Periplasmic glucose-binding protein from Pseudomonas putida CSV86 – identification of the glucose-binding pocket by homology-model-guided site-specific mutagenesis

Authors

  • Arnab Modak,

    1. Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Mumbai, India
    Search for more papers by this author
  • Prasenjit Bhaumik,

    1. Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Mumbai, India
    Search for more papers by this author
  • Prashant S. Phale

    Corresponding author
    1. Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Mumbai, India
    • Correspondence

      P. S. Phale, Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Powai, Mumbai 400 076, India

      Fax: +91 22 2572 3480

      Tel: +91 22 2576 7836

      E-mail: pphale@iitb.ac.in

    Search for more papers by this author

Abstract

Glucose transport in Pseudomonas putida CSV86 is mediated via a periplasmic glucose-binding protein (GBP)-dependent putative glucose ABC transporter. Here we describe a homology model and functional characterization of GBP from CSV86 (ppGBP). A whole-cell [14C]-glucose uptake study revealed that glucose is transported by the high-affinity intracellular phosphorylative pathway. ppGBP was cloned, over-expressed in Escherichia coli and purified to apparent homogeneity. The purified ppGBPs from both E. coli and CSV86 were found to be specific for glucose. A homology model of ppGBP was constructed that resembles the class II family of periplasmic binding proteins. The model showed highest structural similarity to GBP of Thermus thermophilus (ttGBP, rmsd 0.64 Å). Structural analysis and molecular docking studies predicted W35, W36, E41, K92, K339 and H379 of ppGBP as putative glucose-binding residues. Alanine substitution of these residues resulted in significantly reduced [14C]-glucose binding activity. Analysis of the operonic arrangement and structural comparative studies suggested that ppGBP and ttGBP probably originated from a common ancestor. Structural adaptations that inhibit binding of di- or trisaccharides at the glucose-binding pocket of ppGBP were also identified.

Abbreviations
ABC

ATP-binding cassette

GBP

glucose-binding protein

IPTG

isopropyl thio-β-d-galactoside

MBP

maltose-binding protein

PBP

periplasmic binding protein

Introduction

A specific hierarchy exists in all organisms for utilization of carbon sources. For example, Escherichia coli utilizes simple sugars such as glucose followed by complex sugars such as lactose, leading to a diauxic growth response [1]. This phenomenon is known as carbon catabolite repression [2]. Unlike E. coli, Pseudomonas sp. preferentially utilize organic acids such as succinate, citrate etc. rather than sugars. This phenomenon is known as ‘reverse carbon catabolite repression’ [3]. One of the major reasons for the inefficiency of pseudomonads in the process of bioremediation, despite their metabolic versatility, may be due to carbon catabolite repression, which is poorly understood [4].

Pseudomonas putida CSV86 (hereafter referred to as CSV86), a soil isolate, utilizes naphthalene, methylnaphthalenes, benzyl alcohol, benzoate and phenylacetate as the sole source of carbon and energy [5-8]. The draft genome sequence of this strain has recently been determined [9]. The unique property of CSV86 is its ability to utilize aromatic compounds preferentially to glucose and co-metabolism of aromatic compounds and organic acids [10]. Aromatic and organic acid-mediated repression of OprB (an outer membrane protein that allows the passage of sugar molecules across the outer membrane) [11], periplasmic glucose-binding protein (ppGBP) [12] and the glucose-metabolizing enzyme Zwf [13] was found to be responsible for this ability. In addition, identification of genes encoding various components of the glucose uptake system (which includes OprB, GBP and three subunits of the ATP binding cassette (ABC) transporter) [9] led us to hypothesize the presence of a high-affinity GBP-dependent ABC transporter in P. putida CSV86. Both GBP and OprB were found to be induced by glucose and suppressed by aromatic compounds and organic acids. These proteins were also found to be repressed in CSV86 in the first log phase of diauxic growth in the presence of aromatic compounds/organic acids and glucose [11, 12]. This allows CSV86 to preferentially utilize aromatic compounds and organic acids over glucose. This property is unique and can be used for efficient bioremediation.

