Membrane Binding and Oligomerization of the Lipopeptide A54145 Studied by Pyrene Fluorescence

Abstract

A54145 is a lipopeptide antibiotic related to daptomycin that permeabilizes bacterial cell membranes. Its action requires both calcium and phosphatidylglycerol in the target membrane, and it is accompanied by the formation of membrane-associated oligomers. We here probed the interaction of A54145 with model membranes composed of dimyristoylphosphatidylcholine and dimyristoylphosphatidylglycerol, using the steady-state and time-resolved fluorescence of a pyrene-labeled derivative (Py-A54145). In solution, the labeled peptide was found to exist as a monomer. Its membrane interaction occurred in two stages that could be clearly distinguished by varying the calcium concentration. In the first stage, which was observed between 0.15 and 1 mM calcium, Py-A54145 bound to the membrane, as indicated by a strong increase in pyrene monomer emission. At the same calcium concentration, excimer emission increased also, suggesting that Py-A54145 had oligomerized. A global analysis of the time-resolved pyrene monomer and excimer fluorescence confirmed that Py-A54145 forms oligomers quantitatively and concomitantly with membrane binding. When calcium was raised beyond 1 mM, a distinct second transition was observed that may correspond to a doubling of the number of oligomer subunits. The collective findings confirm and extend our understanding of the action mode of A54145 and daptomycin.

Introduction

The lipopeptide antibiotic A54145 (1) is homologous to daptomycin, a clinically important drug that is used in the treatment of infections with Grampositive bacteria (2). Both antibiotics bind to the bacterial cell membrane and cause its depolarization (3, 4), and with both molecules, this involves the formation of membrane-associated oligomers. Mixtures of both antibiotics even give rise to hybrid, partially functional oligomers (5), which suggests a closely similar mode of action. The two compounds differ, however, in two criteria that are important for clinical application. A54145 is too toxic for application in humans; on the other hand, it is not inhibited by lung surfactant. Daptomycin, which is much less toxic, is strongly inhibited by surfactant, which limits its usefulness for treating pneumonia (6). The creation of hybrid derivatives that combine the useful traits of both has been attempted (7).

Although the interaction of daptomycin with cell membranes is not yet fully understood, it is known that specific lipids have important roles in membrane susceptibility and resistance. In model liposomes, phosphatidylglycerol (PG) was found to be essential for oligomer formation and membrane permeabilization (8, 9), and in bacteria, increased conversion of PG to lysyl-PG correlates with resistance to daptomycin (10). The evidence regarding a role of cardiolipin in bacterial daptomycin resistance is ambiguous (11, 12), but it is noteworthy that cardiolipin can inhibit membrane permeabilization in a liposome model (13).

In a recent study, we explored the ternary interaction of daptomycin and A54145 with PG and calcium using isothermal calorimetry and fluorescence (14). When Ca²⁺ concentrations were raised successively from 0 to 1 mM and beyond, two successive calcium-dependent transitions were observed, which correspond to the binding of two Ca²⁺ ions to each lipopeptide molecule. Fluorescence results indicated that both peptides bound to the membrane concomitantly with binding of the first Ca²⁺ ion, whereas oligomer formation was apparent only when the second ion had bound also. It must be noted, however, that all the fluorescent probes used in that study were part of or attached to the peptide moiety (cf. Fig. 1), whereas the behavior of the fatty acyl tail was not observed.

Figure 1.

Structure of native A54145 Factor D (1) and of the pyrene-butanoyl label that was substituted for its native iso-decanoyl residue (blue) to produce Py-A54145. Nonstandard acronyms are as follows: Sar, sarcosine; MeO-Asp, methoxy-aspartate; HO-Asn, hydroxyasparagine. Details of the semisynthesis and purification of Py-A54145 are given in the Supporting Material. To see this figure in color, go online.

With daptomycin, the behavior of the fatty acyl tail had been studied earlier by incorporating a perylene moiety into it (15). Perylene molecules in close proximity will form excited dimers (excimers), and the labeled lipopeptide indeed produced a strong excimer signal upon oligomerization, indicating that the acyl residues are in close proximity within the oligomer.

A more established fluorescence probe of molecular proximity is pyrene (16). Compared to perylene, pyrene excimers display stronger fluorescence that also is spectrally better separated from that of the monomer. Daptomycin contains a kynurenine residue, which interferes with pyrene fluorescence (15). However, A54145 does not contain a kynurenine residue, and we here used pyrene fluorescence to probe the membrane interaction of this lipopeptide.

