Short Communication

Improving the Specific Synthetic
Activity of a Penicillin G Acylase Using DNA Family Shuffling

ZHOU Zheng, ZHANG Ai-Hui, WANG Jing-Ru¹,
CHEN Mao-Lin¹, LI Ren-Bao, YANG
Sheng¹*, YUAN Zhong-Yi*

(
Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological
Sciences, the Chinese Academy of Sciences, Shanghai 200031, China; ¹
Institute of Plant Physiology and Ecology, Shanghai Institutes for
Biological Sciences, the Chinese Academy of Sciences, Shanghai 200032, China )

Abstract        Penicillin
G Acylas (PGA) of Providencia rettgeri (ATCC 25599) was evolved using a
modified DNA family shuffling method. The identity of pga genes from Escherichia
coli, Kluyvera citrophila and Providencia rettgeri ranges
from 62.5% to 96.9%. The pga genes from above three species were
recombined and shuffled to create interspecies pga gene fusion
libraries. By substituting assembled chimaeras for corresponding region of
pETPPGA, different recombinants were constructed and expressed in E. coli JM109(DE3).
Mutants with obvious β-lactam synthetic activity were selected from the plates
and the ratios of synthesis to hydrolysis (S/H) were determined subsequently.
It was shown that the primary structures of selected positives exhibited
significant diversity among each library. The best mutant possessed 40% higher
synthetic activity than the wild type enzyme of PrPGA. It was further proved in
this study that the domain of α subunit contributed much more to improve the
specific activity of synthesis. Results showed a recombinant PGA with higher
synthetic activity was acquired by the method of DNA shuffling.

Key words     penicillin G acylase;
DNA family shuffling; S/H ratio

Penicillin G
acylase (PGA EC 3.5.1.11) is an important hydrolase which catalyzes the
deacylation and acylation reactions, and can be commercially used as a
transferase to catalyze the synthesis of condensation products such as β-lactam
antibiotics[1]. Because of the high value, PGAs had been studied extensively
and the crystal structures of several PGAs had been clarified[2,3]. Combination
of rational protein design and site-directed mutagenesis was very effective in
elaborating improved PGAs. Interesting progresses such as elucidation of PGA
maturation mechanism[4], alteration of kinetic properties and substrate
specificity[5] as well as improvement of stability[6] etc., had been
reported[7]. Recently, Alkema et al.[8] reported a new procedure to
enhance PGA synthetic capability. However, most attempts at redesigning enzymes
had failed. In rational redesign, precise changes in amino acid sequence should
be predetermined based on a detailed knowledge of protein structure, function
and mechanism, and then introduced using site-directed mutagenesis, but the
knowledge needed were not always sufficiently obtained. On the other hand, the
function of enzyme was determined by not only the amino acid sequence in the
vicinity of the active centers, but also relied on the correct overall conformation
of the protein[9]. Recently, method of DNA family shuffling supplied
researchers an alternative for developing PGA[10]. DNA family shuffling
mimicked and extended classical breeding techniques by recombining more than
two parental genes, or genes from different species, in a single DNA shuffling
reaction[11]. In this approach, neither structural information nor mechanistic
roadmap was required to guide the directed evolution experiment. Moreover,
homologous sequences used in DNA family shuffling had been functional evolved
during millions of years of natural selection. As a driving force for in vitro
evolution via recombination among family genes, DNA family shuffling became a
powerful tool for rapidly evolving genes for desired properties[12－14].

The PGA
producing bacterium Providencia rettgeri (Pr), Escherichia coli
(Ec) and Kluyvera citrophila (Kc) belonged to Gram-negative bacteria
family. PGAs from these species were evolutionarily related. PrPGA, EcPGA and
KcPGA genes were three most closely related pga genes (62.5%－96.9% DNA identity) in this highly
divergent family. Therefore DNA family shuffling among these three genes could
facilitate establishing chimeric pga libraries.

Here we described the cloning of a mutant pga
gene from P. rettgeri (ATCC 25599) and its expression in E. coli.
A modified DNA family shuffling method was developed to create different
interspecies chimeric pools. Mutants were screened, assayed, and then analyzed
for the synthesis/hydrolysis(S/H) ratio. The strategy has led to the finding of
mutants with evidently improved synthesis capability. It also provides an
approach to investigate which domain of PGA is crucial to maintain its
catalytic specificity.