The periplasmic binding protein (PBP)-dependent ABC transporters comprise an important class of active transport systems for uptake of ions and nutrients by prokaryotes. PBPs confer specificity to these importers by binding to the substrate and delivering it to the cognate membrane transport assembly located in the inner membrane. PBPs themselves form a protein superfamily that includes members that are integral part of other functional complexes such as prokaryotic two-component regulatory systems [14], prokaryotic tripartite ATP-independent periplasmic transporters [15, 16], and ligand-gated ion channels [17]. The PBPs are monomeric (molecular mass range 25–70 kDa). Although there is little sequence similarity among PBPs, their spatial structural fold is highly conserved, suggesting that they have originated from a common ancestor [18]. The core of all PBPs consists of two structurally conserved globular domains (mainly α/β type), with a central β-sheet of five β-strands flanked by α-helices, connected by two or three segments of polypeptide (a hinge region), usually in extended conformation [19]. PBPs were classified based on the topology of the β-sheets core in the domains. The sheet topology in type I is β2β1β3β4β5, whereas that in type II is β2β1β3βnβ4, where βn represents the strand after the first cross-over from the N-terminal domain to the C-terminal domain and vice versa [18]. However, based on structural alignment of available crystal structures, PBPs were re-classified into six clusters: A–F [20]. In the absence of ligand, PBPs exist largely in the open conformation, with both domains separated, with possibly a small fraction in a closed un-liganded state [21, 22]. Interaction with the ligand(s), as reported for most of the PBPs, leads to significant movements in domains that stabilize the closed conformation, with the ligand(s) bound in the cleft between the two domains [23, 24]. This process is termed as the ‘Venus flytrap’ mechanism [25]. Periplasmic GBP (44 kDa) was purified from Pseudomonas aeruginosa and was found to be specific to glucose (Kd 0.35 μm) [26]. However, information regarding the structure of GBP from Pseudomonas sp. is not available.

Here we report, for the first time, the results of biochemical and structure–function studies on ppGBP from Pseudomonas sp. The periplasmic GBP from P. putida CSV86 was cloned, purified and biochemically characterized. Further, the 3D structure was modeled, and putative glucose-binding residues were identified and validated by site-directed mutagenesis, followed by functional analysis.

Results

Glucose uptake in Pseudomonas putida CSV86 is mediated by a high-affinity transporter

[14C]-glucose uptake by whole cells revealed the Km and Vmax values to be 0.81 μm and 6.9 nmol·min−1·mg−1 dry weight, respectively, suggesting that glucose uptake in the cells is mediated by a high-affinity transport system (Fig. 1). Similar values have been reported for glucose transport in other Pseudomonas species: P. chlororaphis (Km 0.3 μm; Vmax 8.8 nmol·min−1·mg−1), P. fluoroscens (Km 1.4 μm; Vmax 29.7 nmol·min−1 mg−1) [27], P. aeruginosa (Km 8 μm) [28] and P. putida U (Km 8.3 μm) [29].

Figure 1.

Glucose uptake rate for Pseudomonas putida CSV86. [14C]-glucose saturation profile for glucose-grown CSV86 cells. Curve fitting was performed using a Michaelis–Menten model.

Cloning, expression and purification of ppGBP

The low yield and purity of ppGBP obtained from CSV86 prompted to clone the gene encoding periplasmic ppGBP into pET28a for over-expression in E. coli BL21(DE3). The constructs were sequenced at Xcelris Labs, Ahmedabad, India, and the gene sequence was deposited in GenBank (accession number JQ585252.1). Upon induction with isopropyl thio-β-d-galactoside (IPTG), the majority of the protein was found to be present in the pellet (Fig. 2A). However, the soluble fraction of this construct showed higher [14C]-glucose-binding activity (20 pmol bound per mg protein, 6–7-fold higher) compared to the soluble fraction of IPTG-induced E. coli BL21(DE3) (3 pmol bound per mg protein) as well as E. coli transformed with empty pET28a vector (2.3 pmol bound mg per protein). The presence of a His or Pro tag at the N-, C- or both termini did not affect the glucose-binding activity of the protein, indicating no alteration in the structures of these proteins.

Figure 2.