Pyrene forms excimers when two molecules come within 10 Å of each other (17). Moreover, the fluorescence intensity and lifetime of the pyrene monomer undergo a large increase upon translocation from water to an apolar and highly viscous environment, such as the interior of lipid bilayers (18, 19, 20, 21, 22, 23, 24). This combination of traits makes pyrene a versatile and informative probe not only of proximity but also of environmental change. In the context of protein structure and function, pyrene has been widely used to study interactions with lipid bilayers, substrates, ions, or denaturants, as well as inter- and intramolecular spatial relationships (see, for example, (16, 22, 24)).

Most previous biophysical applications of pyrene have relied on the analysis of the steady-state fluorescence, whereas the fluorescence decays of the pyrene-labeled molecules were deemed too complex and never quantitatively analyzed. In contrast to these earlier reports, we here show how the quantitative analysis of time-resolved fluorescence can be used to obtain detailed information on the distribution of the pyrene-labeled A54145 (Py-A54145) between water and lipid membranes, as well as between different states of aggregation within the membrane. By performing this analysis at different calcium concentrations, we show that calcium-induced binding of monomeric Py-A54145 to the membrane is immediately followed by the formation of membrane-embedded oligomers. Although this process is complete at 1 mM calcium, a well-defined second transition is observed when the calcium concentration is raised further. This second transition is apparent through increased excimer and reduced monomer fluorescence, respectively. It is consistent with an increase in the number of oligomer subunits, possibly arising from the alignment of two oligomers in opposite membrane leaflets (13).

Our study was carried out using the recently introduced model-free analysis (MFA), which involves a global, simultaneous analysis of both monomer and excimer fluorescence decays (25, 26). The successful dissection of the complex behavior of A54145 illustrates that the MFA can be a powerful tool for the study of membrane-binding and/or oligomerizing peptides and proteins.

Materials and Methods

Chemicals

The phospholipids 1,2-dimyristoyl-sn-glycero-3-phosphocholine (PC) and 1,2-dimyristoyl-sn-glycero-3-phospho-rac-(1′-glycerol) (PG) were purchased from Avanti Polar Lipids (Alabaster, AL) and used as received. A54145-Factor D (A54145) and polymer-supported Actinoplanes utahensis deacylase were gifts from Cubist Pharmaceuticals (Lexington, MA). The structures of A54145 and Py-A54145 are illustrated in Fig. 1; details of the Py-A54145 synthesis are given in part A of the Supporting Material.

Preparation of liposomes

PC and PG at an equimolar ratio, or PC alone, were dissolved in a 3:1 mixture of chloroform/methanol. The solvent was evaporated under a stream of nitrogen in a round-bottom flask, followed by evacuation for 3 h. The lipid film was resuspended with HEPES buffer (20 mM HEPES/NaOH, 150 mM NaCl, pH 7.4), and the suspension was extruded 10–15 times through 100 nm polycarbonate membranes using a nitrogen-pressurized device (27). The total lipid concentration for the fluorescence measurements was 250 μM and the liposome solutions were kept at 30°C.

Antibacterial activity of A54145 and Py-A54145

A54145 and Py-A54145 were serially diluted in LB broth supplemented with 5 mM CaCl₂. Each serial dilution was inoculated with 1% (v/v) of an overnight culture of Bacillus subtilis ATCC 1046, and the culture tubes were incubated at 37°C overnight with shaking. The growth was evaluated visually. The minimum inhibitory concentration was taken to be the lowest concentration without any visible turbidity.

Fluorescence measurements

All fluorescence experiments were carried out at 30°C with Py-A54145 (2 μM, unless stated otherwise) in buffer or liposome suspensions containing either 125 μM each of PC and PG or 250 μM of PC only. The samples were not deaerated. Samples were incubated for ≥5 min at the same temperature before the measurements were begun.

Steady-state fluorescence spectra were acquired on a Photon Technology International (Edison, NJ) LS-100 spectrofluorimeter equipped with a continuous Ushio (Cypress, CA) UXL-75Xe xenon arc lamp and a PTI 814 photomultiplier detection system (Photon Technology International). The excitation wavelength was set at 344 nm and the fluorescence spectra were acquired from 350 to 600 nm. The monomer fluorescence intensity (I_M) was determined by taking the integral under the fluorescence spectrum between 372 and 378 nm. Similarly, integration of the fluorescence signal between 500 and 530 nm yielded the fluorescence intensity of the excimer (I_E).

