Identification of the essential histidine residue for high-affinity binding of AlbA protein to albicidin antibiotics

Xanthomonas albilineans, the bacterial pathogen of sugarcane leaf scald disease, produces a family of potent antibiotics and phytotoxins known as albicidins. The major antibacterial compound produced in rich culture media has been partially characterized, with a molecular mass of 842 Da and a structure containing three to four aromatic rings (Birch & Patil, 1985a). Several genes encoding the enzymes involved in albicidin biosynthesis have been cloned and characterized, including a methyltransferase (Huang et al., 2000a), a phosphopantetheinyl transferase (Huang et al., 2000b) and a polyketide-peptide synthetase (Huang et al., 2001).

Albicidins are bactericidal at nanomolar concentrations against a range of Gram-positive and Gram-negative bacteria (Birch & Patil, 1985a). The exquisite sensitivity is due to illicit uptake of albicidins through a nucleoside uptake channel (Birch et al., 1990). Inhibition of DNA replication in bacteria and plastids is the primary mechanism of action (Birch & Patil, 1985b, 1987). Several albicidin-resistance mechanisms have been identified, including albicidin-binding proteins (Walker et al., 1988; Basnayake & Birch, 1995) and an esterase which hydrolyses and detoxifies albicidin (Zhang & Birch, 1997). Expression of the albicidin esterase, either in the pathogen or in transgenic sugarcane, prevents development of chlorotic disease symptoms and systemic invasion in inoculated plants (Zhang & Birch, 1997; Zhang et al., 1999).

The albA gene from Klebsiella oxytoca encodes a protein that binds albicidin with strong affinity and selectivity. This is of interest as a potential phytotoxin- and disease-resistance mechanism, and as a novel ligandmatrix combination for biotechnological applications (Walker et al., 1988; Zhang et al., 1998b). Kinetic and stoichiometric analyses indicate the presence of a single high-affinity binding site with a dissociation constant of 6·4x10^-8 M in the AlbA protein (Zhang et al., 1998b). Albicidin-binding capacity decreases by 30 % when pH is shifted from 6 to 4, which is the range for side-chain ionization of histidine residues.

To assess the role of histidine residues in the AlbAalbicidin interaction, we first cloned and resequenced the albA gene from K. oxytoca ATCC 13182^T, and we report here the revised albA DNA and peptide sequences. We show, by stoichiometric analysis and site-directed mutagenesis, that His₁₂₅ is uniquely involved in albicidin binding.

Bacterial strains and cultivation.
The bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli DH5α, used as an albicidin biosensor and as the host strain for DNA cloning and expression of the GSTAlbA fusion protein, was grown at 37 °C in LuriaBertani medium. X. albilineans XA13, used for albicidin production, was grown at 28 °C in SP medium (Birch & Patil, 1985b). Broth cultures were aerated by shaking at 200 r.p.m. on an orbital shaker.

Table 1. Bacterial strains and plasmids used in this study

DNA manipulation.
Routine recombinant DNA and protein techniques were performed as described by Sambrook et al. (1989). The albA gene from K. oxytoca ATCC 13182^T was PCR-amplified by using the primers albA-f (5'-CTC GGA TCC ATG AAA ATG TAC GAT CGC TGG) and albA-r (5'-GAT GAA TTC TTA TTC GGC GGC AGG CCC C). After electrophoresis, the DNA band was excised, purified and cloned into vector pBluescript II SK(+). Albicidin-resistant colonies were selected on LB agar supplemented with albicidin. Sequencing was performed on both strands by using the ABI Prism dRhodamine Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems).

Protein sequence analysis and prediction.
Protein sequence analysis and secondary structure prediction were performed using the PROTEAN module in DNASTAR (Plasterer, 2000).

Site-directed mutagenesis.
Site-directed mutagenesis of albA in pGST-AlbA was performed by using a QuikChange Site-Directed Mutagenesis Kit according to the manufacturer's recommendations (Stratagene). The four histidines of AlbA were converted to glycine or other amino acids using the primers listed in Table 2. Each mutated clone was sequenced to verify the desired mutation and assayed for albicidin-binding activity.

Table 2. PCR primers for site-directed mutagenesis Only forward primer sequences are listed. Each underlined codon indicates the corresponding substitution of the histidine residue as indicated in the oligonucleotide designation.

Purification of AlbA variants and albicidin.
AlbA and its variants were expressed separately as glutathione S-transferase (GST) fusion proteins, from which the recombinant AlbA and the derivatives were released by incubation with thrombin. After confirmation of their purity using SDS-PAGE, the proteins were freeze-dried and stored at -20 °C (Zhang et al., 1998b). The protein concentration was determined by means of the DC protein assay (Bio-Rad) with BSA as the standard. Albicidins were produced in culture by X. albilineans as described previously (Birch & Patil, 1985b; Zhang et al., 1998a). The major albicidin peak after C18 HPLC was crystallized before use in kinetic and stoichiometry studies. For other experiments, the mixture of albicidins obtained after HW-40(s) chromatography was used.