Expression and purification of ppGBP. (A) SDS/PAGE analysis of ppGBP in Escherichia coli BL21 (DE3) cells when un-induced (UI) or induced (I) with IPTG. ppGBP in the pellet (P, inclusion bodies) and soluble protein fraction (S) is indicated by an arrowhead. M indicates standard protein molecular mass markers. (B) SDS/PAGE analysis of immobilized metal affinity-purified ppGBP from E. coli before (lane 1) and after (lane 2) thrombin cleavage. (C) SDS/PAGE analysis of partially purified ppGBP from Pseudomonas putida CSV86 cells. (D) Gel filtration protein elution profile for thrombin-treated ppGBP (E. coli, open circles) and corresponding [14C]-glucose binding activities (open triangles). Inset: plot of log Mr (molecular mass) versus Velution/Vvoid (Ve/V0) for ppGBP (filled circles, thrombin-treated) purified using Sephacryl S-200 HR gel-filtration chromatography. The column was pre-equilibrated with binding buffer and calibrated with standard molecular mass marker proteins (open circles): alcohol dehydrogenase [150], bovine serum albumin [66] and carbonic anhydrase [29].

ppGBP was purified to apparent homogeneity from E. coli (Fig. 2B, yield 1.09 mg·g−1 cells, wet weight) and partially purified from CSV86 (Fig. 2C, yield 0.05 mg·g−1 cells, wet weight). ppGBP from E. coli was treated with thrombin to remove the N-terminal His6 tag, resulting in a molecular mass of ~ 43 kDa on SDS/PAGE (Fig. 2B). Gel-filtration chromatography of the thrombin-treated protein showed a single peak (Fig. 2D, native molecular mass ~ 47 kDa), which coincided with the [14C]-glucose binding activity peak.

In the text below, ppGBP purified from P. putida CSV86 and the recombinant protein purified from E. coli are referred as ppGBP (CSV86) and ppGBP (E. coli), respectively.

Substrate specificity of ppGBP (Escherichia coli)

A 100-fold molar excess concentration of unlabeled glucose displaced [14C]-glucose binding to ppGBP (E. coli) by ~ 95%. However, other unlabeled sugars (fructose, xylose, mannose, galactose, arabinose and ribose), sugar alcohols (mannitol and glycerol), organic acids (gluconate, succinate and pyruvate), or aromatic compounds (salicylate, benzyl alcohol and naphthalene) did not show any significant displacement of [14C]-glucose binding to ppGBP (E. coli, Fig. 3). This indicates that the ppGBP is specific for glucose.

Figure 3.

Substrate specificity of ppGBP (Escherichia coli). Inhibition of [14C]-glucose binding (%) to ppGBP (E. coli) in the presence of 100-fold molar excess of various unlabeled compounds was measured by a membrane filtration assay. The binding activity of GBP to [14C]-glucose in the absence of unlabeled compounds was considered as 100%.

Molecular modeling and identification of putative residues involved in glucose binding

The structure of GBP from Pseudomonas species is not available. To understand the nature of interaction between the residues at the binding pocket and glucose, the structure of ppGBP was modeled using the I-TASSER server (Fig. 4A). The best-fit model yielded a C score of −0.81 and a TM score of 0.61 ± 0.14. The overall structure of ppGBP was found to comprise two globular domains (α/β) of similar tertiary structures/topology. Each domain contains a central β-sheet of five β-strands flanked by α-helices. The two domains are connected by a three-stranded hinge region. The glucose-binding site was buried in the cleft between the two domains.

Figure 4.

Homology model of ppGBP and its structural comparison. (A) Cartoon representation of the homology model of ppGBP. The helices and β-strands are shown in cyan and pink, respectively. Glucose at the binding pocket is represented by a ball-and-stick model (yellow). The β-strands of the N-terminal domain are numbered to represent the structural fold similar to that observed in class II PBPs. (B) Superimposition of the Cα backbone of ppGBP (cyan) onto class I PBPs such as ecGBP (green) and tmGBP (red) to show overall structural divergence. (C) Superimposition of the Cα backbone of ppGBP (cyan) onto class II PBPs such as ttGBP (green), ecMBP (magenta) and tlMBP (yellow) to show the overall structural similarity. (D) Superimposition of the glucose-binding pocket of ttGBP (green) on that of ppGBP (yellow). Amino acid residues important for glucose binding are labeled in the corresponding color, and the bond distance (Å) is indicated by broken lines (magenta). The oxygen atoms in the glucose molecule are labeled as O1–O6 (red). (E) Comparison between the substrate-binding pockets of ppGBP and ecMBP to show the structural differences between mono- and disaccharide binding sites. Loop 2 (red, amino acids 375–382) and the α-helix (cyan, amino acids 190–206) of ppGBP that obstruct maltose binding (green) but favor glucose binding (magenta) are shown. The loop (blue arrow, amino acids 11–16) of ecMBP that is very close to the glucose (magenta) binding site was found to be wide enough for binding in ppGBP (red arrow, amino acids 35–41).

