B (pH 8.0). Puri ed protein was concentrated by an Ultra-15 centrifugal lter tube equipped with an Ultracel-10 membrane, and exchanged with buffer A. The freshly prepared protein was used for cysteine labeling and activity assay.
Modi cation of the free cysteine residues in the puri ed TEVp constructs The free cysteine residues of the TEVp constructs were labeled with AMS as the described method [19].
Puri ed TEVp variants was incubated with 150 µM CuCl 2 as the oxidative agent or 1 mM dithiothreitol (DTT) for 1 h at 25 °C , precipitated by trichloroacetic acid. After it was washed and re-suspended with buffer A twice, the precipitated protein was labeled with AMS, a maleimidyl reagent speci cally alkylating free SH-group of cysteine to increases the molecular weight in about 0.5 kDa. When the reaction was nished, the mixture was centrifuged and washed with buffer A. The labeled TEVp variants were incubated with loading buffer in absence of DTT, and separated by 12% SDS-PAGE under non-reducing condition.

Quantitative analysis of in vitro activity
In vitro cleavage activity was assayed using the GST-eDAL puri ed by Ni-NTA [13]. The mass ratio was 30:1 for the puri ed protein substrate and soluble extracts containing the recombinant TEVp construct, and 50:1 for puri ed protein substrate and the protease in the buffer A. The cleavage was reacted at 30°C for 1 h, and the activity was determined by coupled assay of the eDAL activity [11]. As a PLP dependent enzyme, eDAL catalyzes L-DAP to pyruvate and ammonia. The amount of pyruvate was measured with 2,4-DNP. The reaction mixture for testing DAL activity contained 50 μM PLP and 10 mM L-DAP, and the crude extracts in a nal volume of 1 ml. After incubated at 37 °C for 5 min, 1 ml of 2 mM HCl plus 0.03% 2,4-DNP was added to stop the DAL catalytic reaction. Following incubation at 4 °C for 5 min, 2 ml of 2 M NaOH was supplemented. After centrifugation, absorbance at 520 nm representing the amounts of pyruvate in the mixture was measured on a U-2001 spectrometer (Hitachi, Japan).

Recovery of inclusion bodies
The induced cells were re-suspended in buffer A and disrupted by sonication. After centrifugation at 4000 g for 10 min, insoluble fraction was collected, washed with buffer A twice and re-suspended with buffer B [30 mM Tris/HCl, 150 mM NaCl, 10% (v/v) glycerol, 0.5% (v/v) Triton X-100, pH 7.5]. The mixture was centrifuged and IBs were re-suspended with buffer B in the absence of Triton X-100.
The washed IBs containing the bEK was solubilized in buffer A containing 8 M urea, 5 mM DTT (V/V) and 10 mM EDTA [20]. The mixture was left for 2 h at room temperature to ensure su cient amounts of IBs for solubilization and reduction of the mismatched disul de bonds, and centrifuged at 18000 g for 30 min to remove the pellet. IBs containing the mPex dissolved in the solution (4.5 M urea, 40 mM Tris-HCl, pH 9.0, 5 mM DTT and 0.2 mM hemin) to a protein concentration of 0.3 mg/mL, with slight modi cation [21]. The resultant suspension was centrifuged at 12000 g for 20 min and the debris was discarded. The solubilized protein was used for refolding.

Matrix-assisted refolding of target proteins
For increasing the cellulose loading yields of target proteins, the regenerated amorphous cellulose (RAC) was prepared based on the published paper [15]. The microcrystalline cellulose was suspended with double-distilled water at ratio of 1:3 (W/V), and slowly added 20 fold volume of 10 ml ice-cold H 3 PO 4 with vigorous stirring, than supplemented 2 fold volume of ice-cold water. After centrifugation, the pellet was suspended with ice-cold water several times for removing phosphoric acid in soluble fraction, and centrifuged. With adding 2 M Na 2 CO 3 to the pellet, the mixture was centrifuged and discarded. After the resin was washed with ice-cold water several times until the pH value in mixture containing RAC was reached about 7.0. The solubilized proteins from IBs were mixed with RAC at mass ratio of about 1:20 to make the urea concentration decreased from 8 M to 6 M. The mixture was stirred for 1 h at room temperature to allow the target protein binding RAC.
Puri cation and activity assay of the refolded tag-free proteins by the TEVp variant digestion The puri ed TEVp variant was incubated with the refolded protein bound to the RAC with mass ratio of 1: 10 at 10 °C for 24 h. After reaction was nished, the Ni-NTA-resin was added and incubated for 2 h at room temperature. The supernatant after centrifugation was subjected to SDS-PAGE analysis. The bEK cleaving the GST tagged sDAL with incorporation of D4K as the bEK recognition sequence was analyzed, based on the coupled assay of sDAL activity. The mPex catalyzes the degradation of H 2 O 2 using OPA as a hydrogen donor, which turns yellow upon oxidation [22]. The freshly prepared mPex was incubated in the mixture (20 mM Tris/HCl, pH 7.5, 50 μg/mL OPA, 10 mM or 30 mM H 2 O 2 ) at 37 °C for 30 min, and absorption at 411 nm was measured.

