Insights into the role of protein molecule size and structure on interfacial properties using designed sequences

Abstract

Mixtures of a large, structured protein with a smaller, unstructured component are inherently complex and hard to characterize at interfaces, leading to difficulties in understanding their interfacial behaviours and, therefore, formulation optimization. Here, we investigated interfacial properties of such a mixed system. Simplicity was achieved using designed sequences in which chemical differences had been eliminated to isolate the effect of molecular size and structure, namely a short unstructured peptide (DAMP1) and its longer structured protein concatamer (DAMP4). Interfacial tension measurements suggested that the size and bulk structuring of the larger molecule led to much slower adsorption kinetics. Neutron reflectometry at equilibrium revealed that both molecules adsorbed as a monolayer to the air–water interface (indicating unfolding of DAMP4 to give a chain of four connected DAMP1 molecules), with a concentration ratio equal to that in the bulk. This suggests the overall free energy of adsorption is equal despite differences in size and bulk structure. At small interfacial extensional strains, only molecule packing influenced the stress response. At larger strains, the effect of size became apparent, with DAMP4 registering a higher stress response and interfacial elasticity. When both components were present at the interface, most stress-dissipating movement was achieved by DAMP1. This work thus provides insights into the role of proteins' molecular size and structure on their interfacial properties, and the designed sequences introduced here can serve as effective tools for interfacial studies of proteins and polymers.

1. Introduction

The behaviour of proteins at air–water interfaces determines their behaviour as foam stabilizers. In order to improve product formulations, it would be desirable to be able to predict functional behaviour from an understanding of interfacial properties, and ideally from knowledge of molecular structure. This link is not well established [1–3], largely owing to the complexity of protein molecules and the interplay between many contributing factors including surface coverage/pressure, adsorption rate, electrostatic and steric repulsion, and interfacial rheology. For example, increasing the ionic strength of solution from which protein foam is formed may have the effect of increasing the measured interfacial elasticity owing to increased intermolecular interaction, however changes in surface packing, interfacial charge and bulk aggregation may, and often do, occur. Discrepancies are, therefore, often observed in the literature as to whether a given property improves foam functionality; for example, in the case of interfacial rheology, some studies observe a positive correlation with foam functionality [4,5], whereas others show negative or no correlation [6–8].

In nature as well as in practical formulations, proteins rarely exist as single components [9], further obscuring understanding. The addition of a smaller, unstructured component (such as a peptide) to a structured protein has been seen to improve foam properties, as is often observed in the case of protein hydrolysates [10,11]. It is generally accepted that this functional improvement is related to the smaller size (better adsorption kinetics) and increased hydrophobicity (greater exposure of hydrophobic residues usually buried in the structured protein) of the peptide component [10,12,13]. At a fundamental level, however, the effect of molecule size and structure on interfacial properties is even less clear than in the case of single-component systems. Improved understanding is important to facilitate the optimized use of naturally occurring proteins, which is currently empirical, as well as the design of synthetic sequences that bring new or enhanced function.

Two issues hinder progress in this area. First, systems composed of natural components are often complex and difficult to characterize. For example, heat treatment of β-lactoglobulin at pH 2 to form fibres results in a mixed system of fibres, monomers, peptides and deamidated variants of each [14]. The exact extent of cleavage to peptides, formation of fibres and deamidation is very difficult to identify and quantify, therefore the bulk composition of the system being studied is at best only qualitatively understood. This type of issue makes even the simplest of experiments difficult to interpret, as in the study by Jung et al. [14], where the observed interfacial tension kinetics could not be clearly explained. Second, the interfacial composition is even more difficult to characterize than that in the bulk. This is not only due to the same issues as discussed for bulk characterization, but also due to the low accessibility and high complexity of experimental techniques that characterize surface composition (such as neutron reflectometry). When the concentrations and structures of components at an interface are unknown, observed properties, such as interfacial tension, viscoelasticity, foam and thin film stability, cannot be directly interpreted. A meaningful study of the roles of molecular size and structure on interfacial properties requires a well-defined model system to allow decoupling of these factors from differences in chemistry.

In this work, we present such a model system, which has known bulk concentrations, known amino acid sequence and structuring in bulk. We study a larger, structured protein (DAMP4), a smaller, unstructured peptide (DAMP1) and mixtures of the two. DAMP4 is a concatameric repeat of the DAMP1 sequence, therefore this system represents the simplest form of a mixed protein–peptide system, where only size and bulk structuring varies between components, and differences in chemistry have been eliminated. DAMP4 [15], MD(PSMKQLADS-LHQLARQ-VSRLEHAD)₄, is four repeats of the DAMP1 sequence, PSMKQLADS-LHQLARQ-VSRLEHAD, with two additional residues at the start, a methionine (required for bioexpression purposes), and an aspartate (for aspartyl–proline bond cleavage studies [16]). Besides these N-terminal residues, DAMP1 and DAMP4 are chemically identical. While the presence of the additional aspartate residue (with its negative charge) has the effect of shifting the theoretical pI slightly from 7.3 for DAMP1 to 6.7 for DAMP4, our earlier study of the foaming and film-forming properties of DAMP1 and DAMP4 [16] indicated that their behaviours were not significantly influenced by the presence of these additional residues. DAMP1 and DAMP4 are based on the surface-active peptide AM1 [17], which was designed to adsorb as a single α-helix at the air–water or oil–water interface (monolayer structure was confirmed by neutron reflectometry [18]). It is expected that DAMP1 will behave similarly. Like AM1 [19], DAMP1 is largely unstructured in bulk (see the electronic supplementary material, figure S1), whereas DAMP4 forms a very stable four-helix bundle [15].