AlbAalbicidin binding assay.
AlbA variants (0·77 µM) in TMM buffer (10 mM Tris/HCl, pH 7·0; 10 mM MgCl₂; 2 mM 2-mercaptoethanol) were mixed with albicidin at concentrations ranging from 0·12 to 1·17 µM, then incubated at 25 °C for 5 min before a quantitative assay of free albicidin as described previously (Zhang et al., 1998b). Briefly, 100 µl of a fresh E. coli DH5α culture, which was adjusted to an OD₆₀₀ value of 1·8, were added to 5 ml of 1 % (w/v) agarose at 50 °C. The mixture was quickly added to a plate containing 20 ml of LB agar. After solidification, wells 4 mm in diameter were punched in the agar, 20 µl of the test solution was added to each well and the plate was incubated at 37 °C overnight. Free albicidin in the test solution was calculated using the formula albicidin (ng ml^-1)=4·576e^(0·315w), where w is the width (in mm) of the zone of growth inhibition surrounding the well. This formula is derived from the linear portion of an albicidin dose-response plot under the same assay conditions (Zhang et al., 1998a). Bound albicidin was calculated by subtracting the amount of free albicidin from the total albicidin added to the reaction mixture.

Inactivation of AlbA by diethyl pyrocarbonate (DEPC).
DEPC was stored desiccated at 4 °C to minimize decomposition by hydrolysis. The stock solution of DEPC was freshly prepared by diluting DEPC with cold absolute ethanol (1 : 19, v/v). Buffered DEPC solutions were prepared immediately before use in 0·1 M sodium phosphate buffer (pH 6·5) and kept on ice. The reaction mixture was prepared by adding 0·1 ml of buffered DEPC solution to 0·9 ml of AlbA in the same buffer to obtain the final concentrations specified below. The mixtures were incubated at 25 °C, and aliquots were taken at the times specified for photometry and assays of albicidin-binding activity.

UV differential spectra and stoichiometry of histidine modification.
UV absorbance spectra were recorded for purified AlbA (3 mg ml^-1), before and after treatment with 13·8 mM DEPC for 10 min at 25 °C. For kinetics and stoichiometry, purified AlbA (0·2 mg ml^-1) was treated with DEPC concentrations ranging from 0·05 to 1·5 mM. The extent of histidine residue modification was calculated from the absorbance change at 242 nm, the absorption coefficient (3200 M^-1 cm^-1) of N-carbethoxyhistidine (Ovadi et al., 1967; Miles, 1977) and the molar concentration of AlbA (M_r=25 852 Da). Immediately after the absorbance measurement, albicidin was added (100 ng ml^-1) and the residual albicidin-binding activity was determined by bioassay.

Substrate protection against AlbA modification by DEPC.
To obtain the specified albicidin to AlbA molar ratios, 40 µl of an albicidin stock in 50 % (v/v) methanol was added to 860 µl of AlbA solution (0·1 mg ml^-1), and incubated at 25 °C for 5 min before the addition of 100 µl of DEPC solution (10 mM). The final concentration of methanol in the reaction mixture was 2 %, which did not affect albicidin binding or DEPC modification. The stoichiometry of N-carbethoxyhistidine residue formation was determined as described above.

Hydroxylamine treatment of inactivated enzyme.
After 10 min incubation of AlbA at 25 °C with 0·25 mM DEPC, hydroxylamine was added to a final concentration of 100 mM. The mixtures were then incubated at 4 °C for 40 min. Absorbance at 242 nm and albicidin-binding activity were determined before and after hydroxylamine treatment.

Cloning of albA from K. oxytoca ATCC 13182^T
While preparing for site-directed mutagenesis of albA in pGST-AlbA, we found that the sequence of albA differed from that reported previously. albA in pGST-AlbA was PCR-amplified from pJP1076, which was used in the original sequencing of the albA gene from K. oxytoca JMP 4505 (Walker et al., 1988; Zhang et al., 1998b). Therefore, we cloned albA from K. oxytoca ATCC 13182^T and confirmed the sequence using clones from five separate PCR amplifications. The sequence of albA from strain ATCC 13182^T was identical to that from pGST-AlbA. It showed 96 % identity at the nucleotide level and 69 % identity at the peptide level with the previously reported albA sequence (Walker et al., 1988). The differences were in GC-rich regions prone to error in manual sequencing, and we interpret these differences as errors in the previously reported sequence of albA. The corrected albA sequence has been deposited in GenBank under accession number AF525464.

