Phytase (myo-inositol hexakisphosphate phosphohydrolase) has been applied for the efficient hydrolysis of phytate and stepwise liberation of the bound phosphate as well as the various chelated metal ions [1]. As an environmentally- friendly enzyme, phytase has been used in the feed and food industries as well as agriculture. Monogastric animals such as swine, poultry and fish are unable to digest phytate [2]. Therefore, phytase is supplemented to their diet to increase the available phosphate content and ease the environmental pollution stemming from the phosphate released as non-digested phytate. However, the enzyme is unstable during the high-temperature palletization process, which greatly limits its application range and value. In view of this, enhancing the enzyme’s thermostability and elevating its industrial robustness have become some of the most urgent demands to resolve in current agricultural biotechnology.

Since phytase was first described in 1907, numerous wild-type and engineered enzymes derived from microbes, plants and animal tissues were reported in the literature [3]. Furthermore, in a recent study, numerous engineered and industrial phytases were applied as in vitro and in vivo catalyst to produce multiple types of high value- added products under high-temperature conditions [4, 5]. Based on these studies, two significantly antagonistic protein engineering strategies were developed, one based on rules gleaned from proteins of natural thermophiles [6], and rational engineering techniques [7]. Numerous non- rational, mostly labor-intensive protein engineering methods such as eqPCR [8], DNA shuffling [9] and mutagenesis [10, 11] were also applied. At the same time, strategies such as metagenome analysis [12] and screening of thermophilic bacteria [13, 14] were also utilized to develop novel heat- resistant phytases. Phytases from well-known micro- organisms such as Aspergillus niger [15], Escherichia coli [16] and Bacillus subtilis [1] have been reengineered by directed evolution employing random mutagenesis and multigene DNA shuffling. Furthermore, bioinformatics analysis and identification of novel thermophilic phytases, such as those from Yersinia mollaretii [17] and Citrobacter amalonaticus [18], also greatly added to the available thermoresistant phytases. However, to cut the labor-associated cost of the high false-positive rate of non-rational methods, a number of rational protein engineering methods were established. Rational strategies such as molecular dynamics (MD) simulation [11, 19] and structural dynamics [20, 21] have been applied to redesign thermophilic proteins. Based on the crystal structure and rational cascade methods, Fei et al. constructed a series of engineered phytases, and the mutants showed a great increase of heat resistance compared to the wild-type [19]. Furthermore, under the guidance of 3D structural information, analysis and calculation of some types of molecular bonds, such as hydrogen bonds [22], salt bridges [23] and disulfide bonds [24], have also been utilized to improve the heat and pH resistance of phytase. In other examples, glycosylation [25] and modification of α/β structure domains [26] were also successfully applied to improve the heat resistance of the protein. In high- resolution X-ray crystallographic studies of the protein, the temperature factor (B-value) is used to reflect the blurring of atomic electron densities and ambiguity of the atomic spatial state in the crystal structure [27]. As reported previously by Manfred [28] and Samuel [29], a reduction of the B-values can increase a protein’s heat stability. In this approach, a reduction of the protein flexibility (entropy) and modification of the surface amino acids can increase the protein’s thermostability to some extent [29]. Thus, the method of protein crystal structure analysis may have a more considerable role in enhancing the stability of thermophilic proteins.

In this study, analysis of temperature factors (B-value) and amino acid surface engineering were applied to enhance the thermostability of the E. coli phytase (GenBank: DQ513832). By combining these methods, thirteen amino acids were selected, categorized into six domains, and used to construct phytase mutant libraries. Via cascade library design, screening and iterative site mutation, the mutant P56214 was obtained, which showed 55% thermostability increasement compared with the wild-type expressed in E. coli. Therefore, this study provides a promising method to increase the thermostability of phytases and promote the development of environmentally friendly enzyme catalysts that are active at increased temperatures.

