Introduction

Phytase is a general term describing an enzyme that initiates the hydrolysis of phosphomonoester bonds from phytate (myo-inositol 1,2,3,4,5,6-hexakisphosphate, InsP6), thereby liberating inorganic orthophosphate (Mullaney and Ullah 2003). Phytases are used as an animal feed additive to improve phosphate bioavailability and to reduce the loss of phosphate and divalent cations from phytate, which represents the principal storage form of phosphorous in plant seeds. According to their pH optimum, acid and alkaline phytases can be distinguished (Oh et al. 2004). Acid phytases evolved from acid histidine acid phosphatases (HAPs, EC 3.1.3.8), which display general phosphomonoesterase activity directed against a broad spectrum of phosphate-containing substrates including phenyl- and naphthyl-ortho-phosphates. A minority of HAPs are able to hydrolyse phytate with varying degrees of efficiency. The phytase enzyme with the highest specific activity presently known is the pH 2.5 acid phosphatase AppA from Escherichia coli (Golovan et al. 2000). In contrast, E. coli acid glucose-1-phosphatase (Agp, EC 3.1.3.10), which, despite its apparent sequence similarity to AppA, primarily cleaves small monosaccharide phosphates such as glucose-1-phosphate, glucose-6-phosphate and fructose-6-phosphate, but it also exhibits phytase activity (Cottrill et al. 2002). Occurrence and characterisation of Agp in bacteria other than E. coli has not been reported so far.

Previous experiments revealed that representatives of the genus Enterobacter can hydrolyse phytate (Sajidan 2002). Here, we have cloned the gene agpE, which encodes an enzyme that hydrolyses glucose-1-phosphate and phytate, from an Enterobacter cloacae VKPM B2254 plasmid library. In addition, a second agpE gene was cloned from strain ASR5 sharing high 16S rDNA sequence similarity with that of E. cloacae. The deduced amino acid sequence of agpE was nearly 80% identical with that of the E. coli agp gene, and the biochemical properties of the expressed recombinant gene product were found to be similar to those reported for E. coli Agp except 3-phytase activity that exceeded that of its E. coli counterpart two times.

Materials and methods

E. cloacae VKPM B-2254 was obtained from S.P. Sineoky, Moscow, and is deposited in the strain collection of the State Institute for Genetics and Selection of Industrial Microorganisms, Moscow, Russia. Enterobacter sp. ASR5 strain was isolated and taxonomically characterised by Sajidan (2002). According to its 16S rDNA sequence strain, ASR5 was found highly related to E. cloacae and was deposited in the DSMZ culture collection as Enterobacter sp. DSM 17269. E. cloacae DNA was partly digested with Sau3A and ligated into BamHI-linearised pUCI9 and cloned into E. coli XL1-Blue. The resulting E. cloacae genome library contained about 10,000 clones with insert sizes ranging from 5 to 8 kb. Transformants were screened for phytase activity on plates as previously described (Zinin et al. 2004). Plasmid DNA of two clones expressing phytase activity was isolated and found to harbour overlapping inserts of 6.1 and 5.3 kb. After appropriate subcloning, the insert DNA containing the glucose-1-phosphatase gene was sequenced in both directions using standard primers [pUC/M13 (forward), pUC/M13 (reverse)] and specific primers (pOB1, forward 1,261–1,280: 5′-acgctggagctgaaaggctg-3′; pOB2, reverse 670–651: 5′-tcatctgctccagcagctta-3′; and pOB3, reverse 261–242: 5′-gacggctcatgattaaaact-3′). DNA sequences were determined with an automatic sequencing system (ALF, Pharmacia). Sequence analysis was performed with the BLAST, ClustalW (Thompson et al. 1994) and phylogenetic analysis using parsimony (PAUP; Swofford 2002) programme packages. To clone the agpE gene in the expression pET 22b(+) vector, the gene was amplified using the primers OB_for1, forward: 5′-agtgaggaattacatatgagaaaagcac-3′, and OB_rev, reverse: 5′-tgaattcgccgcgttattcatcac-3′. The obtained DNA fragment was digested with NdeI and EcoRI, and inserted between the NdeI and EcoRI sites of the vector; ligation products were transferred into E. coli C41(DE3) (Miroux and Walker 1996) to test for enzyme activity. C41(DE3) transformants were cultured at 37°C in TBY containing ampicillin (50 μg ml−1). At OD600=0.6–0.8, phosphatase expression was induced by addition of isopropyl-β-d-thiogalactopyranoside (IPTG, final concentration 1 mM), and the cultures were further incubated at 37°C for 5 h. After centrifugation, glucose-1-phosphatase was purified from the supernatant by affinity chromatography with Ni-NT agarose (Electronic Supplementary Material, Fig. S1). Phytase and phosphatase activities were assayed at 37°C in 0.1 M sodium acetate (pH 5.0) as previously described (Sajidan et al. 2004). One unit of activity was defined as the amount of enzyme that liberates 1 μmol phosphate in 1 min at 37°C. Protein concentrations were determined by the method of Bradford (1976) using bovine serum albumin as a standard. Specific enzyme activities were defined in units per milligram protein and determined as previously described (Sajidan et al. 2004). The molecular mass of native AgpE was determined by FPLC-column chromatography (column TOSOHAAS TSK gel G2000 SWXL, 7.8×300 mm, particle size 5 μm) in 100 mM Tris–HCl, pH 7.5, and in 100 mM sodium acetate buffer, pH 6.0, supplemented with 150 mM sodium chloride. Products of phytate hydrolysis were determined as previously described (Sajidan et al. 2004). The coding sequence of the agpE gene from E. cloacae VPKM B2254 has been deposited in the EMBL nucleotide sequence data bank under accession number AJ783768.