Time-resolved fluorescence decays were obtained (using the time-correlated single photon counting technique) on an IBH (Glasgow, UK) time-resolved fluorimeter equipped with an Integrated Behavioral Health 340 nm NanoLED as well as excitation and emission monochromators. The Py-A54145 solutions were excited at 344 nm and the fluorescence decays of the pyrene monomer and excimer were acquired at 375 and 510 nm, respectively. On the emission side, the instrument was additionally fitted with 370 and 495 nm cuton filters, respectively, to prevent residual stray light scattered by the liposome suspension from reaching the detector. The fluorescence decays were acquired over 1024 channels with 40,000 and 20,000 counts at the decay maximum for the pyrene monomer in buffer and the pyrene excimer in PC/PG membranes or 40,000 counts at the decay maximum of the pyrene monomer in PC or PC/PG membranes.

Analysis of the fluorescence decays

The fluorescence decays were analyzed separately with sums of exponentials or globally with MFA (25, 26). The preexponential factors and decay times were optimized using the Marquardt-Levenberg algorithm (28). The fits were considered satisfactory when the χ² was <1.3 and the weighted residuals and autocorrelation of the residuals were randomly distributed around zero. A detailed description of these analyses is provided in the Supporting Material. All parameters retrieved from the analysis of the fluorescence decays are listed in Tables S1–S8.

Results

Antibacterial activity of Py-A54145

The minimum inhibitory concentration of both native A54145 and Py-A54145 against Bacillus subtilis was 0.75 μg/mL, indicating that Py-A54145 retains full activity and is a valid model for studying the activity of A54145. The observation agrees with previous observations on daptomycin, which also tolerates substitutions of its native fatty acyl residue with other hydrophobic moieties (15, 29).

Phosphatidylglycerol facilitates A54145 membrane binding and oligomerization

Fig. 2 shows the effects of calcium on the steady-state fluorescence spectrum of Py-A54145 without liposomes and with PC or PC/PG liposomes. All spectra exhibit a series of sharp peaks between 370 and 420 nm, which are typical of the pyrene monomer. With PC/PG liposomes, a broad, structureless emission band centered at 480 nm is also apparent that is characteristic of the pyrene excimer. In the absence of liposomes, the fluorescence intensity of the pyrene monomer decreased by ∼20% when the Ca²⁺ concentration was increased from 0 to 100 mM. This observation suggests that calcium binds to A54145 in solution, which has previously been documented with daptomycin (30). The absence of excimer emission indicates that Py-A54145 does not aggregate in solution at low micromolar concentrations, which agrees with prior observations on daptomycin using FRET (8).

Figure 2.

Fluorescence spectra of 2 μM Py-A54145 in the presence of Ca²⁺ incubated with buffer (A), PC liposomes (B), or PG/PC liposomes (C and D). In (A)–(C), the intensity observed without Ca²⁺ at 375 nm was arbitrarily set to 100, and all other fluorescence spectra were scaled accordingly. In (D), 0.5 μM Py-A54145 was incubated, with or without a fivefold excess of unlabeled native A54145, with PC/PG membranes and 5 mM Ca²⁺. The peak intensity of the pure Py-A54145 sample was set to 100. The excitation wavelength was 344 nm throughout. To see this figure in color, go online.

In the presence of PC liposomes, the intensity of the Py-A54145 monomer increases rather than decreases upon addition of calcium. The increase in monomer intensity suggests that pyrene enters a more hydrophobic environment, which is consistent with binding of Py-A54145 monomers to the membrane. The fluorescence intensity at 480 nm is somewhat higher than in buffer, indicating some measure of excimer formation. This would be expected from the concentrating effect of membrane binding, which would promote diffusional encounters between excited and ground-state pyrene monomers. However, excimer fluorescence remains very low relative to monomer fluorescence.

With PC/PG liposomes, the monomer intensity increases much more readily and at much lower Ca²⁺ concentrations than is observed with PC membranes. A strong excimer signal emerges concomitantly, which indicates much closer interaction between the Py-A54145 molecules. A straightforward explanation for the strong excimer signal is that Py-A54145 undergoes oligomerization. Within an oligomer, the distances for diffusional encounters between pyrene labels are much smaller than those between molecules diffusing independently in the membrane, and therefore, the likelihood of excimer formation increases.