Among the available PBP structures from other organisms, the ppGBP model showed highest structural similarity to the glucose/galactose-binding protein of Thermus thermophilus (ttGBP), which belongs to class II, with a calculated rmsd of 0.64 Å (Table 1). Both these proteins shared 27% identity at the amino acid sequence level. Intriguingly, the ppGBP model showed little or no similarity to class I proteins such as the glucose/galactose-binding protein from E. coli (ecGBP), Salmonella typhimurium (stGBP) and Thermotoga maritima (tmGBP) (Fig. 4B). Interestingly, ppGBP was found to be more similar to the maltose-binding protein from Thermococcus litoralis (tlMBP) and E. coli (ecMBP, Fig. 4C).

Table 1. Structural comparison of ppGBP (421 amino acids) with other reported PBP structures.
PDB IDProteinClassNumber of amino acidsRMSD (Cα backbone, Å)Sequence identity (%)
2B3BGlu/Gal-binding protein of T. thermophilusII3920.6427.4
1EU8ATrehalose MBP of T. litoralisII4072.2717.3
3MBPMBP of E. coli K-12II3702.3817.9
3GBPGlu/Gal receptor of S. typhimuriumI3073.776.0
2HPHGlu/Gal-binding protein of E. coliI3163.956.4
2H3HGlucose-binding protein of T. maritimaI3134.436.1

Superimposition of the ppGBP model onto the crystal structure of ttGBP suggested involvement of W35, W36, E41, K92, K339 and H379 from ppGBP in the putative glucose-binding pocket (Fig. 4D). This pocket was found to be highly similar to that of ttGBP in terms of the amino acid residues (W8, W9, E13, A42, H66, D278, K312 and H348) that interact with the glucose molecule. In ppGBP, E41 and K339 interact with the hydroxyl groups present on the C4 and C6 carbon, respectively; W36 interacts with the C3 hydroxyl group and K92 and H379 interact with the hydroxyl groups present on the C1 and C2 carbon of glucose, respectively, via hydrogen bond interactions. The sugar molecule is further stabilized by stacking interaction by W35 and W36 (Fig. 4D). These residues form an annulus-like structure similar to that reported for ttGBP, wherein the hydroxyl groups of glucose molecule were placed equatorially, forming hydrogen bonds with these polar amino acids [23].

Further, in silico docking of d-glucose to the ppGBP model was performed using AutoDock. The best fit of the interaction between the ligand and the receptor is conventionally chosen on the basis of the top-ranked cluster according to binding affinity or the cluster with the maximum number of poses. Docking studies of d-glucose onto ppGBP resulted in 250 binding poses. Analysis revealed that the top-ranked cluster with the highest binding affinity also has the maximum number of bound poses. These poses are clustered at the interface of the two globular domains (α/β), involving W35, W36, E41, K92, K339 and H379 (data not shown). This observation is in accordance with the results obtained from superimposition analysis.

At the glucose-binding pocket of ttGBP, two loops (loop 1, amino acids 40–44; loop 2, amino acids 344–351) and an α-helix (amino acids 166–181) were found to occlude the wide groove that accommodates a disaccharide in tlMBP (see Fig. 7C of [23]). Compared to ttGBP, in the ppGBP model, loop 2 (amino acids 375–382) and the α-helix (amino acids 200–206) were found to be intact and located at a similar position, while loop 1 was distorted. Superimposition of ppGBP model onto the structure of ecMBP suggested that the loop of ecMBP (amino acids 11–16) that hampers glucose binding was wide enough in ppGBP (amino acids 35–41) to accommodate a glucose molecule (Fig. 4E).

Site-directed mutagenesis of ppGBP

Based on molecular docking and structural comparison studies, six single (W35A, W36A, E41A, K92A, K339A and H379A) and one double (E41A/K92A) alanine-substituted mutants of ppGBP were generated, sequence-confirmed, over-expressed and purified. These mutants were analyzed for their [14C]-glucose binding activity. All single mutants except H379A showed ~ 95% loss of glucose binding ability (Table 2). The double mutant (E41A/K92A) did not show any further loss of binding activity compared to either E41A or K92A. Further, the far-UV CD spectra of purified ppGBP (E. coli) and its mutants were found to be similar except for H379A (Fig. 5), suggesting no major differences in the secondary structural elements. Thus the decrease in activity may be attributed to involvement of these residues in binding of the ligand.