Results and discussion

The agpE gene was isolated from a plasmid library prepared from genomic DNA from E. cloacae VPKM B2254, as described in Materials and methods. Sequence analysis revealed that an open reading frame with 78% nucleotide identity and 79% amino acid identity to E. coli periplasmic glucose-1-phosphatase (Agp) was present in both cloned DNA fragments. The agpE gene was flanked by two open reading frames, orf1 and orf2. The deduced amino acid sequences of orf1 and orf2 displayed 81% identity to the Trp-repressor binding protein from Salmonella typhimurium LT2 and 70% identity with a transcriptional regulatory protein from Bradyrhizobium japonicum, respectively (Electronic Supplementary Material, Fig. S2). The agpE gene product is a 413-amino acid protein with notable homology to HAPs. The 391-amino acid sequence of the putative mature protein corresponds to a molecular mass of 44 kDa, and it contains a variation of the active site signature motif RHNXRXP instead of RHGXRXP, which is common in HAPs and phytases (Mitchell et al. 1997).

Amino acid sequence alignment with available bacterial HAP/phytase sequences was performed to generate a phylogenetic tree, which was used as a starting point for maximum likelihood and maximum parsimony heuristic searches with bootstrap support. The topology of the resulting phylogenetic tree, which was very similar to that of the tree obtained by the neighbour joining method, is supported by the results of bootstrap repetition tests (Fig. 1). E. cloacae AgpE and E. coli Agp, together with the putative acid Providencia rettgeri glucose-1-phosphatase and the enterobacterial S. flexeri and S. typhimurium homologs, form a separate Agp branch distinguished by the unique RHNXRXP active site motif (Electronic Supplementary Material, Fig. S3). Other bacterial HAPs/phytases previously grouped as PhyC phytases (Oh et al. 2004) can be subclassified as AppA-related phytases, along with a more heterogenous group consisting of Klebsiella spp. and Pseudomonas syringae phytases, termed PhyK. AgpE shares about 30% identical residues with E. coli AppA and Klebsiella PhyK. E. cloacae AgpE and E. coli Agp, together with a few open reading frames identified in the completed genomes of several enterobacteria, constitute a separate branch clearly separated from the “true” phytases belonging to the AppA or PhyK groups (Fig. 1).

