Identification and functional analysis of fructosyl amino acid-binding protein from Gram-positive bacterium Arthrobacter sp.


  • A. Sakaguchi-Mikami,

    1. Graduate School of Bionics, Computer and Media Sciences, Tokyo University of Technology, Hachioji, Japan
    Search for more papers by this author
  • S. Ferri,

    1. Department of Biotechnology, Graduate School of Engineering, Tokyo University of Agriculture and Technology, Tokyo, Japan
    Search for more papers by this author
  • S. Katayama,

    1. Department of Biotechnology, Graduate School of Engineering, Tokyo University of Agriculture and Technology, Tokyo, Japan
    Search for more papers by this author
  • W. Tsugawa,

    1. Department of Biotechnology, Graduate School of Engineering, Tokyo University of Agriculture and Technology, Tokyo, Japan
    Search for more papers by this author
  • K. Sode

    Corresponding author
    1. Department of Biotechnology, Graduate School of Engineering, Tokyo University of Agriculture and Technology, Tokyo, Japan
    • Graduate School of Bionics, Computer and Media Sciences, Tokyo University of Technology, Hachioji, Japan
    Search for more papers by this author


Koji Sode, 2-24-16 Nakacho Koganei 185-8588 Tokyo, Japan. E-mail:



Fructosyl amino acid-binding protein (FABP) is a substrate-binding protein (SBP), which recognizes fructosyl amino acids (FAs) as its ligands. Although FABP has been shown as a molecular recognition tool of biosensing for glycated proteins, the availability of FABP is still limited and no FABP was reported from Gram-positive bacteria. In this study, a novel FABP from Gram-positive bacteria, Arthrobacter spp., was reported.

Method and Results

BLAST analysis revealed that FABP homologues exist in some of Arthrobacter species genomes. An FABP homologue cloned from Arthrobacter sp. FV1-1, FvcA, contained a putative lipoprotein signal sequence, suggesting that it is a lipoprotein anchored to the bacterial cytoplasmic membrane, which is a typical characteristic for SBPs from Gram-positive bacteria. In contrast, FvcA also exhibits high amino acid sequence similarity to a known Gram-negative bacterial FABP, which exists as a free periplasmic protein. FvcA, without the N-terminal anchoring region, was then recombinantly produced as soluble protein and was found to exhibit Nα-FA-specific binding activity by intrinsic fluorescent measurement.


This study identified a novel FABP from a Gram-positive bacterium, Arthrobacter sp., which exhibited Nα-FA-specific binding ability. This is the first report concerning an FABP from a Gram-positive bacterium, suggesting that FABP-dependent FA catabolism system is also present in Gram-positive bacteria.

Significance and Impact of the Study

The novel FABP exhibits the ability to specifically bind to Nα-FA with a high affinity. This selectivity is beneficial for applying FABP in HbA1c sensing. The successful preparation of water-soluble, functionally expressed Gram-negative bacterial FABP may make way for future applications for a variety of SBPs from Gram-positive bacteria employing the same expression strategy. The results obtained here enhance our understanding of bacterial FA catabolism and contribute to the improved development of FABP as Nα-FA-sensing molecules.


Glycation is a series of nonenzymatic reactions between reducing sugars and amino groups on free amino acids or proteins (Hodge and Rist 1953). In the early stages of glycation, a free amino group reversibly reacts with the sugar to produce a Schiff base that may then undergo an essentially irreversible Amadori rearrangement producing Amadori products such as fructosyl amino acid (FA). Amadori products can undergo several further chemical reactions leading to the formation of advanced glycation end products (Singh et al. 2001), which are recognized as indicators of diabetes mellitus, for example, Nα-glycated haemoglobin or haemoglobin A1c [HbA1c] (Schnedl et al. 2000; Tanaka et al. 2001) and glycated albumin (Iberg and Fluckiger 1986), and are assumed to be associated with diabetic complications and ageing (Ulrich and Cerami 2001; Konova et al. 2004). Glycated compounds also accumulate in food products during processing and storage, and they are considered to influence taste, appearance and overall quality. Therefore, it is very important to accurately, and easily, detect glycated compounds, such as FA, for diagnostic purposes as well as for the quality control of foods during processing and storage (Ames 1998; Borrelli et al. 2003; Foerster and Henle 2003; Yamagishi et al. 2007).