The assumed correlation of excimer formation and A54145 oligomerization is further supported by the finding that the excimer fluorescence of a fixed amount of Py-A54145 can be suppressed by the addition of unlabeled A54145. If excimers arose from collisional encounters of membrane-bound monomers only, their formation should depend solely on the concentration of membrane-bound Py-A54145 and be unaffected by the presence or absence of unlabeled A54145. On the other hand, if the labeled and unlabeled species form hybrid oligomers, many of the Py-A54145 molecules contained in these hybrids should be surrounded by unlabeled neighbors and thus be unable to form excimers. Monomer fluorescence should be increased, and excimer fluorescence diminished, as is indeed observed (see Fig. 2 D).

The behavior of Py-A54145 described thus far is consistent with results reported previously for daptomycin using perylene excimer fluorescence (15). However, a novel observation is that, on PC/PG membranes, the response of Py-A54145 to increasing Ca²⁺ levels occurs in two distinct phases. This biphasic response is seen in Fig. 2 C and, more clearly, in Fig. 3. Fig. 3 A shows the steady-state intensity of the pyrene monomer as a function of Ca²⁺, whereas Fig. 3 B shows the number-average monomer lifetime, which was obtained by fitting individual monomer decays with sums of exponentials (see Eq. S1 and Tables S1–S4). Both rise rapidly toward a maximum at 0.5–1 mM Ca²⁺. Increasing the Ca²⁺ concentration beyond this point causes them to drop back again. In contrast, the excimer intensity, which is shown relative to the monomer intensity in Fig. 3 C exhibits another steep rise and approaches a plateau at ∼10 mM Ca²⁺.

Figure 3.

Calcium effects on the steady-state intensity (A) and the fluorescence lifetime (B) of Py-A54145 monomers, and on the ratio of steady-state excimer to monomer emission (C). The y axes in (A) and (B) show the monomer intensity and lifetime, respectively, at a given Ca²⁺ concentration relative to that obtained without Ca²⁺. Monomer and excimer intensities were taken at 375 and 480 nm, respectively. Lifetimes in (B) are weighted averages obtained from a three-exponential fit (see the Supporting Material). Lines are visual aids without physical meaning. To see this figure in color, go online.

We also note that, in buffer or with PC membranes, all changes to any of the parameters are minor and occur only at higher Ca²⁺ concentrations. These observations corroborate the dominant role of PG in the membrane interaction of Py-A54145.

Time-resolved pyrene monomer fluorescence (Figs. 3 B and 4) agrees with the steady-state experiments. The decay time is longer with PC/PG membranes than with Py-A54145 in buffer or on PC membranes, and it is yet longer for hybrid oligomers of Py-A54145 and unlabeled A54145. These findings reinforce the notion that A54145 interacts differently with PC and PC/PG membranes.

Figure 4.

Time-resolved monomer fluorescence decays for Py-A54145 (2 μM) under different conditions. The sample labeled “hybrids” was incubated with PC/PG membranes and Ca²⁺ but additionally contained a ninefold molar excess of unlabeled A54145. Lifetime parameters for the various decays are listed in the Supporting Material (Tables S1–S3 and S8). To see this figure in color, go online.

Quantitative analysis of time-resolved monomer and excimer fluorescence

The second calcium-dependent transition that is observed on PC/PG membranes raises the question of how exactly the two states of A54145 before and after this transition differ from one another. To address it, we applied the analytical method MFA that we have described previously (25, 26).

Assumptions

To adapt MFA to the A54145/membrane system, we made the following assumptions, which are illustrated in Fig. 5:

• 1.Py-A54145 can reside in water or in the lipid bilayer.
• 2.Pyrene moieties that reside in water exhibit only monomer fluorescence. This is supported by the data shown in Fig. 2 A.
• 3.Pyrene moieties of membrane-bound Py-A54145 can emit excimer fluorescence or monomer fluorescence. Excimers can form through diffusional encounters between excited and ground-state monomers or from aggregates of two ground-state molecules.
• 4.Dissociation of excimers to excited-state monomers is neglected. This assumption is reasonable for temperatures below 35°C (25, 31).

Figure 5.

