Molecular and Conformational Basis of a Specific and High-Affinity Interaction between AlbA and Albicidin Phytotoxin

Abstract

The albA gene of Klebsiella oxytoca encodes a protein of 221 amino acids that binds the albicidin phytotoxin with a high affinity (dissociation constant = 6.4 × 10⁻⁸ M). For this study, circular dichroism (CD) spectrometry and an alanine scanning mutagenesis approach were used in combination to investigate the molecular and conformational mechanisms of this high-affinity protein-ligand interaction. CD analysis revealed that AlbA contains a high-affinity binding site, and binding of the albicidin ligand to AlbA in a low-ionic-strength environment induced significant conformational changes. The ligand-dependent conformational changes of AlbA were specific and rapid and reached a stable plateau within seconds after the addition of the antibiotic. However, such conformational changes were not detected when AlbA and albicidin were mixed in the high-ionic-strength buffer that is required for maximal binding activity. Based on the conceptual model of protein-ligand interaction, we propose that a threshold ion strength allows AlbA to complete its conformational rearrangement and resume its original stable structure for accommodation of the bound albicidin. Mutagenesis analysis showed that the replacement of Lys¹⁰⁶, Trp¹¹⁰, Tyr¹¹³, Leu¹¹⁴, Tyr¹²⁶, Pro¹³⁴, and Trp¹⁶² with alanine did not change the overall conformational structure of AlbA but decreased the albicidin binding activity about 30 to 60%. We conclude that these residues, together with the previously identified essential residue His¹²⁵, constitute a high-affinity binding pocket for the ligand albicidin. The results also suggest that hydrophobic and electrostatic potentials of these key amino acid residues may play important roles in the AlbA-albicidin interaction.

Albicidin is a potent antibiotic and phytotoxin produced by the plant bacterial pathogen Xanthomonas albilineans. The toxin, which blocks DNA replication in bacteria as well as in plastids of the sugarcane plant (2, 3), plays an important role in systemic invasion and symptom development in sugarcane leaf scald disease (4, 20, 21). It is bactericidal at concentrations between 1 and 100 ng ml⁻¹ for a range of gram-positive and gram-negative bacteria. The compound, which has been partially characterized, has a molecular mass of 842 Da with a structure containing several aromatic rings (2). It is likely a new member of the rapidly expanding polyketide antibiotic family; genetic data show that the xabB gene, which encodes a polyketide-peptide synthetase, is essential for the production of albicidin by X. albilineans (8). The methyl transferase and phosphopantetheinyl transferase encoded by xabA and xabC, respectively, are also involved in albicidin biosynthesis (9, 10).

Two classes of albicidin resistance mechanisms have been found in the natural environment. The albA and albB genes, identified in Klebsiella oxytoca and Alcaligenes denitrificans, respectively, encode proteins which inactivate albicidin by high-affinity binding (1, 4, 17, 18, 22). The AlbD protein from Pantoea dispersa is a potent albicidin hydrolase belonging to the esterase family (20). There is no significant homology between these three albicidin-resistant proteins. It has yet to be determined whether they share a conserved mechanism of substrate binding. Since albicidin is a potent antibiotic and phytotoxin, these resistance mechanisms may have significant potentials in biotechnology. It was shown previously that the expression of albicidin hydrolase, either in the pathogen or in transgenic sugarcane plants, blocked the development of chlorotic disease symptoms and the systemic invasion of inoculated plants (20, 21).

AlbA is a small protein of 221 amino acids (aa) with an isoelectric point at 4.9 (18). Its normal function in K. oxytoca remains unknown. Notably, its peptide sequence shows about 25% identity to the DNA binding domains of NifA and NtrC, two transcription factors from Klebsiella pneumoniae (5). NifA functions as the Nif-specific activator of transcription, while NtrC is the bifunctional regulatory protein involved in nitrogen control. Kinetic and stoichiometric analyses indicated that AlbA contains a single high-affinity binding site with a dissociation constant of 6.4 × 10⁻⁸ (22). The maximal binding ability of AlbA was maintained from pH 6 to 9, with a sharp decrease from pH 5 to 4, which is in the pK_a range of a histidine side chain. Subsequent biochemical and molecular analyses identified that His¹²⁵ is the essential residue for high-affinity binding of the AlbA protein to albicidin (18). The mutation of His¹²⁵ to either alanine or leucine resulted in an approximately 32% loss of binding activity, and the deletion of His¹²⁵ totally abolished the binding activity. Hence, His¹²⁵ may play a key role either in the electrostatic interaction between AlbA and albicidin or in the maintenance of the local conformational structure at the albicidin binding site (18). However, the other amino acid residues involved in albicidin binding have not yet been identified.