Glycated compounds are also found in the natural habitat of micro-organisms. Agrobacterium tumefaciens causes crown gall tumours on higher plants through the genetic transfer of its tumour-inducing plasmid (Vaudequin-Dransart et al. 1995; Palanichelvam et al. 2000; Kim et al. 2001; Baek et al. 2003; White and Winans 2007). Expression of the transferred genes results in the synthesis of opines, which are specifically utilized as nutrient sources by Ag. tumefaciens that have genes for the appropriate opine catabolism systems. Some strains, such as Chry5 and EU5 , have pTi-encoded genes that induce the production of Amadori compounds in the host plant (Palanichelvam et al. 2000). Santhopine is one Amadori product found in crown galls and is one opine produced as an intermediate of the mannityl opine-type strain. Recently, Beak et al. reported that Ag. tumefaciens C58 lacking pTi can also assimilate santhopine, and it contains the santhopine catabolism operon (soc operon) on the At plasmid (Kim et al. 2001; Baek et al. 2003). Soc operon consists of four open reading frames, socA, B, C and D. We determined that socA encodes a new class of substrate-binding proteins (SBPs) with fructosyl amino acid-binding protein (FABP) as the key component of the ABC transporter (Sakaguchi et al. 2005). Considering its high affinity and ligand specificity, we propose the application of FABP as the ligand in an HbA1c diagnostic kit/system. SocB and SocC are assumed to encode ABC transporter and oxidoreductase, respectively. SocD has been identified as bacterial fructosyl amino acid oxidase (FAOD) (Horiuchi et al. 1989; Ferri et al. 2005). These proteins are assumed to constitute a santhopine catabolism system.

Fructosyl amino acid oxidases catalyse oxidative degradation of FA generating corresponding amino acid, glucosone and hydrogen peroxide. FAODs have been isolated from a number of different micro-organisms, including bacteria, filamentous fungi and marine yeast (Ferri et al. 2009). Bacterial FAODs have been isolated from soil micro-organisms, such as from Corynebacterium sp. (Horiuchi et al. 1989) and Arthrobacter sp. FV1-1 (Ferri et al. 2005). Interestingly, both proteins are capable of utilizing FA as the sole carbon or nitrogen source. On the basis of the primary structural information of these bacterial FAODs, we identified that the socD gene of Ag. tumefaciens encodes FAOD, and its homologues are widely distributed in the genome of Agrobacterium strains (Hirokawa and Kajiyama 2002). Owing to the highly conserved FAOD genes among the discovered bacterial FAODs, it is suggested that they were distributed through horizontal gene transfer at some time during their evolution. To demonstrate its capability of utilizing FA, we reported the presence of FAODs in various strains in the genus Arthrobacter (Ferri et al. 2005).

In this study, we investigated the FABP from Arthorbacter spp. Only a few Gram-positive bacteria-derived SBPs and no FABP from Gram-positive bacteria have been reported. Assuming that the N-terminal region of FABP from Gram-positive bacteria is expected to differ from those of Gram-negative bacteria (Tam and Saier 1993), we performed BLAST analyses of the genus Arthrobacter by using SocA-encoded FABP without its membrane-anchoring region as a query. As a result, SocA-like protein (SolA) was found from four strains of Arthrobacter genomic information, which is assumed to encode FABP. The PCR amplification of the corresponding gene from Arthorbacter sp. FV1-1 and its characterization of recombinant product indicated that solA encodes the first Gram-positive bacterial FABP.

Materials and methods

Chemicals, enzymes and bacterial strain

Glycated products were synthesized as previously described (Sakaguchi et al. 2005). Restriction endonucleases, AmpliTaq Gold and DNA ligation kit were purchased from Takara (Kyoto, Japan). Arthrobacter sp. FV1-1 was isolated in our laboratory as an alpha-glycated amino acid-assimilating bacterium, as previously described (Ferri et al. 2005).

