Applied Microbiology and Biotechnology

, Volume 79, Issue 1, pp 69–75

Enhancing thermostability of Escherichia coli phytase AppA2 by error-prone PCR


  • Moon-Soo Kim
    • Department of Animal Science and Graduate field of Food ScienceCornell University
    • Department of Medicinal Chemistry and Molecular PharmacologyPurdue University
    • Department of Animal Science and Graduate field of Food ScienceCornell University
    • Department of Animal ScienceCornell University
Biotechnologically Relevant Enzymes and Proteins

DOI: 10.1007/s00253-008-1412-7

Cite this article as:
Kim, M. & Lei, X.G. Appl Microbiol Biotechnol (2008) 79: 69. doi:10.1007/s00253-008-1412-7


Phytases are used to improve phosphorus nutrition of food animals and reduce their phosphorus excretion to the environment. Due to favorable properties, Escherichia coli AppA2 phytase is of particular interest for biotechnological applications. Directed evolution was applied in the present study to improve AppA2 phytase thermostability for lowering its heat inactivation during feed pelleting (60–80°C). After a mutant library of AppA2 was generated by error-prone polymerase chain reaction, variants were expressed initially in Saccharomyces cerevisiae for screening and then in Pichia pastoris for characterizing thermostability. Compared with the wild-type enzyme, two variants (K46E and K65E/K97M/S209G) showed over 20% improvement in thermostability (80°C for 10 min), and 6–7°C increases in melting temperatures (Tm). Structural predictions suggest that substitutions of K46E and K65E might introduce additional hydrogen bonds with adjacent residues, improving the enzyme thermostability by stabilizing local interactions. Overall catalytic efficiency (kcat / Km) of K46E and K65E/K97M/S209G was improved by 56% and 152% than that of wild type at pH 3.5, respectively. Thus, the catalytic efficiency of these enzymes was not inversely related to their thermostability.


ThermostabilityPhytaseError-prone PCRMelting temperatureEnzyme


Phytases (myo-inositol hexakisphosphate phosphohydrolase) catalyze the stepwise hydrolysis of phytate into myo-inositol and inorganic phosphate. These enzymes are added to animal diets to improve their phosphorus absorption and to reduce their phosphorus excretion. Among many phytases, Escherichia coli AppA phytase has a great potential for industrial applications due to an acidic pH optimum, high specific activity for phytate, and resistance to pepsin digestion (Greiner et al. 1993; Lei and Stahl 2001; Rodriguez et al. 1999b, 2000; Wyss et al. 1999). The second E. coli phytase gene, appA2 has 95% sequence homology to appA gene (Rodriguez et al. 1999a). The 1.3-kb appA2 gene encodes a protein of 432 amino acids, with three putative N-glycosylation sites and a molecular mass of 46.3 kDa (Rodriguez et al. 1999a). The crystal structure of E. coli phytase (Lim et al. 2000) consists of a conserved α/β-domain and a variable α-domain. It shares a very similar overall structure with rat prostatic acid phosphatase, Aspergillus niger PhyA phytase, and A. niger pH 2.5 acid phosphatase, despite low sequence homology (Lim et al. 2000). The similarity in overall structure is also seen between two homologous fungal phytases from A. niger and Aspergillus fumigatus that shares only 66% sequence identity (Xiang et al. 2004). However, the two phytases display remarkably different thermostability that probably stems from their structural deviations.

Analyses of structural basis for protein stability have illustrated general factors governing the stability of proteins (Kumar et al. 2000; Querol et al. 1996; Ragone 2001; Vieille and Zeikus 1996, 2001; Yip et al. 1995). They include increase in hydrogen bonds and ionic interactions, reduction of conformational strain, improvement of the packing of the hydrophobic core, and enhanced secondary structure propensity. Using rational design, Rodriguez et al. (2000) added potential glycosylation sites to improve the thermostability of E. coli AppA phytase by site-directed mutagenesis. As a semirational approach, the consensus concept was generated to improve phytase thermostability and catalytic efficiency (Wyss et al. 1999). Structure-based chimeric enzymes were developed as an alternative to directed evolution to improve the thermostability of Aspergillus terreus phytase (Jermutus et al. 2001).