The effect of molecule size and structuring on interfacial adsorption is probed using DAMP1 and DAMP4 at different ratios of both components. The use of known components overcomes aforementioned issues regarding the difficulty of characterizing mixed systems. The kinetics of interfacial tension is probed by drop-shape analysis, whereas the equilibrium adsorption and interfacial structure are studied by neutron reflectometry. Specifically, the relationship between the bulk and interfacial concentrations of DAMP1 and DAMP4 is determined using neutron reflectometry, with deuterated DAMP4 used to contrast it from DAMP1. This approach overcomes the issues discussed above regarding unknown interfacial composition. In the final section of this work, the response of the DAMP4 and DAMP1 mixed system to an applied compression and extensional strain is studied. As the interfacial composition and structure are known, interesting insights into the effect of molecule size on interfacial mobility and stress response are gained.

The approach of using a simple, defined peptide and its concatamer removes the complexity and factor coupling which usually complicates studies of protein mixed systems. As DAMP1 and DAMP4 have the same sequence and, therefore, differ only in their size and bulk structuring, insights are gained into the role of molecule size and structuring on interfacial adsorption and rheology that would not be easily possible with naturally occurring protein systems and their hydrolysates.

2. Material and methods

2.1. Materials

DAMP4 (11 116.5 Da) was produced biologically and purified as described previously [15]. DAMP1 (2731 Da) was synthesized by GenScript Corporation (Piscataway, NJ), and quantified by high sensitivity amino acid analysis (Australian Proteome Analysis Facility, Sydney). Purity of both DAMP4 and DAMP1 was greater than 95 per cent by HPLC. A biologically-deuterated form of DAMP4 (d-DAMP4) was produced by the National Deuteration Facility (ANSTO, Lucas Heights), and purified using the same procedure as for non-deuterated DAMP4. The extent of deuteration of d-DAMP4 was calculated by mass spectroscopy to be 91.1 per cent on a non-exchangeable hydrogen basis (see the electronic supplementary material, figure S2).

The buffer condition for all samples was 25 mM sodium 4-(2-hydroxyethyl)-1-piperazine ethanesulphonate (HEPES), 200 µM ethylenediamine tetraacetic acid (EDTA), at pH 7.4. The concentrations tested were 8 µM DAMP4, DAMP1, or a mix of the two at 25, 50, 75 and 90 per cent DAMP4 on a mass basis, 8 μM total concentration (table 1). Interfacial tension measurements were also performed at varying concentrations of the individual components, as described in the text. Milli-Q water was used for all solutions (Millipore, North Ryde, Australia).

Table 1.

Experimental concentrations of DAMP1 and DAMP4.

mass basis				molar basis
% DAMP4	(DAMP4), µg ml−1	(DAMP1), µg ml−1	total µg ml−1	% DAMP4	(DAMP4), µM	(DAMP1), µM	total µM
100	89	—	89	100	8	—	8
90	62	6.6	69	67	5.6	2.4	8
75	36	13	49	42	3.2	4.8	8
50	18	18	35	19.5	1.6	6.4	8
25	6.7	20	27	7.5	0.6	7.4	8
0	—	22	22	0	—	8	8

2.2. Interfacial tension

A DSA-10 drop-shape analysis unit (Krüss GmbH, Hamburg, Germany) was used to measure interfacial tension kinetics. The sample (8 ml) was held in a quartz cuvette (Hellma GmbH, Mülheim, Germany). Bubbles were formed through a U-shaped stainless steel capillary of known diameter fed by a glass syringe operated manually. Cleanliness and operation of the system was checked by forming an air bubble in milli-Q water, and confirming a constant interfacial tension of 72.8 mN m⁻¹ for 10 min. To measure the interfacial tension kinetics, the cuvette was filled with the sample of interest, a bubble of about 10 µl formed, and interfacial tension as extracted by the software (via images of the bubble collected by a connected camera) was monitored at a rate of about 1 measurement per second.

2.3. Calculation of experimental and theoretical characteristic diffusion times

The experimental diffusion times were estimated from interfacial tension data, as the approximate time it takes to reach around 90 per cent of the final interfacial tension [20]. Theoretical characteristic times of diffusion for DAMP1 and DAMP4 were calculated by the method previously used for similar peptides Lac28 and Lac21 [21]. First, the Polson equation [22], D = 2.85 × 10⁻⁵ M^−1/3 cm² s⁻¹ (where M is molecular mass) was used to calculate the molecular diffusion coefficients, D, in bulk, giving D_DAMP4 = 1.28 × 10⁻¹⁰ m² s⁻¹ and D_DAMP1 = 2.04 × 10⁻¹⁰ m² s⁻¹. Characteristic diffusion time constants, t_d, were then calculated assuming a linear isotherm [20]: t_d = 1/D(Γ_max/C_∞)², where C_∞ is the bulk concentration and Γ_max is the maximum surface coverage, calculated based on the method for Lac21 and Lac28 [21], where 35 Ų is the area assumed per exposed hydrophobic residue, and two hydrophobic residues per heptad are assumed for an α-helix. This gives a Γ_max equal to 1.5 × 10⁻⁷ mol m⁻² for DAMP4 and 6.8 × 10⁻⁷ mol m⁻² for DAMP1.

2.4. Neutron reflectometry

Information on the interfacial thickness and coverage of DAMP1/DAMP4 systems was collected using the Platypus time of flight neutron reflectometer (ANSTO, Lucas Heights; [23,24]), which is fed by a cold neutron beam (2.8 Å ≤ λ ≤ 18.0 Å) from the OPAL 20 MW research reactor. The neutron reflectivity (the ratio of reflected and incident intensities) was measured as a function of momentum transfer, Q, where Q = 4π sinθ/λ, where λ is the neutron wavelength and θ is the angle of incidence of the collimated neutron beam onto the air–liquid surface. A Q-range of 0.013–0.33 Å⁻¹ with a constant Q resolution of 8 per cent was used.