1    Materials
and Methods

1.1   Strains
and plasmids

P. rettgeri (ATCC 25599) was purchased from American Type Culture Collection.
Plasmids harboring genes of EcPGA and KcPGA were kindly provided by Prof. Gong
Yi (Shanghai Research Centre of Biotechnology, shanghai, China) and Dr.
Ignatova (Department of Biotechnology II, Technical University of Hamburg,
Hamburg, Germany) respectively. pBluescript SK(+) and pUC18 were used as the
cloning vectors. E. coli XL1-Blue, TG1 and JM109, the host strains for
gene cloning, were cultivated with aeration at 37 ℃ in Luria broth (LB). Ampicillin
(100 mg/L) and kanamycin (50 mg/L) were added to the medium when strains
harboring different recombinant plasmids were selected. The E. coli
JM109(DE3) engineering strain was applied for the expression in the medium
consisting of 1.5 g/L yeast extract, 0.5 g/L NaCl and 5 g/L glycerol.

1.2   Reagent

The restriction
enzymes, Pfu polymerase, dNTP, DNase I, calf intestinal alkaline
phosphatase (CIAP), and T4 DNA ligase were purchased from TaKaRa(Dalian,
China); all primers were synthesized by SBS(Beijing, China);
isopropylthio-beta-D-galactoside (IPTG) was from Sangon (Shanghai,
China); 6-nitro-3-phenylacetamino-benzoic acid (NIPAB) was purchased from
Dongfeng Reagent Factory (Shanghai, China); the DNA sequencing reactions were
conducted by using PRISMTM dye terminator cycle sequencing ready reaction kit
(PE Applied Biosystems). Data were collected and analyzed with a ABI PRISMTM
310 DNA sequencer. All the reagents were of AR grade.

1.3   Isolation
of PrPGA gene and construction of expression plasmid

P. rettgeri (ATCC 25599) was cultured for 24 h in LB medium for the isolation
of chromosomal DNA. Two sets of primers Pr1, Pr3 and Pr2, Pr4 derived from the
partial sequence of P. rettgeri (ATCC 31052) were used to amplify
segments I and II of pga gene (Table 1). Pr1 contained an initial codon
(ATG), whereas Pr4 contained a translational stop coden (TAA). The PCR program:
94 ℃ 3 min
followed by 30 cycles of 94 ℃ 30 s, 55 ℃ 30 s, 72 ℃ 3min, followed by 10 min at 72 ℃. PCR products of segment I and II were blunt-end and ligated into a
SmaI digested pBluscript SK(+) vector to construct pSKI and pSKII, and
then sequenced. Expression plasmid pETPPGA containing complete pga gene
was constructed as following (Fig.1): firstly, pSKII was digested with HindIII
and BamHI，the 0.8 kb
fragment was cloned into pUC18, which had been cleaved with the same enzymes.
After digestion with NdeI and SalI, a 0.2 kb fragment of this resulting plasmid
pUCII was substituted by the 1.5 kb NdeI/SalI fragment of pSKI to
form pPPGA. Full-length pga gene of pPPGA was sequenced and submitted to
GenBank and the accession number was AF499615. Plasmid pPPGA was digested with NdeI
and BamHI, and the 2.5 kb fragment containing pga gene was
inserted into pET24a(+) to form the expression vector pETPPGA.

Table 1   DNA
sequence of used primers

Primer
derived from the partial pga gene sequence of the GenBank M86533. Restriction
sites were underlined. Annealing positions of the primers were illustrated in
Fig.1(C).

1.4   Enzyme
activity assay of PGA

PGA activity was
assayed by modified colorimetric method[2]. One unit of enzyme was defined as
the amount of enzyme catalyzing the hydrolysis of 1 micromole of NIPAB per
minute at 37 ℃.

