Three-Dimensional Structure of the Antimicrobial Peptide Cecropin P1 in Dodecylphosphocholine Micelles and the Role of the C-Terminal Residues.

Cecropin P1 (CP1) isolated from a large roundworm Ascaris suum, which is found in pig intestines, has been extensively studied as a model antimicrobial peptide (AMP). However, despite being a model AMP, its antibacterial mechanism is not well understood, particularly the function of its C-terminus. By using an Escherichia coli overexpression system with calmodulin as a fusion partner, we succeeded in the mass expression of recombinant peptides, avoiding toxicity to the host and degradation of CP1. The structure of the recombinant 15N- and 13C-labeled CP1 and its C-terminus truncated analogue in dodecylphosphocholine (DPC) micelles was determined by NMR. In this membrane-mimetic environment, CP1 formed an α-helix for almost its entire length, except for a short region at the C-terminus, and there was no evidence of a hinge, which is considered important for the expression of activity in other cecropins. Several NMR analyses showed that the entire length of CP1 was protected from water by micelles. Since the loss of the C-terminus of the analogue had little effect on the NMR structure or its interaction with the micelle, we investigated another role of the C-terminus of CP1 in its antimicrobial activity. The results showed that the C-terminal region affected the DNA-binding capacity of CP1, and this mechanism of action was also newly suggested that it contributed to the antimicrobial activity of CP1.

Introduction

Recently, due to the misuse of antibiotics, the continuous emergence of several kinds of drug-resistant bacteria has occurred,[1] and a steady and efficient substitution of antibiotics to combat new and existing microbes is required. For this reason, antimicrobial peptides (AMPs) have been the focus of attention among scientists and have become a popular topic for research.[2]

Cecropin P1 (CP1) was once thought to originate from pig intestines but was later found to originate from the parasitic nematode Ascaris suum.[3] Once isolated, the 31-residue cationic AMP was studied extensively. CP1 has strong antimicrobial activity toward pathogenic Gram-positive or Gram-negative bacteria[4] and plays an important role in antiviral and anticancer treatments.[5,6] CP1 is a relatively early discovery among AMPs, and it is derived from lower animals. Although it is not derived from mammals, it has been used as a model AMP in many structural and mechanism studies. After its initial isolation, the three-dimensional structure of the chemically synthesized CP1 was investigated via 1H nuclear magnetic resonance (NMR).[7] The molecule has an α-helical structure that is closely related to its antibacterial activity. Subsequently, several studies have been conducted on its structure in different membrane-mimetic environments.[8−10] The structure and activity of the major members of the cecropin family from the lepidopteran insect Hyalophora cecropia (cecropia moth) have been more extensively studied, with the most promising targets being the cell membranes of Gram-negative bacteria, although their specific antibacterial mechanisms are also not fully understood.[11,12] Furthermore, since CP1 is an AMP from non-lepidopteran insects, it has different structural and active properties from them, and it is important to study its antimicrobial mechanism. While most typical cecropins have a structure of two α-helices connected by a hinge region, which is considered important for the expression of activity, CP1 is reported to be formed from only one long α-helix.[13−16] The development of methods to prepare larger quantities of CP1 is becoming more important to help elucidate the mechanism of action and for application purposes. Chemical synthesis is a common method for obtaining small peptides. However, with an increase in the number of residues, synthetic efficiency and cost often become limiting factors. Even for relatively short AMPs, 15N- and 13C-labeled samples for NMR experiments are difficult to obtain via chemical synthesis. Recombinant expression is an effective solution to these problems.[17,18]

The traditional recombinant expression system using Escherichia coli (E. coli) can at times be ineffective in expressing AMPs. Various proteases in the cell can pose a threat to short peptides with a simple structure. Additionally, the microbial host cell cannot protect itself from antimicrobial toxicity resulting from the recombinant products. Thioredoxin (Trx), a traditional fusion partner protein, successfully solved a series of problems and enabled the expression of various AMPs, such as human cathelicidin LL-37, which is also an α-helical AMP.[19,20] Unfortunately, we obtained very low yields, probably due to the toxicity of CP1, even with Trx fusion.[21] To address the problems of expression of recombinant AMPs, we have recently developed a fusion expression system using calmodulin (CaM) as a fusion partner. CaM can bind to amphiphilic, positively charged peptide regions in various target proteins with high affinity and regulates their function.[22−24] AMPs, which are amphiphilic and positively charged, have a high affinity with CaM because of their sequence similarity to the target peptides of CaM.

In this study, we applied this CaM fusion expression system to obtain active CP1, which can be used in NMR analysis. We performed a three-dimensional structure analysis of CP1 and its analogue with high accuracy in a membrane-mimetic environment, in addition to clarifying its mode of interaction with the membrane. Based on these results, we succeeded in obtaining new insights into the membrane interactions of the C-terminal region of CP1 and its antimicrobial activity via DNA binding.