Strains, Plasmids, and Culture Conditions

The strains, plasmids and primers used in this study are listed in Tables 1, 2, and 3, respectively. The strain DH5α was used for plasmid construction and amplification. The strain BL21 (DE3) was used for screening of the phytase mutant library and characterization of the engineered mutants. LB was used to culture engineered E. coli. Antibiotics (kanamycin sulfate 50 μg/ml and zeocin 30 μg/ml) were added if needed. Agarose was added to a final concentration of 2% for solid-state cultivation.

Table 1 . Strains used in this study..

Strain	Description	Source
DH5α	Escherichia coli (F- φ80lacZΔM15 Δ(lacZYA-argF) U169 recA1 endA1 hsdR17(rk-, mk+) gal- phoA supE44 λ- thi-1 gyrA96 relA1)	Invitrogen
BL21(DE3)	Escherichia coli (F- ompT hsdS(rB-mB-) gal dcm(DE3))	Invitrogen
BLP	BL21(DE3) containing plasmid pET28a	This study
BLP10	BL21(DE3) containing plasmid p10	This study
BLP11	BL21(DE3) containing plasmid p11	This study
BLP12	BL21(DE3) containing plasmid p12	This study
BLP13	BL21(DE3) containing plasmid p13	This study
BLP14	BL21(DE3) containing plasmid p14	This study
BLP15	BL21(DE3) containing plasmid p15	This study
BLP16	BL21(DE3) containing plasmid p16	This study
BLP56	BL21(DE3) containing plasmid p56	This study
BLP52	BL21(DE3) containing plasmid p52	This study
BLP526	BL21(DE3) containing plasmid p526	This study
BLP5261	BL21(DE3) containing plasmid p5261	This study
BLP52613	BL21(DE3) containing plasmid p52613	This study
BLP52614	BL21(DE3) containing plasmid p56214	This study
BLP526143	BL21(DE3) containing plasmid p562143	This study

Table 2 . Plasmids used in this study..

Plasmid	Description	Source
pET28a	E. coli expression vector, KanR	Invitrogen
p10	pET28a containing the original phyE gene, KanR	This study
p11	p10 containing mutation library 1, KanR	This study
p12	p10 containing mutation library 2, KanR	This study
p13	p10 containing mutation library 3, KanR	This study
p14	p10 containing mutation library 4, KanR	This study
p15	p10 containing mutation library 5, KanR	This study
p16	p10 containing mutation library 6, KanR	This study
p56	p15 containing mutation library 6, KanR	This study
p52	p15containing mutation library 2, KanR	This study
p526	p52 containing mutation library 6, KanR	This study
p5261	p526 containing mutation library 1, KanR	This study
p52613	p5261 containing mutation library3, KanR	This study
p52614	p5261 containing mutation library 4, KanR	This study
p526143	p52614 containing mutation library 3, KanR	This study

Table 3 . Primers used in this study..

