Targeted Engineering of the Antibacterial Peptide Apidaecin, Based on an In Vivo Monitoring Assay System

Abstract

Seven mutant forms of the antibacterial peptide apidaecin with increased activity were created by combinatorial mutagenesis targeted to the three N-terminal amino acid residues that had previously been identified as a nonessential region. An in vitro MIC assay revealed that the amino acid substitutions in the functionally variable region were effective in improving differential activity toward the four gram-negative bacteria tested, while a gram-positive bacterium was unaffected.

Antibacterial peptides (ABPs) are effector molecules that form a first line of host defense against pathogens, are involved in innate immunity, and are found in different groups of organisms (1, 8, 15). In particular, insects having large repertories of ABPs are amazingly resistant to bacterial infections and thus are prominent cell factories for genuine combinatorial chemistry of ABPs acquired during the long-term evolution of immune defense systems against environmental pathogens. Apidaecin (AP) is a member of the proline-rich ABP family (13, 16). Honeybee-derived AP exists in four isoforms that are each 18 amino acids long, commonly including three arginine and six proline residues (3). ABPs such as AP have been recognized as potential therapeutic alternatives to antibiotics because of their immediate effect, their apparent nontoxicity toward eukaryotic cells, and the fact that there is little or no bacterial resistance (6, 11). Although it does not display membrane-disrupting activity (bacteriolytic action), AP exhibits bacteriostatic action specifically toward gram-negative bacteria such as Escherichia coli (2, 4). To date, a mechanism for the action of AP was proposed in which AP binds initially to lipopolysaccharide in the cell membrane and subsequently to the heat shock protein DnaK and related chaperones of E. coli in a specific manner (14, 16).

In this study, we used combinatorial mutagenesis to create APs with improved activities toward gram-negative bacteria, including E. coli. To this end, the region of AP targeted for engineering was the three N-terminal residues. These residues were chosen based on the structure-activity relationships of naturally occurring derivatives (AP family members) (4) or artificially mutated derivatives (5, 7, 17, 21, 22, 24). Previously, we proposed the first in vivo system for monitoring the antibacterial activities of the products of the randomly mutagenized genes for AP and thanatin (20) based on antibacterial activity-associated growth inhibition (21). This “suicide” system utilizes an ABP-sensitive strain of E. coli as the host for expression of PCR-mutagenized AP molecules. When a successful (i.e., activity-enhancing) PCR-generated random mutation is introduced into the structural gene for AP, diminished growth of isopropyl-β-d-thiogalactopyranoside (IPTG)-induced recipient cells is measured via either reduced colony size (agar cultures) or reduced optical density (liquid cultures). However, in principle, severe growth inhibition caused by mutations with highly enhanced activity may be a constraint for screening mutants by colony formation on a plate. The plate assay based on activity-dependent colony size was used for primary screening, and the liquid assay profiling growth inhibition caused by antibacterial activity was used for more precise estimation of the effects of mutations. Indeed, with these assays, amino acids indispensable for activity identified by this system were easily confirmed by measuring the MICs of synthetic artificial mutant peptides (22) (Fig. 1). The results obtained in our study were also supported by those obtained from site-specific amino acid substitution experiments (5, 7, 26). Figure 1 shows a multiple-sequence alignment of natural homologs and the mapping of mutant forms of AP causing complete or partial activity loss. The C-terminal region (amino acids 12 to 19) and many of the proline and arginine residues have been found to play a crucial role in the antibacterial activity of AP (5, 7, 17, 22, 26). Therefore, for targeted engineering, we specifically targeted the N-terminal region consisting of the first three amino acid residues that are structurally nonconservative among AP homologs and are functionally tolerant of amino acid substitutions. Together with this mutation strategy, the in vivo monitoring assay system was expected to function efficiently in high-throughput screening for AP molecules with increased activity.

FIG. 1.

Multiple-sequence alignment of natural proline-rich homologs with APs. This sequence alignment is based on the previous report of Casteels et al. (4). A dash indicates that the residue at that position is identical to the one in the primary sequence. A space denotes a gap inserted into the amino acid sequence to optimize alignment. Consensus sequences are boxed. Mutation-based functional mapping results are overlapped with the sequence alignment. Characters above amino acids in the top line represent effects caused by mutation experiments as follows: *, activity silent; @, activity lost or lower (21, 22). The three N-terminal positions are specified as a structurally and functionally variable region and are boxed with dotted lines.