Results and Discussion

Expression of CP1 from Different Fusion Protein Constructs

To investigate suitable fusion partner proteins for CP1 overexpression, two types of pET vectors, i.e., CaM fusion (CaM-CP1) and Trx fusion (Trx-CP1), were designed for use in E. coli as a host (Figures S1 and S2). To analyze the toxicity of both constructs toward host cells, we evaluated the OD600 of cells with the constructs and monitored the growth curves in LB media (Figure ). After induction with isopropyl β-d-thiogalactopyranoside (IPTG), the OD600 was measured for at least 4 h. The expression of CP1 fused with CaM was compared for the Trx fusion, CaM without any target, Trx without any target, and CP1 without any fusion expression systems. The results showed that except for the Trx-CP1 construct, the growth curves for all expression systems reached an OD600 of ∼1.0 (Figure A). The growth curves increased even higher without IPTG addition. The expression of Trx-CP1 inhibited the growth of cells after 1 h of induction, reaching an OD600 of ∼0.6. In the absence of any fusion protein, the growth curve of CP1 also increased gently. The overexpression of fusion CP1 via both constructs was confirmed by tricine–SDS-PAGE (Figure B). CaM-CP1 was expressed in the soluble fraction at comparable levels to CaM alone. In contrast, the expression of Trx-CP1 was very low compared to that of Trx alone. In the absence of the fusion partner protein, CP1 expression was not confirmed compared to the standard sample.

Figure 1

(A) The effect of CP1 expression and fusion protein expression using the CaM and Trx fusion systems on the growth of the E. coli BL21 (DE3) host cells. The growth curves of IPTG-induced expression obtained for CaM-CP1 (red line), CaM (black line), Trx-CP1 (blue line), Trx (gray line), and CP1 without any fusion (green line) are shown as solid lines. E. coli cells were cultivated in 5 mL of LB medium at 37 °C and induced with 1.0 mM IPTG at an OD600 of ∼0.60. IPTG was added at 0 min. For comparison, the growth curves of each construction cultivated without IPTG are shown as dashed lines. (B) Tricine–SDS-PAGE showing the expression of different constructs after 4 h of IPTG induction at 37 °C. The leftmost lane shows the molecular mass marker. The odd-numbered lanes show the supernatant after sonication of IPTG-induced culture, and the even-numbered lanes show the supernatant after sonication of IPTG-free culture. Lanes 1 and 2, Trx; lanes 3 and 4, Trx-CP1; lanes 5 and 6, CaM-CP1; lanes 7 and 8, CaM; lanes 9 and 10, CP1; lane 11, CP1 standard (90 ng). The bands for Trx and CaM expressed with and without CP1 are indicated by arrows. The band for CP1 is also marked by an arrow. (C) Tricine–SDS-PAGE showing the large-scale expression and purification of CP1. The leftmost lane shows the molecular mass marker. Lane 1 shows the supernatant of the cell lysate after induction. Lane 2 shows the flow-through of IMAC after loading all the samples. Lane 3 shows the collection of IMAC after injecting the wash buffer. Lane 4 shows the peak fraction eluted from IMAC.

These results indicate that Trx cannot effectively control the toxicity of CP1, unlike what has been reported for other AMPs from the cecropin family.[25] Other α-helical AMPs, such as LL-37 that is toxic to E. coli cells, enable host cells to grow normally with the Trx fusion system.[26] However, the toxicity of CP1 could not be efficiently controlled by the Trx fusion protein, even though it is a common α-helical AMP. In contrast, the growth of E. coli expressing CaM-CP1 indicated that CaM can reduce the toxicity of CP1, as was observed for the expression of other AMPs.[22] In the absence of the fusion protein, the expression level of CP1 was very low, which may have resulted from insufficient synthesis or degradation in the host cell and may not have been highly toxic to the host cell. It is hypothesized that CaM can protect amphiphilic AMPs that form an α-helical structure by binding to them in an enveloping manner,[22] which may have resulted in the efficient expression of CaM-CP1.

Purification of Fused CP1

Large-scale purification of CP1 was performed using the CaM-CP1 construct. The constructs were effectively expressed in the large-scale culture (Figure C); CaM-CP1 showed good solubility and remained in the cell lysate supernatant after a high-speed centrifugation. The N-terminal (His)6-tag of the CaM partner-enabled purification of immobilized metal affinity chromatography (IMAC) with Ni2+-column chromatography could be achieved (Figure C).

