Interspecies Bombolitins Exhibit Structural Diversity upon Membrane Binding, Leading to Cell Specificity

Abstract

Bombolitins, a class of peptides produced by bees of the genus Bombus, target and disrupt cellular membranes, leading to lysis. Antimicrobial peptides exhibit various mechanisms of action resulting from the interplay between peptide structure, lipid composition, and cellular target membrane selectivity. Herein, two bombolitins displaying significant amino-acid-sequence similarity, BII and BL6, were assessed for antimicrobial activity as well as correlated dodecylphosphocholine (DPC) micelle binding and membrane-induced peptide conformational changes. Infrared and circular dichroism spectroscopies were used to assess the structure-function relationship of each bombolitin, and the results indicate that BII forms a rigid and helically ordered secondary structure upon binding to DPC micelles, whereas BL6 largely lacks secondary structural order. Moreover, the binding affinity of each peptide to DPC micelles was determined, revealing that BL6 displayed a difference in binding affinity by over two orders of magnitude. Further investigations into the growth-inhibitory activity of the two bombolitins were performed against Escherichia coli and Saccharomyces cerevisiae. Interestingly, BII specifically targeted S. cerevisiae, whereas BL6 more effectively inhibited E. coli growth. Overall, the antimicrobial selectivity and specificity of BII and BL6 are largely dependent on the primary as well as secondary structural content of the peptides and the membrane composition.

Introduction

Insect-derived antimicrobial peptides (AMPs) are short (15–40 amino acid) amphiphilic peptides that act as defensive molecules, and their interactions with certain cellular membranes are composition dependent (1, 2, 3, 4). In aqueous solution, most AMPs are primarily intrinsically disordered, but upon target membrane interaction, secondary structure conformational changes can occur that lead to membrane binding (5, 6, 7). The primary AMP mode of action is postulated to involve the formation of membrane pores, which results in ion leakage and subsequent cell lysis (1, 8). Although many experimental and theoretical studies have been able to shed insight into the kinetics associated with binding and folding for select AMPs, fine mechanistic details governing peptide-membrane association and secondary-structure-lytic activity are still of interest to the broader AMP community. For instance, the role of folding and the extent of the secondary structural content in the binding and insertion processes are still somewhat ambiguous. Some experiments have suggested sequential steps of folding, binding, and insertion, whereas other studies have demonstrated more of a cooperative interplay between the two phenomena (9, 10, 11, 12, 13). AMPs are found in all multicellular species and exhibit a wide spectrum of functionality, including antibacterial, antifungal, antiviral, and antitumorigenic effects (6, 8, 14, 15, 16, 17). Therefore, there is substantial interest in developing AMPs as novel therapeutics for the treatment of a variety of diseases.

Although several AMP pore formation mechanisms have been proposed, such as the barrel stave, carpet, toroidal pore, and sinking raft, sufficient understanding of this process is lacking (8, 14, 18, 19, 20, 21, 22, 23, 24). Each of these proposed mechanisms suggest that AMPs first undergo a conformational transition from disordered to ordered in the presence of membrane environments that triggers target membrane association and integration, leading to cooperative peptide aggregation that results in the formation of a membrane pore (19, 21, 25, 26, 27, 28, 29, 30). However, the details pertaining to AMP amino acid side-chain participation in folding and membrane insertion, the temporal nature of these events, and how target membrane specificity arises remain unclear. Furthermore, even less is known about AMP dynamics in living cells, where the mechanism of in vivo AMP pore formation differs greatly from investigations in vitro in key aspects, such as stability and lifetime of pore-forming events (20, 21).

Bombolitins are bumblebee-derived AMPs that are similar to the well-studied honeybee (Apis melifera) AMP melittin. Over 250 bumblebee species have been identified, but only a handful of bombolitin analogs have been discovered and investigated in detail. Previous work has demonstrated that Megabombus pennsylvanicus produces a cocktail of at least five unique bombolitin AMPs, and various biophysical studies have demonstrated that these AMPs interact with synthetic membrane systems to form α-helical structures that integrate into lipid bilayers (31, 32, 33, 34, 35, 36). A survey of bombolitin peptide sequences in the AMP Database (http://aps.unmc.edu/AP/main.php) revealed that different Bombus species produce different AMP cocktails, despite the fact that the primary sequence of these peptides retains a high degree of amino acid conservation. This sequence diversity may arise from evolutionary or ecological pressures that lead to bombolitin functional diversity, but the specific antimicrobial and structural properties of these bombolitin sequences has not been investigated in detail.