1.5   DNA
family shuffling and construction of pools containing different chimeric
sequences

PrPGA, EcPGA and
KcPGA genes were amplified by PCR. DNase I was used to fragment 10 μg of such PCR products; 100－250 bp fragments were isolated from
2% agarose gel as Zhao et al.[15] described. Then PCR reassembly without
primers was carried out as program: 3 min 94 ℃ followed by 45 cycles of 94 ℃ 30 s, 37 ℃ 30 s, 72 ℃ 1 min + 3 s/cycle (the duration of the elongation step was
increased by 3 s after each cycle), followed by 10 min elongation at 72 ℃. The PCR products were
subsequently subjected to 25 cycles of PCR with five primer pairs. The entire
chimerical coding region of PGA(CH1-AB), α subunit coding region (CH1-A), β subunit coding region (CH1-B), 5’-half of the CH1-B (CH1-B1) and
3’-half of the CH1-B (CH1-B2) were amplified respectively. After digested by
corresponding restriction enzymes, these fragments substituted for relative
segments of pETPPGA resulting in various chimeric libraries.

1.6   Screening
for chimeric enzymes with PGA activity

For colonies
screen, the filters paper were saturated with 2% NIPAB and dried beforehand.
The plasmids containing various chimeric genes were introduced into JM109 (DE3)
by electroporation. Agar plates containing 50 mg/L kanamycin were plated with
transformants, incubated at 37 ℃ for 18 h and imprinted with NIPAB filter paper. The filter paper
was removed from plate after being soaked, and then was kept at 37 ℃ for 10 min. The visible yellow
halos on the filter paper indicated the colonies expressing active PGA. Positive
colonies were selected and cultured. IPTG was added when A600=0.6 and the
fermentation continued for 3 h. The mutant enzymes were isolated by a modified
osmotic shock method and the activities were assayed[2]. Colonies with higher
activity were selected and the nucleotide sequences of the inserts were
determined.

1.7   Assay
the synthesis to hydrolysis (S/H) ratio

Enzymatic
conversions of cephalexin were carried out with D-phenylglycinamide and
7-amino-deacetoxycephalosporanic acid (7-ADCA) as substrates. S/H ratios were
determined according to the previous paper[16].

2    Results

2.1   Isolation
and characterization of the gene

Two
oligonucleotides Pr1 and Pr4’ were synthesized according to pga sequence
(GenBank M86533) derived from P. rettgeri (ATCC 31052). Unfortunately,
it was not successful to amplify full-length pga sequence of P. rettgeri
(ATCC 25599). To improve PCR efficiency, the internal primer pairs Pr2/Pr3 were
designed to contain a SalI site and used in following segment
amplification. Two truncated segments, designated I and II[Fig.1(A)]
respectively, were successfully amplified, and both sequences showed high
similarity with published pga[17]. However, it was found that Pr4’ was
not complementary to the DNA sequence of 3′-terminus of segment II, although
Pr2 complemented to both the 5′-and 3′-terminal DNA sequence of segment II. In
another word, Pr2 was the upstream primer as well as the downstream one in
amplifying segment II. Therefore, a new PCR primer Pr4 was synthesized by adding
a translational stop codon and a BamHI site at 3′-terminus of Pr2. The
newly generated segment II was ligated with segment I at SalI
restriction site to construct the plasmid pPPGA. Traditional expression vectors
contain a β-lactamase gene which interferes with β-lactam acylase assay.
Therefore, plasmid pETPPGA containing a kanamycin resistance marker was
constructed and transformed into E. coli JM109(DE3)[Fig.1(B)]. The
engineering organism showed to produce the active enzyme.

The open reading
frame of 2514 bases encoded an 837 amino-acid protein of PGA. The full sequence
of the enzyme exhibited high similarity with the published PGA of P.
rettgeri (ATCC 31052). These enzymes share 92.8% protein sequence identity
(59 different amino acids) and 90.2% DNA sequence identity[18].The failure in
gene amplification with primer Pr1 and Pr4’ indicated that C-termini of these
two genes were significantly different. DNA and amino acids sequence alignments
of α and β subunits from EcPGA, KcPGA with
those of the generated PrPGA showed that the generated PrPGA shares a high
identity with EcPGA (62.6% in protein sequence and 61.8% in DNA sequence) and
KcPGA (60.6% in protein sequence and 60.0% in DNA sequence).