CaM-CP1 eluted from IMAC was collected for overnight dialysis in enterokinase (EK) reaction buffer, and the purified proteins were subjected to EK protease cleavage. The results of the analysis for the optimal conditions for EK digestion were confirmed by tricine–SDS-PAGE (Figure S3). The fusion protein was incubated in the reaction buffer at 25 °C, using 1 U/mL EK for 4 h. We found that although the fusion protein could be fully cleaved by long-duration incubation, the production of CP1 decreased. Because of the possibility that nonspecific cleavage may have occurred, the short digestion yielded the most CP1, even though some of the fusion protein remained. Previous studies using CaM expression systems have used tobacco etch virus (TEV) cleavage sites (ENLYFQ/G) for purification.[22] However, after digestion, an undesirable glycine, which is not present in the original sequence, will remain on the N-terminus of the target peptide, and it is unclear whether the extra amino acid will affect the structure and function of the target peptide.[27] Therefore, we changed the TEV site to an EK cleavage site (DDDDK) to avoid excess residues at the N-terminus. Isolated CP1 was separated from CaM and EK proteases by reverse-phase high-performance liquid chromatography (RP-HPLC) using a C18 column (Figure A), and the purified products in each collected fraction were analyzed by tricine–SDS-PAGE (Figure B). CP1 was present in a single peak in the RP-HPLC profile that confirmed that it was separated with good purity (Figure S4). Purified CP1 after mixing each fraction of peak 1 of RP-HPLC was further confirmed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) analysis (Figure C). The measured average mass of the recombinant CP1 was 3337.44 Da, which was within the error range of the expected result for the chemically synthesized sample, which was used as the standard at 3336.61 Da and the theoretical average mass value of 3338.9 Da.

Figure 2

(A) RP-HPLC chromatogram for the purification of CP1 after EK protease digestion from CaM-CP1. The arrows indicate the (1) purified CP1, (2) cleaved CaM, and (3) remaining undigested CaM-CP1 fusion protein. The black line shows the absorbance at 280 nm. The gray line indicates the acetonitrile gradient. (B) Tricine–SDS-PAGE of fractions from each peak using RP-HPLC. The leftmost lane shows the molecular mass marker. Lanes 1–3 show peak 1 including three collected fractions: 21–21.5, 21.5–22, and 22–22.5. Lanes 4–6 show peak 2 including three collected fractions: 24.5–25, 25–25.5, and 25.5–26. Lanes 7 and 8 show peak 3 including two collected fractions: 26–27 and 27–28. (C) MALDI-TOF MS spectrum of isolated CP1 after HPLC. The strongest peak represents CP1, and the mass-to-charge ratio is indicated.

The yield estimated from the absorbance at 280 nm for purified CP1 using the CaM fusion protein system was 2.7–4.7 mg from 1 L of LB media. For comparison, the same purification treatment was performed using the Trx-CP1 construct (Figure S5), and 0.03 mg of CP1 was obtained using the Trx fusion protein system. The yield of CP1 compared with the Trx fusion protein system was increased using the CaM fusion construct, which was similar to that of other AMPs expressed using CaM.[22] Previous studies on CP1 have been based on naturally isolated or chemically synthesized CP1.[13,28] In this study, we have succeeded in establishing a method to obtain sufficient amounts of recombinant CP1 for various experiments, including NMR.

NMR Chemical Shift Assignment and Three-Dimensional Structural Analysis

13C- and 15N-labeled CP1 was expressed using the CaM fusion system in M9 minimal media. After the same purification process described above was used, ∼2 mg of CP1 was obtained from 1 L of media. The structure of chemically synthesized CP1 has been shown using 1H NMR as a random coil structure in water and an α-helical structure in an aqueous solution with 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP), although the study was relatively old and detailed data were not registered in the PDB.[7] The structure of the recombinant CP1 produced by the CaM fusion system was evaluated using 1H, 13C, and 15N NMR in this study, and we used dodecylphosphocholine (DPC) micelles as a membrane-mimetic environment to understand the mechanisms underlying the reaction of CP1 with the bacterial membrane. The chemical shift assignment was carried out without ambiguity (BMRB: 36394) by using the standard triple resonance NMR experiments (Table S1A). All backbone amide chemical shifts were assigned and are shown in the 1H–15N HSQC spectrum (Figure A). Because of the presence of Pro30, multiple conformations were present from Gln27 to Arg31 and the peaks for these minor components were also assigned.[29,30] Cγ chemical shifts of the minor and main components of Pro30 illustrated that the ratio of cis conformation to trans conformation was 1 to 4 (Figure S6). The large change in the chemical shift from the random coil values of the Cα and Cβ resonances of CP1 in water (Figure S7) and in DPC micelles (Figure B) indicated the formation of an α-helical structure between residues Trp2 and Gln27 in DPC micelles. The three-dimensional structure calculated by CYANA is shown in Figure C (PDB code: 7DEH), and the structural calculation statistics are summarized in Table S2. A single, well-converged α-helix was formed, of which the N-terminal side up to Glu20 formed an amphiphilic helix (Figure D), and the C-terminal side from Gly21 formed a hydrophobic helix (Figure S8). The two ends of the helix were relatively disordered. The information on the backbone dihedral angles using the Ramachandran plot illustrated that CP1 preferred an α-helical structure in the DPC environment.