An investigation of the key residues and the binding mechanism of AlbA is important to allow for more efficient use of this potent ligand binding protein in biotechnological applications. The potential exists to pyramid genes for different mechanisms in transgenic plants to protect plastid DNA replication from inhibition by albicidins (22). The present study reports circular dichroism (CD) spectroscopic and mutagenesis analyses of the AlbA-albicidin protein-ligand interaction with an aim to elucidate the molecular and conformational bases that determine the high-affinity interaction between AlbA and albicidin. We investigated whether a conformational change is important for albicidin binding and determined the speed and specificity of the protein-ligand interaction. Furthermore, by alanine scanning mutagenesis, we identified the key amino acid residues that are involved in the AlbA-albicidin interaction and conducted CD spectroscopic analyses of AlbA and its variants in order to understand the roles of these key residues in the protein-ligand interaction.

MATERIALS AND METHODS

Bacteria and cultivation.

The bacterial strains and plasmids used for this study are listed in Table 1. Escherichia coli DH5α, used as the host strain for DNA cloning and as the indicator strain for the bioassay of albicidin, was grown at 37°C in Luria-Bertani medium. K. oxytoca ATCC 13182 and X. albilineans strain XA13 were grown at 28°C in sucrose and peptone medium (2).

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Genotype or phenotypea	Reference or source
Strains
E. coli DH5α	recA-1 endA-1 hsdR-17 supE-4 gyrA-96 relA-1 Δ(lacZYA-argF) U169 (φ80dlacZΔM15), Albs	Laboratory collection
X. albilineans XA13	Wild-type albicidin producer from sugarcane (Australia), Apr	Laboratory collection
K. oxytoca ATCC 13182	Type strain of K. oxytoca	American Type Culture Collection
Plasmids
pGEX-2T	Cloning vector to generate GST gene fusion, Apr	Pharmacia
pGST-AlbA	albA gene fused in frame to the GST gene in pGEX-2T, Albr Apr	22
pGST-K106A	Lys106 of AlbA in pGST-AlbA was replaced with alanine	This study
pGST-W110A	Trp110 of AlbA in pGST-AlbA was replaced with alanine	This study
pGST-Y113A	Tyr113 of AlbA in pGST-AlbA was replaced with alanine	This study
pGST-L114A	Leu114 of AlbA in pGST-AlbA was replaced with alanine	This study
pGST-H125A	His125 of AlbA in pGST-AlbA was replaced with alanine	18
pGST-Y126A	Tyr126 of AlbA in pGST-AlbA was replaced with alanine	This study
pGST-P134A	Pro134 of AlbA in pGST-AlbA was replaced with alanine	This study
pGST-W162A	Trp162 of AlbA in pGST-AlbA was replaced with alanine	This study
pGST-K106AH125A	Lys106 and His125 of AlbA in pGST-AlbA were replaced with alanines	This study
pGST-W110AH125A	Trp110 and His125 of AlbA in pGST-AlbA were replaced with alanines	This study
pGST-Y113AH125A	Tyr113 and His125 of AlbA in pGST-AlbA were replaced with alanines	This study
pGST-L114AH125A	Leu114 and His125 of AlbA in pGST-AlbA were replaced with alanines	This study
pGST-H125AY126A	His125 and Tyr126 of AlbA in pGST-AlbA were replaced with alanines	This study
pGST-H125AP134A	His125 and Pro134 of AlbA in pGST-AlbA were replaced with alanines	This study
pGST-H125AW162A	His125 and Trp162 of AlbA in pGST-AlbA were replaced with alanines	This study

^aAp, ampicillin; Alb, albicidin.

Preparation of albicidin.

The purification of albicidins from X. albilineans XA13 has been described previously (2, 20). The albicidin product obtained after HW-40(s) chromatography was used for this study. Albicidin was quantified by use of the formula albicidin (ng ml⁻¹) = 4.576 e^0.315W, where W is the width, in millimeters, of the zone of growth inhibition surrounding each well. This formula was derived from a dose-response plot of E. coli DH5α to albicidin purified as a single peak by C₁₈ high-performance liquid chromatography following HW-40(s) chromatography, with a correlation coefficient (R²) of 0.993 (23).