Primer	Sequence	Description
PhyE-F	AAAGCGATCTTAATCCCATTTTTATCTCTTCTGATTCCGTTAACCCCGCAATCTGCTGAAGCTCAGAGTGAGCC	WT phyE
PhyE-R	TTACTACAAGGAACAAGCTGGGATTCTAG	amplification
pPHy-ZF	GAGAAAAGAGAGGCTGAAGCTCAGAGTGAGCCTGAGTTGAAAC	ISMVII library amplification
pPHy-ZR	GTCTAAGGCTACAAACTCAATGATGATGATGATGATGCAAGGAACAAGCTGGGATTCT
pET-F	TGGGATTAAGATCGCTTTCATGGTATATCTCCTTCTTAAAG	pET28a linearization
pET-R	GCTTGTTCCTTGTAGTAACTCGAGCACCACCACCACCACCACTGAGATCCGGCT
MU1-F	AGACGCTTGGCCAACCTGGDBHKBYWNCTGGGTGAATTGACACCTAGAGG	Library 1
MU1-R	CAGGTTGGCCAAGCGTCTGGGGTGACACATTGCATAAG
MU2-F	TAAGTGTGGTTGTCCACAANNKGGTCAAGTAGCTATTATTGC	Library 2
MU2-R	TGGACAACCACACTTAGGCAACAATTCGTCGGCAAC
MU3-F	CCAACTGTTGCCTTAAGYHYHBKAAGVHDGACGAATCCTGTTCCTTGACTCAAGC	Library 3
MU3-R	CTTAAGGCAACAGTTGGATTGTGGGAAGTTAAGAACTC
MU4-F	GCTTTGACTCCTCACCCACCTDNKNNYCAAGCCTACGGTGTTACCTTGCC	Library 4
MU4-R	CAAGTCCAAC
MU5-F	TTCGAAAGATGGCGTAGACTANNKGATAACTCTCAATGGATTCAGGTTTC	Library 5
MU5-R	CTACGCCATCTTTCGAAAACGAGCTCACCACCTGGT
MU6-F	TGACCTTGGCTGGATGTNHWNDYVMKAATGCTCAGGGTATGTGTTC	Library 6
MU6-R	CATCCAGCCAAGGTCAATTTGACTTCTCCTGGAGGC
ISM1-F	ACCTGGTGGTCTCTCCCTGGTGAATTGACACCTAGAGG	Iterative mutant library 1-6
ISM1-R	CAGGAGAGACCACCAGGTTGGCCAAGCGTCTGGGGTGAC
ISM2-F	TGTGGTTGTCCAGACATTGGTCAAGTAGCTATTATTG
ISM2-R	AATTTGTGGACAACCACACTTAGGCAACAATTCGTCGG
ISM3-F	AGTCTAGTAACGCACGAGGTTCCTGTTCCTTGACTCAAGC
ISM3-R	TCCGTCTGCCTTACTAGACTTAAGGCAACAGTTGGATTGT
ISM4-F	CACCCATGGGATTACAGGGCCTACGGTGTTACCTTGCCC
ISM4-R	TTGGTAATCAGGTGGGTGAGGAGTCAAAGCAGTCTTGA
ISM6-F	CTGGATGTGCCGTGGCTAATGCTCAGGGTATGTGTTC
ISM6-R	AGCCACGGCACATCCAGCCAAGGTCAATTTGACTTCTC

Note: the bold letters represent the different degenerate primer sequences..

Determination of the Phytase Mutation Domains

To design the appropriate mutant libraries, crystal structure files and sequences of phytases from the histidine acid phosphatase (HAP) family were downloaded from the PDB database, and the amino acid sequences aligned using Clustal [30] and SMS (Sequence Manipulation Suite). MEGA 6 was used to construct the phylogenetic tree of the HAP phytases. The online software SWISS-MODEL was used to construct the models of the phytase mutants. PyMOL was used for the modification of the structures 1DKL and 1DKQ and analysis of mutant sites. Protein surfaces were analyzed using Swiss PDB viewer. The B-values of amino acid residues were calculated using B-FITTER [7]. The degenerate primer sequences were designed and mutant library pools were optimized using Evolution Tools CASTER [28].

Phytase Mutant Library Construction

To construct the different mutants, sequences were introduced into different original vectors using the designed primers (Table 3) by whole-plasmid amplification. The mutation efficiency was confirmed by sequencing (GENEWIZ, China). The whole- plasmid amplification samples were digested with DpnI (New England Biolabs, NEB) and introduced into E. coli BL21 (DE3). Q5 DNA Polymerase (New England Biolabs, NEB) was utilized for gene amplification and plasmid mutation in all experiments.

Screening, Iterative Mutagenesis and Characterization of Phytase Mutants

After B-value analysis and amino acid surface engineering, a total of 13 amino acids were obtained, located in six domains. A phytase mutant library was constructed for every domain. Each library had 90% coverage, with a minimum capacity of 100 and maximum capacity of 3,000. To realize the screening of the different mutant libraries, a modified auto-induction medium was used as reported by Studier et al. [31]. Individual colonies were transferred into 96-deep-well plates containing 600 μl medium per well and cultured in a Microtron microplate shaker (INFORS, Switzerland) at 37 °C and 220 rpm for 24 h. After that, the samples were centrifuged at 8,000 g and 4°C for 10 min and then used for protein expression level analysis and the phytase activity assay.