Figure 3

(A) 1H and 15N HSQC NMR spectrum of CP1 in DPC with assignments. W2Nε, N12Nδ, R17Nε, Q27Nε, and R31Nε are peaks from side chains. The labels with numbers and apostrophes indicate peaks of the minor component in the multi-form. (B) Chemical shift deviations from random coil values for the Cα and Cβ resonances of CP1 in the DPC micelle. The secondary structure element is indicated at the top of the figure by a black box. (C) NMR structures of the main chain of CP1 in the DPC micelle: an overlay of the ensemble of 20 final energy-minimized CYANA structures. The left-hand side is the N-terminus. (D) Helical-wheel diagrams of CP1 using HeliQuest: the helix area is from Trp2 to Gln27. Positively charged amino acid residues are shown in blue, negatively charged residues are in red, and hydrophobic residues are in yellow. Ala and Gly are shown in gray. The arrow indicates the helical hydrophobic moment.

HFIP is a solvent that contributes to the formation of secondary structures in peptides. DPC has been widely used to study the conformation of peptides in membranes because the stabilization of the preferred peptide structure is facilitated by them.[31] The use of DPC can result in the formation of micelles that have a hydrophobic core and a hydrophilic surface, which better simulates a membrane-mimetic environment compared to HFIP.[32,33] The structures of several peptides in DPC and HFIP have been studied previously, and at times the structures in the two solvents were found to be similar and at other times were different.[34] A previous study using chemically synthesized CP1 in HFIP showed the prevalence of a typical α-helical structure almost the entire length of the peptide from residues Leu3 to Gly29,[7] which is similar to our results from Trp2 to Gln27 using DPC micelles in this study. Unfortunately, detailed structural information on the structure of CP1 in previous HFIP environments is not available in public databases, making a direct detailed comparison with the present results difficult. The results of this study confirm that CP1 is likely to form a single long helical structure over its entire length, even in a real membrane environment.

Cecropin A, another model AMP in the cecropin family, has two helical regions joined by a hinge, extending from residues 5 to 21 and from residues 24 to 37.[14] Gly23 in cecropin A is involved in forming the hinge region that is common in the cecropin family (Figure S9).[13,15,16] Cecropin B, which has the strongest antibacterial activity in the cecropin family,[35] also contains a helix–hinge–helix structure, and the N-terminal helix affects its antibacterial and anticancer activities by accelerating the breakdown of the membrane structure.[36−38] In contrast, the corresponding glycine, Gly21, as a part of the helix in CP1, does not show any structural variability in DPC micelles, which is consistent with previous results using HFIP.[7] A Pro is prevalent behind Gly in the hinge region, which is highly conserved in cecropins A, B, and D. However, CP1 lacks this Pro, which may explain why CP1 has a continuous helix structure without a hinge area. Our results using DPC micelles do not exclude the existence of a potentially mobile region due to the presence of Glu20–Gly21 in CP1, which was suggested as a possibility in previous studies using HFIP.[7] However, these results are useful for evaluating the relationship between the structure and function of typical cecropin family peptides and CP1. Combined with phylogenetic analysis, CP1–4 seems to be in a separate cluster compared with other AMPs from the cecropin family.[39]

We have reported the structure of CP1 in the lipopolysaccharide (LPS) micelle showing completely different structural features compared with the results in this study using DPC micelles, interestingly. LPS is a component of the outer membrane of Gram-negative bacteria with negative charges, building a hydrophobic bilayer with a thickness of 22 Å.[40] The gyration radius of the LPS micelle is considerably larger, 105 nm, when the LPS concentration is higher than the apparent critical micelle concentration.[41] The results of analysis using Tr-NOE showed that CP1 in LPS formed a hydrophobic α-helix from residues Lys15 to Gly29 alone, whereas the N-terminal side did not show a convergent conformation.[21] Taken together with the helix results at full length in this study, the helix formation only at the C-terminal side may be important only for interactions with the outer membrane. The lack of helix structure formation at the N-terminal side may be important for interactions with flexible glycans present in the outer membrane and for functions such as the need to interact with the inner membrane without strong interaction with the outer membrane.