The expression plasmid vector pOSΔB-AP1 (21, 22) was used to generate a random AP gene mutant library. pOSΔB-AP1 was previously constructed by fusing the chemically synthesized AP gene with that for a bacterial cell stable protection partner, namely, the Streptomyces subtilisin inhibitor (19, 23). Expression of the AP fusion gene can be stringently regulated by the addition of a fine-tuned concentration of the transcription inducer IPTG, depending on the growth inhibition of the host cells (21, 22). pOS1 (23) with the Streptomyces subtilisin inhibitor gene alone was used as a control expression vector. Mutant APs were generated by PCR with four forward primers that included mutagenic sequences that encoded all 20 of the possible amino acids in the three N-terminal positions of AP (Table 1). The entire region of the AP gene was amplified with individual forward primers and a Tag reverse primer. PCR was performed with 100 μl of a reaction mixture (2.5 nM pOSΔB-AP1, 0.01 U of KOD+ [Toyobo], 1 μM primers, 0.2 mM each deoxynucleoside triphosphate, 10 mM Tris-HCl [pH 7.0], 50 mM KCl, 1.5 mM MgCl₂) with a program consisting of 30 cycles of 95°C for 20 s (denaturation), 47°C for 30 s (annealing), and 67°C for 90 s (elongation), followed by an extension reaction consisting of 72°C for 7 min and finally termination by cooling down to 4°C on an iCycler (Bio-Rad). After amplification, the mutagenized DNA fragments, including heterogeneous mutant AP genes, were religated at the EcoRI and BamHI sites into the original plasmid, pOSΔB-AP1, predigested with the same restriction enzymes to generate a mixture of the plasmids that encode mutant APs (designated pOSΔB-APN). The mixture of resultant plasmids was introduced into E. coli JM109 cells that were then plated on Luria-Bertani agar plates supplemented with 0.02 mM IPTG and 50 μg/ml ampicillin. Plasmid pOSΔB-APX carrying the wild-type AP gene was also used as a control for accurate screening.

TABLE 1.

Primers used for PCR-mediated mutagenesis^a

Primer	Sequence
Tag forward	5′-AATTCATCGAGGGCCGCATGGGCAACAACCGCCCGGTCTACATCCCGCAGCCGCGGCCGCCGCACCCGCGCCTGTAGTAG-3′
Tag reverse	5′-GAAGTTAGTCGATAAAGCGTTG-3′
Tag A	5′-TTCGAATTCATCGAGGGCCGCATGANNNNNNNNCGCCCGGTCTAC-3′
Tag G	5′-TTCGAATTCATCGAGGGCCGCATGGNNNNNNNNCGCCCGGTCTAC-3′
Tag C	5′-TTCGAATTCATCGAGGGCCGCATGCNNNNNNNNCGCCCGGTCTAC-3′
Tag T	5′-TTCGAATTCATCGAGGGCCGCATGTNNNNNNNNCGCCCGGTCTAC-3′

^aFour forward primers were used for PCR-mediated mutagenesis targeting the three N-terminal amino acids of AP together with a reverse primer. Each bold letter is the first letter of the codon encoding the N-terminal residue of AP. The EcoRI site is underlined. The Tag forward primer contains the sequence encoding wild-type AP.

Theoretically, there are 20³ possible combinations that could be generated by substitution of all 20 of the possible amino acids at the three positions. With the plate assay method developed here, from a total of 5.7 × 10⁴ clones obtained, 431 clones were primarily selected as candidates, as judged by colony size, which is closely related to the levels of antibacterial activity of the mutated AP. Finally, we isolated 95 clones by repeated screening based on the plate assay method. Next, all clones were subjected to the liquid culture-based growth inhibition assay to quantitatively estimate antibacterial activity. Induction of the expression of mutant genes was performed by the addition of IPTG to the Luria-Bertani medium containing transformant cells with an optical density of 0.05 at 600 nm. Finally, we obtained seven mutants that clearly exhibited enhanced growth inhibition. Figure 2 shows the representative growth curves in liquid medium for transformant strains harboring plasmids carrying the wild-type AP gene and the seven mutated AP genes, with pOS1 as a control. Mutant 1C-20 exhibited the greatest growth inhibition. Nucleotide sequence analysis revealed that two or three amino acid substitutions were found in the target region in all seven of the mutants, as shown in Fig. 2. Notably, a single or double substitution of arginine occurred at the first and/or third position of the target region. Mutant 1C-20 had arginine residues at the first and third positions.