Table 2. [14C]-glucose binding activity of purified ppGBP (Escherichia coli) and its mutants. Binding activity was measured using 0.25 μm [14C]-glucose.
MutantSpecific activity (pmol bound per mg protein)Relative activity (%)
Wild-type378.8100
W35A00
W36A7.222.1
E41A236.8
K92A12.623.78
K339A00
H379A83.4324.6
E41A/K92A23.446.9
Figure 5.

CD spectroscopic analysis of ppGBP (Escherichia coli) and its mutants. The far-UV CD spectra of ppGBP (E. coli, solid line) and its mutants (W35A, W36A, E41A, K92A, K339A and H379A) are shown.

Discussion

PBPs play a crucial role in the high-affinity ABC transport systems. In addition to transferring solute molecules to its cognate transporter, PBPs also stimulate the ATPase activity of the nucleotide-binding subunits of ABC transporters through trans-membrane signaling [30]. This activation is mediated via conformational changes in the trans-membrane subunits that lead to an alternate access mechanism for the ABC transporter [31]. The major hindrance for study of the GBP from P. putida CSV86 was its partial purification and low yield. Therefore, periplasmic GBP from CSV86 was cloned into E. coli, over-expressed, purified and characterized. Further, the 3D structure was modeled and validated by site-directed mutagenesis and functional analysis.

The reported apparent affinities of PBP-dependent ABC transport systems (in vivo) towards their ligands are generally close (within an order of magnitude) to those of the cognate periplasmic binding proteins. This suggests that the binding protein is primarily responsible for the affinity of the transport system towards its substrate(s), and correlates with the fact that the binding specificity of PBP confers substrate specificity to the transport system [32]. The affinity constant of the whole cell of CSV86 was found to be in the micromolar range (0.81 μm), and is in accordance with previous reports for other PBP-dependent ABC transport systems in Gram-negative bacteria [32]. ppGBP was found to be one of the largest proteins among various reported sugar-binding proteins, and, like other members of PBP superfamily, is a functional monomeric protein. The presence of a His or Pro tag at the N-, C- or both termini did not affect the glucose binding activity of the protein. ppGBP (E. coli) shows high specificity towards glucose in the presence of 100-fold molar excesses of other sugars, organic acids and aromatic compounds. Together, these observations suggest that the topology of the ppGBP is unaffected by expression in E. coli.

Periplasmic binding proteins are ideal candidates for investigation of the evolutionary plasticity of proteins, as they share common features of spatial organization and patterns of ligand binding despite large sequence length variations and low sequence identity. Although the sequence identity among PBPs is not sufficient enough to draw any conclusion about their origin and history, the gene arrangement of PBP-dependent ABC uptake systems provides strong evidence for their common origin. In Pseudomonas, genes encoding various components of PBP-dependent ABC transporters are organized together in an operonic arrangement in the order: PBP–permease(s)–ATP-binding protein, which suggests that they evolved by duplication of an ancestral operon [18]. The organization of genes involved in glucose uptake in Paeruginosa was reported to be similar to that in T. thermophilus, except that the ATP-binding subunit is absent in the latter [23]. The constructed model of ppGBP showed highest structural similarity to the crystal structure of ttGBP, as well as to ecMBP and tlMBP. Intriguingly, the ppGBP model showed very little structural similarity to ecGBP and stGBP. This observation suggests that ppGBP and ttGBP may have originated from a common ancestor with minimum evolutionary divergence.

Superimposition of the ppGBP model onto the crystal structure of ttGBP revealed that the overall structural fold and the glucose-binding pocket in these two proteins are well conserved. The glucose-binding pocket in ppGBP comprises W35, W36, E41, K92, K339 and H379. Involvement of these residues in forming the glucose-binding pocket was also supported by a molecular docking study. Compared to the wild-type GBP, alanine substitution mutants showed a significant loss (~ 95%) in [14C]-glucose binding activity. The double mutant (E41A/K92A) did not show any additional loss of glucose binding activity. E41 and K92 are the key residues that anchor the glucose molecule at the pocket by series of hydrogen bond interactions, while W35, W36, K339 and H379 stabilize the anchored glucose molecule by stacking and hydrogen bond interactions. Mutational studies indicate that alanine substitution of these residues resulted in a decrease in the affinity of GBP, but not complete loss of binding ability to glucose.

