A novel alkaline protease from alkaliphilic Idiomarina sp. C9-1 with potential application for eco-friendly enzymatic dehairing in the leather industry

Alkaline proteases have a myriad of potential applications in many industrial processes such as detergent, food and feed production, waste management and the leather industry. In this study, we isolated several alkaline protease producing bacteria from soda lake soil samples. A novel serine alkaline protease (AprA) gene from alkaliphilic Idiomarina sp. C9-1 was cloned and expressed in Escherichia coli. The purified AprA and its pre-peptidase C-terminal (PPC) domain-truncated enzyme (AprA-PPC) showed maximum activity at pH 10.5 and 60 °C, and were active and stable in a wide range of pH and temperature. Ca2+ significantly improved the thermostability and increased the optimal temperature to 70 °C. Furthermore, both AprA and AprA-PPC showed good tolerance to surfactants and oxidizing and reducing agents. We found that the PPC domain contributed to AprA activity, thermostability and surfactant tolerance. With casein as substrate, AprA and AprA-PPC showed the highest specific activity of 42567.1 U mg−1 and 99511.9 U mg−1, the Km values of 3.76 mg ml−1 and 3.98 mg ml−1, and the Vmax values of 57538.5 U mg−1 and 108722.1 U mg−1, respectively. Secreted expression of AprA-PPC in Bacillus subtilis after 48 h cultivation resulted in yield of 4935.5 U ml−1 with productivity of 102.8 U ml−1 h−1, which is the highest reported in literature to date. Without adding any lime or sodium sulfide, both of which are harmful pollutants, AprA-PPC was effective in dehairing cattle hide and skins of goat, pig and rabbit in 8–12 h without causing significant damage to hairs and grain surface. Our results suggest that AprA-PPC may have great potentials for ecofriendly dehairing of animal skins in the leather industry.

Other than the heavy use in laundry detergents as cleansing additives, alkaline proteases have become increasingly important in dehairing animal skin and hides in the leather industry 29 . The conventional dehairing process is performed with a saturated solution of lime and sodium sulfide, which is known for high chemical oxygen demand (COD), biological oxygen demand (BOD), total suspended solid (TSS), total dissolved solids (TDS) and sulfide. As a consequence, the conventional process has been estimated to contribute 60-70% of the total pollution load associated with leather industrial processing 29,30 . Furthermore, the extensive use of sulfide is harmful to the health of the workers 31 . In addition, chemical processing is notorious for damaging leather besides weak performance in dehairing. Therefore, the enzyme-based dehairing process using alkaline protease, which holds promise to substantially reduce or replace the conventional lime and sodium sulfide effluent in addition to an improvement of leather quality, is being intensively pursued as an eco-friendly and viable alternative 31,32 .
Despite the advantages and increasing interests and demands, compared with laundry detergent, only a few alkaline proteases have been reported for dehairing in leather industry. These alkaline proteases are mostly from Bacillus 4,30,33,34 , Paenibacillus 35 , along with Vibrio 31 , Pseudomonas 36,37 , and Aspergillus 38,39 . There are still significant limitations in exploiting these alkaline proteases for industrial used, which include high cost for manufacturing, instability over wide range of pH and temperature, low tolerance to chemical reagents in leather processing, poor performance on dehairing, and significant damage to collagen 31,34,36 . Therefore, the search and development of novel alkaline proteases with high activity and complementary properties suitable for leather processing remain a challenge 31 .
In the present study, we isolated alkaliphilic bacteria from a soda lake in Inner Mongolia, China. An alkaline protease gene (aprA) from one of the strains, Idiomarina sp. C9-1, was cloned and expressed in Escherichia coli and Bacillus subtilis. We purified and characterized the biochemical properties of the recombinant enzyme and evaluated its ability in dehairing cattle hide and skins of goat, pig and rabbit. Our results showed that several features of this alkaline protease made it a promising enzyme for leather processing industry. To our knowledge, this is the first study on alkaline protease from a microorganism of genus Idiomarina.

Results
Screening and identification of protease-producing strains. After two enrichment-cultivation cycles, 41 strains showing protease activity were isolated from a soda lake using isolation agar plate. Through secondary screening in the format of shaking flask fermentation, 16 strains showing relatively higher activity (>50 U ml −1 ) were selected for further characterization. Crude enzymes from all 16 strains showed maximum activity at 50-60 °C and pH10.0-11.0, and were stable at temperature below 60 °C. Five strains termed C9-1, 3A-1, B4-1, DA1-1 and N1 were selected for further activity assessment and strain identification.
To identify the isolated strains, we sequenced 16 S rRNA genes and blasted against sequence database. As shown in Table 1, the five strains showed 99% sequence identity to Idiomarina sp. ST3, Pseudidiomarina sp., Halomonas campisalis, Vibrio metschnikovii, and Bacillus pseudofirmus, respectively. Next, we obtained the crude enzymes and examined their activity at different temperature and pH. The crude enzymes from all five strains showed optimal temperature at 60 °C except for N1, whose optimal temperature was 50 °C. The crude enzyme from 3A-1 showed maximum activity at pH11.0, whereas that from C9-1 was most active at pH10.5. The crude enzyme from all other three strains had an optimal pH of 10.0. In addition, we tested the stability of the crude enzymes. The crude enzymes from C9-1, 3A-1 and DA1-1 were more stable than the enzymes from B4-1 and N1. All five crude enzymes had good tolerance to JFC-2 and Peregal-O which usually occurs in leather industrial processes.