To distinguish DAMP4 from DAMP1, deuterated DAMP4 was used as a second contrast condition. Each sample condition (besides 8 µM DAMP1) was, therefore, measured at two contrast conditions, with deuterated DAMP4 (d-DAMP4) and with non-deuterated DAMP4 (h-DAMP4), totalling 11 experiments: four mix conditions (2 contrasts) + 1 × 100% DAMP4 condition (2 contrasts) + 1 × 100% DAMP1 condition (1 contrast). All samples were prepared in the same buffer condition as for interfacial tension tests (25 mM HEPES, 200 µM EDTA, pH 7.4), but in null-reflecting water (8.1% w/w D₂O) instead of Milli-Q to eliminate reflection from the aqueous phase. Thirty-five millilitres of sample was held in a Teflon sample trough, within a sealed chamber with single-crystal quartz windows to allow passage of the neutron beam. At least 2 h aging was allowed prior to first measurement of each sample; however, owing to sample sequencing 3–7 h aging was often achieved.

The Motofit analysis program was used to fit the data [25]. This program describes the interface of interest with an arbitrary number of layers. Each of these layers is described by four parameters, a thickness, scattering length density (SLD), solvent penetration and roughness. For these systems, a single-layer model proved suitable for all datasets. Each dataset was fitted simultaneously with its corresponding contrast variant (e.g. the 50% h-DAMP4 data were fitted simultaneously with the 50% d-DAMP4 data). The layer roughness, thickness and solvent penetration values were linked for the h- and d- contrast conditions, and allowed to vary. The scattering lengths of each component were calculated based on the molecular formulae of h-DAMP4, DAMP1 and d-DAMP4 (knowing the extent of deuteration on a non-exchangeable hydrogen basis is 91.1%, see the electronic supplementary material, figure S2), and using tabulated neutron scattering lengths for the atoms C, H, D, O, N and S [26]. The effect of 8.1% w/w D₂O in the solution on scattering length was accounted for assuming 8.1 per cent replacement of exchangeable H atoms by D. Molecular volumes were estimated at 13919.2 ų for DAMP4 (both h- and d-) and 3408.7 ų for DAMP1 using the sum of individual amino acid volumes reported elsewhere [27], as previously done for the related sequence AM1 [28]. Dividing the scattering lengths by molecular volume allowed the SLDs to be estimated: SLD_h-DAMP4 = 1.95 × 10⁻⁶ Å⁻², SLD_d-DAMP4 = 5.96 × 10⁻⁶ Å⁻², SLD_DAMP1 = 1.95 × 10⁻⁶ Å⁻². As DAMP1 and h-DAMP4 have equal SLDs, the SLD for all h-conditions was fixed at 1.95 × 10⁻⁶ Å⁻². The SLD for the 100 per cent d-DAMP4 data was fixed at 5.96 × 10⁻⁶ Å⁻². In the case of mixed d-DAMP4 and h-DAMP1 layers, the SLD parameter depends on the relative proportions of d-DAMP4 and h-DAMP1, so was allowed to vary.

2.5. Langmuir-trough compression tests

A commercially available Langmuir trough (9.2 × 5 cm, Nima Technology, Coventry, UK) was used to study interfacial pressure during a compression and expansion cycle. Prepared samples (40 ml of 8 μM DAMP4, 8 μM DAMP1, or 1.6 μM DAMP4 + 6.4 μM DAMP1 in 25 mM HEPES, 200 μM EDTA, pH 7.4) were poured into the trough, covered and allowed to age for 3 h. After aging, a compression–expansion cycle was performed by linear movement of the Langmuir barriers (at the maximum speed of 114 cm² min⁻¹), from the maximum area of 46 cm² to the minimum of 11 cm² and back to 46 cm². Interfacial pressure during the cycle was monitored with a filter-paper Wilhelmy plate attached to a microbalance, which had been pre-zeroed on the surface of a 40 ml sample of milli-Q water.

2.6. Interfacial extensional rheology

The extensional rheology of adsorbed films at the air–water interface was measured using a custom-made apparatus, the Cambridge interfacial tensiometer, which has been previously described [29]. Samples of interest (6.5 ml) were pipetted into a PTFE trough. Located at the sample air–water interface were two optical fibre T-pieces, one connected to a piezoelectric motor which imposed a strain on the interface, and the other which measured the stress developed in response to this strain via connection to a force transducer. Upon filling of the trough, the moving T-piece was programmed to automatically run once per minute from 0 to 1 per cent linear strain and then back to 0 per cent (at a speed of 10% strain per second). The stress response was monitored, and from this time course an indication of the time taken for mechanical equilibration of the interfacial system was obtained, without destruction of the structure. After 3 h aging and monitoring 1 per cent strain response, a large-strain test to at least 60 per cent strain was performed. The true data frequency is around 100 points per second; however, this is greatly reduced during analysis to enhance graphing clarity.

3. Results and discussion

3.1. Interfacial tension behaviour

The air–water interfacial tension (IFT) kinetics to 5 min (300 s) for DAMP1 solutions at concentrations 2.4, 4.8, 6.4, 7.4 and 8 µM are shown in figure 1a. At 2.4 µM DAMP1, a significant lag time of about 1 min was observed prior to a rapid decrease in IFT to a final value of 53.8 ± 0.2 mN m⁻¹. At all other concentrations, the lag time was not so significant and rapid IFT decrease was complete in less than a minute. The plateau IFT values reached were similar, between 52 and 53 mN m⁻¹, and the effect of concentration on rate of IFT decrease was not substantial over the observed timescale. The interfacial tension kinetics of DAMP1 resemble closely those of AM1 under similar conditions [19], with most of the change occurring in the first 50 s, and reaching statistically identical final values (51.6 ± 0.4 mN m⁻¹ for AM1 [19] and 52.0 ± 0.4 mN m⁻¹ for DAMP1).