FIG. 2.

Liquid culture-based growth inhibition assay with E. coli JM109 cells. Representative cell growth patterns are presented with a summary (right side) of amino acid substitutions and mutated codons (in parentheses) of mutant APs. Amino acids are represented by the one-letter code. See the text for details. The standard deviations calculated for three trials are indicated by error bars. OD₆₀₀, optical density at 600 nm.

Enhancement of the antibacterial activity of the mutated APs was not fully judged based on the growth inhibition assay, in which AP was produced intracellularly, because in practical use, extracellularly added AP exhibits antibacterial activity. Therefore, we chemically prepared the seven selected APs as follows. Peptide synthesis was carried out with a manual peptide synthesis system (CCS-600V; EYELA) according to the standard 9-fluorenylmethoxycarbonyl solid-phase peptide synthesis protocol. The peptide sequence was extended on leucine-loaded Wang resin (90-μmol scale), which provided C-terminal peptide carboxylic acid upon cleavage. Protected amino acids were coupled with 3 equivalents of 9-fluorenylmethoxycarbonyl amino acids, 3 equivalents of N,N′-diisopropylcarbodiimide, 3 equivalents of 1-hydroxy-7-azabenzotriazole, and 6 equivalents of diisopropylethylamine in N,N-dimethylformamide. Amino acid side chains were protected as follows: tert-butyl for Tyr; trityl for Asn, Gln, and His; and 2,2,5,7,8-pentamethylchroman-6-sulfonyl for Arg. Cleavage of the peptide from the resin and deprotection of all side chain protecting groups were accomplished by treating the resin with a fresh mixture of trifluoroacetic acid (TFA), H₂O, and triisopropylsilane (95:2.5:2.5) for 2.5 h. The solution was filtered, and the resin was rinsed with the cocktail. Diethyl ether was added to the filtrate to precipitate the peptide. Purification of the crude peptide by high-performance liquid chromatography was carried out on a C₁₈ reverse-phase column (5 μm, 10 by 250 mm) with solvent A (0.1% TFA in water) and solvent B (80% acetonitrile in 0.1% TFA in water) as the mobile phase. Elution was effected with a linear gradient of 20 to 50% B over 20 min and then 50 to 100% B over 10 min at a flow rate of 4.0 ml/min (monitoring at 225 nm). The desired product was recovered from the appropriate fractions by lyophilization. The molecular masses of the peptides were analyzed by matrix-assisted laser desorption ionization-time of flight mass spectrometry with a Voyager-DE STR-H (Applied Biosystems) with α-cyano-4-hydroxycinnamic acid as the matrix. The matrix-assisted laser desorption ionization-mass spectrometry (M+H)⁺ results were as follows: wild type, 2,108.2 (calculated) and 2,108.1 (determined); 1C-20, 2,234.3 (calculated) and 2,234.4 (determined); 1A-11, 2,175.3 (calculated) and 2,175.3 (determined); 1A-20, 2,135.2 (calculated) and 2,134.8 (determined); 1A-39, 2,190.3 (calculated) and 2,190.3 (determined); 1C-21, 2,206.3 (calculated) and 2,206.8 (determined); 1C-3, 2,189.3 (calculated) and 2,189.1 (determined); 1G-17, 2,177.3 (calculated) and 2,177.4 (determined).

Synthesized wild-type and mutant AP peptides were subjected to a MIC assay. The MIC was defined as the lowest concentration of the peptide needed for the inhibition of bacterial growth. The MIC was determined by a microtitration assay with test strain cells in wells of multititer plates according to the procedure reported previously (21, 22). Briefly, the assay was performed by adding 20 μl of each purified peptide sample to 80 μl of PB medium (1% Bacto tryptone, 0.9% NaCl) inoculated with 1,250 cells (defined by CFU counting) of all of the test strains, followed by culturing for 15 h at 30°C under shaking conditions in 96-well multititer plates. The five bacterial strains used as test targets (listed in Table 2) were E. coli JM109 and BL21(DE3), the gram-negative bacteria Pseudomonas putida GPp104 and Ralstonia eutropha H16, and the gram-positive bacterium Bacillus subtilis 168 (data not shown). To check for bacterial contamination in the peptide sample, medium containing the peptides alone was used as a negative control. Experimental reproducibility was confirmed by performing three trials.