2.2   Family
shuffling with randomly cleaved fragments

The probability of homo-duplex formation in
family shuffling was much higher than that of hetero-duplex formation because
of the high divergence between parents. In order to improve the efficiency of
chimeric formation in family shuffling, hetero-primers were designed to
generate chimeric sequences in different PCR. As described in Table 1 and
Fig.1(C), Pr1, Ec1 and Pr4, Ec2 were external 5′ and 3′ primers with
restriction sites of NdeI and BamHI, respectively. Pr2 and Ec3
were internal 5′ primers, whereas Pr3 and Pr5 were internal 3′ primers. Each of
them contained a unique restriction site of SalI or KpnI,
respectively. After random fragmentation and then reassembly, chimeric
sequences were generated by PCR from a pool of reassembly using primers Ec1/Pr4
(Fig.2). The chimeric sequences were inserted into pETPPGA by digestion of NdeI
and BamHI. Other chimeric sequences were generated and cloned in similar
way. It should be noted that a chimeric primer of Ec3 was designed. Its 5’ half
had the same sequence as that of PrPGA containing a KpnI site, and its
3’ half had the same sequence as that of Ec/KcPGA. This design was based on the
fact that there are identical amino-acid sequences in near β subunit of above
PGAs. Thus the KpnI site could facilitate the following DNA
manipulation.

Fig.1       Strategy of
primer design, gene cloning and expression vector construction

(A)
Cloning strategy for PrPGA. (B) Construction of the expression plasmid pETPPGA.
For details to see materials and methods. (C) Sketch map of various pga
genes and corresponding primers.

Fig.2       DNA
shuffling procedure to generate full-length gene of CH1-AB

1, fragmentation; 2, reassembly; 3,
amplification with primers Ec1 and Pr4; 4, reamplification with the same
primers; M, the molecular masses of the marker DL2000.

By
electroporation, ligation product containing chimeric gene were introduced into
JM109(DE3) with a transformation efficiency about 105. However, the efficiency
was far higher when JM109 or TG1 was applied as host strains. Unfortunately,
the latter libraries have to be first retrieved and transformed JM109(DE3) to
express PGA. Because repeated transformation might decrease the diversity of
chimeric pools, direct transformation was employed in this paper.

2.3   Construction
and analysis of chimeric libraries

Generally, α and
β subunits of PGA are structurally and functionally independent. A covalent
internal spacer forces the α subunit (A) and β subunit (B) to bind
together(Fig.1). In order to achieve high gene diversity, the chimeric sequence
libraries of CH1-AB, CH1-A and CH1-B were constructed separately and then
screened for PGA activity (Table 2). There are few active mutants from these
libraries except for CH1-A. Since homologous PGAs differ in gene lengths and
gaps, DNA crossover events occurring among gene gaps will have a big chance to
generate mutants with shifted reading frame. In order to avoid this problem, we
created shortened versions of CH1-B, CH1-B1 (5′-half of β subunit) and CH1-B2 (3′-half of β subunit). As shown in Table 2,
library of CH1-B2 domain generated much higher percentage of active mutants
than the others. This indicated this domain was probably less sensitive to
mutagenesis and could tolerate high mutation densities

Table 2   Construction
and analysis of different chimeric libraries

2.4   Analysis
of β-lactam antibiotic synthetic activity and gene diversities

85 chimeric PGAs
were assayed for S/H ratio. As shown in Fig.3(A), these mutants exhibited
marvelous diversity. The best one with the S/H ratio of 10.74 possessed a
synthetic activity 40% higher than PrPGA (7.65). It was observed that mutants
from library CH1-A1 were more favorable than those from CH1-B1 and CH1-B2. 66%
of colonies (25/39) in library CH1-A1 had higher S/H ratios than that of PrPGA.
On the contrary, only 30% colonies of CH1-B1 (5/17) and CH1-B2 (9/30) libraries
had higher S/H ratios than that of PrPGA[Fig.3(B)].

Fig.3       Screen of
the chimeric libraries reveals improved clone

(A)All of the active clones were screened for S/H ratio. ○, enzyme mutant; ●, PrPGA. (B)
Evaluation of different libraries. Mutants from ○, CH1-A1; □, CH1-B1; △, CH1-B2. Symbols
in darkness represented PrPGA.