In this study, two bombolitins, BII from M. pennsylvanicus and BL6 from Bombus lapidarius, were chemically synthesized to investigate the structural and functional differences between these AMPs isolated from geographically distinct species. Our results demonstrate that these peptides assume a relatively disordered conformation in solution. However, in the presence of membrane mimics, they take on unique secondary structure conformations that result in both differential binding affinities to dodecylphosphocholine (DPC) micelles as well as distinct and specific antimicrobial properties. This study sheds light on the structural properties that contribute to AMP functional diversity.

Materials and Methods

Peptide synthesis and purification

Trifluoroacetic acid, phenol, thioanisole, piperidine, and 1,2-ethanedithiol were purchased from Sigma-Aldrich (St. Louis, MO). Fluorenylmethyloxycarbonyl-protected amino acids, Knorr resin, N,N-diisopropylethylamine, and 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate were purchased from Creosalus (Louisville, KY). Triisoproylysilane was obtained from Acros Organics (Waltham, MA). Ultrapure water was used for all samples. High-performance-liquid-chromatography-grade acetonitrile was purchased from Fischer Scientific (Hampton, NH).

All peptides were synthesized using an Aapptec Focus XC solid-phase peptide synthesizer (Louisville, KY), which utilizes standard fluorenylmethyloxycarbonyl solid-phase peptide chemistry. Peptides were cleaved before reverse-phase high-performance liquid chromatography purification. Purified peptides were validated by mass spectrometry using a Bruker (Billerica, MA) Microflex matrix-assisted laser desorption ionization time of flight mass spectrometer (see Supporting Materials and Methods for more details).

Spectroscopic methods

Peptide concentrations were determined using the extinction coefficient of tryptophan at 280 nm (5690 cm⁻¹ × M⁻¹) via ultraviolet–visible spectroscopy (λ 25; Perkin Elmer, Waltham, MA). Fluorescence spectra were obtained at room temperature using a Fluor-max 3 fluorimeter (Horiba Scientific, Kyoto, Japan). Samples were excited at 280 nm with emission and excitation slits set to 5 nm.

Circular dichroism (CD) spectral analysis was performed using an Aviv Biomedical (Lakewood, NJ) CD spectrometer model 202-01. Serial dilutions were performed using the same method described for the fluorescence binding curves. Fourier transform infrared (FTIR) spectra (Nicolet 6700; Thermo Scientific, Waltham, MA) of the short model peptides were collected with 1 cm⁻¹ resolution and a nitrogen-cooled mercury cadmium telluride detector. The optical densities for the samples were between 0.05 and 0.20 (see Supporting Materials and Methods for more details).

Automated microbial growth analyses

Saccharomyces cerevisiae 288c and Escherichia coli DH5α strains were propagated on solid yeast peptone dextrose (1% (w/v) yeast extract, 2% (w/v) peptone, 1% (w/v) glucose) and Luria broth media, respectively. Single S. cerevisiae colonies were used to inoculate 10 mL yeast peptone dextrose cultures, which were grown for 16 h at 30°C. The resulting saturated culture was diluted 100-fold and separated into 1 mL aliquots. This diluted culture was treated with the indicated concentration of MBII or BL6 synthetic peptide. Diluted cultures of untreated S. cerevisiae served as positive controls (see Supporting Materials and Methods for more details). Data were analyzed in GraphPad Prism v6. Additionally, area under curve measurements of growth curves were analyzed and quantified in GraphPad Prism v6.