Alanine scanning mutagenesis and generation of double-site mutants.

Mutagenesis was performed by use of a QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's recommendations, using Pfu Turbo DNA polymerase and complementary single-stranded pairs of mutagenic primers (12, 15). The primers were designed to separately replace each of the amino acid residues from Glu⁸¹ to Gln¹⁷⁸ in wild-type AlbA with alanine, using the plasmid pGST-AlbA as the template. Eighty-three pairs of oligonucleotide primers were used to introduce single amino acid point mutations (data not shown). PCRs were performed with 10 ng of pGST-AlbA in 50-μl reaction mixtures containing 20 mM Tris-HCl, a 200 μM concentration of each deoxynucleoside triphosphate, 1.5 mM MgCl₂, and 2.5 U of Pfu Turbo DNA polymerase. The following PCR conditions were used: 16 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 1 min, and extension at 68°C for 12 min. The PCR products were incubated with 10 U of the DpnI restriction enzyme at 37°C for 1 h, and then 1 μl of each DpnI-treated DNA was transferred into supercompetent cells. Three independent clones of each mutation were separately assayed for albicidin binding activity, and those with a changed binding activity were sequenced to verify the desired mutation. For double-site mutagenesis, the plasmid pGST-H125A (18) was used as the template, and the same approach was used.

Protein purification.

E. coli DH5α cultures harboring pGST-AlbA and variant proteins were grown at 37°C overnight. The starter cultures were inoculated separately at a 1:100 ratio into Luria-Bertani liquid medium containing ampicillin (100 μg ml⁻¹). The cultures were grown at 30°C with shaking until they reached an optical density at 600 nm of 0.7. Isopropyl β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 0.1 mM, and the cultures were allowed to grow overnight. Bacterial cells were collected, and the total soluble proteins were released by ultrasonification. The recombinant AlbA protein and its variants, which contained four extra amino acids (Gly, Ser, Met, and Lys) at the N terminus compared to the native AlbA protein (22), were released from the fused glutathione S-transferase (GST) protein by incubation with thrombin dissolved in 1× phosphate-buffered saline (PBS; 140 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, pH 7.3) according to previously described procedures (18, 22). The collected AlbA derivatives were subjected to a second round of glutathione-Sepharose 4B affinity chromatography to minimize impurities. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis confirmed that the purity of AlbA and its variants was about 99.5% after the purification procedures described above (22). The protein solutions were concentrated by ultrafiltration to about 3 mM, divided into aliquots, and stored at −80°C prior to use. The enzyme concentrations were determined by UV spectrophotometry at 280 nm based on their corresponding molar extinction coefficients (for example, ɛ_AlbA = 52,160 M⁻¹ cm⁻¹). These purified recombinant AlbA and variant proteins were used for all binding activity and CD analysis experiments.

AlbA-albicidin binding assay.

AlbA and its variants (0.5 μM), diluted in water or buffer as indicated, were mixed with albicidin at three concentrations (0.1, 0.5, and 1.0 μM) and then incubated at room temperature for 5 min before quantitative determinations of free albicidin were performed as described previously (18, 22, 23). The amount of albicidin bound by AlbA and its variants was calculated by subtracting the amount of free albicidin from the total amount of albicidin added to the reaction mixture.

CD spectrometric analysis.

Far-UV CD spectra were recorded on a JASCO J-810 spectropolarimeter in the wavelength range between 185 and 260 nm under a constant nitrogen flush, with 1-mm-path-length quartz cells. Protein samples were diluted to 25 μM with buffer or water, as indicated. The spectra were derived from averages of three scans recorded at 50 nm min⁻¹ along with a 1-s response time. Each spectrum was blank corrected, smoothed, and analyzed with the software package provided by JASCO. The fraction of secondary structure was calculated by the method described by Yang et al. (19). The time course of the AlbA-albicidin binding interaction was monitored by use of a JASCO ATS-429S automatic titration accessory. Measurements were recorded at 222 nm with a 1-nm bandwidth and a response time of 64 ms. The data represent the averages of five injections. The instrument was routinely calibrated with ammonium d-(+)-10-camphorsulfonate as specified by the manufacturer.

RESULTS

AlbA is an α-helix-rich protein with a stable conformational structure.

