Insight into the self-assembly and gel formation of a bioactive peptide derived from bovine casein

Highlights

• •Capping and uncapping a food-derived bioactive peptide altered its self-assembling properties.
• •Capped FFVAPFPEVFGK formed a self-supporting gel made of a dense fibrillar network.
• •Peptide concentration and incubation time influenced the gel mechanical properties of capped FFVAPFPEVFGK.
• •Uncapped FFVAPFPEVFGK demonstrated fibrillation potential.

Structured abstract

Abstract

The self-assembling and gelation properties of a bioactive peptide derived from bovine casein (FFVAPFPEVFGK) were studied in the peptide's natural form (uncapped, uncapFFV) and capped with protecting groups added to both termini (capped, capFFV). Although the natural peptide (uncapFFV) did not demonstrate self-assembly, the capped peptide (capFFV) spontaneously self-assembled and formed a self-supporting gel. Variations in peptide concentration and incubation time influenced the gel's mechanical properties, suggesting the peptide's properties could be tuned and exploited for different applications. These results suggest that food-derived bioactive peptides have good potential for self-assembly and therefore utilisation as gels in functional foods and nutraceuticals.

Background

Self-assembly is a natural phenomenon that occurs in many fundamental biological processes. Some peptides can self-assemble and form gels with tunable properties under given conditions. These properties, along with peptide bioactivity, can be combined to make unique biomaterials. Instead of synthesising the self-assembling bioactive peptides, we aim to extract them from natural sources. In order to use these peptides for different applications, it is essential to understand how we can trigger self-assembly and optimise the assembly conditions of these peptide gels.

Scope

The self-assembling and gelation properties of a bioactive peptide derived from bovine casein (FFVAPFPEVFGK) were studied in the peptide's natural form (uncapped, uncapFFV) and capped with protecting groups added to both termini (capped, capFFV).

Major conclusions

Although the natural peptide (uncapFFV) did not demonstrate self-assembly, the capped peptide (capFFV) spontaneously self-assembled and formed a self-supporting gel. Variations in peptide concentration and incubation time influenced the gel's mechanical properties, suggesting the peptide's properties could be tuned and exploited for different applications.

General significance

These results suggest that food-derived bioactive peptides have good potential for self-assembly and therefore utilisation as gels in functional foods and nutraceuticals.

Statement of significance.

Self-assembly is a natural phenomenon that occurs in many fundamental biological processes. Some peptides can self-assemble and form gels with tunable properties under given conditions. These properties, along with peptide bioactivity, can be combined to make unique biomaterials. Instead of synthesising the self-assembling bioactive peptides, we aim to extract them from natural sources. In order to use these peptides for different applications, it is essential to understand how we can trigger self-assembly and optimise the assembly conditions of these peptide gels. This paper describes how parameters such as the peptide concentration, net charge, and incubation time can influence the resulting properties of the peptide gels and nanostructures.