Figure 1.

Interfacial tension versus time for (a) DAMP1, (b) DAMP4, (c) mixtures of DAMP4 and DAMP1 in 25 mM HEPES (200 µM EDTA, pH 7.4).

Figure 1b shows the interfacial tension kinetics of DAMP4 at concentrations of 0.6, 1.6, 3.2, 5.6 and 8 µM up to a time of 5 min. It can immediately be seen that DAMP4 adsorption is much slower than that of DAMP1, and the effect of concentration more pronounced. As concentration increases, the rate of interfacial tension decrease (adsorption) is also increased. Unlike DAMP1, which at all concentrations tested completes interfacial adsorption within 5 min, DAMP4 does not reach equilibrium within this time frame for the concentrations tested. Compared with 8μM DAMP1, which takes about 1 min to reach its final value, 8 μM DAMP4 takes more than 5 min. This concentration is on a per-molecule basis and would be four times greater on a per-monomer basis. (i.e. the mass in 8 μM of DAMP4 is four times that in 8 μM DAMP1, table 1). To shed light on the rates of interfacial tension decrease, the approach previously described for similar peptides Lac21 and Lac28 [21], was used to calculate theoretical diffusion time constants, t_d, for DAMP4 and DAMP1, and compared with those experimentally observed (table 2). Adsorption of DAMP1 appears to be diffusion controlled, as the theoretical t_d values are close to those observed experimentally. Adsorption of DAMP4 is much slower than predicted by the diffusion time constants, indicating a significant energy barrier to adsorption exists for DAMP4 that is not present for DAMP1. DAMP4 folds into a very stable four-helix bundle in bulk [15], therefore the slow adsorption rates observed are likely to be due to the significant energy barrier associated with the unfolding of this four-helix bundle in order for adsorption to occur, as observed previously for Lac28 compared with Lac21 [21].

Table 2.

Theoretical and approximate experimental diffusion time constants (t_d) for DAMP4 and DAMP1^a.

DAMP4				DAMP1
concentration, μM DAMP4	concentration, μM monomer equivalent	theoretical td (s)	approx. experimental td (s)	concentration, μM DAMP1	theoretical td (s)	approx. experimental td (s)
0.6	2.4	488.3	>1000	2.4	391.2	150
1.6	6.4	68.7	>800	4.8	97.8	50
3.2	12.8	17.2	>500	6.4	55.0	40
5.6	22.4	5.6	>300	7.4	41.1	30
8	32	2.8	>300	8	35.2	25

^aExperimental diffusion time constants were estimated from interfacial tension data shown in figure 1.

The interfacial tension kinetics at the air–water interface for a DAMP1 and DAMP4 mixed system are shown in figure 1c. The mix conditions chosen corresponded to 90, 75, 50 and 25 per cent DAMP4 on a mass basis, as shown in table 1. The total molar concentration was kept constant at 8 µM under all conditions. The interfacial tension kinetics under all mix conditions are similar, dropping rapidly within the first 30 s to final values of 52–53 mN m⁻¹, which is much faster than the kinetics of the individual components at each mix condition. This is seen most clearly by comparing the kinetics of the 75 per cent DAMP4 condition (3.2 µM DAMP4 + 4.8 µM DAMP1) with that of the individual components (3.2 µM DAMP4 and 4.8 µM DAMP1), which are much slower than when combined. The major reason for the increased adsorption kinetics is the change of total bulk mass concentration of the mixed system. Owing to the larger size of DAMP4 in comparison to DAMP1, an increase in the molar proportion of DAMP4 in the systems in figure 1c results in an increased bulk mass concentration (table 1), which is larger than DAMP1 alone. This increased mass concentration is expected to increase kinetics of protein adsorption at interfaces [30]. Indeed, control experiments show that the DAMP1-only system and the mixture of DAMP1 and DAMP4 have similar dynamic interface tensions when the total concentration is equivalent or higher than 12.8 µM DAMP1 (see the electronic supplementary material, figure S4). While interfacial tension measurements give important information on adsorption kinetics, it should be noted that measurements of interfacial tension cannot be directly interpreted as indications of interfacial adsorption because of nonlinear relationship between interfacial tension and interfacial coverage, as evidenced by direct measurements of surface coverage such as neutron reflectometry [31], and measurement of surface excess using radioactive surfactants [32]. Neutron reflection experiments have thus been carried out to determine composition of the interfacial layer as discussed in §3.2.

3.2. Composition of the interfacial layer (neutron reflectometry with deuterated DAMP4)

To identify the components adsorbed at the interface at equilibrium, neutron reflectometry was used. This technique exploits the fact that the neutron reflectivity of an air–water interface is modulated by adsorbed molecules, and with enough prior knowledge of the system, can provide information such as adsorbed amount and layer thickness.