Fig. 1
figure 1

Phylogenetic tree constructed by random stepwise parsimony using the PAUP programme package of homologs of the Enterobacter cloacae acid glucose-1-phosphatase (Agp). The bar represents one substitution per ten amino acids. Bootstrap values (%) from analysis of 1,000 bootstrap replicates are given at the respective nodes. Homologs with experimentally verified phytase activity are in bold letters. Histidine acid phosphatase (HAP) from Klebsiella pneumoniae ASR1-AphA Klpn (gi:18092533) was used as out-group. The abbreviations, source, gi accession numbers for experimental data, if available, of proteins are Agp Eclo (gi:50284400)—G1Pase E. cloacae; Agp EcoK12 (gi:145218)—G1Pase Escherichia coli K12; Agp Eco O157:H7 (gi:13360618)—G1Pase E. coli O157:H7; Agp Shifle (gi:30062539)—putative G1Pase Shigella flexneri genome; Agp Salty (gi:16764475)—putative G1Pase Salmonella typhimurium genome; Agp Prore (gi:45772)—putative G1Pase Providencia rettgeri; PhyX Xaax (gi:21244849)—phytase Xanthomonas axonopodis; PhyX Xaca (gi:21230234)—phytase Xanthomonas campestris; Phy Pssyr (gi:26280977)—putative phytase Pseudomonas syringae MOK1; PhyE Ercar (gi:50123118)—putative phytase Erwinia carotovora; PhyK Rater (gi:32451230)—phytase Raoultella terrigena; PhyK Klpn (gi:29465764)—phytase K. pneumoniae; AppA Yerpe (gi:45441589)—probable HAP Yersinia pestis; PhyO Obpro (gi:37266291)—phytase Obesumbacterium proteus; Phy Cbfreu (gi:40748277)—phytase Citrobacter freundii; AppA EcoK12 (gi:145285)—phosphoanhydride phosphohydrolase E. coli K12; AppA Eco O157:H7 (gi:25350609)—phosphoanhydride phosphohydrolase E. coli O157:H7; AppA Shifl (gi:24112393)—probable phosphoanhydride phosphohydrolase S. flexneri; AppA EcoCF (gi:26247006)—probable phosphoanhydride phosphohydrolase E. coli CFT073

Full size image

Using sequence-specific primers derived from the agpE sequence, we amplified another agp gene from the soil bacterium Enterobacter sp. strain ASR5 (DSM 17269), which was isolated from the rhizosphere of an Indonesian rice field, demonstrating that the gene is well conserved among different members of the genus Enterobacter and that it might be responsible for the phytase activity previously observed in this strain (Sajidan et al. 2004).

The apparent molecular mass of AgpE as determined by native gel filtration chromatography in 100 mM Tris–HCl buffer, pH 7.5, 150 mM NaCl, indicates that the enzyme has a dimeric form (Electronic Supplementary Material, Fig. S4b,d). Protein fractions corresponding to about 82.5 kDa cross-reacted with His-tag-specific antibodies and displayed glucose-1-phosphatase activity (Electronic Supplementary Material, Fig. S5). E. coli Agp was also reported to occur in a homodimeric state (Pradel and Bouquet 1988). Notably, pH conditions seem to affect the assembly of AgpE monomers. Complexes with a higher molecular mass of 125 kDa indicative of a homotrimeric AgpE were detected by native gel chromatography performed in sodium acetate buffer, pH 6 (Electronic Supplementary Material, Fig. S4a,c).