To characterize the selected mutants, they were grown in 5 ml LB and transformed into 250-ml shake flasks containing 50 ml 2×YT medium. When the OD 600 reached 0.7, 1 mM IPTG was added and the cells were cultured for another 24 h for the enzyme activity assay and protein expression level analysis. The protein expression levels were analyzed by SDS-PAGE and western blotting [32].

Phytase Activity Determination and Analysis of Kinetic Parameters

The phytase activity was determined using the ammonium molybdate method [33]. One unit of phytase activity was defined as the amount of enzyme that releases 1 μmol of inorganic phosphate from sodium phytate per min at 37°C pH 5.5. To test the thermostability, the protein samples were incubated at 90°C for 5 min, immediately cooled on ice for 10 min, and used for the activity assay. Phytase activity without being heated was defined as 100%. To study the enzyme kinetics, the phytases’ activity was assayed in a substrate concentration range of 0.1 mM to 10 mM sodium phytate in 5 mM sodium acetate buffer, pH 5.5 at 37°C for 30 min. The kinetic parameters were calculated by plotting the initial velocities measured at various substrate concentrations according to the Lineweaver-Burk plot [34].

Screening the Potential Mutation Domains to Improve the Thermostability of Phytase

E. coli phytase is a HAP family phosphohydrolase that has been broadly used in the feed industry and agriculture in past decades [3]. However, its low heat stability has severely limited its application. According to the multiple sequence alignment results (Fig. 1) and a previous literature report [35], none of the E. coli phytases from the PDB database are able to tolerate extreme heat treatment. Consequently, the two crystal structures 1DKQ and 1DKL, which share 96% and 95% sequence similarity with the original phytase shown in the phylogenetic tree and sequence alignment (Fig. 1), were used as templates for protein temperature factor calculation and surface loop engineering analysis. Firstly, the B-value of each amino acid of the two crystal structures was extracted and calculated using B-FITTER (Fig. 2). Nine amino acids (P41, V42, K43, R181, E182, Q285, K286, E384, R385) with the highest B-values were obtained and chosen as the mutation sites. Moreover, the method of protein surface engineering was also implemented to further enhance the thermostability of phytase. According to previous reports that the modification of the protein surface amino acids can cause two effects, one being the reduction of protein flexibility (entropy) [36] and the other an increase of electrostatic interactions [29, 37] in engineered proteins to enhance the robustness against factors such as pH and temperature. Reetz et al. found some overlap between B- value analysis and protein surface engineering in enhancing the thermostability of lipase [28]. Based on this, after analyzing the model 1DKQ using Swiss PDB viewer, another four amino acids (S80, Q184, S342, E383), located in the surface loop, were selected. Thus, a total of 13 amino acids were selected and applied to construct and screen thermostable phytase mutants.

Figure 1. The sequence alignment and phylogenetic tree of the HAP family phytases. (A) Multiple sequence alignment of HAP phytases constructed using Crustal X; the yellow box indicates the conserved catalytic site motifs RHGRXP and HD. (B) Phylogenetic tree of HAP family phytases constructed using MEGA 6. The models 1DKL and 1DKQ were used to target residues on the PHLJ phytase (red) evolved in this study.

Figure 2. The B-values of every amino acid of the models 1DKL and 1DKQ, calculated using B-FITTER. The residues with the highest B-values in the two model were chosen as the target sites to construct the mutant library.