The predicted AlbA peptide has 221 aa residues, an isoelectric point at 4·9 and four histidine residues (His₇₈, His₁₂₅, His₁₄₁ and His₁₈₉). It shows a low level of sequence similarity to the DNA-binding proteins encoded by ntrC from Acidithiobacillus ferrooxidans (Salazar et al., 2001) and nifA from several bacterial species (Monteiro et al., 1999; Gu et al., 2000; Souza et al., 2000). There is no significant similarity beyond that previously noted at the N terminus between AlbA and AlbB from Alcaligenes denitrificans, which also encodes an albicidin-binding protein (Basnayake & Birch, 1995).

Effect of DEPC on albicidin-binding activity of AlbA
We used DEPC, a histidine-specific reagent, to probe the role of the histidine residues of AlbA in binding to albicidin, as a previous study had suggested that histidine residues could be involved in the proteinligand interaction (Zhang et al., 1998b). After incubation of AlbA with 0·05 mM DEPC for 13 min, about 55 % of the albicidin-binding activity was lost. In 1·5 mM DEPC, about 95 % of the binding activity was lost within 5 min (Fig. 1a).

(23K): Fig. 1. Probing the function of the histidine residues by DEPC treatment. (a) Time-course of reduction in albicidin-binding capacity of AlbA, after addition of DEPC at 0·05 mM () or 1·5 mM (). (b) UV absorption spectra of AlbA before and after DEPC treatment. Spectra from top down: AlbA after DEPC treatment (post); AlbA before DEPC treatment (pre); subtracted UV spectrum (post minus pre). (c) Relationship between the number of DEPC-modified histidine residues per AlbA molecule and albicidin-binding capacity. (d) Substrate protection against DEPC modification of histidine in AlbA.

Spectral changes accompanying inactivation
Inactivation of AlbA by DEPC was followed by monitoring absorbance from 220 to 300 nm (Fig. 1b). The differential spectrum shows a substantial increase in absorbance at 240244 nm, which is indicative of histidine modification (Ovadi et al., 1967; Miles, 1977; Rua et al., 1995). In rare cases, DEPC can also modify tyrosine residues, which decreases absorbance at 278 nm. The differential spectra of DEPC-treated and untreated AlbA are almost identical at 278 nm, indicating that inactivation is due to modification of histidine residues. We further tested whether hydroxylamine, a reagent widely used to deacylate N-carbethoxyhistidine, could reverse DEPC modification of the histidine residue. About 65 % of the lost binding activity of DEPC-treated AlbA was restored by incubation with 100 mM hydroxylamine at pH 7·0 for 40 min. Absorbance at 242 nm was correspondingly decreased (data not shown).

Substrate protection against histidine residue modification by DEPC
The AlbAalbicidin complex is very stable, with albicidin released only under protein denaturing conditions (Zhang et al., 1998b). When albicidin was mixed with AlbA at different molar ratios before addition of DEPC, the percentage of histidine residues alkylated was inversely proportional to the albicidin to AlbA molar ratio (Fig. 1d). Thus, bound albicidin blocked the reaction between DEPC and the active His residue.

Stoichiometry of inactivation
AlbA was treated with DEPC concentrations ranging from 0·05 to 1·5 mM for 5 min. The reaction was stopped by adding albicidin. The relationship between the extent of histidine residue modification and the residue-binding activity of AlbA was calculated from the change in the OD₂₄₀ value and bioassay of binding activity of DEPC-treated AlbA. There are four histidine residues in AlbA; the stoichiometry data show that only one histidine residue has been modified by DEPC (Fig. 1c). There was no further increase in the number of histidine residues alkylated at higher DEPC concentrations or extended incubation times. Albicidin-binding activity decreased linearly with modification of the histidine residue (Fig. 1c).

Prediction of the active histidine residue in AlbA
We assumed that the three histidine residues in AlbA that were not alkylated by DEPC might be located in regions whose structural features prevent access by DEPC. Secondary structure predictions for AlbA from the GARNIERROBSON program indicate that His₇₈, His₁₄₁ and His₁₈₉ are located in α-helix regions, whereas His₁₂₅ lies at the immediate front of a turn region (Fig. 2). The EMINI program indicates that His₁₂₅ is most likely exposed, with a surface probability (SP) index of 2·4. The other histidine residues have SP indices of less than 1·4 (Fig. 2). Thus, His₁₂₅ is likely to be the active histidine residue of AlbA that interacts with DEPC and albicidin.

(24K): Fig. 2. Predicted secondary structures surrounding the four histidine residues of AlbA and the corresponding histidine residue surface probability (SP) indices from the DNASTAR PROTEAN module. The position of each histidine residue is indicated. The bottom line of amino acid sequence shows the predicted change in the secondary structure after deletion of His125. Open squares represent amino acid residues in an α-helix region; open circles represent amino acid residues in a turn region. Numbers inside the squares or circles indicate the position of the residue in the peptide chain of AlbA. The SP indices of corresponding histidine residues are listed; NA, not applicable.