Results and Discussion

Bombolitin peptide model systems

AMPs possessing comparable length and charge to bombolitins have been studied previously. For example, melittin, an AMP found in honey bee venom, exhibited pore formation in phosphatidylglycerol (PG) vesicles. Melittin also undergoes a structural transition from intrinsically disordered to a primarily α-helical structure in the presence of increasing PG vesicle concentration (37, 38). Magainin, an AMP found in the Xenopus laevis epidermis, also exhibits similar structural properties to bombolitins, and this peptide shows broad antimicrobial activity as well as antitumor activity (39, 40, 41, 42, 43). Based on their peptide length, hydrophobic content, and cationic nature, we postulated that the differences in BII and BL6 primary sequences may give rise to differences in solution and bound peptide structure, membrane binding affinity, and target membrane selectivity.

BII and BL6 were chosen for several reasons, including the habitats of the specific species, the similarity of peptide sequences, and the lack of prior studies on BL6. M. pennsylvanicus (BII) and B. lapidarius (BL6) are found in the United States and Europe, respectively. Despite this geographical isolation, the analog peptide sequences are still well conserved. A total of seven amino acids are conserved between the two sequences, which corresponds to 47% amino acid identity. The primary sequence of each bombolitin is shown in Fig. 1, with conserved amino acids between the two peptides highlighted in red. The nonconserved residues are postulated to confer target membrane selectivity, which is further described below. All peptides were synthesized via solid-state peptide synthesis (see Supporting Materials and Methods for details).

Figure 1.

(Top) Structure of dodecylphosphocholine (DPC) surfactant that was utilized as a membrane mimic and antimicrobial peptides (AMPs) of interest, BII and BL6. To see this figure in color, go online.

Determination of bombolitin binding affinities for DPC micelles

Numerous AMPs have been shown to bind membranes and undergo conformational changes leading to membrane lysis (37, 44, 45). Several binding studies have shown that melittin has a binding affinity to 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine/1-palmitoyl-2- oleoyl-sn-glycero-3-phosphatidylglycerol (POPC/POPG) vesicles with an association constant of 6 × 10⁴ M⁻¹ or a dissociation constant of ∼2 × 10⁻⁵ M, based on the model used by Seelig and co-workers (37, 45, 46).

To examine whether the primary sequence differences between BII and BL6 resulted in membrane binding differences, the binding affinities of BII and BL6 to DPC (Fig. 1) micelles were measured using fluorescence spectroscopy of intrinsic tryptophan probes, which exhibit a characteristic increase in emission signal after binding to the synthetic membrane (47, 48). The tryptophan residue in the native BL6 sequence was exploited for these measurements, but BII lacked this residue and therefore required N-terminal tryptophan addition. This tryptophan-containing variant of BII was referred as modified BII (MBII).

Fluorescence spectra of both MBII and BL6 in the presence and absence of DPC micelles indicated a blue shift in the emission maximum and an increase in the quantum yield of the tryptophan probe fluorescence in the presence of DPC micelles (Fig. 2). This result indicates that the peptides are initially bound to the micelles with their tryptophan residues sequestered into the micelle membrane. As the micelle concentration is decreased via serial dilutions with a stock peptide solution, fewer peptides become bound to the micelles, and the quenching effects of water become apparent as the equilibrium (Eq. 1) characteristically shifts such that fewer peptides are bound to the micelles.

Figure 2.

Fluorescence binding curves of BL6 (45 μM) (green dots, top) and MBII (35 μM) (blue dots, bottom) were plotted as a function of DPC concentration and fitted to the fractional binding model (purple and red dashed lines, respectively). (Inset) Normalized fluorescence spectra of 30 μM MBII with varying concentrations of DPC (0–7 mM) are shown. The data presented are representative of three independent trials. To see this figure in color, go online.