Evaluation of the Enzyme Activity of Phytase Mutants

From the above analysis, thirteen potential amino acids (Table 4) that are located in six different domains of phytase were extracted, and a separate phytase mutant library was constructed for each domain. After the first round of screening, four engineered strains were chosen from every mutant library. After shake-flask fermentation and sequencing of the mutants, one engineered strain with different thermostability was obtained from each mutant library (Table 1 and Fig. 3). Although the enzyme activities of the mutant strains BLP11-BLP16 without heat treatment were lower than that of the wild-type, after being heated at 90°C for 5 min, the mutant strains all exhibited a thermostability improvement. This was especially true for BLP15, which retained 34.1% of the initial activity, representing a 14.1% thermostability improvement over the wild-type, which retained 20.0% of the initial activity. Consequently, BLP15 was chosen for further iterative mutations.

Table 4 . Enzyme kinetics of wild-type PhyE and the thermostable mutant P56214..

Variant	Vmax a	KM (mM) a	Kcat (s-1)	Kcat/KM
WT PhyE	1.51±0.03	1.56±0.13	768.68±4.32	439.68
P56214	1.30±0.12	2.58±0.02	578.91±8.12	224.38

^*
p<0.05, Student’s t-test; calculated using GraphPad Prism 6..

Figure 3. The enzyme activity and relative residue activity of wild-type phytase and the single mutants at 90°C for 5min. The gray bars represent the activity of the enzyme after heating at 90°C for 5 min. Phytase activity of each mutant without heat treatment was defined as 100%. The pink bars represent relative residual activity. The details of the strains’ genetics and phenotypes are listed in Table 1. The error bars represent the standard deviations from triplicate experiments. The data were analyzed using GraphPad Prism 6. based on the work by Manfred et al. was applied to further elevate the heat resistance of the single mutant strains [38].

Evaluation of the Enzyme Activity of Different Iterative Mutation Libraries

Each mutant library yielded one engineered strain with different thermostability. Furthermore, an iterative algorithm A series of engineered mutant phytases from the first generation of P15 (library 5) were constructed and tested in E. coli, and an evolutionary pathway of gradual thermo-stability performance increase was defined in the linear order of P15 (library 5) →P56 (library 5+6)/P52 (library 5+2) →P526 (library 5+2+6) →P5261 (library 5+2+6+1) →P52613 (library 5+2+6+1+3)/P52614 (library 5+2+6+1+4), as shown in Fig. 4A. After five rounds of iterative mutation and screening, the engineered mutant named P56214 with 55% thermostability improvement over the wild-type phytase was constructed. Unfortunately, compared with the original phytase, the K_M value of this mutant increased and the K_cat value decreased (Table 4; p < 0.05). Thus, the increasement in thermostability of P56214 came at the expense of kinetic and catalytic efficiency of the enzyme. This result was consistent with the argument raised in the past decades that protein thermostability and enzyme activity improvement cannot be realized simultaneously in a single variant [39].

Figure 4. A, the iterative order of the different mutant libraries. WT and P15-P562143 represent the wild-type phytase and the different iterative mutant libraries, respectively. The details of the strains’ genetics and phenotypes are listed in Table 1. Only the positive iterative roadmap towards enhanced heat resistance of engineered phytase was shown in the graph. B, SDS-PAGE and western blot of the engineered phytase mutant P56214. The arrows represent the phytase band.

Furthermore, we found that introducing the mutations from library 3 or 4 into P56214 or P56213, could not produce any further positive effect on the heat resistance of the variants (Fig. 4A). Considering the features of library 3 and 4, we found that the composition of substituted amino acids was biased towards aspartic acid or serine, similar to the study by Predrag et al, which found that enrichment for such amino acids does not help reduce the disorder of the engineered protein [40]. Arias et al. found that the Ser position was the most frequent site of mutations in mesophilic to thermophilic substitution [41]. When the mutations from library 6 or 2 were inserted into the initial library 5, regardless of the order, two thermostable engineered variants were obtained (Fig. 4A), with a 15% and 18% thermostability increase, respectively. Simul- taneously, compared with the original phytase, the final engineered strain BLP56214 kept almost the same level of protein expression, as shown in Fig. 4B, albeit with a small difference in enzyme catalytic performance. Therefore, compared with the conventional methods, the combined methods used in this study can more quickly and efficiently realize the goal of transforming the mesophilic phytase into a thermophilic enzyme under the direction of rational physical parameters and experimental pathway design.