NMR Analysis of the Terminal Region of CP1

To further investigate the structure of CP1 and its interaction with the membrane, we focused on the N- and C-termini of the long helix structure. Trp2, located on the N-terminal side of CP1, is highly conserved in the cecropin family and is thought to play a pivotal role in membrane interactions. Reportedly, mutation of Trp2 results in a significant loss of cecropin activity.[42] In contrast, there are few reports on the role of residues on the C-terminal side of CP1. Interestingly, even though the toxicity of full-length CP1 was not suppressed in the Trx fusion expression system, truncation from the C-terminal side of CP1 greatly increased the growth curves (Figure S10). Recombinant CP1 (1–30) with the C-terminal residue removed considerably eliminated growth inhibition of host E. coli in which one C-terminal residue was removed, and full growth recovery and large-scale expression of the fusion protein were observed in CP1 (1–29). Therefore, we prepared large-scale preparations of CP1 (1–29) for further NMR studies. The CP1 (1–29) structure in solution in DPC micelles was analyzed using NMR. Details are shown in Tables S1B and S2. The structure has been deposited in the PDB with the code 7VOZ (Figure A,B). The chemical shift information was deposited in the BMRB with the code 36449. CP1 (1–29) had an essentially similar α-helical structure as CP1 (WT), except for the deleted C-terminus. Heteronuclear steady-state {1H}–15N NOE values explained that the helix segments of CP1 (1–29) were constant before Gly21 and CP1 (WT). The values began to decrease after Gly21, which meant that the flexibility started to increase. The C-terminal residues like Gly29 and Arg31 of CP1 (WT) and Gly28 and Gly29 of CP1 (1–29) were highly mobile (Figure C). It has been argued that the disruption of lipid membranes due to increased peptide flexibility is responsible for the disruption of membrane structure, and there may be no significant difference between CP1 (WT) and CP1 (1–29) in this regard.[43]

Figure 4

(A) 1H and 15N HSQC NMR spectrum of CP1 (1–29) in DPC with assignments. W2Nε, N12Nδ, R17Nε, and Q27Nε are peaks from side chains. (B) NMR structures of the main chain of CP1 (1–29) in the DPC micelle: overlay of the ensemble of 20 final energy-minimized CYANA structures. The left-hand side is the N-terminus. (C) Comparison of the heteronuclear steady-state {1H}–15N nuclear Overhauser effect (NOE) values for CP1 (WT) (red) and CP1 (1–29) (blue). The error bars are shown as a signal-to-noise value.

To compare the interaction modes of CP1 (WT) and CP1 (1–29) with DPC micelles via NMR, the interaction between CP1 in micelles and surrounding water molecules was investigated by examining the residues that cause cross-saturation when selective irradiation of water is used.[44] In CP1 (WT), the main chain amide of Trp2 at the N-terminus and the side chain of Arg31 at the C-terminus showed significant peak attenuation to below 50%, probably due to the interaction with water molecules (Figure A). Our structural calculation results by NMR data showed that the average length of CP1 (from Ser1 to Arg31) in the DPC micelle was 49.4 Å. DPC micelles are not regular spheres, although they are reported to have 56 aggregation numbers and a mean radius of 19.5 Å.[45] This result suggests that the long α-helical structure formed by CP1 (WT) in the presence of DPC micelles may have been well covered by the micelles so that only the main chain NH of Trp2 and the side chain NH of Arg31 were exposed to water, even though they were longer than the average micelle diameter. It has been reported that the increase in the concentration of DPC can increase in aggregation number and the size of micelles will also increase.[46] Another study showed that in the presence of peptides, the aggregation number of micelles increased and the radius increased accordingly.[47] Taking these reports into account, it seems reasonable that CP1 (WT) was protected from DPC micelles for almost its entire length, except for the terminal residue. Notably, in contrast to the amide in the main chain of Trp2, which is considered to be exposed to water, the amide in the side chain did not show peak attenuation and was considered to be buried in the micelle. Trp, which acts as an anchor point at the lipid membrane interface, reportedly plays an important role in many AMPs,[15,48] including those of the cecropin family, and is thought to play a similar role in CP1. The same trend of water exposure was also observed in CP1 (1–29) for the residues that formed a long helix with Trp2 (Figure B). The average length of CP1 (1–29) (from Ser1 to Gly29) in the DPC micelle was 40.6 Å. On the C-terminal side, where two residues were deleted, no resonance decay was observed, unlike in CP1 (WT), indicating that the whole peptide except the N-terminus was buried in the DPC micelle. Due to its hinge structure, cecropin A does not penetrate the membrane and is distributed on the surface of the membrane, though it has 37 amino acids with a longer helix.[43] Some amino acids in the middle part of cecropin A were exposed to water. Considering the results of this experiment, it is expected that, unlike other cecropins, CP1 interacts with DPC micelles in a covered form.