The structure of ttGBP showed striking similarity with class II PBPs such as tlMBP [23], which typically bind to larger ligands such as di- and tri-saccharides or peptides (except class II PBPs that bind ions) [18]. The class I PBPs ecGBP and stGBP have a comparatively small binding pocket that limits the interaction with monosaccharides [33] and amino acids [34]. Structural comparison of ttGBP with tlMBP revealed the adaptations that inter-convert mono- and disaccharide binding sites [23]. Comparison of the structure of ecMBP with the putative model of ppGBP suggests the probable structural adaptations that accommodate glucose in ppGBP obstruct maltose binding in the class II fold [23]. Similar strategies appear to have been adopted for inter-conversion between class II ion- and sugar-binding proteins [35]. In addition, a loop in ecMBP was identified as very close to the superimposed glucose molecule in its substrate-binding pocket. However, the corresponding loop in ppGBP, as predicted in the model, is wide enough to accommodate the glucose molecule at the binding pocket.

In summary, the structure of ppGBP was predicted by homology modeling, and putative key residues in the glucose-binding pocket were confirmed by site-directed mutagenesis followed by functional analysis. X-ray crystal structure elucidation of ppGBP will provide detailed information on structural features to help establish its evolutionary relationship.

Experimental procedures

Micro-organisms and culture conditions

The bacterial cultures used in this study were P. putida CSV86 [5] and E. coli strains DH5α and BL21(DE3) (Novagen, Madison, WI, USA). Strain CSV86 was grown on 150 mL minimal salt medium [6] in 500 mL capacity baffled Erlenmeyer flasks at 30 °C on a rotary shaker (200 rpm) supplemented aseptically with glucose (0.25%). E. coli strain DH5α and BL21 (DE3) were grown in Luria–Bertani (LB) medium (10 g peptone, 5 g yeast extract and 10 g NaCl per liter of distilled water) at 37 °C [36]. Agar (1.5%) was used to prepare solid medium.

Whole-cell [14C]-glucose uptake assay

Uptake of [14C]-glucose (universally labeled, specific activity 140 mCi·mmol; BRIT, Mumbai, India) by CSV86 cells was studied using a modified membrane filtration assay [27]. Briefly, late-log phase cells, grown on glucose (0.25%), were harvested by centrifugation (7800 g for 10 min at 4 °C), washed twice with 25 mL of ice-cold sterile minimal salt medium for 10 min, and re-suspended in minimal salt medium (attenuance at 540 nm of 0.2). Pre-warmed cell suspension (1.0 mL, 30 °C for 10 min) was incubated with varying concentrations of [14C]-glucose at 30 °C for 30 s in a water bath, and rapidly filtered through a pre-moistened mixed cellulose esters filter (0.45 μm; Millipore Co., Bedford, MA, USA). The filters were washed twice with sterile minimal salt medium (1 mL), air-dried and mixed vigorously in scintillation cocktail (0.4% PPO [2,5-diphenyloxazole] and 0.025% POPOP [1,4-bis(5-phenyloxazolyl)benzene] in toluene) (SRL, Mumbai, India). The radioactivity was measured using a liquid scintillation counter Rackbeta LKB1209 (Pharmacia, Turku, Finland). The radioactivity of the scintillation cocktail alone as well as of the reaction mixture without cells was measured and subtracted. To obtain the dry weight, the cell suspension (1 mL, attenuance at 540 nm of 0.2) was centrifuged (20 000 g for 20 min), dried at 37 °C for 5 h, and its weight recorded. The binding is expressed as nmol glucose min−1·mg−1 dry cell weight. Affinity constants were determined by fitting the experimental data to theoretical Michaelis–Menten model using the sigmaplot 11.0 software package (Systat Software Inc., Chicago, IL, USA).