Alt-text: Unlabelled box

1. Introduction

Self-assembling peptides have been under the spotlight for a few decades now and have been used in a wide range of applications, including cell culture [1], tissue engineering [2], and drug delivery [3]. In addition to their short length, biocompatibility, and ease of synthesis, one key characteristic of self-assembling peptides is their capacity to form structures such as hydrogels [2,[4], [5], [6]]. Most self-assembling peptides are chemically synthesised, which can sometimes be associated with high production costs and therefore limits their use as functional ingredients. Finding a source of readily available peptides that can be manufactured at scale is therefore a useful contribution. A potentially viable route to manufacturing gels without the need for expensive synthesis is to source peptides from food, particularly from secondary food processing streams. Food is a well-known source of bioactive peptides and potentially self-assembling ones [7]. In a recent review paper, we discussed the potential for combining peptide self-assembly properties with bioactivity to create interesting systems, termed self-assembling bioactive peptides (SABP), that can be used for diverse applications, such as oral delivery [7]. The resulting system has the potential to avoid the need for delicate entrapment of a drug or natural bioactive compound [7]. Moreover, one advantage of using SABP as oral delivery systems would be a higher resistance to the harsh conditions of the gastrointestinal tract making the bioactive less accessible for breakdown resulting in a delayed release of the bioactive [7]. It could then be used at lower concentrations than conventional treatments as the structure would protect the bioactive helping it reach the lower intestine less degraded. In order to find such SABP in food, it is possible to perform an in silico analysis using online databases and predictive tools. Databases such as the BIOPEP-UWM are fruitful resources for bioactive peptides that are of interest not only for self-assembly purposes but for the discovery of new functional foods [8,9]. Predictive tools such as PeptideCutter and PeptideMass have proven useful to predict the release of peptides from the cleavage of a given protein using specific enzymes [10]. Another useful predictive tool is Gel Predictor, developed by Gupta et al., predicting the gelation potential of given molecules [11]. Thus, by combining such tools, one can identify peptides under specific criteria, e.g. bioactivity, extraction method, gelation properties, and protein source. In our case, we examined food-derived bioactive peptides that can be extracted using enzymatic digestion with common digestive enzymes such as trypsin, pepsin, and chymotrypsin, and with predicted gelation properties. The gelation potential of the peptides can then be experimentally determined with synthetic peptides before continuing with the extraction of the peptides from the food protein. This paper describes the self-assembly and gelation potential of a synthetic bioactive peptide derived from bovine casein: FFVAPFPEVFGK identified with in silico tools. In particular, a focus was brought on the influence of peptide concentration, incubation time and net charge on the self-assembling properties of the synthetic peptide. In addition, the influence of end-capping (acetyl and amide groups) was assessed by using two synthetic peptides, either capped (capFFV) or uncapped (uncapFFV). The nanostructures formed were studied using a combination of spectroscopic and microscopic techniques to try to unravel the mechanism behind the peptide self-assembly process. Furthermore, the gel formation was monitored, and the rheological properties of the resulting networks were analysed with oscillatory rheology.

2. Materials and methods

2.1. Materials

Synthetic capFFV and uncapFFV peptides were purchased from Mimotopes (Australia) with a purity $\ge$ 95%. CapFFV comprised an acetyl group (-CH₃CO) “Ac-” at the N-terminus and an amide group “-NH₂” at the C-terminus. UncapFFV did not contain any end-capping. All chemicals, reagents, and solvents were of analytical grade and were purchased from Sigma-Aldrich Corporation, BDH Laboratory Supplies or Invitrogen.

2.2. Computational approach to predict gelling potential

Prior to the experiments described in this paper, FFVAPFPEVFGK peptide was identified with an in silico analysis. In particular, the peptide was obtained from the BIOPEP-UWM database and selected for its gelation potential, extraction method (enzymatic digestion with common digestive enzymes) and origin (food protein). The gelation potential was estimated by the computational approach developed by Gupta et al. [11]. A score above the arbitrary threshold of 50% was necessary for the peptides to be further investigated. The algorithm calculates the probability that a peptide sequence will form a gel within a domain of applicability. The authors defined the “applicability domain” as the chemical space in which the predictive model can be used with confidence. This domain of applicability implies that the molecules tested when making a prediction are chemically similar to what the model has encountered previously, that is dipeptides [11]. The model developed by Gupta et al. was based on dipeptides rather than larger peptides, and it was therefore expected that the peptides would fall outside of this domain of applicability as they are larger than dipeptides. Nevertheless, a molecule that lies outside the applicability domain does not indicate an incorrect prediction but instead helps provide researchers with confidence for selecting a given molecule for further experiments. For this reason, the domain of applicability was considered but was not eliminatory.

2.3. Gelling experiments

The pH of the synthetic peptides was acidic upon receipt and was thus readjusted to near-neutral pH, between 7 and 7.5, using 0.2 M ammonium hydroxide. The peptide solutions were then freeze-dried and stored at −20 °C until use. The lyophilised peptides were dissolved in buffers of varying pH at different concentrations. The peptide solutions were vortexed, sonicated for 10 min or more to ensure complete dissolution of the powder, then left at room temperature. Over the next three days, the gelation capacity was visually determined at 2 h-intervals. Gelation was indicated when the solution did not exhibit free-flowing properties [12,13]. The buffers were as follows: A (glycine-HCl) and B (citric acid/sodium citrate) pH 3.5, C (citric acid/sodium citrate) pH 4, D (phosphate) and E (Tris–HCl) pH 7.5, F (glycine-NaOH) pH 10, G (glycine-NaOH) pH 10.5, H (Na₂HPO₄—NaOH) pH 11.5, and I (KCl-NaOH) pH 12.5. All buffers were made to a final concentration of 50 mM.