Gene cloning and sequence analysis of the AprA protease. A genomic DNA library containing approximately 30,000 clones of Idiomarina bacterium C9-1 was successfully constructed. Of these, approximately 10,000 clones were screened and one positive clone with protease activity was obtained. An open reading frame (ORF) of 1,878 bp termed gene aprA was obtained by DNA sequencing. The aprA gene was predicted to encode a 625-amino-acid protein, which we termed protease AprA. The phylogenetic tree of AprA and some other highly homologous (identity ≥55%) proteases was generated using MEGA 7.0 based on amino acid sequence. As shown in Fig. 1, the AprA showed significant relatedness to peptidase from Idiomarina bacteria: they are localized at the same branch in the phylogenetic tree, indicating a common evolutionary origin.
The  44 with 56%, 55%, 55%, 50% and 48% identity, respectively. Based on the phylogenetic tree and sequence identity analysis, we concluded that AprA is a serine protease that belongs to the S8A subfamily as classified by the MEROPS peptidase database (http://merops.sanger.ac.uk/). Based on the sequence alignment ( Fig. 2), we proposed that the conserved catalytic triad of AprA could be Asp213-Thr-Gly (Asp213 as the active site), His275-Gly-Thr-His (His275 as the active site), and Gly-Thr-Ser456-Met-Ala-Ala-Pro (Ser456 as the active site). A typical amino-terminal signal peptide with 23 amino acids ( Fig. 2) was identified using the SignalP 4.1 server (http://www.cbs.dtu.dk/services/SignalP/). Like some other peptidases, protease AprA also has an N-terminal domain (R59-Y186), a catalytic domain (G204-G502), and a pre-peptidase C-terminal (PPC) domain (E540-V608) (Fig. 2). The N-terminal sequencing and Nano-LC-MS/MS analysis showed that the N-terminal amino acid sequence of the mature AprA was AFQRSMGLPN (Fig. 2). This N-terminal sequence was different from other reported bacterial alkaline serine proteases including the 3 highly similar alkaline proteases from Pseudoalteromonas sp. 2-10, Pseudoalteromonas sp. A28 and S. maltophilia BBE11-1, which have a conserved five N-terminal amino acids of S/A/LTPND 40,41,44 . The modeled structure of full-length AprA was built (Fig. 3) based on the structure of a subtilisin-like serine protease (ProN-TK-SP) from Thermococcus kodakaraensis (PDB ID: 3AFG) which exhibited the best sequence coverage and about 32% total sequence identity with AprA in PDB database. The modeled AprA structure confirmed that there were also three functional domains and a linker comprising residues from 514 to 538 between the catalytic and the C-terminal domain. According to a previous report 45 , truncation close to a catalytic domain can result in the loss of activity. Therefore, Ala536 located inside the linker region but slightly distant from catalytic domain (Fig. 3) was mutated to generate the PPC domain truncation form of AprA.
The ProN-TK-SP contains two calcium ion binding sites located in PPC domain: Ca-I and Ca-II 46 . In the Ca-I binding site, the Ca 2+ is coordinated with the side chain of Pro397, Ile400, Asp474 and Try475. In the Ca-II binding site, the Ca 2+ is coordinated with the side chain of Asp460, Leu461, Asp462, Glu484 and Thr478. Despite the similarity in modeled structure, however, no similar calcium ion binding sites were found in AprA PPC domain (Fig. 2). Interestingly, a catalytic domain of a subtilisin-like protease AprV2 from Dichelobacter nodosus (PDB ID: 3LPA), which exhibiting a higher sequence identity (50%) to mature AprA, was also found in PDB database. In this structure, three calcium ion biding sites are showed 47 . The Ca 2+ is coordinated with Asp4, Asp48, Val115, Asn118, Ile120 and Val122 in the Ca-I binding site, whereas Asp70, Gly71 and Asp73 are coordinated with Ca 2+ in the Ca-II binding site. In the Ca-III binding site, Ca 2+ is coordinated with Asp58, Asp68 and Asp75. As shown in Fig. 2, the corresponding conserved residues in Ca-II (Asp243, Gly244, Asp246) and Ca-III (Asp231, Asp241 and Asp248) binding sites were also found in the catalytic domain of AprA. Moreover, most catalytic domains of proteases that exhibit sequence similarity to AprA in the PDB database have 2-6 Ca 2+ binding sites (data not shown). Based on above analysis, we speculated that there might be at least two Ca 2+ binding sites in AprA catalytic domain. However, the expression of AprA and AprA-PPC genes without signal peptide sequence was poor and almost no activity was detected after 20 h induction. These results indicated that the signal sequence might be important for expression of AprA in E. coli but might not contribute to its secretory expression, and also suggested that the PPC domain plays no roles in AprA secretion. Interestingly, protein degradation was observed in the cell lysis supernatant after boiling processing without addition of the protease inhibitor PMSF (Phenylmethylsulfonyl fluoride) (Fig. 4A), indicating that AprA and AprA-PPC exhibited proteolysis activity towards other proteins. The recombinant AprA and AprA-PPC were purified to homogeneity with 10.2-and 8.1-fold purification, respectively. As shown in Fig. 4B, the mature AprA and AprA-PPC had a molecular mass of about 56 and 44 kDa, respectively, in line with the calculated molecular weight deduced from the amino acid sequence. Interestingly, the purified protein samples prepared without adding PMSF were partly degraded as shown on SDS-PAGE (Fig. 4B), indicating an autohydrolytic activity. Similar autohydrolysis phenomena have also been observed in other proteases 42,45,[48][49][50] . However, unlike those proteases, autohydrolysis of AprA only occurred at high temperature (>60 °C), whereas it was stable at and below room temperature (data not shown).