Cloning of the gene encoding periplasmic GBP

The gene encoding GBP of CSV86 was amplified by PCR using primers GBPF (5′-CCGGAATTCCATATGAATGCGATCACCCGTCTCG-3′; NdeI and EcoRI sites underlined) and GBPR (5′-CCGGAATTCCTACCTGGCCGCCTTGATCGC-3′; EcoRI site underlined) and CSV86 genomic DNA as the template. The PCR-amplified fragment (~ 1.2 kb) was first cloned at the EcoRI site of pBSKS(+) vector (Novagen), yielding pBSKS-GBP, followed by sub-cloning at the NdeI/EcoRI site of the pET28a expression vector (Novagen), yielding pET28-GBP. The gene without a stop codon was also cloned at NdeI/EcoRI sites (pET28-GBP*) to obtain GBP variants with as His tag at the N-terminus and a Pro tag at the C- terminus. Clones were confirmed by DNA sequencing at Xcelris Labs, Ahmedabad, India as well as by SciGenom, Cochin, India.

Over-expression and purification of ppGBP

A single colony of pET28-GBP-transformed E. coli BL21 (DE3) was grown overnight in LB medium (10 mL) supplemented with kanamycin (30 μg·mL−1) at 37 °C. The culture (1% v/v) was re-inoculated into LB medium (800 mL) containing kanamycin (30 μg·mL−1), grown at 37 °C to an attenuance at 600 nm of 1.0, and induced by addition of IPTG (100 μm) for 4 h. Cells were harvested (8000 g for 10 min), and re-suspended in binding buffer (10 mm Tris/HCl pH 7.5, 1 mm MgCl2). Cell-free lysate was prepared by sonication (three cycles per gram of cells, 1 s pulse, 1 s interval, cycle duration 30 s, output 15 W), followed by centrifugation (20 000 g for 20 min). Expression of the protein was assessed by SDS/PAGE (12%) as described previously [37].

N-terminally His-tagged ppGBP was purified using immobilized metal affinity chromatography. Cell-free lysate (10 mg·mL−1 of matrix) was loaded onto a pre-equilibrated Ni-NTA column (10 mL), followed by washing with five column volumes of binding buffer. The protein was eluted with a linear gradient of imidazole (0–200 mm, flow rate 30 mL·h−1, fraction size 1 mL) in binding buffer. GBP was eluted in the range 100–120 mm imidazole. The specific activity is expressed as pmol [14C]-glucose bound per mg protein. Fractions with higher specific activity were pooled and dialyzed against binding buffer for 4 h at 4 °C.

The N-terminal His tag was removed by treating the purified protein with thrombin agarose matrix according to the manufacturer's instructions (thrombin CleanCleave™ kit; Sigma Aldrich, St Louis, MO, USA) followed by purification of the GBP using Ni-NTA column chromatography. The native molecular mass of the purified ppGBP was determined by gel-filtration chromatography on a Sephacryl S-200 HR column (Sigma Aldrich, St Louis, MO, USA) calibrated with standard protein molecular mass markers. The molecular mass of GBP was determined from a plot of log Mr (molecular mass) versus Velution/Vvoid (Ve/V0).

Extraction and purification of ppGBP (CSV86)

Periplasmic proteins from CSV86 were isolated using the cold shock method as described previously [12, 38]. ppGBP (CSV86) was partially purified from total periplasmic protein fraction by gel-filtration chromatography using a Sephacryl S-200 HR column (Sigma Aldrich) as described previously [12]. Protein was estimated by the Bradford method [39], using bovine serum albumin as the standard.

[14C]-glucose binding assay and substrate specificity of ppGBP

[14C]-glucose binding activity was measured as described previously [12]. In brief, ppGBP (2.5 μg, from CSV86 or E. coli) was incubated with [14C]-glucose (0.5 μm) for 5 min at 30 °C. After incubation, the mixture was rapidly filtered through pre-moistened poly(vinylidene difluoride) membranes (0.45 μm; Pall Life Sciences Corp., New York, NY, USA). The radioactivity of the [14C]-glucose bound to GBP retained on the filter paper was measured using a liquid scintillation counter (Rackbeta LKB1209). A binding reaction mixture without ppGBP was used as the control, subtracted from the experimental values and expressed as pmol [14C]-glucose bound per mg protein.