2.4. Thioflavin T assay

The thioflavin T (ThT) assay was used to detect the presence of amyloid fibrils and followed in-house methods based on Medini et al. [14] Briefly, ThT solution was made fresh before each use by dissolving 2.5 mM ThT in ThT buffer (50 mM Tris-base, 100 mM NaCl at pH 7.5). The solution was filtered (0.22 μm pore size) and stored in the dark until use. Samples and ThT solutions were added into 96-well, clear bottom NuncTM fluorescence well plates (Thermo Fisher Scientific). Samples of capFFV at 2% w/v and in triplicate were incubated with ThT dye for 6 min in the dark, and fluorescence emission intensity was measured with excitation/emission filters (430 and 485 nm, respectively) with a SpectraMax iD3 (Molecular Devices, USA). Bovine insulin amyloid fibrils were used as a positive control, due to the confirmed presence of β-sheet structures [15], and were prepared as described in [16]. Briefly, bovine insulin (Sigma-Aldrich) was diluted in 100 mM NaCl, 25 mM HCl, pH 1.6 buffer to obtain a final concentration of 5.8 mg/mL and heated at 60 °C for 24 h [16].

2.5. Transmission electron microscopy

TEM samples from capFFV at 0.6, 1, and 2% w/v were prepared using Formvar-coated copper TEM grids (400 mesh). Grids were prepared by glow discharge as described by Medini et al. [14]. TEM micrographs were taken using a Tecnai F20 (FEI Company, Oregon, USA) operating at 200 kV and a Morgagni 268D TEM (FEI Company, Oregon, USA) operating at 80 kV.

2.6. Attenuated total reflectance fourier-transform infrared spectroscopy

Attenuated total reflectance Fourier-transform infrared spectroscopy (ATR-FTIR) spectra were recorded using a Bruker Vertex 70 FTIR spectrometer equipped with the platinum diamond Micro-ATR unit. The ATR-FTIR spectra were obtained in the spectral region of 400– 4000 cm⁻¹, using a resolution of 4 cm⁻¹ and 16 scans for each sample from capFFV peptide. All spectra were normalised and baseline-corrected with Opus v7.8 software.

2.7. Atomic force microscopy

Gel samples from capFFV in buffer A were diluted to 1% w/v (T_0min and T_2h) and 0.5 w/v (T_7days) in buffer A and applied to freshly cleaved silicon square (∼ 1 cm²) surfaces prior to AFM analysis. The squares were washed with milliQ water and the excess of water was removed with filter paper. The samples were then allowed to air-dry prior to imaging. Each sample was prepared in triplicate. AFM imaging was performed in air using repulsive tapping mode with an Asylum Research Cypher ES (Oxford Instruments, US) and TAP-150AL-G probe coated with aluminium reflex probe (Budget Sensors, Innovative Solutions Bulgaria Ltd). The images were acquired with a resonance frequency of 150 kHz and a force constant of 9.5136 N/m, at scan speed of 2.44 Hz. The images were flattened, and the height profiles were obtained with Gwyddion analysis software [17]. For the 7-day incubated gel, lengths of at least 30 fibrils were measured.

2.8. Oscillatory rheology

Oscillatory rheology measurements of the gels from capFFV were carried out with an CR302 rheometer (Anton Paar) operating in cone-plate mode (cone angle: 0.979°, diameter: 24.954 mm, truncation: 48 μm) with a gap height of 0.048 mm. A solvent trap was used to minimise evaporation. All measurements were acquired at 20 °C. Dynamic time sweeps (DTS) were performed on a freshly prepared peptide solution at 2 w/v to capture the gelation process, a 7-day incubated gel at 2% w/v, and 24-hour incubated peptide solutions at 1% w/v and 0.6% w/v, prepared as described above. The storage modulus (G’) and loss modulus (G’’) were monitored over 1 or 2 h with a frequency of 1 rad/s, and a constant shear train of 0.001. The DTS experiment was followed by amplitude sweep measurements to study the viscoelastic properties of the gels. Amplitude sweep measurements were performed within a shear strain range of 0.1–1000% and an angular frequency of 1 rad/s.