Effects of pH and temperature on the activity and stability of purified recombinant AprA and AprA-PPC. Ca 2+ was previously considered to be important for the stability of some serine proteases 16,51 .
Therefore, in our study, the effects of pH and temperature on recombinant AprA and AprA-PPC was determined using casein as substrate with or without adding 5.0 mM CaCl 2 . As shown in Fig. 5A, both AprA and AprA-PPC showed the optimal activity at pH 10.5, the same as that of the crude enzyme from the original Idiomarina sp. C9-1, but higher than that of similar alkaline proteases characterized before [40][41][42][43][44] . Over 60.0% of the activity was maintained at the pH range from 8.0 to 11.0. The activity declined rapidly at pH greater than 11.0. Ca 2+ showed no significant influence on optimal reaction pH (data not shown). Both of AprA and AprA-PPC showed good  stability at a wide pH range from 7.0 to 11.0 with or without CaCl 2 , and over 80.0% of the original enzyme activity was retained after 2 h at 37 °C ( Fig. 5B,C). However, the stability of both enzymes at pH 12.0 was dramatically improved by the addition of 5.0 mM Ca 2+ (90% of activity retained compared with complete inactivation). This indicates that Ca 2+ is important for AprA and AprA-PPC to maintain their stability and activity under highly alkaline condition. Next we characterized the protease activity and thermostability of AprA and AprA-PPC under different temperatures. For both enzymes, more than 60.0% of the activity was maintained in the range of 40-70 °C (Fig. 6A,B). The optimal temperature for both enzymes was 60 °C without Ca 2+ , the same as that of crude enzyme from the original Idiomarina sp. C9-1, but higher than those for the five similar alkaline proteases characterized before. With 5.0 mM Ca 2+ , the optimal temperature increased to 70 °C. The upward shift of optimal temperature by Ca 2+ has been reported for SPTC protease from Trametes cingulate 52 . In addition, adding Ca 2+ also broadened the temperature range where more than 60.0% of activity was maintained. Furthermore, the relative activity of both enzymes in the presence of Ca 2+ was obviously higher than that in the absence of Ca 2+ . The thermal stability assay showed that the half-life of AprA was 70 min at 65 °C, whereas only 21.0% and 8.0% of the original activity was retained after 30 min incubation at 70 °C and 10 min incubation at 75 °C, respectively (Fig. 6C). However, with 5.0 mM Ca 2+ , more than 90.0% of the original activity was retained after 70 min incubation at 65 °C, and approximately 65.0% and 68.0% of the original activity was retained after 60 min incubation at 70 °C and 10 min incubation at 75 °C, respectively (Fig. 6C). For AprA-PPC, the half-life was 120 min at 65 °C, whereas only 38.0% and 36.0% of the activity was retained after 30 min incubation at 70 °C and 10 min incubation at 75 °C, respectively ( Fig. 6D). With 5.0 mM Ca 2+ , no activity loss was found after 120 min incubation at 65 °C, and approximately 70.0% and 60.0% of the original activity was retained after 120 min incubation at 70 °C and 30 min incubation at 75 °C, respectively (Fig. 6D). Meanwhile, the residual activity was slightly activated by approximately 11.7% and 4.1% upon incubation for 10 min at 65 °C and 70 °C, respectively. These results indicated that Ca 2+ can improve not only the thermal activity of AprA and AprA-PPC but also the stability at high pH and temperature. In addition, the half-life of AprA-PPC was 120, 25, and 8 min at 65 °C, 70 °C and 75 °C, respectively, whereas that of AprA was 70, 15, and 4 min at the corresponding temperature. Therefore the thermal stability of AprA-PPC was better than that of AprA under the same condition regardless of the presence of Ca 2+ , which indicated that the PPC domain negatively affected the thermal stability of the AprA protease.

Effects of metal ions and chemical reagents on the activity of the purified recombinant enzymes.