TABLE 2.

In vitro assays of mutant AP activity toward various gram-negative bacteria

Peptide	Amino acid sequence	MIC (μM)a
		E. coli JM109	E. coli BL21(DE3)	P. putida GPp104	R. eutropha H16
Wild type	GNN	25	1.25	25	50
1C-20	RVR	2.5	0.313	3.13	10
1A-11	RPV	6.25	0.313	6.25	10
1A-20	RGV	6.25	0.313	6.25	25
1A-39	RNP	6.25	0.313	6.25	25
1C-21	RLN	10	1.25	12.5	50
1C-3	RLP	25	1.25	12.5	50
1G-17	VVR	12.5	1.25	25	50

^aThe MIC was judged as the level at which no growth occurred after 16 h of incubation for E. coli JM109 and BL21 and P. putida GPp104 and 24 h of incubation for R. eutropha H16. The broth microdilution method was used.

Table 2 is a summary of the MIC assay results. Most of the mutants exhibited activity comparable to or higher than that of wild-type AP toward E. coli JM109, which was used in the in vivo monitoring assay system as the host for AP gene expression. Interestingly, it was found that mutant 1C-20 had acquired approximately 10-fold-improved activity over that of wild-type AP. This MIC result correlated well with that obtained in the growth inhibition assay. Target spectrum specificity is also an attractive objective for the engineering of ABPs. The positive correlations between in vitro and in vivo assays were found in the other three bacteria, although sensitivity to the APs was different in the different bacterial species. Thus, activity was stronger toward E. coli BL21(DE3) (up to 20-fold over wild-type AP) and weaker toward R. eutropha H16 (down to one-quarter of the activity of wild-type AP). Thus, the in vivo monitoring assay system has been proven useful for the rapid screening of mutants. However, only 1G-17 showed a lack of correlation between the results observed in the two assays. This contradictory relationship might be a result of the reduced membrane-permeating ability of the VVR mutant, since the in vivo assay system is conducted by expression of the AP gene from the inside of the bacterial cells. With regard to the gram-positive bacterium B. subtilis, no antibacterial activity was detected as it was not intrinsically sensitive to wild-type AP (data not shown).

Thus, construction of an AP mutant library will allow treatments to consist of either a single molecule or a mixture with other types of antibacterial molecules such as conventional antibiotics, depending on the practical demands. There are several functional-mapping and engineering strategies that can be applied to ABPs, such as chimeragenesis (17, 18), site-specific glucosidation (10), chemical modification (10, 12), and amino acid substitution (5, 7, 20-22, 26). In this study, we have demonstrated the effectiveness of targeted engineering in a region that is dispensable for the antibacterial activity of AP. In addition, the in vivo monitoring assay system has proven to be a powerful tool for efficient screening of AP derivatives with increased activity in conjunction with the targeted mutagenesis approach. The functional requirement of arginine for gain of activity can be accounted for by the mutation pattern. High diversity in the codon encoding arginine (AGG, CGT, CGC, and CGG) was found in the first position of six of the mutant forms, with the exception of 1G-17 (Fig. 2). Also, the contribution of arginine at the third position to enhanced activity can be considered by the study of two mutant forms, 1C-20 and 1G-17.

Recent studies have indicated that arginine-rich peptides such as the human immunodeficiency virus type 1 Tat protein can cross the cytoplasmic membrane and enter mammalian cells (9, 25). Although the mechanism of this translocation is not well understood, arginines in these peptides play an important role in facilitating their translocation. This notion suggests that arginines in the N-terminal region of mutant APs allow efficient translocation through the bacterial membrane, which is consequently reflected in higher antimicrobial activities. Actually, cell penetration efficiency is dramatically reduced when an N-terminal cationic cluster of bactenecin, a proline-rich ABP like AP, is removed (24). Thus, the charged N-terminal region might be a key structural element of the membrane translocation ability of mutant peptides. Interestingly, 1C-20 acquires the highest activity by a cumulative effect of arginine substitutions. Also, a proline and aliphatic residues (valine or leucine) seem to be potential contributors to activity enhancement. Furthermore, the frequent occurrence of basic amino acids and proline was also found in the corresponding target region of natural or artificial AP derivatives (Fig. 1). These findings may provide us with a more rational sequence design strategy to further improve the antibacterial activity and alter the activity spectrum of AP and related peptides.