Figure 5

The solvent-exposed area verification experiment for (A) CP1 (WT) and (B) CP1 (1–29). The y-axis shows the ratio of 1H–15N HSQC peak intensity of each amino acid residue at pulse on 4.7 ppm to pulse on −5.3 ppm. W2s, R17s, R31s, and R31’s represent the side chains. R31’ means the minor component of R31.

Comparison of the Activity of CP1 (WT) and CP1 (1–29)

To investigate the similarities and differences between CP1 (WT) and CP1 (1–29) observed in the NMR profiles for micelle interactions that can affect the activity, we measured the antimicrobial activity of CP1 and some commercially available antibiotics. The results of the minimum bactericidal concentration (MBC) assay showed that CP1 (1–29) was slightly weaker than CP1 (WT) but had almost the same activity (Table ). Especially at lower concentrations, CP1 (WT) showed stronger antibacterial activity (Figure S11). The activity of CP1 (1–29) and CP1 (WT) was stronger against Gram-negative bacteria than Gram-positive bacteria, which is in agreement with previous reports.[49] This MBC assay is more suitable for the antibacterial agents targeting the cell membrane, so ampicillin did not have a good antibacterial effect. Nisin, another antimicrobial peptide-type antibiotic, also has the same antibacterial mechanism by destroying the bacterial membrane,[50] and its antibacterial activity was not as good as CP1, especially against Gram-negative bacteria. These results indicated that Pro30 and Arg31 of the C-terminus of CP1 had a limited effect on the antibacterial activity of CP1. This result is consistent with the results of NMR analysis, which showed no obvious difference between CP1 (WT) and CP1 (1–29). This suggests that the difference in toxicity in overexpression by Trx fusion between CP1 (WT) and CP1 (1–29) may be due to other factors.

Table 1

MBCs of CP1 (WT), CP1 (1–29), Ampicillin, and Nisin against Selected Bacteria

microorganism	CP1 (WT) (μM)	CP1 (1–29) (μM)	ampicillin (μM)	nisin (μM)
Listeria innocua	2.4	2.4	>64,000	8
E. coli (BL21)	0.025	0.05	>64,000	8000
E. coli (ML35)	0.2	0.2	>64,000	4000

The DNA-binding ability of AMPs is a mechanism by which AMPs exhibit antimicrobial activity after the permeabilization of the cell membrane.[51] We tested the DNA-binding ability of CP1 (WT) and CP1 (1–29) by a DNA electromobility shift assay using DNA incubated with peptides. CP1 (WT) showed stronger DNA-binding ability than CP1 (1–29) (Figure A). CP1 (WT) showed a clear DNA-binding-induced band shift starting at a peptide-to-DNA molar ratio of 64,000:1. In contrast, CP1 (1–29) showed an effect on the band only after reaching a fourfold higher concentration, 64,000:1. These results demonstrate the importance of charge strength for the DNA-binding ability of AMPs. The DNA-binding ability of AMPs varies greatly depending on their type.[52,53] For example, even within helix-forming AMPs, it has been reported that magainin 2 and LL-37 of the cathelicidin family have distinct DNA-binding ability that plays to their antimicrobial activity, whereas cecropin A shows weaker DNA-binding ability.[54] The DNA-binding capacity of CP1 was slightly lower than that of magainin 2 (Figure B). To the best of our knowledge, this is the first report showing that CP1 has distinct DNA-binding ability. The DNA-binding ability of CP1 provides new insight into the mechanism of CP1 activity.

Figure 6

DNA-binding affinity of CP1 (WT) and CP1 (1–29). (A) Agarose electrophoresis of DNA was incubated with different amounts of CP1 (WT) and CP1 (1–29) at 25 °C. Lanes 1–5 are CP1 (WT). The molar ratios of the peptide to DNA from left to right are 64,000:1, 32,000:1, 16,000:1, 8000:1, and 4000:1. Lanes 6–10 are CP1 (1–29). The molar ratios from left to right are 64,000:1, 32,000:1, 16,000:1, 8000:1, and 4000:1. Lane 11 is only DNA as a control. (B) Agarose electrophoresis of DNA incubated with different amounts of magainin 2 and CP1 (WT) at 25 °C. Lanes 1–3 are magainin 2. The molar ratios from left to right are 32,000:1, 16,000:1, and 8000:1. Lanes 4–6 are CP1 (WT). The molar ratios from left to right are 32,000:1, 16,000:1, and 8000:1. Lane 7 is only DNA as a control. (C) Agarose electrophoresis of DNA incubated with different amounts of CaM-CP1 and CP1 (WT) at 25 °C without EDTA. Lanes 1–5 are CaM-CP1. The molar ratios from left to right are 64,000:1, 32,000:1, 16,000:1, 8000:1, and 4000:1. Lane 6 is CP1 (WT). The molar ratio is 64,000:1. Lane 7 is only DNA as a control.