The effect of metal ions and various chemicals on the activity of AprA and AprA-PPC was also evaluated and the results were shown in Table 2. For both AprA and AprA-PPC, Ca 2+ , Cu 2+ , Ba 2+ , Mn 2+ , Mg 2+ , and Co 2+ increased whereas Ag + , Zn 2+ , Fe 2+ , Fe 3+ , Ni 2+ , Pb 2+ and Hg 2+ substantially decreased the enzyme activity. Other tested metal ions including K + , Na + , Al 3+ and Li + did not markedly affect the activity. The serine protease inhibitor PMSF completely inactivated both enzymes, indicating that AprA is a typical serine protease. The inhibiting effect of EDTA (Ethylene diamine tetraacetic acid) and EGTA (Ethylene glycol tetraacetic acid) gradually increased as their concentrations increase (Table 3). This might have occurred because of the chelation on Ca 2+ . The inhibitory effect of EDTA and EGTA suggested a requirement of metal ions for optimal activity. Some serine proteases have two Ca 2+ binding sites, and Ca 2+ removal by chelators would result in a significant decrease in thermal stability 53 . It is also interesting to note that AprA-PPC tolerated higher concentrations of EDTA and EGTA better than AprA.
Tolerance to surfactants, oxidizing and reducing agents are important for industrial application of proteases 29 . Our results showed that most of the tested reagents only slightly inhibited AprA activity except the surfactant SDS (Sodium dodecyl sulfate), which decreased the activity by approximately 25.0% (Table 3). SDS is an amphiphilic organosulphate that normally interacts with amino acid residues, causing protein unfolding therefore loss of enzymatic activity 16 . Similar effect was also found for AprA-PPC. However, AprA-PPC showed better tolerance to JFC-2, Peregal-O, β-Mercaptoetanol and high concentrations of EDTA and EGTA than AprA (Table 3). Both   Table 3). The good tolerance to surfactants as well as oxidizing and reducing agents make AprA potential enzyme for industrial application.
Substrate specificity and kinetic parameters of the purified recombinant enzymes. With respect to substrate specificity, both AprA and AprA-PPC showed specific activity towards casein, keratin, skim milk, BSA (Bull serum albumin) and gelatin to different extent. The activity of AprA towards these substrates, when normalized to casein as 100%, was 51.3% for keratin, 47.2% for skim milk, 17.3% for BSA and 8.0% for gelatin (Table 4). Conversely, only 2.5% activity on collagen was found. Almost the same results were obtained for AprA-PPC. AprA-PPC demonstrated the highest activity toward casein with 99511.9 U mg −1 , which was higher than most previously reported alkaline proteases but lower than those from B. clausii I-52 (3.9 × 10 5 U mg −1 ) 11 ,  Table 5, the activity and cell density in the medium containing glucose were the highest likely due to the fact that glucose is helpful to the cell growth, which then resulted in better AprA-PPC expression. The highest extracellular activity (3124.8 U ml −1 ) was obtained in SRG medium after 48 h cultivation. We also evaluated the effect of culture temperature on the expression of AprA-PPC. A higher activity (4935.5 U ml −1 ) was obtained when the strain was cultured at 30 °C after 48 h, which indicated that lower temperature was better for AprA-PPC expression in B. subtilis. The yield was also 65 times of the native enzyme activity from the original Idiomarina sp. C9-1.
Animal skins dehairing evaluation by alkaline protease. The dehairing ability of crude AprA-PPC from WB600A on cattle hide and goat skins were studied. As shown in Fig. 7, almost complete and uniform hair removal was observed on cattle hide and goat skins after 10-12 h treatment with alkaline protease AprA-PPC. Compared with chemical treatment, the enzyme treated skins were smoother, whiter, softer and slightly thinner. The recovered hair from enzymatic dehairing was intact owing to the absence of hair destructing sulfide, whereas the hair recovered from chemical dehairing was normally pulped with damage (Fig. 7). The SEM (Scanning electron microscope) was also used for morphological studies to compare the conventional chemical-treated with enzyme-treated skin samples. As shown in Fig. 8, the conventional pelt sample showed a white surface and several particles, which indicated the deposition of lime. In contrast, the enzyme-treated sample showed a clear surface without any deposition of foreign particles or grain damage. Although almost all the hair pores of conventional cattle hide sample were clear without unremoved hair root, there were many foreign particles around the hair pores (Fig. 8A). Chemical-treated goat skin sample showed the presence of unremoved hair root on the hair pore, indicating incomplete depilation (Fig. 8C). In contrast, enzyme-treated cattle hides and goat skins showed clean surface without any major deposition of foreign particles (Fig. 8B,D). We also tested the enzymatic dehairing effect of AprA-PPC on pig and rabbit skins. As shown in Fig. 9, almost complete hair removal was achieved after only 8 h of treatment with AprA-PPC. The effective dehairing without using damaging chemicals thus makes AprA-PPC an eco-friendly alkaline protease for leather industry.