Figure 2a shows the neutron reflectivity profiles of the same systems as in figure 1c, in the same buffer conditions (25 mM HEPES, 200 µM EDTA, pH 7.4), but in null-reflecting water (8.1% w/w D₂O) instead of Milli-Q in order to eliminate reflection from the solution subphase. Non-deuterated DAMP4 (h-DAMP4) was used. As DAMP4 is a repeat of the DAMP1 sequence, DAMP1 and h-DAMP4 have the same SLDs (1.95 × 10⁻⁶ Å⁻²), and are, therefore, indistinguishable with neutrons. The neutron reflectivity profiles in figure 2a contain information about the overall interfacial film structure, and not the individual components. As the reflectivity curves are almost identical, qualitative observation of figure 2a indicates that the overall structure of the interfacial film formed does not vary with changes in DAMP1 : DAMP4. Given that figure 2a shows equivalent neutron reflectivity profiles at equilibrium for 100 per cent DAMP1, 100 per cent DAMP4 and for mixtures of these two molecules, these results indicate that the surface excess for DAMP1 and DAMP4 are essentially the same despite the fact that the bulk mass of DAMP1 added to the subphase was 1/4 that of DAMP4. No Kiessig fringes were observed in these reflection data owing to the very thin interfacial layers, and the relatively high background resulting from incoherent neutron scattering from the largely protonated aqueous subphase (approx. 92% H₂O).

Figure 2.

Neutron reflectometry profiles of the same systems as in figure 1, with (a) non-deuterated DAMP4 (h-DAMP4), and (b) with deuterated DAMP4 (d-DAMP4). The buffer conditions (25 mM HEPES, 200 µM EDTA, pH 7.4) are the same as figure 1, but in null-reflecting water (8.1% w/w D₂O). DAMP1 and DAMP4 have almost identical neutron SLDs, so the overlap of profiles in (a) shows that there is no difference in the interfacial structure (thickness and surface coverage) for the conditions tested. With increased d-DAMP4 in bulk (b), the reflectivity increases, indicating the presence of d-DAMP4 at the interface. The black lines show the fits to the data. (Online version in colour.)

In order to determine the composition of the interfacial layer, the ‘visibility’ with respect to neutrons of one of the components must differ compared with the other. The conditions in figure 2a were, therefore, repeated in figure 2b using deuterated DAMP4 (d-DAMP4), which has a much higher SLD than DAMP1 (5.96 × 10⁻⁶ compared with 1.95 × 10⁻⁶ Å⁻²), and, therefore, provides contrast variation between the two molecules. Immediately it can be seen that the total reflectivity is greater than that in figure 2a, which indicates that d-DAMP4 is present in the equilibrium interfacial film. Reflectivity increases with increasing bulk d-DAMP4 concentration, which indicates that the proportion of DAMP4 at the interface increases as its bulk concentration is increased. This is a result that could not be determined by interfacial tension profiles (figure 1) which suggest that, on the timescale of minutes, DAMP1 dominates the interface. Here, from neutron reflectometry data, it is evident that both DAMP4 and DAMP1 are co-populating the interface once equilibrium has been established (on the timescale of hours).

A single-layer model fits the neutron profiles very well (black solid lines in figure 2a,b). Previous neutron reflectivity studies showed that a related peptide, AM1, adsorbs as a single monolayer at the air–water interface, about one α-helix width thick (approx. 15 Å) [18]. AM1, and subsequently DAMP4 and DAMP1, are designed with hydrophobic and hydrophilic residues placed strategically to drive helical interfacial structuring [15,17]. DAMP4 and DAMP1 are, therefore, expected to form a similar monolayer structure. For thin layers, the layer thickness and overall SLD of the layer (a volume fraction weighted average of the solvent SLD and protein SLD) can be inversely correlated in the analysis as there are no Kiessig fringes. However the overall excess of SLD (surface excess) at the surface can be determined accurately. This is the case for these measurements; as only one solvent condition (8.1% w/w D₂O) was tested, and the films are thin, it is not possible to determine layer thickness with high precision. However, the fitted thickness values for all datasets were in the range of 10–14 Å, which is approximately the thickness observed for AM1, indicating that a monolayer structure is being formed (figure 3). In the case of DAMP4, this means unfolding of the four-helix bundle existing in bulk (as indicated by CD spectra previously collected [15]) to expose the hydrophobic core, and adsorption to the air–water interface as a connected ‘chain’ of four DAMP1 monomers.

Figure 3.

Adsorption at the air–water interface of unstructured DAMP1 and structured DAMP4. DAMP1 most likely folds into a single helical unit at the air–water interface, whereas DAMP4 unfolds from a four-helix bundle in bulk into a chain of four connected DAMP1 monomers. (Online version in colour.)

The fitting process yielded SLD profiles for the non-deuterated dataset that were integrated to obtain total scattering length/unit area at each experimental condition. As the scattering length per molecule is known, the area per DAMP1 monomer at each experimental condition was calculated (table 3). In general, the values obtained are close, but slightly lower than the area/molecule of 390 Ų found for the similar molecule, AM1 [18], under the same buffer conditions. The pI of AM1 is higher than that of DAMP4 and DAMP1 (pI 8.5 versus 6.7 and 7.4, respectively), therefore the higher area/monomer is likely to be due to the increased charge and therefore lower surface ‘packing’ of AM1 at pH 7.4 compared with DAMP4 and DAMP1. Further, the area per monomer for DAMP1 (350 ± 2.2 Ų) is slightly lower than that for DAMP4 (366 ± 0.7 Ų), which is consistent with their differences in pI. The area per monomer values are similar (ranging from 360 ± 2.2 to 374 ± 1.4 Ų) under all of the mixture conditions.

Table 3.

Parameters from fitting of neutron reflectometry data (figure 2).

DAMP4 bulk mass fraction	area/α-helix (Å2) (from fits of figure 2a)
0	350 ± 2.2
0.25	374 ± 1.4
0.50	362 ± 2.0
0.75	369 ± 2.4
0.90	360 ± 2.2
1.00	366 ± 0.7

The scattering length/area values extracted from fitting of the deuterated dataset were multiplied by corresponding area/monomer values to obtain the average scattering length/helix (SLD_av) at each condition. The fraction of total area accounted for by DAMP4 at each condition, calculated using the formula DAMP4 interfacial fraction = (SLD_av – SLD_DAMP1)/(SLD_d-DAMP4 – SLD_DAMP1) (i.e. assuming a linear relationship between area/helix and SLD) is shown in figure 4. Interestingly, the amount of surface area accounted for by each component is very similar to the mass fractions in the original bulk solution (figure 4). The linear relationship DAMP4 interfacial mass fraction = DAMP4 bulk mass fraction fits well, yielding a R² value of 0.985 (figure 4, dotted line).