Directed evolution has emerged as a successful alternative to rational design for genetic engineering of enzymes (Kuchner and Arnold 1997; Williams and Berry 2003). This approach does not require detailed information on structures and accurate predictions on amino acid substitutions at the proper sites (Kim et al. 2003). It involves generating a vast library of the gene of interest by random mutagenesis such as error-prone polymerase chain reaction (PCR) or DNA shuffling, followed by screening mutants for desired properties. This approach has been particularly successful in improving the thermostability of proteins. The half-life of subtilisin S41 at 60°C was increased by 1,200-fold and melting temperature of the mutant increased by 25°C over the wild type after eight successive rounds of error-prone PCR and in vitro recombination (Wintrode et al. 2001). The thermostability of a fungal peroxidase was enhanced by 110-fold by combining mutations from error-prone PCR and in vivo shuffling with those from site-directed mutagenesis (Cherry et al. 1999). These successful studies prompted us to employ directed evolution to generate thermostable E. coli AppA2 variants in the present research.

Materials and methods

Error-prone PCR and mutant library construction

Error-prone PCR was performed using plasmid pYAα2 containing the wild-type appA2 gene cloned into a Saccharomyces cerevisiae expression vector, pYES2 (Lee et al. 2005), as a template. A 100 μl of reaction mixture contained 10 mM Tris–HCl (pH 8.3), 50 mM KCl, 7 mM MgCl2, 0.2 mM deoxyadenosine triphosphate and deoxyguanosine triphosphate, 0.6 to 1 mM deoxycytidine triphosphate and deoxythymidine triphosphate, 10 ng of template, 0.1 to 0.3 mM MnCl2, and 5 units (U) of Taq polymerase (Fisher). Two primers used were E2 (5’-GGA ATT CCA GAG TGA GCC GGA-3’) and X2 (5’-GGT CTA GAT TAC AAA CTG CAC G-3’). PCR was carried out at 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min for a total of 30 cycles. The PCR products were inserted into pYES2 vector at EcoR I and Xba I sites, followed by transformation into S. cerevisiae INVSc1 by electroporation using an ECM 600 Electro Cell Manipulator (Genetronics, BTX Instrument Division, San Diego, CA, USA). The transformed cells were plated onto synthetic complete (SC; Ura-) minimal medium containing 2% glucose for the selection of positive transformants and incubated at 30°C for 3 days.

Screening for improved thermostability

Thermostability was determined as residual phytase activity after incubation of the enzyme at 85°C for 15 min. Single colonies of transformants were transferred into 96-well plates containing 20 μl of SC (Ura-) minimal medium per well. The 96-well plates were incubated at 30°C, 220 rpm for 24 h, followed by addition of 100 μl of yeast extract–peptone–galactose medium (1% yeast extract, 2% peptone, and 2% galactose). After being incubated for 36 h, enzymes from culture supernatants were diluted in 0.2 M glycine–HCl, pH 3.5, and transferred into new 96-well plates. One out of two replica plates was incubated at 85°C for 15 min and chilled on ice for 15 min. Both of the replica plates were assayed for phytase activity as previously described (Han et al. 1999) with modification to be suitable for a 96-well plate. One phytase unit is defined as the amount of activity that releases 1 μmol of inorganic phosphorus from sodium phytate per minute at 37°C. Mutants with improved thermostability were verified by DNA sequencing.

Protein expression and purification

The selected mutants were further expressed in Pichia pastoris X-33 as described previously (Kim et al. 2006; Rodriguez et al. 1999b). The selected transformants were cultured in yeast extract–peptone–dextrose media (1% yeast extract, 2% peptone, and 2% dextrose) at 30°C from 48 to 72 h. The expressed phytase enzymes in culture supernatant were sequentially purified by ultrafiltration and Macro-Prep high S cation exchange chromatography (Bio-Rad Laboratories, Hercules, CA, USA). The proteins were eluted in 25 mM glycine–HCl buffer, pH 3.2 with a linear gradient of NaCl. Pooled peak fractions were dialyzed in pH 3.5 glycine buffer, followed by further purification through the same column. Purified protein concentrations were obtained from the absorbance at 280 nm using an extinction coefficient (ɛ = 50,460 M−1 cm−1). Protein purity was checked by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. A single protein band was discerned and protein was approximately 95% pure.