The DNA-binding ability to meditate the toxicity of CP1 (WT) was completely repressed in the CaM fusion expression system (Figure C), which may result in the remarkable difference in expression levels observed. At the same molar ratio of peptide to DNA, CP1 showed obvious DNA-binding ability, but DNA migration was not affected by the presence of CaM-CP1 even at high concentrations. These results suggest that the CaM fusion expression system, which can protect AMPs in an enveloping manner, is effective in preventing the toxicity during recombinant AMP expression.

Conclusions

We designed two expression systems for the overexpression of CP1 with CaM and Trx fusion proteins. Using a simple purification procedure, we obtained approximately a hundred times greater amount of CP1 than the Trx overexpression system. The formation of the α-helical structure of CP1 in DPC micelles and their interactions with membranes was confirmed by NMR. NMR analysis suggested that CP1 was covered by DPC micelles for almost its entire length and formed an α-helix, suggesting that it has a different form of interaction with the membrane than other cecropins. Noting the lack of significant differences in the mode of interaction of CP1 and its analogues with DPC, we found that CP1 has a strong DNA-binding affinity, which may contribute to its antimicrobial activity. The DNA-binding capacity of CP1 was sufficiently suppressed by the CaM fusion, suggesting that the action of CaM has led to increased overexpression of the recombinant fusion protein by suppressing the toxicity of CP1 in the host cell. In this study, there are many unknowns regarding the mode of binding of CP1 to DNA, and further clarification is expected. Our results demonstrate the DNA-binding capacity of CP1, which provides a new direction for the study of the mechanism of CP1 antimicrobial activity.

Methods

Protein Expression and Purification

The pET15b CaM-EK vector used in our experiments was based on a previous study that used the pET15b CaM-TEV vector.[22] In our experiments, the TEV cleavage site in the original plasmid was replaced with the EK cleavage site via inverse PCR, and the CP1 gene was introduced simultaneously (Figure S1). Competent E. coli BL21 (DE3) cells (BioDynamics Laboratory Inc., Japan) were transformed with the pET15b-CaM-CP1 construct. To compare the overexpression of CP1 using this system, fusion proteins pET32b-CP1 and pET32b-CP1 (C-terminal truncated analogues) were also constructed. pET32b is commercially available and already contains the Trx fusion tag. As a control, we designed plasmids with fusion proteins without any target AMPs and target CP1 without any fusion proteins. The details of the various constructs are listed in the Supporting Information (Figure S2).

Transformed E. coli BL21 (DE3) cells were grown in Luria–Bertani (LB) medium containing 50 mg/L ampicillin (FUJIFILM Wako Pure Chemical Corporation, Japan) at 37 °C. To obtain 13C- and 15N-labeled CP1, fusion CaM-CP1 protein was produced using minimal M9 media with 2 g/L 13C-glucose and 2 g/L 15NH4Cl. When the OD600 value reached 0.6, the fusion protein was induced with 1.0 mM IPTG for 4 h at 37 °C. The cells were harvested by centrifugation at 6000 rpm for 10 min at 4 °C. The precipitate was resuspended in lysis buffer (20 mM Tris–HCl, 150 mM NaCl, pH 8.0). After sonication (Insonatoe201M, Kubota, Japan) and centrifugation at 7500 rpm (6700 × g) for 20 min (rotor: TOMY TLA-11, Japan), the supernatant was passed through a filter (Millex-HV 0.45 μm, Merck, USA). All sample solutions were applied onto a Ni-NTA column (GE Company, USA) equilibrated with IMAC washing buffer (20 mM Tris–HCl, 300 mM NaCl, pH 8.0). The column was washed with 5× column volumes of IMAC washing buffer. The target protein was eluted using 5× column volumes of IMAC elution buffer (20 mM Tris–HCl, 300 mM imidazole, 150 mM NaCl, pH 8.0). All fractions were analyzed at an absorbance of A280 and using Coomassie Brilliant Blue staining after tricine–sodium dodecyl sulfate-polyacrylamide gel electrophoresis (tricine–SDS-PAGE). The eluate was dialyzed with dialysis buffer (20 mM Tris–HCl, pH 8.0) for 20 h at 25 °C. The recovered sample was incubated with EK digestion buffer (50 mM Tris–HCl, 1 mM CaCl2, 0.1% Tween-20, pH 8.0) and 1 U/mL EK protease (Thermo Fisher Scientific) at 25 °C. After 3 h, the digested solution was passed through a filter (Millex-HV 0.22 μm, Merck, USA) and mixed with trifluoroacetic acid (TFA, Nacalai Tesque Inc., Japan) to result in a pH of 2.0–3.0 for RP-HPLC. CP1 was purified using a Cosmosil 5C18AR-300 column (Nacalai Tesque Inc., Japan). The fractions were eluted using a gradient from solution A (0.1% TFA in filtered water) to solution B (0.1% TFA in HPLC-grade acetonitrile). The peak fractions containing CP1 were collected and lyophilized. The contents of each fraction were evaluated using 15% tricine–SDS-PAGE. The purity and molecular weight of CP1 were confirmed by MALDI-TOF MS (Bruker Autoflex, USA).