Figure 4.

The relationship between interfacial and bulk DAMP4 as determined by fitting of deuterated neutron reflectometry data in figure 2.

3.3. Langmuir-trough compression tests

To investigate interfacial adsorption of these two components further, interfacial pressure isotherms were measured using a Langmuir trough. Figure 5 shows the compression–expansion isotherms for 100 per cent DAMP4, 50 per cent DAMP4/DAMP1 and 100 per cent DAMP1, after each of these adsorbed films were allowed to age for 3 h to reach equilibrium. The starting interfacial pressure for DAMP4 is 25.7 mN m⁻¹, higher than that of DAMP1 and the mixed system, which had starting interfacial pressures of 20.9 and 22.8 mN m⁻¹, respectively. The higher interfacial pressure of DAMP4 corresponds to the low interfacial tension values it reaches after several hours of aging (shown for 3.2 μM DAMP4, electronic supplementary material, figure S3). The area under the curve is related to the energy required for compression. It can be seen that DAMP4 has a greater energy barrier to desorption than DAMP1, consistent with DAMP4 being a larger molecule. The compression isotherm for the mixed system (50% DAMP4) has an area under the curve, which is intermediate to that of DAMP4 and DAMP1. Upon expansion, the slope of the DAMP4 isotherm is much greater than that of DAMP1 and reaches a lower final interfacial pressure. This is due to the slower adsorption kinetics of DAMP4 compared with DAMP1 (observed in figure 1); the slower the interface is repopulated, the more ‘drop’ in interfacial pressure should be observed. In the case of the mixed system, the slope of the curve in the expansion region is between that of DAMP4 and DAMP1. Once the area was returned to the original value, it took the DAMP4 system about 1 h to reach the original interfacial pressure while the DAMP1 system re-reached equilibrium in a few seconds.

Figure 5.

Compression–expansion cycle, performed using a Langmuir trough, for the conditions 100 per cent DAMP4, 100 per cent DAMP1 and 50 per cent DAMP4. Rate of compression was 114 cm² min⁻¹.

3.4. Adsorption behaviour

From the information gathered using interfacial tension measurements (figure 1), neutron reflectometry (figures 2 and 4), and Langmuir-trough tests (figure 5), an energy–state diagram for DAMP1 and DAMP4 from the bulk to the interface can be proposed (figure 6). The slower adsorption of DAMP4 to the interface compared with DAMP1 (figure 1 and expansion region figure 5) indicates that DAMP4 has a much higher energy barrier to overcome. This is shown in (figure 6a) as the greater difference between the bulk and transition states of DAMP4 compared with that of DAMP1 (figure 6b). As previously mentioned, a significant factor which most likely contributes to this large energy barrier slowing DAMP4 adsorption is the need for unfolding of the four-helix bundle. This structure was observed to be very stable as very little helical structure is lost even when the bulk temperature is increased to 90°C [15]. The neutron reflectometry results indicated that the free energy of adsorption of the two components is similar. This is shown in figure 6 as similar changes in energy between the bulk and final interfacially adsorbed states for both DAMP1 and DAMP4. This finding indicates that in this case the free energy of adsorption is a function only of the molecule chemistry, not size. This is interesting from the perspective of competitive adsorption of surface-active molecules, which is not well understood even qualitatively [1]. The compression cycle of the Langmuir-trough experiment (figure 5) confirms that the energy of desorption is higher for DAMP4 than for DAMP1, as depicted in figure 6.

Figure 6.

Possible reaction coordinate diagrams for (a) DAMP4, and (b) DAMP1 adsorbing at the air–water interface.

While DAMP4 has shown a two-state adsorption mechanism in this work, it should be noted that larger proteins may have complicated adsorption kinetics. Adsorption of larger protein often involves multi-state unfolding kinetics, and a true equilibrium is rarely attained on observable timescales [33]. For a mixed system of larger proteins, the interaction of two components can further complicate their adsorption. For example, previous work by Le Floch-Fouéré et al. [34] reported that foaming properties of lysozyme was enhanced by the addition of a small amount of ovalbumin, suggesting possible synergistic adsorption of a mixture of model proteins.

3.5. Influence of molecule length on interfacial rheology

Having confirmed the presence of and quantified the interfacial concentrations of both components, the influence of molecular size on interfacial rheology was then investigated. Figure 7a shows the stress response of DAMP1 and DAMP4 (after aging for 3 h to reach equilibrium) to an interfacial extensional strain deformation of 10 per cent per second. In the initial region of the stress strain curve (less than 5% strain), DAMP1 responds with a somewhat higher stress than DAMP4. It is conceivable that this is due to the slightly greater ‘packing’ of DAMP1 compared with DAMP4 at pH 7.4, as evidenced by its slightly lower area per molecule (table 3). It has been previously observed that the low-strain elasticity of adsorbed proteins at the air–water interface is highest around the pI [35–37], which is likely due to increased packing density and, therefore, ‘jammed system’ behaviour [1]. For peptides similar to the systems reported here, there is a clear correlation between the pH of testing and the measured moduli. Lac21E displayed an extensional elastic modulus of approximately 5 mN m⁻¹ when measured 3.7 pH units away from its pI, but a modulus of approximately 430 mN m⁻¹ close to its pI [18]. AM1 shows a low modulus of less than 30 mN m⁻¹ at approximately 1.1 pH units away from its pI [17], and Lac21 behaves similarly [19]. The stress response of DAMP1 reaches a plateau of 5 mN m⁻¹ at around 10 per cent strain, at which point it is exceeded by that of DAMP4, which continues to increase to a maximum value of approximately 10 mN m⁻¹. This large difference in maximum stress achieved between DAMP1 and DAMP4 is likely to be due to the longer molecule length of DAMP4. The effect of interfacial networking on interfacial rheology of DAMP4 and DAMP1 has been investigated in previous work, by deliberately inducing intermolecular interactions via metal ion–histidine bridging [38]. In the absence of such intermolecular interactions (i.e. in the absence of metal ions), it is unlikely that interfacial networking plays a significant role in the systems reported here. This is supported by the fact that the extensional moduli of DAMP4 and DAMP1 in the presence of histidine-bridging metal ions, as reported in the previous work, is much greater than in its absence, reported in this current work.