Site-directed mutations and plasmids construction
To improve the TEVp variant to prohibit obvious loss of activity under the oxidative conditions, we mutated one to three cysteine residues, including C19, C110 and C130 in the TEVp 5M variant were mutated to serine ones. Three variants were constructed including M1-M3 with one cysteine residue replaced with serine one. The variants M4-M6 contained the combined two mutations in M1-M3, and M7 contain all three mutations (Tab. 1).

Production of the TEVp variants in different E. coli strains
The constructed seven variants in soluble and insoluble fractions from E. coli BL21 (DE3) cells were varied, as shown on the SDS-PAGE gel and by Western blotting (Fig. 1a). Four variants including the M1, M4, M5 and M7 containing the C19S mutation displayed poor solubility. In contrast, other three variants in soluble fractions were detected. Inhibition of the background expression in the BL21(DE3) pLysS strain decreased production of all variants (Fig. 1b). With augment of rare tRNA levels in the Rossetta (DE3) strain, soluble production of all variants were not signi cantly increased (Fig. 1c). Since C19S mutation decreased protein solubility, we expressed other four TEVp variants in the Origami (DE3). Except for M7, three certain variants displayed the enhanced soluble production (Fig. 1d). The results indicated that the C110S and/or C130S mutations conferred the TEVp 5M with augment of protein solubility, different from the C19 mutation, and suggested that both mutations failed to recover the C19 mutation on impairing the TEVp 5M folding.

Quantitatively analyzing the soluble amounts of the TEVp constructs
Using the C-terminal EmGFP reporter, soluble production of the TEVp variants were quantitatively analyzed. Introducing C110S and C130S mutations to the TEVp 5M rendered the protease variant with different soluble productivity in various strains, but higher soluble amounts in the Origami (DE3) strain than the TEVp 5M (Figs. 2a-2d). Soluble production of the fusion constructs were also detected by SDS-PAGE (Figs. S2, a-d). The current data re ected that soluble productivity of the various TEVp variants were dependent with the E. coli strains used in this study. Combination of the C110S and C130S further enhanced soluble production of the TEVp 5M in the Origami (DE3).

Cleavage activity of the TEVp variants
The puri ed GST-tagged eDAL was cleaved into two parts by TEVp 5M , M2, M3 and M6 variants in the crude extracts, whereas the other variants had less cleavage e ciency, as detected by SDS-PAGE (Figs. S3, a-d). The changed cleavage activity among the TEVp variants was probably attributed to soluble amounts. By the coupled assay, the M2 and M3 variants produced in the BL21(DE3) cells showed higher cleavage activity than the TEVp 5M , but two variants produced in the recombinant BL21(DE3)pLysS and Rossetta (DE3) cells had less e ciency on cleaving the protein substrate (Figs. 3a-3d). The M6 variant from the Origami (DE3) exhibited higher activity, but the M3 variant displayed almost equal activity to the TEVp 5M . The results suggested that mutations of the C110 and C130 increased soluble production and cleavage activity of the TEVp variants in the oxidative environment, probably resulted from the avoidance of the mis-matched disul de bond(s) formation.