Thermostability assay

Purified enzymes were diluted in 25 mM glycine–HCl buffer, pH 3.5 to 5 μg protein per ml. The diluted enzymes were incubated for 10 min at each of the following temperatures: 50°C, 60°C, 70°C, and 80°C. Immediately after the heat treatment, the enzymes were placed on ice for 30 min (Han and Lei 1999; Han et al. 1999). The remaining phytase activity was measured at 37°C and pH 3.5 as described previously (Han et al. 1999).

Differential scanning calorimetry

Melting temperatures (Tm) of wild-type and mutant phytases were determined with a DSC Q10 (TA instruments, New Castle, DE, USA) differential scanning calorimeter (DSC) equipped with refrigerated cooling system and Thermal Advantage™ for Q Series™ software. Protein samples were concentrated to 40 mg/ml in 25-mM glycine–HCl buffer, pH 3.5. The scanning condition was modified from a previously published method (Garrett et al. 2004). The proteins were scanned from 30°C to 100°C at a heating rate of 10°C/min. A pan containing 25 mM glycine–HCl buffer, pH 3.5 was used as a reference.

pH profile and temperature optimum

The pH profile of phytase was determined at 37°C using three different buffers: 0.2 M glycine–HCl buffer for pH 2.0–3.5, 0.2 M sodium citrate buffer for pH 4.0–6.5, and 0.2-M imidazole–HCl buffer, pH 7.0. Purified enzymes were diluted with each buffer of different pH to give an activity of 0.2 U/ml. The optimal temperature was determined in 0.2 M glycine-HCl, pH 3.5 at 37°C, 45°C, 55°C, 60°C, 65°C, 75°C, and 85°C.

Determination of kinetic parameters

Purified enzymes were diluted with 0.2 M glycine–HCl buffer, pH 3.5 to a final concentration of 0.2 U/ml. Phytase assay was performed at 37°C using sodium phytate as substrates at six different concentrations ranging from 50 to 1,250 μM. Six parallel reactions were carried out with different phytase hydrolysis reaction times from 0, 2, 4, 6, 10, to 15 min. Initial rates were determined at six different substrate concentrations. To determine kinetic parameters (Vmax and Km), data were fitted to the Michaelis–Menten equation, using KaleidaGraph (v. 3.5, Synergy Software).


Error-prone PCR mutagenesis of appA2

Approximately 5,000 clones were screened for increased thermostability. At the initial screening using a 96-well plate, mutants showing a 20% higher residual activity than the wild-type enzyme were selected. Subsequently, thermostability of 80 selected mutants was assayed in a test tube. Considering enzyme activity and thermostability, six best mutants were finally chosen for further purification and characterization. Sequence analysis of the selected mutants showed that the number of amino acid substitutions varied from one to four per mutant (Table 1). Specific activity of mutants was lower than that of the wild type, except K65E/K97M/G103S/G344D (Table 1).
Table 1

Specific activities of wild-type and mutant AppA2 enzymes


Specific activity a (U mg−1)


1,003 ± 8


742 ± 4


209 ± 3


900 ± 7


274 ± 3


905 ± 11


1042 ± 3

aValues represents mean ± standard error (n = 3). The measurement of specific activity was performed at 4.9 mM sodium phytate concentration.

Thermostability profiles and melting temperatures (Tm)

Figure 1 shows that selected mutant enzymes expressed in P. pastoris had improved heat stability over the wild-type enzyme, as the incubation temperature increased. K46E and K65E/K97M/S209G displayed highest residual activity with a 25% improvement (P < 0.05) over that of the wild-type enzyme after being heated at 80°C for 10 min. Because these two mutants were most promising, their melting temperatures were determined by DSC scanning. The mid-point of the thermal unfolding (Tm) was increased by 7°C for K65E/K97M/S209G and 6°C for K46E, compared to wild-type AppA2 (Fig. 2). Optimal temperature for phytase activity for each mutant remained unchanged as that of wild type (data not shown).
Fig. 1

Residual phytase activity of wild-type and mutant AppA2 enzymes. Means at a given temperature point without sharing a common superscript letter indicate differences (P < 0.05)
Fig. 2