The recombinant AlbA protein, which contains four extra amino acids at the N terminus (22), was purified to near homogeneity for CD and activity analyses (see Materials and Methods). Figure 1A shows the far-UV CD spectrum of AlbA in 20 mM sodium phosphate buffer (pH 7.0). The spectrum of AlbA showed that there are two negative maxima, at 208 and 222 nm, and a strong positive band at 193 nm, indicating a large α-helix contribution. The secondary structure contents of AlbA were estimated to be 53.4% α-helix, 21% β-sheet, and 25.6% unordered structure, calculated according to the method developed by Yang et al. (19). Further titration experiments showed that there was no significant change in the CD spectrum of AlbA in sodium phosphate buffer at different ion concentrations from 4 to 100 mM. We also tested the conformational stability of AlbA in other buffers and solutions, including 1× PBS buffer (pH 7.4), water, and 10 to 100 mM NaCl solutions. No significant change was noticed, regardless of the ion concentration and species, suggesting that under these conditions, ions do not affect the conformation of the AlbA protein in the absence of the albicidin ligand, and that AlbA has a stable protein conformation.

FIG. 1.

Far-UV CD spectrum of AlbA and albicidin-induced conformational changes in AlbA. (A) Far-UV CD spectrum of AlbA in 20 mM sodium phosphate. (B) Albicidin-induced conformational changes of AlbA in water or 4 mM phosphate buffer (almost identical CD spectra); the molar ratio of albicidin to AlbA was 0 (○), 0.36 (□), or 1.08 (▵).

Using the same conditions, we also determined the CD spectrum of albicidin. The ligand displayed a flat CD spectrum with molar ellipticity close to zero in the far-UV range from 200 to 250 nm. Increases in the ionic strength of the sodium phosphate buffer from 4 to 100 mM did not affect the molar ellipticity in the far-UV range (data not shown). The absence of CD absorbency interference from albicidin may therefore facilitate investigations of the molecular and conformational bases of the high-affinity AlbA-ligand interaction.

An AlbA-albicidin interaction accompanies a drastic conformational change in the AlbA protein.

Figure 1B shows that the interaction of the albicidin ligand with AlbA induced drastic conformational changes in the protein, particularly in the region between 205 and 235 nm, regardless of whether the protein was in water or 4 mM phosphate buffer. At a molar ratio of 0.36, albicidin increased the proportions of α-helixes and β-turns of AlbA from 53.4 and 0% to 67.4 and 13.8%, respectively, at the expense of the β-sheets and unordered structures (decreased from 21 and 25.6% to 0 and 18.8%, respectively). When the ligand-to-protein molar ratio was increased to 1.08:1, the β-sheets and unordered structures of AlbA were totally diminished, while α-helix and β-turn components were further increased to 83.3 and 16.7%, respectively. It appears that the β-sheets and unordered structures of AlbA were converted to α-helix and β-turn structural components during the process of ligand-protein interaction. The data suggest that the α-helix conformational components are most likely involved in the AlbA-albicidin protein-ligand interaction (7) and that tight binding of the ligand fixed AlbA in a conformation with large proportions of α-helixes and β-turns.

AlbA conformational change is pH dependent.

It was shown previously that the AlbA-albicidin interaction is pH dependent, as shifting the buffer pH from 6 to 4 resulted in a drastic decrease in albicidin binding activity (22). A CD analysis showed that AlbA displayed a clear pattern of pH-dependent conformational change (Fig. 2A). In agreement with the effect of pH on the AlbA binding activity (22), no significant variations in the CD spectrum of AlbA were noticed when the buffer pH was changed from 9 to 6, a slight change occurred when the pH decreased from 6 to 5.4, and an evident conformational change happened when the pH shifted from pH 5.4 to 5, which is close to the isoelectric point of the AlbA protein (4.9). A calculation of the secondary structure contents of AlbA at pH 5 showed that the acidic pH condition eliminated the β-sheet component but increased the α-helix content of AlbA to 77.2%. Obviously, such a conformational structure is less than optimal for effective albicidin binding. The AlbA solution at pH 4.4 became cloudy and the protein was precipitated, probably due to protein denaturation. Figure 2A shows that the molecular ellipticity of AlbA at pH 4.4 is close to zero, indicating that extensive protein unfolding occurred, which resulted in a loss of the asymmetric conformational structure.

FIG. 2.

Effect of pH on conformational change of AlbA. (A) CD spectra of AlbA in citric acid-sodium phosphate buffer at different pHs. (B) Time course of albicidin-induced AlbA conformational change. The molar ellipticity of AlbA in water was monitored for 60 s at 222 nm after the addition of albicidin at a molar ratio of 1:1 with AlbA.