Assumed fluorescent states and transitions of Py-A54145 in the presence of PC/PG membranes. A54145 molecules that reside in water do not aggregate, and accordingly, the attached pyrene moieties (Py_W) emit only monomer fluorescence. In the membrane, excited pyrene moieties can form excimers by diffusion (Py_diff); the rate of this process is governed by the function f(t). Membrane-bound pyrene moieties can also form dimeric aggregates before excitation (Py_agg), which will contribute to excimer fluorescence only. Angular brackets signify the use of multiple lifetime components for the state in question. To see this figure in color, go online.

The existence of ground-state pyrene aggregates (Py_agg in Fig. 5) in the PC/PG membrane can be demonstrated by fitting the time-resolved excimer emission with a sum of exponentials. In such a fit, diffusional excimer formation (Py_diff) will be captured by negative preexponential factors, whereas excimer decay will give rise to positive preexponential factors. If excimers formed by diffusion only, the sums of negative and positive preexponential factors should match (32, 33). However, the negative preexponential factors amount to only $-73\pm 4\text{\%}$ of the positive ones (see Table S4); the difference must arise from direct excitation of ground-state pyrene aggregates.

Since our analysis is based on a set of assumptions, one might ask exactly what is meant by referring to it as model-free. The term denotes the approach to analyzing the rate of diffusional excimer formation, which is named $f\left(t\right)$ in Fig. 5. The conventional approach is to first formulate a physical model that predicts $f\left(t\right)$ from assumptions about the spatial distribution and motional freedom of pyrene monomers (see, for example, (31, 34, 35, 36)), and then to obtain the parameters of the chosen model by numerically fitting it to measured fluorescence decays. In contrast, MFA does not assume a specific physical model; instead, it simply approximates $f\left(t\right)$ numerically with an appropriate number of exponential terms. Physical interpretations of the parameters obtained in this manner are only considered after the fact.

The advantage of this agnostic approach is that it is simple enough to allow for a simultaneous global analysis of monomer and excimer decays. From the scheme in Fig. 5, it is clear that the exponential terms that represent $f\left(t\right)$ must be present—with opposite signs, since excimers are gained as monomers are lost—in the decays of both the monomer and the excimer. MFA enforces this constraint while simultaneously fitting both decays, which reduces the statistical indeterminacy of the final result.

Experimental constraints

As depicted in Fig. 5, we assume multiple populations of peptide-bound pyrene, which moreover may decay with multiple lifetimes. This results in a rather large number of variable parameters over all. To make the analysis more stringent, some of these parameters must therefore be constrained experimentally. This was done as follows:

• 1.The fluorescence decay of pyrene moieties in water (Py_W) was measured on a sample of Py-A54145 in buffer, in the absence of any liposomes. This decay was fitted with three exponential terms. The lifetimes and their associated preexponential factors thus obtained were treated as a single linear component when fitting all other monomer decays in the MFA; that is, the contribution of Py_W was represented in the global fit by only a single free parameter.
• 2.
  The lifetime of the membrane-bound pyrene monomer (denoted as τ_M,m in Fig. 5) was determined on a sample of Py-A54145 with a ninefold excess of unlabeled A54145 that was incubated with PC/PG membranes (cf. Fig. 4). In the hybrid oligomers that will form under these conditions, most Py-A54145 molecules will be surrounded by unlabeled A54145 neighbors, and the pyrene moieties will thus be unable to form excimers. Therefore, the monomer decay of such a sample should be dominated by τ_M,m.

  • In a three-exponential fit of the monomer decay, a lifetime of 170 ns accounted for 55% of the total preexponential weight (see Table S8). This lifetime was used as the value of τ_M,m of all samples in the MFA.
• 3.The time course of excimer fluorescence is fairly complex, as excimers form and decay on similar timescales. To constrain the rate of excimer formation (named f(t) in Fig. 5), monomer and excimer decays were acquired on a sample of Py-A54145 that was incubated with PC/PG liposomes and high Ca²⁺ (20 mM). Under these conditions, Py_W was taken as negligible, so that the monomer decay will solely be defined by the previously determined τ_M,m and f(t).

Monomer and excimer decay curves were fitted globally according to Eqs. S9 and S10. In these fits, f(t) was represented by three variable exponential terms, which were imposed on both monomer decay and excimer formation. The fit also allowed for two variable exponential terms for excimer decay. The lifetimes obtained for the two excimer decay terms were then fixed in all other MFA runs.