Discussion
Alkaliphilic microorganisms, derived mostly from alkaline environments such as soda lakes, constitute the main sources of alkaline proteases. In this study, we isolated several alkaline proteases producing alkaliphiles from soda lake in Inner Mongolia, China. Among them, Idiomarina sp. C9-1 produces an alkaline protease AprA with the highest activity and best tolerance to surfactant reagents. Genus Idiomarina comprises gram-negative, mesophilic and aerobic bacteria, and most reported strains were moderately halophilic or/and haloalkaliphilic 55 . To our knowledge, no protease from bacteria in this genus has been characterized. Our discovery of alkaline protease AprA from Idiomarina sp. C9-1 thus was the first case. AprA has low sequence similarity to other reported proteases (with the highest identity of 63.0% to directly submitted sequence in NCBI database), and is thus a novel alkaline serine protease. AprA exhibits higher optimal pH (10.5) and better stability under alkaline conditions than many previously reported alkaline proteases from Bacillus 4,6,8,9,15,16,27,33,51,[56][57][58][59][60][61] (Table S1). Furthermore, AprA also retains high activity and stability over a wide range of pH (7.0-11.5) and temperature (40-70 °C), several metal ions, surfactants and some oxidizing and reducing agents. These characteristics render AprA suitable for a variety of applications. In the leather manufacturing process, the pH and temperature varies depending on the region and season, and substantial amounts of Ca 2+ , Na + , and surfactants such as JFC-2 and Peregal-O are involved 69,70 . Therefore, AprA is especially suitable for leather manufacturing. AprA and AprA-PPC are also very stable in the presence of the oxidizing agent H 2 O 2 . In fact, their activities were increased by approximately 20.0% in the presence of 2.0% (v/v) H 2 O 2 ( Table 3). Tolerance to H 2 O 2 has also been found for protease rBLAP from B. lehensis 16 and AprX-SK37 from Virgibacillus sp. SK37 64 . Activation by low concentration of H 2 O 2 has also been shown in protease SBcas3.3 71 and SV1 72 . Generally, subtilases are inactivated by H 2 O 2 because of methionine oxidization next to the catalytic serine, which then inhibits the formation of a tetrahedral intermediate during proteolysis 64,73 . AprA also has a methionine residue (Met457) that is located just after the catalytic serine (Ser456). However, the study of the oxidant-stable protease KP-43 showed that the oxidation of Met was not a fatal modification 74 . The structure analysis suggested that the rate of Met-oxidation in KP-43 was lower than those for other subtilases, probably due to the longer distance between the Met residue in the catalytic vicinity and the oxyanion hole 75 . Therefore, the oxidant-stable ability might depend on the intrinsic conformation and structural integrity 64 . In addition, the Met-oxidation in KP-43 also altered its substrate specificity. Therefore, we speculated that the activation of AprA by H 2 O 2 might also be attributed to the effect of free radicals and perhydroxyl anions on its substrate in the catalytic center. Alkaline proteases exhibiting resistance to oxidizing agents and alkaline conditions are suitable for use as detergent additives 16 . Our study suggests that AprA could also serve as a promising candidate additive in the detergent industry.
Proteases are usually classified as a multi-domain enzyme. The PPC domain is generally found at the C-terminus of certain secreted bacterial peptidases such as some metalloprotease families and serine protease family S8, and usually is cleaved after secretion although prior to protease activation 76,77 . As shown in Fig. 2, the alignment of the C-terminal PPC domain from AprA and other homologies showed many hydrophobic residues such as valine, proline and the aromatic residue phenylalanine in the C-terminal domain. Previous structural homology modeling analyses showed that PPC domain comprised a parallel beta-sheets domain that was coupled to a catalytic domain through a loop linker [44][45][46] , which was also confirmed by structural modeling of AprA (Fig. 3). This might be a site prone to aggregation. The actual function of PPC domain remains unclear and there were only a few studies so far. For example, the PPC domain of the metalloprotease from V. vulnificus was essential for efficient attachment to insoluble substrates and erythrocyte membranes 78 . Proteases AprI and AprII displayed lower activity than the PPC domain truncated enzymes 42 . Studies on protease MCP-03 showed that the PPC domain decreased the catalytic activity but improved the thermostability 50  PPC domain stabilized protease Pro-TK-SP 46 . Additional research showed that the PPC domain truncation of protease HP70 and StmPr1 could improve the expression level of active protease in E. coli and the specific activity compared with the native enzyme 44,45 . In the current study, the specific activity of the PPC truncated enzyme, AprA-PPC (99511.9 U mg −1 ), was approximately 2.3-fold that of the parent enzyme AprA (42567.1 U mg −1 ), consistent with some studies 42,44,45,50 . However, opposite to ProN-TK-SP 46 and MCP-03 50 , AprA lacking PPC domain shows enhanced thermostability (Fig. 5C,D). Moreover, the tolerance of AprA-PPC to JFC-2, Peregal-O, β-Mercaptoethanol and high concentrations of EDTA and EGTA is improved (Table 3). Combining previous and our current studies, we speculate that the PPC domain is related to the following functions: attachment to insoluble protein substrates (positive effect), thermal stability (positive or negative effect), catalytic efficiency and activity (negative effect), pH and surfactants