Albicidin induces a rapid and specific AlbA conformational change.

Since the binding of albicidin to AlbA induces drastic changes in molar ellipticity in the region from 207 to 223 nm in a low-ionic-strength solution (Fig. 1B), we employed CD to determine the AlbA binding specificity and the dynamics of the protein-ligand interaction. The time course of albicidin-induced AlbA conformational change was monitored by CD at 222 nm (Fig. 2B). The results showed that albicidin, at a molar ratio of 1:1 with AlbA, induced a rapid conformational change in AlbA and reached a plateau within about 2.5 s.

The nucleotide ATP could transiently block the binding of albicidin to AlbA when used at a molar ratio of 1,000:1 against the protein (17). Since large proportions of ATP interfere with the CD analysis of AlbA, we tested whether a 1:1 or 10:1 molar ratio of ATP to AlbA could block the albicidin ligand binding to the protein. The bioassay showed that while the small proportion of ATP had no effect, 10 times as much ATP as AlbA (and albicidin) reduced the AlbA binding activity about 18%. However, no noticeable conformational change of AlbA was detected when ATP was added at a 1:1 or 10:1 molar ratio to the protein (data not shown). It is not clear at this stage how the nucleotide interferes with albicidin binding to AlbA.

AlbD, an albicidin hydrolytic enzyme, not only inactivates albicidin phytotoxin but also degrades several common esterase substrates, including p-nitrophenyl acetate, p-nitrophenyl butyrate, p-nitrophenyl hexanoate, and α-naphthyl butyrate (20). We thus tested whether these chemicals could also induce a conformational change of AlbA. A CD spectroscopic analysis showed that the addition of these substrates to AlbA under the same conditions of the albicidin assay did not cause any significant change in the molar ellipticity of AlbA from 200 to 250 nm (data not shown), suggesting that AlbA and AlbD differ in their mechanisms of substrate binding.

Identification of a potential albicidin binding pocket.

So far, only His¹²⁵ of AlbA is known to be essential for the protein-ligand interaction (18). More amino acid residues may be involved in the binding of AlbA to albicidin. By using an alanine scanning mutagenesis approach, we replaced all of the amino acid residues of AlbA from Glu⁸¹ to Gln¹⁷⁸ with alanine, except for the existing 14 alanine residues in the region. These variants were then tested for albicidin binding ability, with E. coli strain DH5α as a negative control and DH5α(pGST-AlbA) as a positive control. The quantitative bioassay identified eight single substitution mutants, i.e., Lys→Ala (K106A), Trp→Ala (W110A and W162A), Tyr→Ala (Y113A and Y126A), Leu→Ala (L114A), His→Ala (H125A), and Pro→Ala (P134A), which were defective in albicidin binding (Table 2). Based on the secondary structure predicted with the GARNIER-ROBSON program of the Protean module (13), these eight key residues are distributed in the central region of AlbA containing three α-helix and two turn regions (Fig. 3A), suggesting the presence of an albicidin binding pocket centralized at the His¹²⁵ residue. This finding is consistent with the recently updated conserved domain database information (http://www.ncbi.nlm.nih.gov/Structure/lexington/lexington.cgi?cmd=rps), which shows that AlbA contains two conserved albicidin resistance domains, corresponding to amino acid positions 26 to 101 and 132 to 207. These substitutions did not appear to affect the protein expression level (Fig. 3B) but resulted in a 22 to 58% reduction in albicidin binding activity (Table 2). The most prominent drop in AlbA ligand binding ability was caused by the change of Tyr¹²⁶ to alanine.

TABLE 2.

Albicidin binding activity of AlbA and its variants

Strain (single-site mutation)	AlbA	K106A	W110A	Y113A	L114A	H125A	Y126A	P134A	W162A
Relative activity (%)a	100 ± 0.0	72.1 ± 2.5	77.8 ± 4.6	72.6 ± 5.6	73.9 ± 4.4	68.6 ± 5.8	41.7 ± 5.7	70.2 ± 2.3	73.9 ± 4.4

Strain (double-site mutations)	AlbA	K106A, H125A	W110A, H125A	Y113A, H125A	L114A, H125A	H125A, Y126A	H125A, P134A	H125A, W162A
Relative activity (%)a	100 ± 0.0	37.1 ± 3.1	48.5 ± 2.6	52.9 ± 4.2	56.7 ± 3.2	21.4 ± 4.6	62.9 ± 2.8	67.4 ± 3.5

^aRelative albicidin binding activities were determined by the bioassay described in Materials and Methods. Data shown are means ± standard deviations.