After applying these constraints, the global fit retains 10 freely variable parameters (see page 19 in the Supporting Material); this is comparable to fitting a single decay using five parameters. Fig. 6 shows an example of an MFA fit carried out in this manner. The decays of both the monomer and the excimer are fitted well, with small and evenly distributed residuals and autocorrelations. The excimer time course exhibits a clear initial rise, which represents the formation of excimers through diffusional encounters of monomers.

Figure 6.

Model-free analysis. Example of a global fit of time-resolved fluorescence of pyrene monomers (λ_ex, 344 nm; λ_em, 375 nm) (A) and excimers (λ_ex, 344 nm; λ_em, 510 nm) (B). Experimental conditions were Py-A54145 in PC/PG liposome suspension with 5 mM Ca²⁺. The overall ${\chi }^2$ of the global fit is 1.06. To see this figure in color, go online.

Resulting parameters

The key parameters retrieved by the MFA include the molar fractions of the various states of fluorescent pyrene moieties (cf. Fig. 5), as well as the average rate constant of diffusional excimer formation (<k>). Note that <k> is not a conventional bimolecular rate constant; instead, it is a pseudounimolecular parameter that measures how readily a given excited monomer will find a ground-state molecule to form an excimer (see Eqs. S6 and S7 for its definition and its relation to f(t), respectively). A change in <k> usually reflects a structural change to the pyrene-labeled macromolecule that affects the mobility or local concentration of monomeric pyrene residues. In contrast, a change in the molar fractions usually indicates a change in the solubility or binding of the labeled molecules (33, 37).

A fuller explanation of the MFA is given in parts C and D of the Supporting Material and details how the above parameters are obtained from the global fits, and part E addresses the experimental error associated with the retrieved parameters and shows that the analysis is reliable. Fig. 7 shows how these parameters vary with the Ca²⁺ concentration. The unbound fraction (Py_W) drops rapidly to near zero with increasing Ca²⁺ (Fig. 7 A); this agrees with the observations from steady-state fluorescence (cf. Fig. 3).

Figure 7.

Parameters retrieved from the model-free analysis, as functions of the Ca²⁺ concentration (cf. Fig. 5 for acronyms). (A) Molar fraction of Py_W. (B) Py_diff and Py_agg as fractions of the overall membrane-bound pyrene moieties. (C) Average rate constant of diffusional excimer formation, calculated according to Eq. S6. Lines represent fits to the Hill equation; the fitted parameters are listed in part H of the Supporting Material. (D) Ratio of excimer to monomer emission as reconstructed from the MFA, overlaid on the steady-state results (cf. Fig. 3C). To see this figure in color, go online.

Fig. 7 B shows the relative abundance of membrane-associated pyrene molecules that form excimers by diffusion (Py_diff) or by aggregation (Py_agg). Py_diff dominates at all calcium concentrations; this resembles previous findings with pyrene-labeled polymers in organic solvents (38), as well as with micelles of pyrene-labeled gemini surfactants (33), and it suggests that pyrene resides inside a locally disordered environment within the apolar core of the membrane. Nevertheless, Py_agg appears to undergo a small step increase at a Ca²⁺ concentration slightly above 1 mM. At virtually the same Ca²⁺ level, we observe a doubling in the average rate of diffusional excimer formation (<k>).

In the steady-state experiments, we had noticed that the excimer fluorescence underwent a pronounced increase just above 1 mM Ca²⁺ (cf. Fig. 3), and the same is also seen when the ratio of excimer to monomer fluorescence is reconstructed using the parameters retrieved from the MFA (Fig. 7 D; details are provided in part D of the Supporting Material). From the latter, we now see that this increase does not involve a major change in Py_diff or Py_agg; instead, it is almost entirely caused by an increase in the rate of diffusional excimer formation within the Py_diff population.

Discussion

To make sense of the collective findings, we must first distinguish between the oligomerization—that is, peptide-peptide interaction—of A54145 on the one hand, and the diffusional association or aggregation of pyrene moieties attached to the A54145 molecules on the other. Pyrene monomer and excimer fluorescence may arise from both monomeric and oligomeric Py-A54145. However, as noted in the section entitled Phosphatidylglycerol Facilitates A54145 Membrane Binding and Oligomerization, Py-A54145 exhibits very little excimer fluorescence both in water and when bound to PC membranes. Like daptomycin (8, 15), A54145 most likely binds to PC membranes as a monomer. Although membrane binding will greatly concentrate the monomeric Py-A54145, the highly viscous membrane environment will counteract the formation of excimers by diffusion. Therefore, efficient excimer formation can only be expected after Py-A54145 has formed oligomers.