The thermogram and melting temperature (Tm) of wild-type AppA2 (a), K46E (b), and K65E/K97M/S209G (c)

pH profiles

All mutants except K46E and K65E/K97M/S209G showed unchanged pH profiles compared with the wild-type enzyme (Fig. 3). Whereas K46E and K65E/K97M/S209G had greater phytase activity than the wild-type enzyme at pH 2.5, their activity decrease accelerated at pH 4.5 and beyond. At pH 4.5, K46E and K65E/K97M/S209G retained approximately 35% and 40% of the activity at pH 3.5, respectively.
Fig. 3

pH activity profiles of wild-type and mutant AppA2 enzymes. Phytase activity at pH 3.5 was defined as 100%

Kinetic characterization

The Km values of K46E and K65E/K97M/S209G for sodium phytate at pH 3.5 were 43% and 62% lower (P < 0.05) than that of wild type, respectively (Table 2). The 26% reduction in specific activity of the mutant K46E which was measured at 4.9 mM sodium phytate was probably caused by a decrease in turnover number. Although the kcat of K46E was lower (P < 0.05) than that of the wild-type enzyme, its overall catalytic efficiency (kcat/Km) was 56% higher (P < 0.05). The mutant K65E/K97M/S209G showed 152% higher (P < 0.05) overall catalytic efficiency (kcat/Km) than that of wild type.
Table 2

Comparison of kinetics of wild-type and two mutant AppA2 enzymesa


Vmax (μmol m−1 mg−1)

Km (μM)

kcat (m-1)

kcat/Km (m−1 μM−1)


1,032a ± 47

132.6a ± 14.8

55,800a ± 2,523

423d ± 27


921b ± 6

75.5b ± 4.0

49,700b ± 295

660b ± 35


999ab ± 21

51.0c ± 5.2

54,000ab ± 1,106

1,065a ± 97

aEnzyme reactions (n = 3) were conducted at 37°C in 0.2 M glycine–HCl buffer, pH 3.5 using various sodium phytate concentrations (50 to 1,250 μM) and 200 mU phytase per milliliter of reaction mixture.

b,c,dValues represent mean ± standard error. Different letters indicate differences (P<0.05) within the column.


Compared with the wild type, all six selected mutants showed enhanced resistance to relatively high temperatures. In particular, the K46E and K65E/K97M/S209G mutants showed a 25% improvement of thermostability at 80°C. Even more, this modest change in thermostability actually rendered an increase in melting temperature (Tm) by 6–7°C. The identified amino acid substitutions in the selected mutants are distributed throughout the structure of a highly homologous E. coli AppA (PDB ID: 1DKL; Lim et al. 2000; Fig. 4(a)). Most of mutations are found in loops or surface regions. The only exception is the substitution V227A that is located on the β-strand of α-domain. Based on the structural predictions to probe the molecular basis of thermostability (Fig. 4(b)), the substitution of K46E introduces two hydrogen bonds to alanine 47 (Ala47), whereas lysine 46 (Lys46) forms only one hydrogen bond. Because Lys46 is involved in substrate binding, the increased thermostability due to the substitution of K46E at the cost of specific activity may be associated with a possible overall charge alterations or interactions between protein stability and function (Shoichet et al. 1995).
Fig. 4

(a) Locations of residue substitutions in the structure of E. coli phytase (PDB ID: 1DKL; Lim et al. 2000). (b) The substituted Glu46 and Glu65 are predicted to introduce hydrogen bonds to Ala47 and to adjacent residues, Leu66 and Trp68, respectively. (c) The substituted Asp344 is predicted to form an additional hydrogen bond to Leu379. The ribbon diagrams of the three-dimensional structure were prepared using the Swiss-Pdb viewer. Dotted lines indicate hydrogen bonds

Likewise, Lys65 forms only one hydrogen bond in the wild-type enzyme. The substitution of K65E introduces three additional hydrogen bonds to adjacent residues: the side chain of glutamine 65 (Glu65) forms one hydrogen bond to the NH group of tryptophan 68 (Trp68) and two hydrogen bonds to leucine 66 (Leu66). Because K65E and Trp68 are located on the α-domain and the α/β-domain, respectively, the newly introduced hydrogen bonds might play an important role in stabilizing the interaction between two domains, which could contribute to overall stability of the mutant. The newly formed hydrogen bonds between K65E and Leu66, as well as between K46E and Ala47, are predicted to stabilize the local structure and then improve thermostability. Locating on the loop, glycine 344 forms hydrogen bonds with Leu340 and Leu379 (Fig. 4(c)). The G344D substitution is predicted to form an additional side chain–side chain hydrogen bond to Leu379 which is located on the neighboring α-helix. The additional side chain–side chain hydrogen bond may contribute to stabilize local interactions within the loop and α-helix.