NMR Analysis

For NMR analysis, 13C- and 15N-labeled CP1 (WT) and CP1 (1–29) were expressed using the CaM fusion system in M9 minimal media containing 2 g of 13C6-glucose and 2 g of 15NH4Cl. Details of M9 minimal media are shown in Table S3. After purification as the same protocols of the unlabeled sample, double-labeled CP1 (WT) and CP1 (1–29) were dissolved in 10% D2O/90% H2O at a final concentration of 1 mM at pH 5.0. To create a simulated cell membrane environment, 40 mM DPC (Cambridge Isotope Laboratories, Inc., USA) micelles were mixed with the labeled sample. All NMR experiments were performed at 25 °C on an Agilent Unity INOVA 600 MHz spectrometer equipped with a TR5 probe with a single-axis z-gradient, a Bruker AVANCE Neo 800 MHz spectrometer equipped with a 5 mm TCI cryoprobe with a single-axis z-gradient, and a Bruker AVANCE III HD 600 MHz spectrometer equipped with a 5 mm TXI probe with the x, y, z gradient. The main chain assignments were obtained using the following spectra: [1H–15N] HSQC, [1H–13C] HSQC, HNCA, HN(CO)CA, HNCACB, and HCCH-TOCSY. Heteronuclear [1H–15N] and [1H–13C] nuclear Overhauser effect (NOE) dynamics data were used for structural calculations. The experimental details are shown in Table S2. The 2D and 3D NMR spectra were processed using NMRPipe and analyzed using the Sparky program. NOE distance constraints were automatically assigned and calculated using seven cycles under the “noeassign” macro of the CYANA 2.1 software package. The following conditions were used for the solvent-exposed area verification experiment using the cross-saturation method: The saturation pulse under these conditions was used to irradiate the pulse to water (4.7 ppm), and the pulse-off condition was used to irradiate the resonance-free region (−5.3 ppm). The recovery time to equilibrium was set to 5 s, and the saturation pulse was irradiated for 2 s. The selective saturation pulse was adjusted to a width of 1 ppm using a chirp-shaped pulse. The ratio of intensity with a pulse on 4.7 ppm relative to the intensity with a pulse on −5.3 ppm was used to obtain the interaction between each proton and water molecule.

Minimum Bactericidal Concentration (MBC)

To analyze whether the CP1 (WT) and CP1 (1–29) produced by the expression system showed antibacterial activity, the Gram-positive strain Listeria innocua was cultured in brain heart infusion (BHI) media, and the Gram-negative strains E. coli BL21 and E. coli ML35 were cultured in tryptic soy broth (TSB) media at 37 °C. When the OD600 value reached 0.4, the cultured bacteria were washed with 10 mM phosphate buffer (pH 7.4) and the number of colonies was adjusted to 1 × 107 CFU/mL. The treated solution of bacteria was mixed with peptides using a gradient concentration, and the mix was coated on BHI and TSB agar plates. After incubation at 37 °C for 18 h, the number of colonies on the plate was counted. As a control, commercially available antibiotic ampicillin (FUJIFILM Wako Pure Chemical Corporation, Japan) and nisin (Sigma-Aldrich, USA) were used. The MBC assay was repeated three times on different occasions.

DNA-Binding Experiments

The experimental methods refer to previous experiments.[52,51] The cyclic plasmid DNA pET16 was diluted to a concentration of 16 nM in the reaction buffer (10 mM Tris–HCl, 1 mM EDTA, pH 8.0). Magainin 2 (Sigma-Aldrich, USA) was used as a control. The peptide was diluted in the reaction buffer. The peptide and DNA were mixed in the corresponding molar ratio, and the volume was adjusted to 20 μL with the reaction buffer. The mixture was incubated for 60 min at 25 °C. The results were analyzed using 1% agarose gel electrophoresis. The DNA bands were stained with ethidium bromide solution and observed under UV light. The presence of EDTA would change the structure of CaM,[55] so EDTA was removed from all reagents and buffers involved in the experiment when DNA-binding experiments of CaM-CP1 were performed.