stability (negative effect), and enzyme secretion (negative effect). Ca 2+ and other divalent ions such as Mn 2+ and Mg 2+ represent additional factors known to increase protease activity and thermostability 65,79 . Our current study showed that both AprA and AprA-PPC were activated by Ca 2+ , Mn 2+ , and Mg 2+ . These ions have been reported to activate proteases from organisms such as Caldicoprobacter guelmensis 75 , Streptomyces koyangensis 25 , T. cingulate 52 , and Bacillus circulans 56 . These studies suggest that the bivalent ions may stabilize enzyme structure, especially the active conformation at high temperature and protect the enzyme against thermal denaturation 5,47,52 . However, Cu 2+ , Ba 2+ , and Co 2+ found to be inhibitory in these studies can stimulate AprA and AprA-PPC. The thermal denaturation experiments by GdnHCl (Guanidine Hydrochloride) inactivation also confirmed the results (Fig. S1). The residual activity of AprA and AprA-PPC after incubation at 60 °C for 10 min in the presence of Ca 2+ , Mg 2+ , Mn 2+ or Cu 2+ was higher than that in the absence of divalent metal ions, which indicated that these four divalent metal ions may stabilize the structure of AprA and AprA-PPC and protect the enzymes against thermal denaturation to various degrees. However, Co 2+ and Ba 2+ did not show the same effect on AprA and AprA-PPC. On the other hand, the activity of AprA and AprA-PPC was significantly but not completely inhibited by certain other bivalent ions including Hg 2+ , Pb 2+ , Ni 2+ , Fe 2+ , Zn 2+ , and the monatomic ion of Ag + . These toxic metallic ions might bind to particular organic ligands resulting in enzyme denaturation 52 . Notably, the inhibitory effect of heavy metallic ions such as Hg 2+ and Pb 2+ has been found to be a common phenomenon in many enzymes including proteases and is well documented in the literature. For example, Hg 2+ is known to react with protein thiol groups such as histidine and tryptophan residues 80 . We also found that the thermostability of both AprA and AprA-PPC was also significantly improved by the addition of Ca 2+ . Similar results were also observed for several alkaline proteases from organisms such as B. lehensis 16 , B. pumilus 57 , A. oryxae 81 , S. koyangensis 25 , and Conidiobolus brefeldianus 79 . Ca 2+ is considered to effect protease thermostability by strengthening the intramolecular interactions of the enzyme and by binding of Ca 2+ to autolysis sites 79 . The result of thermal denaturation above in the presence or absence of Ca 2+ also is in line with previous results.
Besides the enzymatic properties, high yield is also critical for industrial application. In the past years, species such as E. coli 27,43 , B. subtilis 58,[82][83][84][85] , Pichia pastoris 12,62,[86][87][88] , Yarrowia lipolytica 23 , Saccharomyces cerevisiae 89,90 and Zygosaccharomyces rouxi 91 have been used as hosts for the heterologous production of alkaline proteases (Table 6). However, very few alkaline protease genes from other genus such as Aspergillus 23,62,86,88,89,91 and Stenotrophomonas 43 have been successfully expressed. B. subtilis probably is more suitable for alkaline protease production in that the highest protease activity (5800.0 U ml −1 was obtained after 72 h incubation) has been reported 57 . In the present study, the AprA-PPC gene from Idiomarina sp. C9-1 was successfully expressed in B. subtilis and the extracellular activity of 4935.5 U ml −1 was obtained after 48 h cultivation with the highest productivity of 102.8 U ml −1 h −1 reported in literature to date (Table 6). This was the first report on heterologous expression of protease gene from genus Idiomarina in B. subtilis. Also, this yield was higher than most of other recombinant strains reported (Table 6). Moreover, this expression level can also be further improved by systematic optimization to a higher yield.
Lime and sulfide constitute the main pollution generated during dehairing process in the leather industry. In recent years, enzymatic dehairing by protease have been reported 32,38,69 , however, many remains using lime and sulfide. Among these processes, enzymes carrying sparingly soluble kaolin or soluble silicates were used for lime and sulfide free enzymatic dehairing, which can still cause a significant increase in COD, BOD, TDS and TSS of the effluent 38,92 . Alkaline protease preparations specifically from organisms such as V. metschnikovii 31 , some Bacillus strains 30,93,94 , and Pseudomonas fluorescens 37 have also been reported for lime and sulfide free dehairing. Nevertheless, these enzymes were generated directly from strain cultures with low activity. Alternatively, in the current study, the alkaline protease AprA-PPC with higher specificity and activity was efficiently expressed in B. subtilis, yielding high activity in culture medium. The crude enzyme was very effective in dehairing animal skins after 8-12 h treatment without any use of lime and sulfide. Also, the dehairing time was less than reported 12 to 24 h for different animal skins in the literature 4,30,37,38,95 , and only longer than the processes by a commercial protease 34 and an alkaline protease from B. subtilis BLBc 11 93 (6 h duration). A shorter dehairing process by alkaline protease not only mitigates damage to collagen 34 , but also reduces constraint of a high degree of control. In addition, the dehairing of animal skins by AprA-PPC yielded intact hair, which may represent a valuable byproduct. Further, approximately 70% of the waste from pretanning processes is resulted from hair rich in nitrogen 34 . Enzymatic dehairing with intact recovered hair would thereby significantly reduce COD in the process.