The substrate specificity of ppGBP (E. coli) was assayed as described previously [12]. Briefly, a 100-fold excess (50 μm) of unlabeled sugars, organic acids or aromatic compounds was added individually to assay mixture (1 mL) containing purified ppGBP (E. coli, 2.5 μg) and [14C]-glucose (500 nm). Stock solutions of substrates were prepared by dissolving them in binding buffer, except aromatic compounds (dissolved in dimethyl sulfoxide). Control reaction mixtures contained [14C]-glucose, GBP in binding buffer, and the appropriate amount of dimethyl sulfoxide. The reaction mixture was incubated for 5 min at 30 °C and rapidly filtered through pre-moistened poly(vinylidene difluoride) membranes. Radioactivity was measured and expressed as percentage inhibition of [14C]-glucose binding.

Molecular modeling of ppGBP

The structure predictions for ppGBP were performed using the I-TASSER server, an online platform for automated protein structure prediction [40] without additional constraints or templates. The structural similarity between the ligand-binding sites of the generated model and the top-ranked template protein was assessed using the secondary structure matching algorithm [41] in Coot [42]. The d-glucose molecule was modeled in the glucose-binding pocket of ppGBP by superimposition of the glucose molecule from the glucose/galactose-binding protein of T. thermophilus (ttGBP, PDB ID 2B3B). All figures describing the structural features were drawn using the PyMOL (The PyMOL Molecular Graphics System, Version 1.5.0.4 Schrödinger, LLC)).

Molecular docking studies

The model obtained for ppGBP was further analyzed for its ability to bind glucose by performing molecular docking using AutoDock version 4.0 [43]. The 3D co-ordinates of the d-glucose were obtained from the structure of glucose-bound ttGBP (PDB ID 2B3B). Clusters were generated with an rmsd of 2 Å using the Lamarckian genetic algorithm. Genetic algorithm (GA) runs (250) were performed for each docking, with 1 500 000 energy evaluations per run. Docked poses were drawn using PyMOL (The PyMOL Molecular Graphics System, Version 1.5.0.4 Schrödinger, LLC).

Site-directed mutagenesis

PCR-based site-directed mutagenesis was performed to generate W35A, W36A, E41A, K92A, K339A, H379A and E41A/K92A mutants using pET28-GBP as the template and the primers listed in Table 3. Mutants were confirmed by DNA sequencing (Xcelris Labs and SciGenom, India). The mutant proteins were over-expressed in E. coli BL21 (DE3), purified using Ni-NTA matrix and assayed for [14C]-glucose binding activity as described above.

Table 3. Primers used to generate site-directed mutants of ppGBP from Pseudomonas putida CSV86. ‘f’ and ‘r’ indicate forward and reverse primers, respectively. Underlined letters indicate the bases substituted.
SubstitutionPrimer nameSequence (5′→3′)
W35AW35A fCGTTCTCCACGCGTGGACCTCC
W35A rGGAGGTCCACGCGTGGAGAACG
W36AW36A fGACGTTCTCCACTGGGCGACCTCC
W36A rGGAGGTCGCCCAGTGGAGAACGTC
E41AE41A fGACCTCCGGCGGCGCAGCCAAGG
E41A rCCTTGGCTGCGCCGCCGGAGGTC
K92AK92A fGCAGATCGCGGGCCCGGATATCC
K92A rGGATATCCGGGCCCGCGATCTGC
K339AK339A fGTTCAACCAGAACGCGGGCTCGC
K339A rGCGAGCCCGCGTTCTGGTTGAAC
H379AH379A fCCGAGCATGGCGGCCAACATGG
H379A rCCATGTTCGCCGCCATGCTCGG

Circular dichroism spectroscopy

Far-UV CD spectra of ppGBP (E. coli) and its mutants in the binding buffer were determined using a JASCO J-810 Peltier spectropolarimeter (Jasco, Gross-Umstadt, Germany) between 198 and 260 nm at 25 °C in a 0.1 cm path length quartz cuvette (volume 200 μL; Hellma GmBH & Co., KG, Müllheim, Germany) with the following parameters: response, 2 s; sensitivity, 100 millidegrees; speed, 100 nm·min−1; average of three scans. Raw data were processed by smoothing and subtraction of spectra obtained using binding buffer alone. Ellipticity values (millidegrees) were recorded as a function of wavelength.

Acknowledgements

We thank Rimi Chakrabarti (Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Mumbai, India) for her help with the molecular docking study. P.B. thanks the Department of Biotechnology, Government of India, for a Ramalingaswami fellowship. P.S.P. thanks the Department of Science and Technology, Government of India, for providing a research grant. A.M. acknowledges a senior research fellowship from Council of Scientific and Industrial Research, Government of India.

Ancillary