Addition of DTT or H 2 O 2 to the culture affecting the enzymatic activity
When DTT was added to the culture at nal concentration of 2 mM or 10 mM, the TEVp 5M or M2, M3 or M6 produced in the BL21(DE3) displayed the increased speci c activity (Fig. 4a). In contrast, addition of 2 mM or 10 mM ( nal concentration) H 2 O 2 to the LB medium decreased the TEVp activity (Fig. 4b).
Supply of reducing or oxidative agent changed the cellular redox in the BL21(DE3) strain, based on the changed excitation at 485 emitted by the produced roGFP1 (Fig. S4). Upon reduction with DTT, the ratio of cell fluorescence excited at 485 and 400 nm (F485/F400) presenting the redox potential in the cells was also increased, different from the cells treated with H 2 O 2 (Fig. 4c). The reducing environment affected the enzymatic activity for TEVp 5M more e ciently than the M2 and M6 variants. Similarly, M2, M3 and M6 mutants displayed less decrease in cleavage activity than the TEVp 5M with supplementary H 2 O 2 to the medium, resulted from probability that the cellular redox affected the TEVp folding, and the mis-matched disul de bond formation.
Free cysteine residues labelling in the puri ed TEVp constructs The TEVp variants were independently expressed in the Rossetta (DE3) with boosting rare tRNAs levels for enhancing soluble production. The main bands were observed on the SDS-PAGE gel for puri ed TEVp constructs under reducing condition (Fig. 5a). The AMS-labelled TEVp 5M exhibited several bands in nonreducing SDS-PAGE (Fig. 5b), resulted from incubation with CuCl 2 facilitating disul de bond formation, and prohibiting AMS labeling. In contrast, the M3 and M6 variant showed less bands than the TEVp 5M , due to one and/or two cysteine residues mutation. Modi cation of the M2 variant was not e cient, and the faint bands were observed. The reason is not known. Under reducing condition, the TEVp 5M displayed the similar bands to the other mutants (Fig. 5b). Presence of DTT caused slower mobility of the labeled TEVp, suggesting that more AMS molecules binding the TEVp upon disruption of the disul de bonds.

Effect of reducing reagent or oxidative agents on the activity of puri ed TEVp constructs
The puri ed TEVp construct cleaved the fusion protein substrate (Fig. 6a). The speci c activity of the constructed TEVp was comparable to the TEVp 5M , except for the M3 showing the slightly decreased activity (Fig. 6b). With addition of 2 mM DTT, activity of the construct increase was increased trivially (Fig. 6c). On the contrary, addition of compound pairs including 5 mM cystine and 0.5 mM cysteine, or 5 mM GSSG and 0.5 mM GSH for facilitating disul de bridge formation during the protein renatuation slightly inhibited the cleavage activity of puri ed TEVp constructs. The activity of the TEVp 5M C110S/C130S variant was increased by about 20%, in contrast to that of the TEVp 5M (Fig. 6c). The data suggested that the engineered TEVp variant was more insensitive to the reagents for changing the redox in the reaction mixture.
Removal of fusion tags in two refolded proteins containing multiple disul de bonds The CBM attached bEK or Mper was produced mainly as IBs (Fig. 7a). Initial attempt for releasing the target proteins from RAC was failed. So, we introduced the exible linker in the fusion protein for enhancing the cleavage e ciency. After refolded, the target protein was detached from the resin using the puri ed M6 variant (Fig. 7b). The refolded bEK cleaved the GST tagged sDAL into two parts, and cleavage activity was assayed based on the released sDAL catalysis (Fig. 7c). With addition the refolded mPex, the H 2 O 2 at 10 and 30 mM concentrations was transformed into yellow compounds in the presence of OPA, in contrast to the heat-inactive enzyme (Fig. 7d). It is noted that auto-oxidation was reacted spontaneously in the mixture containing 30 mM H 2 O 2 as the control, based on the color change. The results indicated that renaturation of the fusion protein for tag-removal to release the tag-free active bEK and Mper was e cient.