Identification of the active histidine residue by site-directed mutagenesis and deletion analysis
Replacement of His₁₂₅ with glycine decreased the albicidin-binding capacity of AlbA by 28·4 %, whereas replacement of His₇₈, His₁₄₁ or His₁₈₉ with glycine had no effect. Substitution of His₁₂₅ with tyrosine and arginine did not affect albicidin binding, but substitution at the same position with alanine or leucine resulted in a 3135 % loss of activity. Deletion of His₁₂₅ from AlbA completely eliminated albicidin-binding activity (Fig. 3). Deletion of His₁₂₅ may minimize the size of the turn in the same region, as predicted by secondary structure analysis (Fig. 2).

(17K): Fig. 3. Effect of DEPC modification and site-directed mutagenesis of His125 on albicidin-binding activity. His125 of AlbA was deleted or mutated to another amino acid as shown. The resulting proteins were assayed for binding activity against three concentrations of albicidin (100, 200 and 1000 ng ml-1). Each concentration was treated as one replicate, and the results are expressed as the mean±SE from the three replicates.

Previous studies have shown that the AlbA protein contains a single high-affinity binding site for albicidin, with a dissociation constant of 6·4x10^-8 M, and that its binding capacity declines with pH values below the pK_a of histidine (Zhang et al., 1998b). Five lines of new evidence show that His₁₂₅ is involved in the binding of albicidin by AlbA. (i) Treatment of AlbA with DEPC, under conditions that result in the selective N-carbethoxylation of histidine, caused loss of albicidin-binding capacity (Fig. 1a). (ii) The effect was substantially reversed by treatment with hydroxylamine, which removes N-carbethoxy groups from histidine (Miles, 1977). (iii) Photometric analysis showed that a single histidine residue was modified by DEPC, and albicidin-binding capacity declined in proportion with the extent of modification of this residue (Fig. 1c). (iv) Pretreatment with albicidin protected against modification of AlbA by DEPC, in proportion to the albicidin to AlbA molar ratio in the reaction (Fig. 1d). (v) Site-directed mutagenesis of His₁₂₅, but not of the other His residues in AlbA, reduced or eliminated albicidin binding.

Histidine residues are involved in substrate binding by many enzymes. The active histidine residue may act as a proton acceptor or donor in proteinligand interactions, chargerelay interactions between amino acids at the active site, conformational changes associated with substrate binding or oligomerization of protein chains (Miller & Ball, 2001; Wiebe et al., 2001).

In AlbA, the deletion of His₁₂₅ abolished albicidin-binding activity. Replacement of His₁₂₅ with non-polar amino acids (glycine, alanine or leucine) caused about 30 % decrease in albicidin-binding activity (Fig. 3). Lowering the pH to 4, which results in protonation of histidine (pK_a=6·0), also decreased albicidin-binding activity by about 30 % (Zhang et al., 1998b). Replacement of His₁₂₅ with amino acids with similar properties (arginine or tyrosine) had little effect on the albicidin-binding capacity of AlbA. These results could indicate the importance of a proton donor at residue 125 for the conformational change of AlbA to form the albicidin-binding region, or for an electrostatic interaction between AlbA and albicidin.

Using the GARNIERROBSON program in the PROTEAN module, deletion of His₁₂₅ was predicted to greatly shorten the turn region between the adjoining α-helices (Fig. 2). DEPC treatment converting a single histidine residue to N-carbethoxyhistidine resulted in an almost-complete loss of albicidin-binding activity (Fig. 1c). The bulky side chain of N-carbethoxyhistidine (Fig. 3) could prevent proper entry of albicidin to the substrate-binding cleft, or hinder a conformational change of AlbA required for albicidin binding.

The full chemical structure of albicidin has not yet been elucidated. NMR and MS analyses on the major antibiotic component produced in rich culture media indicates a structure with 38 carbon atoms including three to four aromatic rings, a probable O-methyl tyrosine and at least one COOH group (Birch & Patil, 1985a; Huang et al., 2001). Biosynthesis involves a multifunctional polyketide-peptide synthetase (Huang et al., 2001), and albicidin is cleaved by the AlbD esterase (Zhang & Birch, 1997). Of the four amino acids in the predicted turn region following His₁₂₅ of AlbA, three have polar side chains (Tyr, Asp and Arg) that could interact with the COOH group or the OH group of O-methyl tyrosine of albicidin through H-bonding.

Computer-based structural analysis is well established for the prediction of active residues in proteins with conserved structural features and sequence motifs. Our results show that computer-based predictions of protein secondary structure and amino acid surface probability can be useful for indicating the residues that are likely to interact with a ligand, even in proteins without evident homology to known structural domains.

This project was partially supported by the Australian Research Council, and by the Agency for Science, Technology and Research, Singapore.