The rate of diffusional encounters between membrane-bound Py-A54145 monomers will vary with their concentration. In contrast, after oligomerization, excimers will form mostly between neighboring subunits of the same oligomer, and the total concentration of membrane-bound Py-A54145 should have little or no effect on the likelihood of excimer formation. With this in mind, we can examine what the pyrene fluorescence tells us about the behavior of A54145 on PC/PG membranes (cf. Fig. 8).

Figure 8.

Proposed interpretation of pyrene fluorescence in terms of A54145 function. (A) Membrane-bound Py-A54145 monomers may form excimers by diffusion. This process will be concentration-dependent and inefficient; it likely occurs on PC membranes. (B) On PC/PG membranes, binding of the first Ca²⁺ ion to Py-A54145 causes formation of small oligomers, which increases excimer fluorescence. Excimer fluorescence is suppressed by mixing Py-A54145 with an excess of unlabeled A54145. (C) Binding of the second Ca²⁺ ion increases the oligomer size, possibly by aligning two small oligomers across both membrane leaflets. This further increases the efficiency of excimer formation. To see this figure in color, go online.

Binding of Py-A54145 to PC/PG membranes

The increase of monomer intensity and lifetime, the decrease in Py_W, and the high fractional contribution of Py_diff to the excimer intensity all indicate that Py-A54145 binds progressively at low Ca²⁺ concentration. The rate of the drop of Py_W with increasing Ca²⁺ (Fig. 7 A) suggests some measure of cooperativity (see Table S9), which would agree with the association of two Ca²⁺ ions to each lipopeptide molecule (14), as well as with the oligomerization of the membrane-bound Py-A54145. However, due to some degree of static quenching, Py_W may slightly underestimate the fraction of unbound Py-A54145. This complicates an accurate analysis of binding, which may be more amenable to calorimetric methods. On the other hand, fluorescence is more suitable for dissecting the behavior of the bound lipopeptide, which was the main focus of this study.

A54145 oligomerizes concomitantly with membrane binding

From the lack of excimer fluorescence of Py-A54145 in solution (cf. Fig. 2 A), it is clear that A54145 does not oligomerize before membrane binding, at least not at the low yet biologically relevant lipopeptide concentrations employed here. This confirms the findings of previous studies on the related lipopeptide daptomycin (8, 39). The question is, then, how tightly membrane binding of A54145 is coupled to the formation of oligomers.

Between 0.1 and 1 mM Ca²⁺, the contribution from Py_W to the overall pyrene monomer fluorescence decreases from ∼75% to ∼5% (cf. Fig. 7 A), which implies a strong increase in membrane-bound Py-A54145. If the bound lipopeptide stayed monomeric, the pseudounimolecular rate constant of excimer formation by diffusion (<k>) should increase in proportion to its concentration in the membrane; however, this rate constant remains virtually unchanged (cf. Fig. 7 C). This finding indicates that the membrane-associated Py-A54145 is already oligomeric at this stage. In keeping with this interpretation, excimer formation on PC/PG membranes far exceeds that observed on pure PC membranes, to which daptomycin binds as a monomer (8); this applies at all Ca²⁺ concentrations (cf. Figs. 2 and 3). Pyrene monomer lifetimes are higher on PC/PG membranes than on PC membranes (see Tables S2 and S3). This effect, as well as lateral phase segregation on PC/PG membranes (40), should cause a somewhat higher rate of excimer formation even between monomeric Py-A54145 molecules; nevertheless, the very high excimer intensity on PC/PG membranes strongly suggests oligomer formation.

Aggregation of fatty acyl residues precedes that of peptide moieties

We recently reported that the membrane interaction of daptomycin and A54145 occurs in two separate steps, each of which is driven by the binding of one Ca²⁺ ion (14). Those experiments were carried out with the fluorescent dye acrylodan, which was attached to amino acid residues 6 and 8 in daptomycin and A54145, respectively. In the first step, acrylodan emission increased in intensity and blue-shifted, which is compatible with membrane binding. In the second step, acrylodan underwent self-quenching, indicating oligomerization.