Genetic manipulation

PCR was carried out using primers (Table 1) as described below to amplify the SocA homologue gene from Arthrobacter sp. Each PCR product was inserted into the pGEM vector (Promega, Madison, WI, USA). Two operons, soc operon and putative soc-like operon from Arthrobacter (sol operon), were compared with design oligonucleotide primers used for the cloning. Two degenerate oligonucleotides, a1 and a2, corresponding to regions on the 5′-terminus or 3′-terminus, which are highly conserved between socA and socA genes – solAs, were synthesized and used for PCR analysis with genomic DNA purified from Arthrobacter sp. FV1-1 as a template. The DNA sequence of the resultant PCR product was then determined. Subsequently, two primers, b2 and b3, designed to the internal sequence of the putative FABP gene from A. sp. FV1-1 were designed on the basis of the DNA sequence of the products and used for PCR of the A. sp. FV1-1 genome. The forward primer corresponds to a conserved region at the upstream of socA and solA genes, another primer, b1, and a reverse primer for the 3′-terminal of FAOD gene from A. sp. FV1-1, b4, respectively. The putative FABP gene was, therefore, amplified using two other oligonucleotides, c1 and c2, designed to correspond to the resulting fvcA's 5′-terminal or 3′-terminal DNA sequence. Consequently, a 4·7-kb region, including the putative FABP-encoding operon and the fructosyl valine catabolism (fvc) operon, was analysed.

Table 1. Oligonucleotides used as primers in this study
NameCorresponding siteSequence
  1. Italic letters indicate restriction enzyme recognition sites.

a1fabp N-terminus conserved sequence site5′-GATAACCCCTACGGCCTGATACAGC-3′
a2fabp C-terminus conserved sequence site5′-CTTCTTCCAGGTGCCGTCCTCC-3′
b1socR conserved sequence site5′-GTACATCAGGAACGAGACATCTGTTGG-3′
b2fvcA internal sequence5′-GATGCACACTTCCTCGATTACGAGG-3′
b3fvcA internal sequence5′-TTTGGAAGACTCGTCGGCCAGC -3′
b4fvcE inner sequence5′-CAACGTCGAAACCCTGTGAAGTTCC-3′
d2fvcA gene 3′-terminal5′-TTGCTCGAGCTTGGCGGCCGTGG-3′

For recombinant expression of the cloned putative FABP from A. sp. FV1-1 (FvcA), 5 ′-terminal region–deleted FvcA gene (Δ1-33fvcA), corresponding to the soluble domain of FvcA (sFvcA), was amplified using the forward primer d1 (Table 1). This primer introduces an NdeI site at the start codon, whereas the reverse primer d2 (Table 1) replaces the stop codon with a serine codon and introduces an XhoI site immediately downstream of the gene. The PCR product was digested with NdeI and XhoI and ligated into pET-30c (Novagen, Madison, WI, USA) digested with the same enzymes.

The nucleotide sequences of the constructed plasmids were confirmed using the ABI Prism BigDye Terminator cycle sequencing kit v3·0 system on an ABI Prism 3100 Genetic Analyzer (Applied Biosytems, Foster city, CA, USA).

Recombinant expression and purification of sFvcA

The sFvcA gene was expressed in Escherichia coli BL21(DE3) in LB medium at 30°C by inducing for 4 h with 0·5 mmol l−1 IPTG. A part of harvested cells were used for preparation of the whole-cell, insoluble, cytoplasmic and periplasmic fractions, according to Novagen_s pET System Manual to analyse the expression pattern of sFvcA. The rest of the harvested cells were resuspended in 8 mol l−1 urea/100 mmol l−1 NaCl/10 mmol l−1 Tris-HCl (pH 8·0), sheared by ultrasonication and centrifuged at 10 000× g for 15 min at 4°C. The supernatant was used for further purification under denaturing conditions. The supernatant was then loaded onto a 2-ml Ni-NTA agarose (Qiagen GmbH, Hilden, Germany) column equilibrated with 8 mol l−1 urea/100 mmol l−1 NaCl/10 mmol l−1 Tris-HCl (buffer A) (pH 8·0). The column was washed with three column volumes of the wash buffer (buffer A at pH 6·3) to remove the unbound proteins. The denatured sFvcA was eluted with one column volume of the elution buffer (buffer A at pH 4·9). The refolded sFvcA was prepared by stepwise dialysis to remove urea, followed by dialysis against 10 mmol l−1 MOPS (pH 7·0).