Discussion
In this study, we engineered one to three cysteine residues in the TEVp 5M to select the variant with the improved activity in E. coli oxidative cytoplasm, and identi ed the various production levels of the recombinant constructs. So far, different GFP variants have been used for monitoring protein folding and solubility in E. coli. A linear correlation between in vivo uorescence and soluble amounts of the engineered target protein variants is yielded by use of the fused enhanced GFP (eGFP) variant [23][24][25]. However, the correlation changes signi cantly for various GFP variants [26]. Certain target proteins affect the GFP folding [27], thus, a robust folding variant of the GFP, called superfolder GFP (SfGFP), is generated [28]. In addition, the metabolic burden of the expressed GFP impacts the reporter e ciency [29].
By use of the GFP reporter, the uorescence intensities were changed irregularly for the different TEVp constructs among the selected E. coli strains. The expressed EmGFP affords the E. coli cell uorescence, color intensity and temperature tolerance more effectively than the eGFP [30]. So, irregular alternation of uorescence from soluble fractions in different E. coli strains is probably caused by the GFP toxicity. The GFP uorophore biosynthesis is accompanied by the generation of H 2 O 2 [31]. High level production of the TEVp 5M fused to the EmGFP at 37°C induction for over 4 h inhibits E. coli cell growth [12]. The P T7 promoter regulating the GFP variant expression at high level decreased cell uorescence after induction at 37°C induction for over 10 h [32]. Besides, the GFP produced as insoluble aggregates emits uorescence [29]. So, uorescence signal emitted from soluble fractions, not from cells, showed improved correlation with the expressed soluble protein amounts. Our current data provided the suggestion that desirable expression level and cellular environment are required for yielding the linear correlation between uorescence and yields of the recombinant target protein using the GFP reporter.
Based on the crystal structure, the C19, C110 and C130 residues are located in the rst, eighth, and tenth β-strands, respectively, and the C130 forms a disul de bridge between TEVp molecules or reacts with βmercaptoethanol [33,34]. Different from the inactive TEVp construct in the eukaryotic cells [3], introduction of the C110S had little impact on protein solubility and activity. The C19 and C110 are buried in the TEVp molecule, the C19A mutation on the protein solubility is needed to be further assessed. Mutations of N23Q, C130S, and T173G introduced in the wild type TEVp for inhibiting N-glycosylation the improving resistance oxidative stress in the ER of mammalian cells afford the active protease, but addition of the C110S inactivates the protease [3]. In this work, we used the TEVp variant improved protein folding for mutation of the cysteine residues. It is proposed that the naturally occurring disul de bridges formation is guided by inherent protein folding [35,36], but little information is available on inherent protein folding resisting formation of the error disul de bridge(s). Fluorescence change of the GFP variants via secretion to the E. coli periplasm provides the indirect evidence. The eGFP forms nonuorescent oligomer via the mixed disul des in the ER, leading to loss of uorescence in E. coli periplasm by the SecYEG transportation pathway. In contrast, the SfGFP is brightly uorescent in the ER and E. coli periplasm [37].
The arti cial two cysteine residues in the roGFP1 are generated for forming a disul de bond, and this reporter is effective for monitoring the intracellular redox in E. coli [38]. Using the SfGFP as a model, the roSfGFP is created to increase cell uorescence, but certain characters of roSfGFP are faintly different from those of the roGFP1 [39,40]. Considering that protein folding contributes the redox sensitive reporter function, here, we used the mGFP5, not the EmGFP for construction of the roGFP. The error disul de bond formation in the NpuDnaE intein produced in E. coli is also detected by non-reducing SDS-PAGE, resulting in decline in the cleavage under non-reducing conditions [41]. In this study, we identi ed that the added DTT to the culture augmented the TEVp activity. This method is simple and cost-effective for increasing soluble production and activity of the other target proteins in E. coli [42,43].
Oriented immobilization of the CBM tag minimizes the contact between target protein molecules, thus preventing their aggregation. The advantage of the CBM tag used for protein refolding is that, even in the presence of 6 M urea, the CBM retains the binding capability [11]. Additionally, this method is economical and time-saving. The target proteins were liberated from the resin after the TEVp variant cleavage.
Moreover, the refolded CBM tagged two enzymes bound to the RAC were used as immobilizates for improving use e ciency and easy extraction from the reaction mixture.
In summary, the multiple mutations afforded the engineered TEVp variant with the enhanced soluble yield, and the decreased self-inactivation, and the improved oxidative stability. This variant is used for removing the fusion partner in the oxidative hyperoxic E .coli strain and in vitro. Removal of the CBM tag with the TEVp variant mediated release of the refolded disul de-rich proteins from the RAC resin in the refolding buffer.

Funding
This study was nancially supported by the Scientific and Technological Major Project of Anhui Province (1803071180). Table   Table 1. The variants constructed in this study Page