FIG. 3.

Relative locations of the eight key residues involved in albicidin binding in AlbA peptide (partial) and the effect of mutations on protein expression in E. coli. (A) Predicted secondary structure of AlbA central region and relative locations of the eight key amino acid residues. Secondary structure prediction was based on the GARNIER-ROBSON program (13). Rectangles and solid horizontal lines represent α-helix and turn regions, respectively. The relative location of each key residue is indicated at the top. The numbers underneath the secondary structure mark the amino acid positions in α-helix regions. (B) Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis of protein expression in E. coli DH5α. Lane 1 shows a protein size marker, and lanes 2 to 10 show protein extracts of E. coli DH5α harboring pGEX-2T, pGST-AlbA, pGST-K106A, pGST-W110A, pGST-Y113A, pGST-L114A, pGST-Y126A, pGST-P134A, and pGST-W162A, respectively. The arrow indicates the position of GST-AlbA fusion proteins.

To assess the potential synergetic effect of these key amino acid residues in the AlbA-albicidin interaction, we used the H125A mutant as a template to generate seven double mutants. Table 2 shows that the double replacement of these key residues led to much more significant reductions in the albicidin binding activity. The most significant decrease in binding activity was caused by the H125A/Y126A double mutant, which maintained only about 21% of the relative albicidin binding activity of wild-type AlbA. This was also a big contrast to its single-replacement parents, i.e., the H125A and Y126A mutants, which had 68.6 and 41.7% relative binding activities, respectively. The other six double mutations also caused about 32 to 63% reductions in relative activity compared with that of wild-type AlbA and about 6 to 35% reductions compared with their corresponding single-replacement parents (Table 2).

Substitution of key amino acid residues of AlbA involved in albicidin binding did not affect the protein's global conformational structure.

To determine the relationship between the conformational structure and the binding activity of AlbA, we conducted detailed CD spectroscopic analyses of AlbA and its variants. The recombinant AlbA variants were purified to near homogeneity by the same protocol that was used for AlbA (see Materials and Methods) (22). Like recombinant AlbA, these purified variants also contained four extra amino acids at the N terminus. The CD spectra of the variants did not differ significantly from that of wild-type AlbA in the absence of albicidin (Fig. 4), indicating that single amino acid substitutions of these eight key amino acid residues in AlbA did not cause a global conformational change or aberrant folding of the protein.

FIG. 4.

CD spectra of AlbA and its variants with and without albicidin. The protein stock solution (3 mM) in 1× phosphate-buffered saline was diluted 120 times in water to a final ion concentration of about 1.2 mM. Albicidin was added to protein samples at a 1:1 ratio.

To study the role of these key amino acids in the protein-ligand interaction, we added albicidin to AlbA and its variants at a 1:1 ratio in a low-ionic-strength buffer and monitored the changes in the CD spectra. Similar to AlbA, the variants showed a reduction in negative molar ellipticity between 202 and 230 nm, in particular at 208 and 222 nm, upon the addition of albicidin (Fig. 4), indicating the formation of protein-albicidin complexes. However, the intensities of the changes in molar ellipticity of these variants with attenuated albicidin binding were significantly different from that of wild-type AlbA. In general, these variants can be grouped into two categories based on the differences in their CD spectra with and without albicidin. The K106A and Y126A variants, which belong to one category, showed a substantially reduced change in molar ellipticity around 193 nm in the presence of albicidin, whereas the remaining variants in the other category displayed much more restricted changes in the region between 202 and 230 nm than did AlbA (Fig. 4). To understand the pattern of conformational change of these variants in the presence of the albicidin ligand, we calculated their secondary structure contents. The common feature of these eight mutants was a reduced β-turn component compared with that in wild-type AlbA (decreased by 3.5 to 12.2%). Except for the K106A, Y126A, and P134A mutants, which contained small proportions of or no unordered structure but had slightly larger α-helix contents than did AlbA (1.4 to 5.3%), the other five mutants also displayed decreased α-helix contents (down by 7.3 to 15.1%) accompanied with increased unordered structures (up to 14.7 to 18.4%). These data indicate that the mutations in these variants do not affect the global structure of AlbA but restrict the albicidin-induced conformational changes, in particular the optimal proportions of α-helix and β-turn components.