Figure 7.

The mechanical response of the DAMP1 and DAMP4 systems to extensional strain. (a) DAMP1 and DAMP4 only, (b) mixed systems. For the DAMP1 and DAMP4-only conditions, only one Maxwell element was required to describe the data to maximum stress (black solid lines, equation (3.1)). In the case of the mixed systems, two Maxwell elements were required (black solid lines, equation (3.2)). (c) Decomposition of the four-parameter Maxwell model for the 90% DAMP4 condition. The dashed lines show the individual contributions of the first Maxwell element (short dashes), and second Maxwell element (long dashes), respectively. (Online version in colour.)

The Maxwell model (equation (3.1)) is the simplest of the spring and dashpot-based mechanical models, which were previously discussed in detail in relation to their application to protein interfaces in the extensional strain model [38].

3.1

where V_M­ is the Maxwell model viscosity parameter, E_M is the Maxwell model elasticity parameter, σ_M is the stress response and t is time. Fitting of the Maxwell model (equation (3.1)) to the DAMP4 and DAMP1 data up to the point of maximum stress (32 and 40% strain, respectively), provided an excellent fit (black lines in figure 7a). The extracted parameters were E_M = 89 mN m⁻¹ and V_M = 104 s mN m⁻¹ for DAMP4, and E_M = 180 mN m⁻¹ and V_M = 48 s mN m⁻¹ for DAMP1. The associated time constants (τ_M = V_M/E_M) are 1.17 s for DAMP4 and 0.27 s for DAMP1, indicating DAMP4 is much slower to respond to extensional strain than DAMP1. This is expected as DAMP4 is much larger, and will take longer to reorient.

The stress response of the mixed DAMP1 and DAMP4 systems previously discussed, as well as two additional conditions, 80 and 95 per cent DAMP4, are shown in figure 7b. There is very little variance of the stress response in the initial strain (less than 5%) region between conditions. From about 5 per cent strain, the curves diverge, and in general, the conditions with higher DAMP4 fraction reach greater maximum stresses.

The two-parameter Maxwell model was not able to provide a satisfactory fit to the data in figure 7b. However, two Maxwell elements in parallel (four-parameter model, equation (3.2)) [38], provided excellent fits to the point of maximum stress for all data (black solid lines).

3.2

This finding suggests there are two relaxation processes occurring in the case of the mixed systems, compared with one when only DAMP1 or DAMP4 are populating the interface. The total response of the four-parameter Maxwell model can be decoupled into the individual contributions of each element, as shown in figure 7c, using the 90 per cent DAMP4 dataset as an example. The short-dotted line represents the first Maxwell element (first term of equation (3.2), characterized by the parameters E₁ and V₁), whereas the long-dashed line represents the second Maxwell element (second term of equation (3.2), characterized by E₂ and V₂). Representing the four-parameter Maxwell model in this way allows the contributions of the two Maxwell elements to be visualized. It can be seen that the first Maxwell element initially contributes almost all of overall stress response, plateauing at around 15–20% strain, at which point the second element begins to dominate.

Figure 8 shows the parameters E_M and V_M obtained from the fits in figure 7a, and E₁, V₁, E₂ and V₂, obtained from fits in figure 7b, as a function of DAMP4 interfacial mass fraction (assuming interfacial DAMP4 mass fraction = bulk DAMP4 mass fraction, as shown in figure 4). Below the 75 per cent condition, the parameters are almost constant. E₂ is almost negligible compared with E₁, and V₁ is greater than V₂, indicating dominance of the first Maxwell element across all strains when less than 75 per cent DAMP4 is present in the interface. Beyond 75 per cent, the influence of the second Maxwell element increases, as evidenced by the increase in E₂, decrease in E₁ and dominance of V₂ over V₁.

Figure 8.

The parameters extracted from fits of the data in figure 7, shown with respect to % DAMP4 at the interface (from figure 4) using equation (3.1) when only one component is present and equation (3.2) when two components are present. (Online version in colour.)

The physical meaning of the first and second Maxwell elements may be hypothesized as being determined by movement in the interfacial plane of DAMP1 and DAMP4, respectively (figure 9). This hypothesis is supported by the fact that in the presence of only one component (pure DAMP1 and pure DAMP4 in figure 7a), only one Maxwell element, not two, is needed. The easier it is for molecules to reorient in response to an applied strain, the less stress will be registered. Such movement will be a function of molecule packing as well as size. This is reflected in figure 7a, where the higher molecule packing (lower area per monomer) of DAMP1 compared with DAMP4 results in a higher initial stress response, then subsequently the larger size of DAMP4 results in a higher maximum stress. With this in mind, figure 8 shows that in response to extensional strain, DAMP4 reorients to redistribute stress very little when DAMP1 is present (figure 9a), except at very high interfacial concentrations (greater than 75%, figure 9b). The essentially unvarying parameters in this region indicate that DAMP1, the smaller, more mobile molecule literally ‘takes the strain’. Beyond this apparent critical concentration, the proportion of DAMP4 becomes significant enough to affect the nature of the response. Transport kinetics from the bulk and subsurface may be expected to play a role on the measured stress response, as previously observed [38], washing-out of the bulk solution caused a higher stress to be registered at strains of greater than 15 per cent for β-lactoglobulin. For the systems studied here however, figure 1c showed that the interfacial tension kinetics were almost identical for all mix ratios, indicating that a factor other than subsurface/bulk diffusion is causing the observed variation in extensional stress response.