To quantify the binding affinity for each peptide, a two-state model is assumed in which the peptide, P, is either bound or unbound to the micelles, M, formed by the surfactant, S_i, which has an equilibrium constant of K_agg (49).

$$\mathrm{P}+\mathrm{M}\begin{array}{c}{\text{k}}_1\\ \mathrm{\leftrightharpoons }\\ {\text{k}}_2\end{array}\mathrm{PM}$$ (1)

$${\mathbf{S}}_{\mathbf{i}-1}+{\mathbf{S}}_1\stackrel{{\text{K}}_{\mathbf{a}\mathbf{g}\mathbf{g}}}{\mathrm{\leftrightharpoons }}\text{M}$$ (2)

The peptide/micelle ratio was initially 1:180, which is indicative of the ratio expected for a fully bound population. Initially, both peptides were fully bound based on the blue-shifted emission maximum and increased quantum yield. Serial dilutions were performed on peptide solutions bound to DPC by diluting with a stock solution of equal peptide concentration. As serial dilutions are performed, the DPC concentration decreases, driving the equilibrium of Eq. 1 toward more unbound peptide in solution. As a result, the tryptophan fluorescence emission maximum redshifts, and the intensity decreases until the peptide is fully unbound (red spectrum shown in insert of Fig. 2). Utilizing the fractional binding method (50), described in more detail in the Supporting Materials and Methods, binding curves were generated by plotting the fraction of peptide bound (using Eq. S2) versus the DPC concentration (Fig. 2). Briefly, using the two-state model consisting of bound (lower wavelength maximum: λ_BL6 = 341 nm and λ_MBII = 340 nm) and unbound (higher wavelength maximum: λ_BL6 = 357 nm and λ_MBII = 352 nm), the fraction of bound peptide was found at discrete concentrations of DPC (Fig. 2 dots). Using Eq. 3, a least-squares fitting analysis was performed to reproduce the binding curve (Fig. 2 dashed lines) in which F, the fraction of bound peptide, and K_d were determined. The peptide concentration and surfactant concentrations—[P] and [S], respectively—were known, and γ is a correction factor (full details in the Supporting Materials and Methods). The lowest concentration of the binding curve was measured at ∼1.1 mM, which corresponds to the critical micelle concentration of DPC (51, 52, 53, 54) and signifies the maximal amount of unbound peptide and/or the release of the bound peptide.

$${\mathit{\gamma }}^2{\mathit{F}}^2{\left[\mathit{P}\right]}^2-\mathit{F\gamma }\left({\left[\mathit{P}\right]}^2+\left[\mathit{P}\right]\times \left[\mathit{S}\right]+{\text{K}}_{\text{d}}\times \left[\mathit{P}\right]\right)+\left[\mathit{P}\right]\times \left[\mathit{S}\right]=0$$ (3)

Binding constants, K_d, were determined from these analyses, yielding values of (3.4 ± 1.7) × 10⁻⁶ M and (1.23 ± 0.13) × 10⁻⁴ M for BL6 and MBII, respectively. These values are similar in magnitude to that of melittin (2 × 10⁻⁵ M) toward POPC/POPG vesicles (45). Remarkably, the binding affinity of BL6 to DPC micelles was two orders of magnitude greater than the MBII-DPC micelle binding affinity. These results demonstrate that subtle changes in bombolitin peptide sequence can translate to significant membrane binding differences. Because membrane binding is a pivotal step in AMP lytic activity, these analog sequences may facilitate a better understanding of target membrane selectivity. Binding in conjunction with any binding-induced secondary structural changes could also reveal a more detailed picture of bombolitin selectivity and activity.

MBII and BL6 exhibit different α-helical contents upon interacting with membranes

Because of the short lengths of many AMPs, the Gibb’s energy of these peptides in an aqueous environment is typically dominated by the entropic contributions rather than by intramolecular enthalpic contributions (i.e., in bulk water, many AMPs are intrinsically disordered) (55, 56, 57, 58, 59, 60, 61, 62, 63). However, limiting the entropic contributions (i.e., the ability to form various conformations via hydrogen bonding with bulk water) can lead to conditions in which secondary structural changes can occur (intramolecular enthalpy dominates) or even aggregation (intermolecular enthalpy dominates). These environmental conditions are expected upon peptide binding with a membrane environment either at the surface or upon insertion where the number of water molecules are somewhat limited (25).