Effect of pH on conformational stability of AlbA variants.

To probe the role of the key amino acid residues in the maintenance of the conformational stability of the albicidin binding pocket, we determined the pH-dependent conformational change by monitoring the molar ellipticity at 222 nm of AlbA and the eight variants. Figure 5 shows that the replacement of Try¹¹³, Pro¹³⁴, Try¹²⁶, and His¹²⁵ with alanine makes the protein significantly more prone to destabilization at low pHs than the wild-type AlbA, while alanine substitution at other key positions has a lesser effect.

FIG. 5.

pH-dependent conformational dynamics of AlbA and its variants. The molar ellipticities of the proteins were monitored at 222 nm.

DISCUSSION

The data in this study demonstrate that AlbA has a high content of α-helical structures and a high-affinity binding site for albicidin (Fig. 1). In low-ionic-strength environments, AlbA undergoes significant conformational changes upon interaction with albicidin ligands, accompanying increased helical secondary contents in the protein (Fig. 1B). Secondary structure content calculations showed that in the absence of ligand, about half of the AlbA structural components are α-helixes (53.4%), and that the remaining components are β-sheets (21%) and unordered structures (25.5%). In the presence of a 1:1 ratio of albicidin, however, both the β-sheet and unordered structure components disappeared, accompanying a drastic increase in α-helixes (83.3%) and the appearance of β-turns (16.7%). We speculate that binding of the ligand to AlbA might promote the conversion of the unordered structures to α-helixes and twist the existing β-sheets into β-turns. The CD spectrometry data suggest that α-helix components of AlbA are important for the protein-ligand interaction, which is supported by the alanine scanning mutagenesis analysis showing that most of the key amino acids are located in α-helix regions (discussed below).

This ligand-dependent conformational change appears to be highly specific, and the binding reaction is amazingly rapid. Several nitrophenolic and naphthyl compounds, the effective substrates of the albicidin detoxification enzyme encoded by albD (20), did not cause noticeable conformational changes in AlbA. ATP, which can transiently block albicidin binding to AlbA at a high concentration (17), was unable to induce conformational changes in AlbA at a molar ratio of 10:1. The CD titration analysis confirmed the observation that AlbA binds rapidly to albicidin, within 2.5 s (Fig. 2B), in agreement with a previous estimation based on ultrafiltration data that the AlbA-albicidin interaction might end within 30 s (4).

AlbA appears to have a stable conformational structure which is not affected by the buffer's ionic strength when the pH is higher than its isoelectric point. However, when the pH was lowered to 5.0, which is near the protein's isoelectric point, AlbA underwent a significant conformational change, with a loss of its β-sheet contents (Fig. 2A). The trend of pH-dependent conformational change was similar to that induced by albicidin (Fig. 1B). Since albicidin is an acidic compound (2), the binding of albicidin to AlbA in a low-ionic-strength buffer will likely disturb the charge balance during the process of diffusion through AlbA into the binding site and thus result in conformational changes (16).

Alanine scanning mutagenesis has been particularly useful for the identification of functional residues. Substitution with alanine removes all side chain atoms past the β-carbon, and thus the role of side chain functional groups at specific positions can be inferred (11). Since the His¹²⁵ residue which is essential for AlbA binding to the ligand is likely located at a turn region and since the modification of His¹²⁵ with the histidine-specific reagent diethyl pyrocarbonate reduced the binding activity up to 95% (22), it is possible that His¹²⁵ may occur at the entry portion of the albicidin binding pocket. Hence, we separately replaced 83 amino acid residues in the regions surrounding the His¹²⁵ residue of AlbA with alanine and identified another seven amino acid residues which are important for high-affinity binding to albicidin (Table 2 and Fig. 3A and 4). Of the eight key amino acid residues, four are potent proton donors (Lys¹⁰⁶, Tyr¹¹³, His¹²⁵, and Tyr¹²⁶) and four are able to form hydrophobic bonds (Trp¹¹⁰, Leu¹¹⁴, Pro¹³⁴, and Trp¹⁶²), highlighting the importance of electrostatic and hydrophobic interactions for the high-affinity binding of the albicidin ligand to the AlbA protein. Except for His¹²⁵ and Tyr¹²⁶, all of the key amino acid residues are located in α-helix regions (Fig. 3A), supporting the finding from CD spectrometry analyses that the α-helix components of AlbA are important for the protein-ligand interaction.