The expression and purification levels of sFvcA were determined by SDS-PAGE. Samples containing 90 ng protein were separated on Gradient 8-25 PhastGels (GE Healthcare, Piscataway, NJ, USA) and detected by silver staining. The protein concentration was determined by the Bradford method by using the Protein Assay kit (Bio-Rad Laboratories, Hercules, CA, USA) and bovine serum albumin as the standard.

Intrinsic fluorescence monitoring

The purified sFvcA was monitored in a 0·5 × 0·5 cm cuvette using an FP-6500 ETC.-273 spectrofluorometer (JASCO, Tokyo, Japan) with a thermostat cell holder. Slits of excitation and emission monochromators were set at 1 and 3 nm, respectively. The excitation wavelength used was 295 nm. Maximum intensity was measured after incubating sFvcA at a final concentration of 1·8 μmol l−1 with each substrate for 2·5 min at 25°C. The relative intensity changes (ΔF/F0) were calculated as the percentage of the total intensity in the absence of substrate and used to evaluate the FABP intrinsic fluorescence changes. Dissociation constants were calculated by a half-reciprocal plot (Scott plot) using the ratio of intrinsic fluorescence intensity (FI) change and substrate concentration.


Homology analysis of SolA proteins from the available Arthrobacter spp. genomic information

SocA is the first reported and sole FABP with a 277-amino acid protein derived from Ag. tumefaciens. The protein contains a 32-amino acid signal sequence at its N-terminus, and the mature SocA (mSocA) is secreted into the periplasmic space after cleavage of the signal sequence by signal peptidase (Sakaguchi et al. 2005). Such a post-translational modification is specific for Gram-negative bacteria; therefore, SocA N-terminal sequence information was not utilized for the analyses of FABP homologues from Gram-positive bacterium. Considering the fact this possibly contained a unique N-terminal sequence, Arthrobacter-derived FABP were analysed using the BLAST tool and the amino acid sequence for mSocA as a query.

Among the six Arthrobacter strains of which whole-genomic information are available, four strains that contained SocA homologues were identified. The sequence identity and similarity with mSocA in these putative proteins were greater than 66% and 78%, respectively. These mSocA homologues from Arthrobacter spp. were, therefore, expected to be a novel FABP.

The PCR cloning towards the genome of the fructosyl valine (f-Val) assimilating strain, Arthrobacter FV1-1, was then carried out. This was performed using degenerated internal probes and those corresponding to the up/downstream conserved regions within socA and its homologues. The obtained gene from FV1-1 encoding the SolA protein was described as fvcA.

A SocA gene homologue, fvcA, was an 858-bp open reading frame encoding a 285-amino acid protein with a molecular mass of 29 837 Da. Figure 1 shows the primary structure alignment of FvcA with SocA and four other SolAs found using the BLAST search tool. According to the 96% similarity in amino acid sequences, FvcA and SolAs were presumed to be identical to SolA derived from the other Arthrobacter strains.

Figure 1.

Amino acid sequence alignment of SocA and its homologues from Arthrobacter strains. Primary structures of SocA, FvcA and SolA from A. aurescens TC1 (SolA-a), A. globiformis strain NBRC 12137 (SolA-g), A. phenanthrenivorans strain Sphe3 (SolA-s) and Arthrobacter sp. strain FB24 (SolA-f) were compared. Confirmed and predicted signal sequences are underlined with solid and dashed lines, respectively, and the cleavage sites are indicated with black triangles. The signature sequence regions for polar amino acid-binding proteins are enclosed in boxes with dashed lines, whereas the residues predicted to contribute ligand binding are enclosed in boxes with solid lines.

FvcA exhibited a 64% amino acid sequence similarity with SocA, and like SocA (Sakaguchi et al. 2005), it was assumed to be a member of the polar amino acid-binding protein (PABP) family (Tam and Saier 1993). The residues predicted to compose the ligand-binding site of SocA were also well conserved in FvcA, suggesting that FvcA exhibits an FA-binding ability (Fig. 1). However, the N-terminal region (1–33 amino acids) of FvcA was predicted to contain a lipoprotein signal peptide (1–21 amino acids) and sortase recognition site (Rahman et al. 2008; Bagos et al. 2009), whereas that of SocA contains a periplasmic signal sequence (Fig. 2a). FvcA is, thus, likely to be a lipoprotein that undergoes a diacylglycerol post-translational modification and is then anchored to the plasma membrane after it is secreted to the outside of the membrane.