Figure 9.

Depiction of interfacially adsorbed DAMP1 and DAMP4 behaviour in response to interfacial extensional strain. (a) At interfacial proportions of less than 75% DAMP4, DAMP1 provides most of the stress-reducing reorientation and movement. (b) Movement and re-orientation of DAMP4 only occurs at high interfacial DAMP4 proportions of greater than 75%. (Online version in colour.)

The question of how a smaller additive will influence the behaviour of a larger molecule is also pertinent in polymer literature [39]. The finding that short chains lower the stress response of long chains is often observed in bulk polymer systems. For example, Kornfield et al. [40] observed that short chains relax first followed by long chains in a polymer bimodal melt. Separate work by Shausberger et al. [41] found that addition of the short-chain polymer to a long-chain polymer in a binary blend had the effect of lowering the plateau modulus and relaxation time of the long-chain molecule alone. As observed in previous work [38], these results further show parallels between polymer bulk mechanical behaviour and protein interfacial mechanical behaviour, and indicate the field of protein interfacial rheology could benefit from application of theoretical frameworks existing in the mature field of polymer rheology. Designed sequences such as those introduced here would serve as effective tools in validating such approaches.

4. Conclusions

Using designed sequences, the effect of molecular length on interfacial rheology and structure was investigated. As the molecule chemistry was unchanged, the contribution of other factors such as altered charge structure was eliminated, allowing the role of molecule size to be isolated and studied in a way which is difficult to do using natural proteins. The rate of adsorption of the larger molecule, DAMP4, was much slower than the smaller molecule, DAMP1. Interfacial tension kinetics of mixed DAMP4/DAMP1 systems were much faster than those of the individual components, indicating a synergistic interaction. The reason for the much slower adsorption of DAMP4 compared with DAMP1 is likely to be due to its bulk structure, a stable four-helix bundle, which requires a significant energy barrier to unfold at the air–water interface. DAMP1 adsorption, however, was diffusion-limited, as it is largely unstructured in bulk.

Using neutron reflectometry, it was found that the interfacial film structure at equilibrium of all conditions tested was a monolayer of one α-helix in thickness, similar to the related molecule, AM1 [18]. DAMP4 is, therefore, likely unfolding into a chain of four DAMP1 monomers at the air interface. This was the expected structure as the molecules are designed to form α-helices when adsorbed interfacially. The area per molecule (related to the inverse of surface coverage) was lower for pure DAMP1 than for pure DAMP4. This was attributed to the higher expected charge, therefore greater molecular repulsion, of DAMP4 owing to its slightly lower pI (pI 6.7 compared with 7.4 for DAMP1). The area per DAMP1 monomer did not vary significantly with changes in composition of the mixed systems. The relative concentrations of DAMP4 and DAMP1 at the interface were close to equal to their original relative mass concentrations in bulk. This finding indicates that the free energy of adsorption is similar for both DAMP1 and DAMP4 and only depends on the molecule chemistry, not size. However, the energy barrier to adsorption and desorption is much higher for the larger molecule DAMP4, as evidenced by the interfacial tension results and Langmuir-trough compression tests.

Insights into the mobility of DAMP1 and DAMP4 molecules at the interface were gained by studying their stress response to a large (60%) extensional strain applied at constant rate. Below 5 per cent strain, DAMP1 displays a higher stress response than DAMP4 or the mixed systems. This is attributed to its higher packing at the pH tested, as evidenced by the area/helix values obtained by neutron reflectometry. The single-component data (i.e. pure DAMP4 and DAMP1) were successfully fitted by single-element Maxwell models (one associated time constant). However, models with two elements (four-parameter Maxwell model) were required to fit the mixed-system data, as two relaxation processes (i.e. movement of both DAMP4 and DAMP1 in response to imposed strain) occur. The relationship between interfacial composition (% DAMP4) and interfacial extensional parameters was able to be plotted using the finding from neutron reflectometry results that the interfacial proportions of DAMP4 and DAMP1 are equal to the bulk mass proportions. It was proposed that the physical reason underlying the two relaxation process is the movement of DAMP1 monomers and movement of the DAMP4 four-monomer chains in the interfacial plane. The contribution of the second Maxwell element (i.e. movement of DAMP4) was only significant at very high interfacial concentrations of DAMP4 (greater than 75%), indicating that DAMP4 stays essentially immobile when enough DAMP1 is present to ‘take the strain’, as it is easier for the smaller molecule to reorient and transfer. These results highlight the qualitative analogy between interfacial protein rheology and bulk polymer rheology.

The approach applied in this work, namely using designed sequences with well-defined size, chemistry and bulk structure combined with contrast-varied neutron reflectometry, accessed information which is difficult to gain with naturally occurring proteins. Designed peptide/protein molecules offer a customizable ‘tool kit’ to enable study into the fundamentals of protein interfacial structure and behaviour which is difficult to access with naturally existing systems. In this work, we demonstrated this for the case of a mixed small, unstructured peptide with a large, structured protein.