In summary, our present study is the first to characterize protease from genus Idiomarina, and the first to express the protease heterologously in B. subtilis. The alkaline protease AprA and its PPC domain truncated enzyme AprA-PPC from Idiomarina sp. C9-1 showed high activity and good stability over a wide range of pH and temperature, and also displayed excellent tolerance to some surfactants as well as oxidizing and reducing agents, which hold promises for broad industrial applications. Its performance in dehairing animal skins is advantageous over the chemical process, making it especially valuable in the leather industry. In addition, our study on the PPC domain provides new information to better understand the function of proteases across different species. NaCl. The isolation medium was made from enrichment medium by replacing tryptone to skim milk and added 2% (w/v) agar. The fermentation medium was obtained from enrichment medium by changing the concentration  Strains, plasmids and materials for gene cloning and expression. The plasmids and bacteria strains used in this study for gene cloning and expression were listed in Table S2. Casein, skim milk, keratin, gelatin, collagen and bovine serum albumin (BSA) were from Sigma-Aldrich (St. Louis, MO, USA). All the enzymes for DNA manipulations were purchased from NEB (Dalian, Liaoning, China). Phenylmethanesulfonyl fluoride (PMSF), isopropyl-β-D-thiogalactopyranoside (IPTG), imidazole, ampicillin, and kanamycin were from Amresco Inc. (Solon, OH, USA). All other chemicals used in this study were of reagent grade.

Gene cloning and expression plasmid construction of alkaline protease AprA. Idiomarina sp.
C9-1, which was isolated from a soda lake in Hulunbuir of Inner Mongolia and exhibited high alkaline protease activity, was used as DNA source for the alkaline protease (AprA) gene cloning. Idiomarina sp. C9-1 was also preserved in the China General Microbiological Culture Collection Center (CGMCC 1.16117). The restriction enzyme Sau3A was employed to obtain randomly digested chromosomal fragments. These 3.0-to 8.0-kb fragments were recovered and purified by the Gel Extraction Kit (OMEGA Bio-tek) and ligated into the BamHI digested pUC118 vector treated with alkaline phosphatase. The ligation product was then transformed into E. coli DH5α cells by electroporation transformation and plated onto the screen medium (pH 8.0) containing 5.0 g l −1 yeast extract, 10.0 g l −1 tryptone, 10.0 g l −1 NaCl, 10.0 g l −1 skim milk, 15.0 g l −1 agar and 60.0 μg ml −1 ampicillin. After incubation at 37 °C for 24 h, the colony with transparent zone was selected for the inserted fragment sequencing by SinoGenoMax Co., Ltd. (Beijing, China).
The AprA-encoding gene (aprA) was obtained by PCR using the primer pair F1 (5′-CATGCCATGGGCATGA AGAATGTTAAAACATT-3′, where the underline indicates the NcoI site) and R1 (5′-ACCGCTCGAGCGGCTG GTAATTTGCTTCAA-3′, where the underline indicates the XhoI site). The enzyme without pre-peptidase C-terminal (PPC) domain (AprA-PPC) encoding gene fragment was obtained by PCR using the primer pair F1 and R2 (5′-ACCG CTCGAGCGCAGATAAGTCCGACACGC-3′, where the underline indicates the XhoI site). The AprA and   Effects of pH, temperature, and reagents on enzyme activity and stability. The optimal pH of the purified proteases was assayed at 60 °C in 50.0 mM Tris-HCl buffer (pH 7.0-8.5), 50.0 mM glycine-NaOH buffer (pH 8.5-10.5), and 50.0 mM Na 2 HPO 4 -NaOH buffer (pH 11.0-12.0) containing 2.0% casein (w/v). The optimal temperature was assayed at 40-100 °C for 10 minutes in 50.0 mM glycine-NaOH buffer containing 2.0% casein (w/v) (pH 10.5) with or without 5.0 mM CaCl 2 . The effect of pH on enzyme stability was assayed by incubating enzyme in the previously described buffers 97 for 2 h at 37 °C. Thermal stability was analyzed by assessing enzyme   The effect of metal ions and PMSF on enzyme activity was analyzed by assaying the relative activity with addition of 5.0 mM of metal ions (Na + , K + , Li + , Ag + , Mn 2+ , Zn 2+ , Cu 2+ , Fe 2+ , Ca 2+ , Co 2+ , Ba 2+ , Fe 3+ and Al 3+ ) and PMSF. Similarly, the effect of ethylene diamine tetraacetic acid (EDTA) and ethylene glycol tetraacetic acid (EGTA) on enzyme activity was also analyzed with addition of different concentration (2.0, 5.0 and 10.0 mM). The effect of some surfactants, oxidizing agents and reducing agents on enzyme activity were also analyzed. 1.0% (v/v) of Tween-20, Tween-40, Tween-80, Triton X-100, Triton X-155, JFC-2, β-Mercaptoethanol, Peregal-O and Sodium dodecyl sulfate (SDS) were added for the relative activity evaluation. Meanwhile, different concentration (1.0%, 2.0% and 5.0%, v/v) of H 2 O 2 was also used for the effect determination on enzyme activity. Substrate specificity and kinetic parameters. The substrate specificity of the purified proteases was assayed at pH 10.5 and 60 °C using 1.0% (w/v) substrates, including casein, BSA, gelatin, keratin, and skim milk. The activity toward casein was set as 100%. The kinetic parameters toward the substrates were determined by incubating appropriate diluted enzyme and substrate with concentration from 0.5 to 20.0 mg ml −1 at 60 °C for 5 min in 50.0 mM glycine-NaOH buffer (pH 10.5). The K m and V max values were calculated by the GraphPad Prism 5.0 software (http://www.graphpad.com/prism/) using non-linear regression. All data are expressed as the means of triplicate measurements.