The increase in hydrogen bonding has been suggested as one of the principal determinants of enhanced thermal stability (Kumar et al. 2000; Vieille and Zeikus 2001; Vogt et al. 1997). The structural predictions shown in Fig. 4(b) indicate that the increased hydrogen bonding interactions are created by an acidic residue which is located on the surface loop regions. This is in agreement with the critical structural differences responsible for the greater thermostability of A. fumigatus phytase than that of A. niger phytase: the former phytase has more hydrogen bonding interactions created by acidic residues and salt–bridge interactions than the latter phytase, despite similar overall structure (Xiang et al. 2004). The critical regions in A. fumigatus phytase are located on the surface-exposed turns and/or loops, which could be correlated with the refolding capability of the protein (Xiang et al. 2004).

In comparison with K46E, the other substitutions of K97M, V227A, and G344D resulted in less benefit to thermostability. The structural predictions indicate that mutations of K97M and V227A, along with S209G, are not involved in hydrogen bonding. The substitution of K97M might remove structural hindrance caused by bulky side chains between Lys97 and neighboring Lys96. The replacement of valine 227 (Val227) with Ala possessing a smaller side chain might eliminate structural hindrance between Val222 and Val227 which face each other in the β-hairpin structure. The substitution of S209G might relieve potential unfavorable interactions between serine 209 with its neighbor cysteine 210 in a close proximity. Thermostability of the triple mutant K65E/K97M/S209G was greater than that of the single mutant K97M. As mentioned above, the substitution of K65E adds three new hydrogen bonds and the other two mutations might improve interactions with the neighboring residues. However, impacts of individual mutations on stability of a given protein are not necessarily additive or synergistic (LiCata and Ackers 1995). In the present study, the quadruple mutant K65E/K97M/G103S/G344D did not show much extra improvement in thermostability at 70°C and 80°C over the single mutants of K97M or G344D. Without having a high-resolution crystal structure of the mutant K65E/K97M/S209G, we cannot tell if the improved thermostability is attributed to the additive benefits and(or) an overall change in charge or folding of the enzyme protein produced by the three individual mutations.

The two thermostable mutants K46E and K65E/K97M/S209G displayed higher overall catalytic efficiency (kcat/Km) than the wild type at pH 3.5. Thus, improvements of thermostability in these mutants did not compromise the enzyme catalytic function at a physiological (stomach)-relevant pH. The removal of a bulky side chain of Lys46 and Lys65 in these two mutants might provide more conformational flexibility toward the substrate binding area, thereby improving catalytic efficiency. Single surface mutations can be accommodated within a protein without compensating structural and/or functional changes (Zhao and Arnold 1999). Simultaneous improvements in thermostability and catalytic activity have also been achieved in other enzymes (Giver et al. 1998; Song and Rhee 2000; Zhao and Arnold 1999).

Although it is unclear to us why the pH–phytase activity profile of the triple mutant shifted downward from the wild type at pH 4.5 and beyond, the same type of change in the mutant K46E may be explained as follows: Lys46 is one of the amino acids involved in phytate binding to the scissile phosphate (Lim et al. 2000), and the negatively charged substituted Glu46 at pH 5.0–5.5 may create repulsive environment for the highly negatively charged substrate phytate, thereby decreasing enzyme activity. In summary, we have demonstrated that directed evolution can be applied to create thermostable phytases. With better overall catalytic efficiency, a couple of the AppA2 phytase variants may survive the high temperature inactivation during feed pelleting and become more desirable and economical source for animal feed supplementation.


This research was supported in part by a Cornell Biotechnology Program grant (to XL). We thank Jeremy Weaver for helpful comment.

Copyright information

© Springer-Verlag 2008