To examine the nature of the putative structural changes taking place upon membrane binding, the secondary structures of MBII and BL6 were evaluated in the presence and absence of DPC micelles using CD spectroscopy. CD utilizes circularly polarized light to monitor the electronic transitions associated with the carbonyls of the peptide backbone (64, 65, 66, 67). Using a similar strategy as the binding experiments, serial dilutions of DPC micelles were performed while maintaining a constant peptide concentration, thus steadily decreasing the surfactant concentration. CD spectra (Fig. S1) were then collected. Using the mean residue ellipticity at 222 nm, the fractional helicity, F_h, was monitored as a function of DPC concentration (Fig. S2) via Baldwin’s method (Eq. 4) (68, 69, 70). Here, [θ]_222 nm is the mean residue ellipticity at 222 nm, and n is the number of residues within the peptide. When fully bound to DPC micelles, MBII and BL6 exhibited 78 and 60% helical content, respectively, as depicted in Fig. 3 (dotted line). As a reference, other bombolitins, BI and BIII, have previously been shown to exhibit between 60 and 70% α-helical content in the presence of sodium dodecyl sulfate micelles (32). It should be noted that the fractional helicities of the peptides matched the maximal fractional helicities of each peptide in 2,2,2-trifluoroethanol (see Fig. S3). Serial dilutions of both peptides resulted in a decrease in helical content, as noted by the decrease in the 190 nm band and increase in the 222 nm band as seen in Fig. S1).

$${\mathrm{F}}_{\mathrm{h}}=\frac{{\left[\mathit{\theta }\right]}_{222\mathrm{nm}}}{\left(-44000\ast \left(1-\frac{2.5}{\mathrm{n}}\right)\right)}$$ (4)

Figure 3.

CD spectra of the fully folded MBII (red, top) and BL6 (blue, bottom) were obtained. The concentration of BL6 and MBII was 170 and 120 μM, respectively. To see this figure in color, go online.

To ascertain the subtle differences between the two actions, binding and folding, folding curves were first generated using the mean residue ellipticity, θ, at 222 nm obtained from CD measurements and then compared to the binding curves obtained via fluorescent measurements, previously presented in Fig. 2. The raw CD data were treated with Eq. 5 to generate the blue dots of Fig. 4, which represents the fraction of peptide folded as a function of DPC concentration. Here, the fraction of folded peptide, α, is simply the ratio of the concentration of folded peptide, [F], to the total peptide concentration, [U] + [F] (71). The fully unfolded peptide would have a value of 0, and the maximal folded peptide would have a value of 1 (see Supporting Materials and Methods for more detail).

$$\mathit{\alpha }=\frac{\left[\mathrm{F}\right]}{\left[\mathrm{U}\right]+\left[\mathrm{F}\right]}=\frac{{\mathit{\theta }}_{\mathrm{obs}}-{\mathit{\theta }}_{\mathrm{U}}}{{\mathit{\theta }}_{\mathrm{F}}-{\mathit{\theta }}_{\mathrm{U}}}$$ (5)

Figure 4.

Comparisons between the fraction of folded peptide, α (blue), observed via CD were made against the binding curve data (red) obtained via fluorescence for both (A) BL6 (170 μM) and (B) MBII (120 μM) as a function of DPC concentration. The data presented are the average of three independent trials with the SD of the mesurements represented as vertical error bars.

These results were directly compared to the fractional binding curves (Fig. 4 red dots) that were generated with Eq. S2. These experiments revealed that the initial change in secondary structure (Fig. 4 blue dots) corresponded well with the critical micelle concentration of DPC (1.1 mM), further indicating that the presence of micelles was necessary to prompt changes in the peptides’ secondary structure.

The curvatures of the MBII binding and folding curves were nearly identical, indicating that these processes were interdependent. However, the BL6 binding and folding curves were much less similar, suggesting that 1) binding is less dependent on the presence of the fully folded conformation of the peptide and 2) there may exist more underlying intermediate secondary structures associated with BL6 than MBII (i.e., BL6 is likely more flexible in structure than MBII). For scenario 2, because of BL6 being potentially more flexible, one would assume that there exist more local minima (stable intermediates) for BL6 in the energy landscape associated with peptide binding and/or induced secondary structural changes. Moreover, it is likely that the distribution of states varies with the amount of bound peptide because of less crowding, allowing larger sampling of the conformation space. Overall, these findings further support that partial folded conformations are capable of binding to the micelles as suggested by several theoretical studies performed on folding dynamics of similar types of peptides (12, 13).