Analysis of Amino Acids in Different Mutation Libraries

From the above analysis, thirteen potential amino acids (Table 5) that are located in six different domains of the phytase were extracted, and a separate phytase mutant library was constructed for each domain. Notably, we found that a total of thirteen enriched amino acids were mainly typically polar and charged amino acids such as glutamic acid (E), lysine (K) and arginine (R) (Fig. 5). This phenomenon was consistent with a previous report by Predrag et al. that the amino acid mainly enriched in thermophilic proteins are buried and slightly depleted in particular charged residues, which means that charged amino acids are mainly found in psychrophilic microbes and longer disordered regions of the proteins [40]. After screening of our mutant libraries, the substitute amino acids were found to be mainly enriched in hydrophobic and aromatic amino acids, with substitutions such as P41W and K286Y. Simultaneously, when aligned with one of the original phytase models, 1DKQ, these potential libraries could be classified into six different domains (Fig. 6). Except for domain 3 (library 3), which is distributed in the small β sheet, the remaining five libraries were distributed equally among the surface or loop regions. This phenomenon to some extent agrees with prior findings that B-value analysis and protein surface engineering shared common effects in protein thermostability improvement. Furthermore, we found that library 1 and 4 were distributed in the α-domain and the other libraries were mainly located in the α/β-domain according to the phytase crystal structure reported by Noorbatcha IA, et al [21]. Thus, in this study, replacing the original low B-value and buried amino acids located in the highly flexible and protein surface regions may result in two positive effects. On the one hand it influences the local protein structure disorder to reduce the entropy [42]. On the other side, the hydrophobic and aromatic amino acids may also have a positive effect in increasing electrostatic interactions and enhancing hydrogen bond formation in the engineered protein [43].

Table 5 . Analysis of amino acids in the mutant library..

Library	1	2	3	4	5	6
Original	P41, V42, K43	S80	R181, E182, Q184	Q285, K286	S342	E383, E384, R385
Mutated	41W, 42S, 43L	80I	181S, 182S, 184A	285D, 286Y	342T	383A, 384V, 385A

Figure 5. The ratios of the selected amino acids. The amino acids were categorized by the R group. The original amino acids were typically polar and charged, while the substitutions were mainly hydrophobic and aromatic.

Figure 6. The spatial structure of E. coli phytase, consistent with a HAP family protein, composed of two structural domains: a small α-domain and a larger α/β-domain. The libraries 1 and 4 were distributed in the α-domain, while all other libraries were in the α/β-domain.

In previous reports, Garrett et al. mutated the phytase gene appA from E. coli by gene site saturation mutation (GSSM) technology and obtained a mutant phytase named Phy9x. After 10 min heat treatment at 85°C, 27% enzyme activity remained [44]. Fei et al. constructed a C-terminal deletion mutant to detect the thermostability of E. coli phytase. The residual activities of the wild-type AppA phytase and C-lose mutant were 31.42% and 70.49%, respectively, after being heated at 80°C for 10 min [22]. Fei et al. also constructed a salt bridge addition mutant Q307D by site-directed mutagenesis, which showed 9.15% thermostability enhancement than the wild-type after being heated at 80°C for 10 min [23]. In this study, compared with the wild-type, the thermal stability of the modified enzyme has been greatly improved. The residual enzyme activity of the mutant was increased from 20% to 75% after incubation at 90°C for 5 min. We provide a promising method to guide the evolution of thermostable proteins and to further improve the generally superior features of enzymes as green catalysts in the future.

This study was supported by the Tianjin Science Fund for Distinguished Young Scholars (17JCJQJC45300), the Natural Science Foundation of Tianjin (CN) (16JCYBJC23500, 15JCQNJC09500), Tianjin Science and Technology project (15PTCYSY00020), and the Science and Technology Service Network (STS) Initiative of the Chinese Academy of Sciences (CAS).

The authors have no financial conflicts of interest to declare.