As shown in
Table 3, totally 7 mutants were selected and sequenced. Among these mutants,
CH1-A-1, CH1-A-2, CH1-B2-1, CH1-B1-1 and CH1-B1-2 were best mutants from each
library. CH1-B2-2 and CH1-B2-3 were randomly selected from CH1-B2 library. Gene
sequences analysis demonstrated all the chimeras had a crossover at different
positions. It was the fact that none of the chimaeras possessed the same gene
length and identical DNA sequence, which might account for such significant
diversity (Fig.4). The chimaeras were constructed using the DNA sequence of
PrPGA as backbone. There were only small percentages replaced by EcPGA or KcPGA
in the whole DNA sequence of PrPGA segment. This implied PrPGA contributed more
to chimaeras properties. This might be the reason that most of mutants had
lower S/H ratios than parent enzymes of Ec (10.39) and Kc (8.56).

Fig.4       Phylogenetic
tree of the wild type and mutant genes

A
dotted line indicated a negative branch length resulted from averaging. DNA
sequences were compared using the DNA-star Megalign program (DNA-Star, Madison,
WI).

Table 3   Chimeric
PGA sequences of selected active colonies

Besides significant efficiency to
form chimeric gene, family shuffling in the study also exhibited remarkably
high efficiency of producing point mutation. The overall mutagenic rate in this
study reached at 0.38% (18/4690). It also showed that the gene recombination
and crossover was prone to occur at the regions with higher similarity. The
length of overlap between parent sequences was crucial to form chimeric genes,
and in this experiment, it was found that a crossover was observed at the
sequences within 8 bases identities.

3    Discussion

For rational
enzyme design, those amino acid residues located close to the active center or
binding pocket etc. are often modified because they are more ‘logical’ to
altering the three-dimensional structure. Evidently, it is more difficult to
demonstrate functions of amino acids located far away from the place where the
reaction takes place. For example, Sizmann et al.[19] found that
maturation process towards alteration at the extreme C-terminus of the E.
coli PGA precursor was highly sensitive. These results indicated that some
‘unimportant’ mutations might prevent PGA proper folding or correct processing.
On the other hand, some ‘unimportant’ mutations would significantly affect the
catalytic properties of an enzyme[20]. In sharp contrast to rational design,
DNA family shuffling permuted blocks of sequence containing conservative amino
acid substitutions that had been selected for function during millions of years
of natural selection. Shuffling of gene families did not require information
about how enzymes structure was related to function and would accelerate the
evolution process.

Family shuffling
always involves the problem of reassembling the gene fragments into parental
gene sequences and prevents the formation of chimeric sequences. Generally, a
highly efficient screening procedure was required to overcome this problem[21].
When attractive selection methods were not available, creating chimeric sequences
by utilizing hereto-primer did become an alternative. In this study, efficiency
of recombination was improved using hetero-primers. This manipulation
eliminated background produced by parental gene. The diversity in the libraries
in this study was overwhelmingly generated by recombination of preexisting
natural sequence diversity in the gene family.

Our results
indicated pga gene could tolerate high mutation densities and derived much more
chimeras from diverse species. To evaluate functions of these segments
exchanges and point mutations, molecular modeling of evolved mutants was being
carried out. Resulted information could be used to minimize the sequence space
that must be searched or to suggest targets for further site-directed
mutagenesis. The strategy described here represented an applicable and
promising route to improving the PGA synthetic activity. Further work could
surely help to increase the understanding on structure/function relationships
of PGA. Increased S/H ratios of mutants to higher levels by multiple-cycle DNA
family shuffling were expected in future research.

References

1     Wu
XJ, Yang S, Wang Z, Zhan XZ, Yuan ZY. Enzyme engineering in the industrial
production of β-lactam
antibiotics. World Sci-Tech R&D, 2001, 23(1): 39－43

2     Kutzbach
C, Rauenbusch E. Preparation and general properties of crystalline penicillin
acylase from Escherichia coli ATCC11105. Hoppe Seylers Z Physiol Chem,
1974, 355(1): 45－53

3     Klei
HE, Daumy GO, Kelly JA. Purification and preliminary crystallographic studies
of penicillin G acylase from Providencia rettgeri. Protein Sci, 1995, 4(3): 433－441

4     Choi
KS, Kim JA, Kang HS. Effects of site-directed mutations on processing and
activities of penicillin G acylase from Escherichia coli ATCC 11105. J
Bacteriol, 1992, 174(19): 6270－6276