Besides glucose-1-phosphatase activity, purified AgpE was active towards other monosaccharide sugar phosphates: glucose-6-phosphate, fructose-1-phosphate and fructose-6-phosphate (Table 1). Activity towards phytate and the general phosphatase substrate p-nitrophenyl phosphate was also detected. Compared with the preferred substrate glucose-1-phosphate, the phytase activity of AgpE was found to be six- to sevenfold lower, corresponding to about 2% to 25% of the activities reported for true HAP phytases such as E. coli AppA (Zinin et al. 2004) and Klebsiella sp. ASR1 PhyK (Sajidan et al. 2004). On the other hand, the specific phytase activity of 32 U mg−1 lies in the same range or slightly higher than that reported for alkaline Bacillus phytases (Tye et al. 2002) and exceeds that of its E. coli counterpart two times. Interestingly, AgpE displayed a K m value for phytate, which is in the same range as that reported for AppA (Table 1). In order to examine the action of the phytate-degrading enzyme on phytate, hydrolysis products were separated by isomer-specific high-performance ion chromatography (HPIC). A marked decrease in the InsP6 (phytate) content of the reaction mixture with a concomitant increase in myo-inositol pentakisphophate content was observed during incubation. In contrast, myo-inositol phosphates with fewer than five phosphate residues were not detected throughout the incubation period (Fig. 2). These data demonstrate that the E. cloacae phytate-degrading enzyme dephosphorylates phytate quantitatively to a single myo-inositol pentakis phosphate isomer. On the contrary, the related E. coli 6-phytase AppA can dephosphorylate InsP6 in a stepwise manner to yield the Ins-2-monophosphate (Greiner et al. 2000). Also, a 3-phytase from Klebsiella ASR1, a representative of the PhyK group, was reported to gradually dephosphorylate phytate to Ins(2)P (Sajidan et al. 2004). The reported 3-dimensional structure of the AppA/phytate complex demonstrates that phytate initially does not fully occupy the flexible AppA binding pocket, but after binding the pocket, perfectly conforms to the substrate allowing its successive degradation (Lim et al. 2000). In contrast, the comparatively stiff and small substrate binding cleft of Agp allows only for the orientation for phytate in which the 3-phosphate is docked at the catalytic site of the enzyme. Leu-24 and Glu-196 in E. coli Agp have been proposed to act as “gating residues” by narrowing access to the substrate (Lee et al. 2003). Both residues are conserved in all Agp-related sequences including AgpE but are absent from the PhyK- and AppA-related homologs (Supplementary Material, Fig. S2). AgpE clearly belongs to the group of HAPs with main activity directed towards glucose-1-phosphate, but its phytase activity is striking higher than that reported for E. coli Agp, demonstrating that there is considerable variability in ratio of both enzymatic activities possibly caused by slight differences in the environment of the active site area. Investigation of further Agps from other bacterial sources might support this hypothesis.

Table 1 Comparison of kinetic parameters of the Enterobacter cloacae AgpE with E. coli Agp with various substrates
Full size table
Fig. 2
figure 2

Hydrolysis of myo-inositol hexakisphosphate by the purified recombinant AgpE. Enzyme and substrate were incubated at pH 5.0 and reaction products were analysed by high-performance ion chromatography (HPIC) (see Materials and methods). Action of purified recombinant glucose-1-phosphate on sodium phytate is indicated as a black line. Peaks of reference myo-inositol phosphates are shown in dotted lines: (1) Ins(1,2,3,4,5,6)P6; (2) Ins(1,3,4,5,6)P5; (3) D/L-Ins(1,2,4,5,6)P5; (4) D/L-Ins(1,2,3,4,5)P5; (5) Ins(1,2,3,4,6)P5; (6) D/L-Ins(1,4,5,6)P4; (7) Ins(2,4,5,6)P4; (8) D/L-Ins(1,2,5,6)P4; (9) D/L-Ins(1,3,4,5)P4; (10) D/L-Ins(1,2,4,5)P4; (11) D/L-Ins(1,2,3,4)P4; (12) D/L-Ins(1,2,4,6)P4; (13) Ins(1,2,3,5)P4; (14) D/L-Ins(1,5,6)P3; (15) D/L-Ins(1,4,5)P3; (16) D/L-Ins(1,2,6)P3, Ins(1,2,3)P3; (17) D/L-Ins(1,3,4)P3; (18) D/L-Ins(1,2,4)P3, D/L-Ins(2,4,5)P3, (19) D/L-Ins(2,4)P2; (20) D/L-Ins(1,2)P2, Ins(2,5)P2, D/L-Ins(4,5)P2; (21) D/L-Ins(1,4)P2, D/L-Ins(1,6)P2

Full size image