Figure 2.

Recombinant expression of putative fructosyl amino acid-binding protein from Arthrobacter FV1-1. The sFvcA was designed on the basis of the FvcA gene structure (a). FvcA had a typical lipoprotein signal sequence containing an N-terminal positive charge (underlined) and hydrophobic (double underlined) regions as well as a C-terminal lipobox (italic) that contains the lipid-modified Cys residue, with a Leu residue at the -3 position found at the N-terminal region. Purity of sFvcA was analysed by SDS-PAGE (b). Samples containing 90 ng of protein or low-molecular-weight marker containing 34 ng of protein were separated on 20% polyacrylamide gels, and the protein bands were detected by silver staining.

The cloned SocA homologue gene from Arthrobacter sp. FV1-1, fvcA, was therefore indicated to encode a novel FABP that exists as a lipoprotein.

Recombinant expression and characterization of soluble FvcA (sFvcA)

To characterize the FABP homologue from Arthrobacter, sFvcA, that is, FvcA without its signal sequence and membrane-anchoring region (Δ1-33 fvcA), was recombinantly produced using E. coli as the host micro-organism.

sFvcA was inserted into the pET-30c vector (pET-sfvcA), with a C-terminal in-frame histidine tag for protein purification. E. coli BL21(DE3) was transformed with the resulting expression vector, and the expression level of each cellular fraction was analysed by SDS-PAGE (data not shown). A 27-kDa band was clearly observed in the whole-cell, soluble and insoluble fractions of IPTG-induced cells, with a much stronger intensity than in pET-30c-induced cells or uninduced pET-sfvcA-harbouring cells. sFvcA was then purified in a single-step process from the whole-cell fraction under denaturing conditions, by affinity chromatography using a Ni-NTA column with batch elution. In this study, a yield of 17 mg purified recombinant sFvcA was achieved from 1 l of culture. The recombinantly expressed sFvcA fused at the C-terminus to a histidine tag is expected to be of 27 kDa. This is in agreement with the size of the 27-kDa protein band observed by SDS-PAGE (Fig. 2b).

Protein–ligand binding can be detected by intrinsic fluorescence measurement when the inherent aromatic residues of the protein are affected by the ligand binding (Engelborghs 2001; Sakaguchi et al. 2005; Sakaguchi-Mikami et al. 2008). The intrinsic florescence of sFvcA, which has two tryptophan residues (Trp186 and Trp193), was thus investigated at an excitation wavelength of 295 nm.

The maximum sFvcA intrinsic FI was observed at approximately 336 nm (Fig. 3a) and increased with an increasing sFvcA concentration (data not shown). sFvcA therefore has a significant intrinsic fluorescence that can be measured by determining the fluorescence spectrum of the protein.

Figure 3.

Characterization of putative fructosyl amino acid-binding protein from Arthrobacter FV1-1. The fluorescence spectra of purified sFvcA (15 μmol l−1) in the presence (solid line) or absence (solid line) of f-Gln was measured to investigate the binding ability of sFvcA to f-Gln (a). The measurements were carried out in 10 mmol l−1 MOPS buffer (pH 7·0) at 25°C. Correlation between the fluorescence intensity and the concentration of f-Gln (filled square, solid line) and f-Val (circle, dashed line) was measured (b). The average maximal intensity for each substrate was determined as in (a). The error bars indicate standard deviation (n = 3).

The maximum FI of sFvcA was then measured after incubation with various substrates (Table 2). The FI increased with increasing Nα-FAs, f-Val and f-Gln (Fig. 3b). No FI change was detected by incubating sFvcA with sugars, amino acids or Nε-fructosyl lysine (Nε-f-Lys). Further investigation revealed that the increased intrinsic FI correlated well with Nα-FA concentration between 0·1 and 1·0 μmol l−1 and was saturated above 10 μmol l−1 Nα-FA (Fig. 3b). The Kd value of sFvcA for f-Gln and f-Val was calculated to be 0·6 and 0·3 μmol l−1, respectively. sFvcA changes its conformation by specifically binding to Nα-FA with high affinity in a concentration-dependent manner.