FTIR spectroscopy reveals MBII and BL6 undergo distinct secondary structural changes upon membrane binding

Although CD is capable of determining global secondary structure, deconvolution of CD spectra to reveal underlying structures is not always trivial. However, FTIR spectroscopy can be used to assess underlying structural changes and offer complementary information to that of CD. The secondary structures of various protein and peptides systems have been evaluated by previous groups using linear and nonlinear infrared (IR) techniques (57, 58, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93).

For further analysis of whether differences in DPC micelle binding affinities were due to differential peptide secondary structural changes induced by peptide binding, we examined the secondary structure of MBII and BL6 via FTIR. The amide I region (1600–1700 cm⁻¹) of these peptides was observed in the presence and absence of DPC micelles (Fig. 5, A and B). The major vibrational modes of this region correspond primarily to the stretching modes of the carbonyls of the peptide backbone (86, 88, 90, 94, 95, 96, 97, 98, 99, 100, 101, 102). Analysis of the MBII and BL6 amide I region revealed that in the presence of DPC micelles, there is no significant shift in the maximal optical density from 1645 cm⁻¹. Rather, band narrowing is observed in the FTIR spectra (Fig. 5, A and B) of both peptides. This band narrowing can be attributed to two phenomena, the first of which is a secondary structural change in which the peptide transitions from an intrinsically disordered conformation to an α-helix, a local ordering of the excitonic band (as suggested by CD). In the α-helical conformation, the peptide would be expected to have fewer vibrational conformations, thus resulting in band narrowing. The second contributing phenomenon to the band narrowing is a dehydration effect of the peptide backbone (i.e., less inhomogeneous broadening). The bandwidth of BL6 and MBII in D₂O were similar: 19.6 and 17.9 cm⁻¹, respectively. Upon introduction of the DPC micelles, the MBII bandwidth exhibits a 16.2% change in bandwidth. Similarly, BL6 exhibits 18.4%. These differences in bandwidth upon binding are similar; however, BL6 exhibits a remarkably greater increase in optical density at lower frequencies, suggesting that the carbonyl backbone of BL6 (Fig. 5 B) is more exposed to water molecules than MBII when bound and therefore potentially possesses more conformations or is more flexible (58, 94, 103, 104, 105).

Figure 5.

(A) FTIR spectra of amide I region of 5 mM MBII and (B) 2 mM BL6 in the absence (blue) and presence (red) of DPC micelles. The DPC concentration ranged between 140 and 160 mM. FTIR difference spectra of (C) MBII and (D) BL6 generated by subtracting the spectra of the peptide without DPC from peptide with DPC are shown. To see this figure in color, go online.

To further examine secondary structural changes, IR difference spectra in the amide I were calculated by subtracting the peptide-D₂O spectra from the peptide-DPC spectra (Fig. 5, C and D). The resulting difference spectra show positive bands corresponding to features that appear as a result of structural changes induced by the presence of DPC micelles. The difference spectrum of MBII (Fig. 5 C) reveals a narrow positive band centered at 1645 cm⁻¹, whereas a much a broader positive band located at 1643 cm⁻¹ was observed for BL6. This significant difference in the bandwidth uncovered here further supports a much larger distribution of secondary structures for BL6, whereas MBII maintains a rigid structure with less variability (102). Another positive band is present at 1591 cm⁻¹ for MBII in micelles, likely attributed to the carboxylate species of aspartic acid (106, 107, 108).