5     Martin
J, Prieto I, Barbero JL, Perez-Gil J, Mancheno JM, Arche R. Thermodynamic
profiles of penicillin G hydrolysis catalyzed by wild-type and Met-Ala168
mutant penicillin acylases from Kluyvera citrophila. Biochim Byiophys
Acta, 1990, 1037(2): 133－139

6     Yang
S, Huang H, Li SY, Ye YZ, Wan L, Zhang FW, Yuan ZY. Enhancing penicillin G
acylase stability by site-directed mutagenesis. Acta Biochim Biophys Sin, 2000,
32(6): 581－585

7     Van
Laan JM, Riemens AM, Quax WJ. Mutated penicillin G acylase genes. US Patent:
006033823A. 2000-03-07

8     Alkema
WB, Dijkhuis AJ, Vries E, Janssen DB. The role of hydrophobic active-site
residues in substrate specificity and acyl transfer activity of penicillin
acylase. Eur J Biochem, 2002, 269(8): 2093－2100

9     Chen
RD. Enzyme engineering: Rational redesign versus directed evolution. Trends
Biotechnol, 2001, 19(1): 13－14

10    Crameri
A, Raillard SA, Bermudez E, Stemmer WP. DNA shuffling of a family of genes from
diverse species accelerates directed evolution. Nature, 1998, 391(6664): 288－291

11    Stemmer
WP. Rapid evolution of a protein in vitro by DNA shuffling. Nature, 1994,
370(6488): 389－391

12    Ness
JE, Welch M, Giver L, Bueno M, Cherry JR, Borchert TV, Stemmer WP et al. DNA
shuffling of subgenomic sequences of subtilisin. Nat Biotechnol, 1999, 17(9):
893－896

13    Chang
CC, Chen TT, Cox BW, Dawes GN, Stemmer WP, Punnonen J, Patten PA. Evolution of
a cytokine using DNA family shuffling. Nat Biotechnol, 1999, 17(8): 793－797

14    Christians
FC, Scapozza L, Crameri A, Folkers G, Stemmer WP. Directed evolution of
thymidine kinase for AZT phosphorylation using DNA family shuffling. Nat
Biotechnol, 1999, 17(3): 259－264

15    Zhao
H, Arnold FH. Optimization of DNA shuffling for high fidelity recombination.
Nucleic Acids Res, 1997, 25(6): 1307－1308

16    Zhou
Z, Zhou LP, Chen MJ, Zhang YL, Li RB, Yang S, Yuan ZY. Purification and
characterization of Alcaligenes faecalis penicillin G acylase expressed in
Bacillus subtilis. Acta Biochim Biophys Sin, 2003, 35(5): 416－422

17    Ljubijankic
G, Konstantinovic M, Glisin V. The primary structure of Providencia rettgeri
penicillin G amidase gene and its relationship to other gram negative amidases.
DNA Seq, 1992, 3(3): 195－200

18    Zhou
Z, Zhang AH, Zhou LP, Yang S, Jiang YM, Yuan ZY. Cloning and expression of
penicillin G acylase gene from Providencia rettgeri in Escherichia coli.
Industrial Microbiology, 2002, 32(3): 1－5

19    Sizmann
D, Keilmann C, Bock A. Primary structure requirements for the maturation in
vivo of penicillin acylase from Escherichia coli ATCC 11105. Eur J Biochem,
1990, 192(1): 143－151

20    Spiller
B, Gershenson A, Arnold FH, Stevens RC. A structural view of evolutionary
divergence. Proc Natl Acad Sci USA, 1999, 96(22): 12305－12310

21    Coco
WM, Levinson WE, Crist MJ, Hektor HJ, Darzins A, Pienkos PT, Squires CH et al.
DNA shuffling method for generating highly recombined genes and evolved
enzymes. Nat Biotechnol, 2001, 19(4): 354－359

Receive: February 19, 2003   Accepted: March 24, 2003

This work was supported by the grants from
the National Natural Science Foundation of China (No. 30100029) and the National
High Technology Research and Development Program of China (863 Program) (No.
2001AA235081)

*Corresponding author:

YUAN Zhong-Yi: Tel, 86-21-54921246; Fax,
86-21-54921011; e-mail, [email protected]

YANG Sheng: Tel, 86-21-64042090-4720;
e-mail, [email protected]