Secreted expression in Bacillus subtilis.
For expression in B. subtilis, the AprA-PPC encoding gene with the original signal peptide was codon optimized for B. subtilis and synthesized and ligated to pMA5 plasmid (pMA5-AprA-PPC) by TsingKe Biotech Co., Ltd. (Being, China). Besides the original peptide, signal peptides including SP lipA , SP lipB , SP amyL , SP amyE , SP aprE , SP nprB , and SP nprE from B. subtilis 168 were also used for secreted expression. The expression plasmids containing these signal peptides were constructed by PCR using the modified Gibson assembly method 98 base on the plasmid pMA5-AprA-PPC. Primers used to amplify the signal peptide fragments for expression plasmid construction were listed in  Table 6. Comparison of secretory expression of alkaline proteases in different strains.
Mix (Tsingke Biotech Co., Ltd, China) was used for PCR amplification. The PCR protocols were as follows: denaturation at 98 °C for 2 min, followed by 30 cycles of denaturation at 98 °C for 20 s, annealing at 55 °C for 20 s, and extension at 72 °C for 3 min, then final extension at 72 °C for 5 min. The plasmid pMA5-AprA-PPC was used as the template. The PCR product was purified by a Cycle-Pure Kit (OMEGA Bio-tek) and then digested by DpnI for 6 h. Then 2 µl of this product and 0.5 µl of Taq DNA ligase were added into 7.5 µl of assembly master mixture 48 , and this mixture was incubated at 50 °C for 1 h. Then the product was directly transformed into competent E. coli DH5α and plated onto LB agar plates containing 50.0 μg ml −1 of ampicillin and incubated at 37 °C overnight. The positive colony samples were selected to be further validated by sequencing (Tsingke Biotech Co., Ltd, China). All the confirmed recombinant pMA5-AprA-PPC and its derivative plasmids with different signal peptide (Table S2) were than transformed into B. subtilis WB600 cells by electroporation to form different recombinant B. subtilis WB600 strains for secreted expression. The medium including LB, LBG (LB containing 10% glucose), 2 × LB, 2 × LBG (2 × LB containing 10% glucose), SR (15.0 g l −1 tryptone, 25.0 g l −1 yeast extract, 3.0 g l −1 K 2 HPO 4 ) and SRG (SR containing 10% glucose) which containing 60.0 μg ml −1 kanamycin and 10.0 μg ml −1 chloramphenicol were used for secreted expression. For flask cultivation, 0.5 ml of the seed culture was inoculated into 50.0 ml of these medium in 500 ml flasks and then incubated at 37 °C with 220 rpm.
Enzymatic dehairing evaluation. Samples of cattle hide, goat, pig and rabbit skins were used for enzymatic dehairing evaluation. The hide and skin samples were cut into appropriate pieces (about 3 × 3 cm) and washed with water several times to remove salt and extraneous matter. An optimized dip method was used for enzymatic dehairing. The dry skin samples were soaked in 10 ml of 50.0 mM NaOH-glycine buffer (pH 9.0) with addition of 0.6 ml 48 h culture supernatant of the recombinant B. subtilis WB600A for dehairing treatment at 40 °C for 8-12 h with shaking at 150 rpm. Then the hide and skins were slightly dehaired by rubbing and washing with flowing water. No enzyme addition was used for negative control and the conventional lime-sulfate treatment (by 10.0% lime and 2.0% sodium sulfide based on soaked weight for overnight) was performed as a positive control. The control experiment was also carried out with the same condition as enzymatic treatment. The dehaired hide and skins were evaluated by visual assessment for quality, such as whiteness, softness, hair removal, hair roots and grain surface. The scanning electron microscope (SEM) Hitachi SU8010 (Hitachi, Minato-ku, Tokyo, Japan) was used for these assessments.

Availability of Data and Materials
The datasets supporting the conclusions of this article are included in the manuscript and additional files.