Replacement of the eight key amino acid residues did not cause any significant conformational change in the absence of albicidin. It is therefore unlikely that these residues are involved in the maintenance of the global conformational structure of AlbA. Rather, their main roles are in the interaction with albicidin, as alanine substitutions of key amino acids resulted in 30 to 60% reductions in the albicidin binding ability (Table 2 and Fig. 4). However, some of these key residues might be required for stabilization of the local conformational structure of the albicidin binding site. The four variants in which a proton donor (Tyr¹¹³, His¹²⁵, or Tyr¹²⁶) or Pro¹³⁴ was replaced with alanine appeared to be more prone to conformational changes in a low-pH environment than the other variants (Fig. 5). The rigid structure of proline has long been known to be important for maintaining proper protein conformational structures. These data confirm the importance of a balanced electrostatic potential for the maintenance of the conformational structure of AlbA.

It appears that the majority of the key amino acid residues contribute more or less equally to the binding of albicidin to AlbA (Table 2). Except for Tyr¹²⁶, alanine substitution of any of the other seven amino acids caused only about a 22 to 30% loss of ligand binding activity. This notion was further strengthened by a double-site mutation analysis in which the sum of binding activity loss of double mutants was almost proportional to that of single mutants. Nevertheless, the data also indicated that the His¹²⁵ and Tyr¹²⁶ pair is the major contributor to the binding of the albicidin ligand; the substitution of both with alanine caused an approximately 79% decrease in binding activity (Table 2). Such multiresidue-dependent ligand binding is not uncommon in protein-ligand interactions. Human Raf-1 kinase contains two ligand binding regions. At least seven amino acid residues are involved in binding to the ligand phosphatidic acid. Similar to the case for AlbA, the mutation of any of the key residues only partially decreased the binding activity, and double mutation further sacrificed the ligand binding activity (6). The involvement of multifunctional residues in the AlbA-ligand interaction is also consistent with the finding that albicidin is a polyketide-type antibiotic with a relatively large molecular size (2, 8, 14).

It will be interesting to determine whether the three albicidin resistance proteins, i.e., AlbA, AlbB, and AlbD, share a conserved albicidin binding motif. No key amino acid residue involved in albicidin binding has yet been identified for AlbB, another albicidin binding protein (1), whereas AlbD, which is an esterase, is known to contain a GXSXG catalytic triad (20). A peptide sequence alignment showed that the albicidin resistance proteins AlbA and AlbB share three identical and three similar key amino acid residues (Fig. 6). Note that the His¹²⁵ residue that contributes to the high-affinity binding of AlbA to the albicidin phytotoxin is replaced with alanine in AlbB. We speculate that AlbB might have a lower albicidin binding activity than AlbA, which deserves further verification. A secondary structure comparison also showed a certain degree of similarity between the two antibiotic resistance proteins around the albicidin binding region, and in particular, both proteins contain a turn region where the essential Tyr¹²⁶ is located. However, the homology between the regions around the AlbA binding pocket and the AlbD enzyme catalytic center (1) is so low that we are inclined to conclude that AlbA and AlbD have different substrate binding mechanisms. This is consistent with the observation that none of the four nitrophenolic and naphthyl substrates of AlbD tested in this study was able to induce conformational changes in AlbA.

FIG. 6.

Partial peptide sequence alignment and secondary structure comparison between AlbA and AlbB. Secondary structure prediction was based on the GARNIER-ROBSON program (13). The predicted α-helixes (horizontal columns) and β-sheet (horizontal arrow) of AlbA and AlbB are shown above the sequences. Conserved key amino acid residues for albicidin binding are boxed, and unmatched key residues are marked by triangles. Identical, very similar, and similar amino acid residues are indicated by stars, colons, and dots, respectively.

In conclusion, the present investigation has identified eight key amino acid residues and provided insight into the structural and functional roles of these amino acids that constitute the high-affinity albicidin binding pocket of AlbA. The data suggest that AlbB, another albicidin resistance protein, might also contain a similar ligand binding pocket. CD spectroscopy analysis coupled with alanine scanning mutagenesis may provide a rich body of information on the molecular mechanism of this protein-ligand interaction, even in the absence of structural knowledge about the ligand. Such information may be valuable not only for understanding the molecular mechanism of antibiotic resistance, but also for designing and using this interesting protein as a highly efficient affinity matrix and as a novel form of disease resistance mechanism (22).