The negative transition located at 1667 cm⁻¹ of the difference IR spectra reveals the presence of random disordered structural distributions associated with both peptides in the absence of DPC (88, 90, 102). Upon incorporation of micelles, a drastic decrease occurs in this band for MBII, whereas this change was much less significant for BL6, further suggesting that BL6 retains a greater proportion of disordered structure than MBII in the presence of DPC. Analysis of the peptide primary sequences reveals that 1) BL6 contains more polar residues and 2) BL6 possesses two glycine residues in close proximity to each other, positioned near the center of the primary sequence, whereas MBII only has one glycine. The greater number of polar residues could induce a greater affinity for interactions with water molecules. Furthermore, because of glycine’s lack of a side chain, glycine can induce helix breaking (109).

Bombolitin peptides exhibit species and target organism-specific growth inhibition activities

To investigate whether the different biophysical properties of MBII and BL6 translated to distinct biological properties, the ability of each peptide to inhibit growth of model bacterial and fungal organisms was assessed. S. cerevisiae and E. coli were chosen as model microbial organisms for these assays because they are well-studied, genetically tractable organisms, and membrane lipid composition between these species varies substantially. Automated growth curve analysis in the presence or absence of each peptide was performed to obtain a quantitative metric of microbial fitness.

MBII and BL6 were assayed for their ability to inhibit S. cerevisiae growth. Diluted cultures of S. cerevisiae S288c were prepared as described in the Materials and Methods, and these diluted cultures were pretreated with 5 μM of either MBII or BL6 synthetic peptides. These cultures were then subjected to automated growth curve analysis, with untreated S. cerevisiae cultures serving as positive controls. As demonstrated in Fig. 6 A, untreated control cells exhibited characteristic sigmoidal growth behavior, with typical lag, log, and stationary phases of growth. Both MBII and BL6 peptides substantially influenced S. cerevisiae growth rates. Cells treated with these peptides exhibited a longer lag phase and increased doubling time compared to untreated controls. Interestingly, MBII-treated cells exhibited slightly slower growth compared to BL6-treated cells. To quantitatively analyze overall growth under each condition, the area under each growth curve was measured as a proxy for total cellular growth under a particular condition. This analysis revealed that BL6 treatment resulted in a 16% total reduction in overall S. cerevisiae growth compared to untreated controls. However, MBII exhibited a 31% inhibition in overall growth, suggesting that MBII is a more potent inhibitor of S. cerevisiae growth.

Figure 6.

Automated growth curve analyses were performed on S. cerevisiae S288C (A) and E. coli DH5α (C) cultures in the presence of 5 μM MBII (blue lines) or BL6 (red lines) peptide. Untreated control growth curves are also shown (black lines). Error bars represent mean ± standard error (SE) (n = 4). These data represent four biologically replicated experiments. These data are represented for S. cerevisiae S288C (B) and E. coli DH5α (D) compared to respective untreated controls. Error bars represent mean ± SE (n = 4). Both data sets were analyzed by one-way ANOVA (^∗∗ = p < 0.01, ^∗ = p < 0.05, ns = not significant by Tukey’s post hoc analysis). Dose-response curves of E. coli DH5α (E) and S. cerevisiae S288C (F) cultures treated with the indicated concentration of MBII (blue line) or BL6 (red line) peptide are shown. Error bars represent mean ± SE (n ≥ 4).

Next, the ability of MBII and BL6 peptides to inhibit E. coli growth was examined. Automated growth curve analyses were performed using the E. coli DH5α strain, as described in the Materials and Methods. Initially, E. coli cells were treated with 5 μM of either MBII or BL6, with untreated cells serving as a positive control. After a short lag phase, untreated control cells grew normally for the 18 h course of the experiment (Fig. 6 C). Interestingly, MBII treatment had essentially no effect of cell growth at a concentration of 5 μM. However, BL6 robustly inhibited E. coli growth at this peptide concentration and effectively prevented any E. coli replication. Additionally, we performed dose-response analysis of each peptide for growth inhibition. These experiments revealed that BL6 exhibited a threefold lower IC50 (half maximal inbitory concentration) compared to MBII for E. coli growth, whereas this trend was reversed for S. cerevisiae growth inhibition. These results strongly indicate that the distinct structural properties and structural transitions exhibited by the MBII and BL6 peptides translate to distinct antimicrobial activity properties against model bacterial and fungal species.