The Ca²⁺ concentrations that will induce each of the two changes in acrylodan emission align very closely with those that were found here to induce the two stepwise increases in pyrene excimer formation. We therefore assume that pyrene and acrylodan report on the same structural changes to the lipopeptide. It is notable, then, that pyrene excimer fluorescence detects oligomerization after binding of the first Ca²⁺ ion, whereas acrylodan self-quenching detects it only after binding of the second one.

The pyrene moiety in Py-A54145 is part of the fatty acyl tail of the molecule (see Fig. 1), whereas acrylodan was attached to the peptide moieties. It therefore appears that the fatty acyl tails associate before the peptide moieties do, and that the initial oligomer must undergo a significant structural transition concomitantly with binding of the second calcium ion.

The second calcium-dependent transition may double the size of the oligomer

With Py-A54145, the second transition manifests itself by a doubling of the rate of diffusional excimer formation (cf. Fig. 7 C). Most likely, after the transition, each excited pyrene monomer now has a greater number of ground-state neighbors to engage with. Of course, the doubling of the rate cannot strictly prove a doubling of the oligomer size; a conformational change that would bring the pyrenyl moieties into greater proximity within the oligomer is another possible explanation. Nevertheless, the observation does resonate with the previous proposal (13) that daptomycin oligomers may contain either four or eight subunits, depending on whether they span only one or both membrane leaflets. It appears possible, therefore, that the second Ca²⁺-dependent transition reported by Py-A54145 corresponds to the alignment of two tetramers across both membrane leaflets, resulting in an octamer that presumably forms the functional transmembrane pore. This possibility needs to be addressed in future experiments.

At this point, one might wonder if the intermediate, presumably singly calcium-saturated state of daptomycin or A54145 has any functional significance in vivo. In the human body, the extracellular level of free calcium is ∼1.25 mM, which suffices to drive daptomycin to the final, presumably doubly saturated state (14). Similarly, in soil, which is the natural habitat of the daptomycin and A54145 producer organisms, calcium concentrations appear to be in the low millimolar range (41). Therefore, although the calcium concentration required for complete saturation may differ depending on the lipid composition of the bacterial target membrane, there is no clear scenario to explicitly suggest that the singly saturated form is functionally important in vivo.

Limitations and approximations inherent in MFA

Although the conclusions reached here by the application of MFA hopefully speak to the usefulness of this method, one should be aware of its limitations. One is inherent in the relinquishment of a physical model for the excimer formation function, f(t). As illustrated here, MFA subsumes the rate of excimer formation in a single parameter (<k>), which may entail the loss of some subtle information that a suitable physical model might have recovered. On the other hand, as is illustrated in Fig. S7, the value of <k> is more robust than that of the individual lifetime components on which it is based. Accordingly, a physical model that assigns meaning to multiple separate lifetimes may be subject to greater experimental uncertainty.

In this study, the excimer fluorescence was represented by two separate lifetime components. This approach is well supported by studies on model compounds that contain two pyrene residues linked by short organic spacer moieties. In such molecules, pyrene excimers may adopt two distinct conformations with different degrees of planar overlap and, accordingly, emission lifetimes (42, 43). In the case of Py-A54145, there may well be more than two discrete excimer conformations. However, two lifetime components proved sufficient to accommodate the observed decays in the global fits. Although MFA is unable to determine the true number of excimer conformations and lifetimes, our interpretation of the results does not hinge on this question. Similar considerations apply to the chosen numbers of lifetimes for pyrene monomers in water and in the membrane.

Conclusions

In summary, this study provides novel insights into the action mode of Py-A54145, a model compound for the therapeutically important antibiotic daptomycin. It supports our previous hypothesis that A54145 interacts with membranes containing PG in two distinguishable Ca²⁺-dependent steps (14), and it further indicates that an oligomeric state is reached already upon completion of the first step. More generally, our study demonstrates the usefulness of pyrene fluorescence to understand the behavior of peptides that interact both mutually and with membranes, particularly when it is used in conjunction with the powerful MFA method.

Author Contributions

T.Z. conducted all the experiments. S.T. synthesized Py-A54145. M.P. and S.T. developed the line of research revolving around the use of fluorescently labelled daptomycin analogues to determine how daptomycin interacts with lipid bilayers. M.P. suggested the use of time-resolved pyrene fluorescence to study the membrane interaction of A54145. J.D. analyzed the fluorescence spectra and decays, and developed the version of the Model Free Analysis used to fit globally the monomer and excimer decays. J.D. and M.P. wrote the manuscript.

Editor: Heiko Heerklotz.