Table 2. Ligand specificity of sFvcA
LigandKd (μmol l−1)
  1. n.d.: not detected.

Amino acid


FvcA isolated in this study is the first FABP found in a Gram-positive bacterium. A number of periplasmic SBPs from Gram-negative bacteria have recently been isolated, engineered and successfully used for the development of fluorescent sensors. However, only few SBPs from Gram-positive bacteria have been described (Tynkkynen et al. 1993; Pearce et al. 1994; Jenkinson et al. 1996; Kempf et al. 1997; Kolenbrander et al. 1998; Seo et al. 2002; Wang et al. 2002; Williams et al. 2004; Paterson et al. 2006), and to the best of our knowledge, no biosensing application has been reported thus far. The characterization of FvcA was carried out using recombinant production by E. coli, expressing FvcA without the N-terminal signal sequence. This region is expected to be involved in membrane anchoring, a trait related to the SBPs found in Gram-positive bacteria. The successful preparation of water-soluble, functionally expressed FABPs may make way for future applications for a variety of SBPs from Gram-positive bacteria employing the same expression strategy.

Gram-positive bacterial SBPs that are anchored to cytoplasmic membranes are elements of ABC transporters as well as periplasmic SBPs of Gram-negative bacteria. According to the sequence homology analysis by the BLAST analysis tool, the soluble domains of SBPs from Gram-positive bacteria were found to exhibit low sequence similarity to SBPs from Gram-negative bacteria despite they have similar scaffolds and ligand-binding specificity and affinity [e.g. oligopeptide-binding protein from Lactococcus lactis (Tynkkynen et al. 1993) exhibits only less than 20% sequence identity to those SBPs from Gram-negative bacteria, and Streptomyces griseus glucose-binding protein (Seo et al. 2002) revealed less than 40% sequence identity to Gram-negative bacterial SBPs for glucose]. However, FvcA exhibits a high similarity towards SocA. The comparison of soc, sol and fvc operon structures suggested that Arthrobacter strains also have Να-FA assimilation systems similar to that from Agrobacterium sp. These results indicate the possibility that the Gram-positive bacteria Arthrobacter spp., which are widely distributed in soil, have acquired an soc operon by the horizontal gene transfer. However, further studies to identify varied SBPs from Gram-positive bacteria, as well as the distribution of Nα-FABP system in soil micro-organisms, are required to fully understand the origin of the bacterial Nα-FA catabolism operon.

sFvcA exhibits the ability to specifically bind to Nα-FA with a high affinity, similar to SocA. This selectivity is beneficial for applying FABP in HbA1c sensing to distinguish the f-Val produced from the degradation of HbA1c from the Nε-f-Lys that results from the degradation of glycated albumin. On the other hand, small differences in ligand preferences and amino acid sequences were found between mSocA and sFvcA. The comparison of their properties could provide important information to engineer the molecule for application to diagnosis. On the other hand, the fructose and amino acid moieties of Nα-FA are predicted to be arranged side by side against the hinge loops of FABP based on its 3D structural model reported previously. This prediction indicates the ability of FABPs to bind with the N-terminal region of HbA1c. The further investigation of the residues to be modified with adequate environmentally sensitive fluorescent probe, which is not affected by the haeme inherent fluorescent property, may allow us the development of HbA1c using FABP. FvcA, as well as SocA, is therefore expected to provide a new, highly sensitive Nα-FA measurement system that is ideally suited for HbA1c, a major diabetes indicator.

This study identified a novel SBP from Arthrobacter sp., which exhibited Nα-FA-specific binding ability. This is the first report concerning an FABP from a Gram-positive bacterium. We also showed that FABP homologues are distributed among the Gram-positive soil bacteria Arthrobacter spp. The finding of the FABP from a Gram-positive bacterium with Nα-FA-binding ability and highly conserved amino acid sequence with that from Agrobacterium's FABP may therefore provide important information for understanding bacterial FA assimilation systems and its evolutional pathway. The Gram-positive bacterial FABP was successfully produced in E. coli as a functionally soluble protein. These results may provide a strategy to engineer a variety of Gram-positive bacterial SBPs for future applications, as well as contributing to the improvement of FABPs as probes for FA measurements, as seen in HbA1c diagnosis.