Conclusions

Herein, we have performed preliminary studies evaluating the differences in cell selectivity and peptide secondary structure between two bombolitins that are derived from two unique Bombus species yet possess 47% similarity in their primary sequences. Our results suggest that MBII and BL6 exhibit unique structural dynamics upon their interactions with membranes and that these dynamics translate to unique antimicrobial and antifungal activity profiles. Overall, we believe that these cell selectivity and lytic activity differences arise because of the unique secondary peptide structure and specific side-chain interactions for each peptide with the membrane models. Specifically, the initial binding of the peptide to a membrane may be governed by attractive/repulsive forces between amino acid side chains and membrane headgroups. In light of this hypothesis, it is noteworthy that the plasma membrane lipid composition of E. coli and S. cerevisiae vastly differ. E. coli plasma membranes consist of 75–80% phosphatidylethanolamine, with the remainder of the bilayer lipids comprised of mono or diphosphatidylglycerol (110). In contrast, S. cerevisiae plasma membrane phospholipids are much more diverse, consisting of 17–20% phosphatidylethanolamine, phosphatidylcholine, phosphatidylserine, and other lipids (111). Therefore, the differences in antimicrobial behavior of MBII and BL6 against these model organisms coupled with their high sequence identity may be informative in future structure-function studies to dissect the amino acid residues that interact with lipid headgroups and control target membrane selectivity. These experiments may also be facilitated by the robust genetic tools that are available for both of these model organisms.

Furthermore, our results suggest that bombolitin lytic activity is governed by the overall secondary peptide structures adopted during the binding process supported by flexibility in structure of BL6, leading to a twofold increase in binding affinity compared to MBII. The BII peptide has been subjected to rigorous NMR and molecular-dynamics-based structural analyses, which suggest that this peptide forms a tilted α-helix that traverses the membrane bilayer (112, 113, 114). These results are in agreement with the biophysical measurements presented here, but the BL6 peptide has not been investigated in such detail. The flexibility in structural content between these peptides is likely due to the specific residues that differ in these analog sequences. Simple analysis of the primary structures reveals that BL6 possesses two glycines in close proximity, more nonpolar residues, and a more positive net charge. Analysis of MBII reveals a greater distribution of nonpolar residues as well as an anionic residue centered between two cationic residues (lysine), which may help stabilize the α-helix. This structural flexibility may be crucial to BL6’s ability to target E. coli and other bacterial species, suggesting that this structural property could be optimized to increase toxicity against bacterial membranes. Further analysis of the primary sequences using HemoPi (http://crdd.osdd.net/raghava/hemopi/design.php) and ToxinPred (http://crdd.osdd.net/raghava/toxinpred/) predict that MBII is more hydrophobic and has a lower net cationic charge than BL6. Additionally, the nonconserved residues seem essential for the selectively of each peptide for a different cell type. The side chains containing more hydrophobic residues and a distribution of negative and positive charge are more favorable for yeast. The side chains containing more polar residues as well as greater net cationic charge are more favorable for E. coli. Lastly, future studies will assess the differences in selectivity of these peptides among other membrane mimics, monolayers, and bilayers to further evaluate the importance of membrane composition as well as the dynamic behavior of the side chains. To our knowledge, this study represents one of the first interspecies comparisons of bombolitin antimicrobial activity, and the distinct antimicrobial activity profiles of MBII and BL6 may suggest that geographically isolated Bombus species produce unique AMPs to address specific environmental or ecological pressures, such as unique microbial contaminants in their environments. These implications on bombolitin interspecies diversity are intriguing, and the relationships between bombolitin structure function and target cell specificity will be further examined over a greater species distribution in the future.

Author Contributions

M.J.T. and I.S.W. designed research. M.G.R. and D.K.S. performed experiments. M.G.R. and S.M.W. synthesized and purified the model systems. M.G.R. performed spectroscopic experiments. D.K.S. performed growth inhibition assays. M.J.T., I.S.W., M.G.R., and D.K.S. analyzed data. M.G.R., D.K.S., I.S.W., and M.J.T. wrote the article.

Editor: Elizabeth Rhoades.