2.9. Comparison with uncapFFV

To compare with the capped peptide capFFV, the self-assembly and gelation properties of the uncapped peptide, uncapFFV, were studied as a function of the net charge by varying the pH of the buffer. The peptide solutions were prepared as described in Section 2.3. The buffers used were as follows: B (citric acid/sodium citrate) pH 3.5, J (citric acid/sodium citrate) pH 4.5, K (phosphate) pH 6.5, L (Tris–HCl) pH 8.5, G (glycine-NaOH) pH 10.5. TEM images of uncapFFV at 2% w/v were also taken and the samples were prepared as described in Section 2.5.

3. Results

3.1. Gelation potential with gel predictor tool

The predictive tool developed by Gupta et al. was used to estimate the gelation potential of the peptides (capFFV and uncapFFV). Both peptides were predicted to gel with a score above the arbitrary threshold.

3.2. Peptide self-assembly and gel formation

The self-assembly and gelation properties of capFFV and uncapFFV peptides were studied as a function of the pH and thus their net charge. CapFFV peptide was insoluble at pH > 3.5 (charge < +1) and no gelation was observed (data not shown). The peptide was soluble in buffers A and B where it formed a self-supporting gel (no flow was observed upon inversion of the vial) within 2 h. The following sections present the results obtained with the peptide that was able to self-assemble and form a gel: capFFV in buffer A. The experiments described in Section 2 were not performed on uncapFFV. This is discussed in Section 4.4.

3.3. TEM

TEM images of the peptide after preparation of the sample, T₀ (Fig. 1, A), after 2 h of incubation (Fig. 2, B) (gelling time point), and after 7 days of incubation (Fig. 1, C) all clearly indicate the formation of a dense network of entangled fibrils from the peptide. This result was expected as peptide gels are known to often be formed from fibrils that, under appropriate conditions, absorb large amounts of water leading to gelation [1,14,[18], [19], [20], [21], [22], [23], [24], [25]]. For the samples at T₀ and T_2h, clusters of fibrils that seemed to be folded back on themselves, and/or with other fibrils were observed. Between these clusters, long and twisted/folded fibrils elongated. The fibrils observed in the 7-day incubated gel were more well-organised, entangled, and straighter (Fig. 1, C) than the fibrils observed at T_0min and T_2h, which were overall shorter and appeared less organised (Fig. 1, A-B).

Fig. 1.

Representative TEM images of fibrils formed from capFFV at (A) T_0min (2% w/v), (B) T_2h (dilution of the gel to 1% w/v), and (C) T_7days (dilution of the gel to 0.5% w/v). Scale bars are 100 nm.

Fig. 2.

TEM images of fibrils formed from fresh solutions of capFFV with final concentrations of (A) 0.6% w/v, (B) 1% w/v and from a diluted gel: (C) 0.6% w/v, and (D) 1% w/v. Scale bars are 100 nm (A, B and D) and 50 nm (C).

The influence of peptide concentration on the ability of capFFV to form fibrils was assessed by preparing fresh peptide solutions at 0.6% w/v and 1% w/v, and by diluting a 2% w/v self-supporting gel to the same concentrations. It must be noted that none of the fresh solutions (Fig. 2, top row) had formed a gel when the TEM images were taken (10 days). Similar nanostructures to the ones obtained at 2% w/v were observed for both concentrations. However, at 0.6% w/v, it was possible to observe nucleation (nuclei), suggesting the peptide self-assembly process was in the lag phase (Fig. 2, A) [26,27]. TEM images of the diluted gel show a dense network of fibrils for both concentrations (Fig. 2, bottom row). Nucleation was also present at 0.6% w/v (Fig. 2, C). A mixture of long and short fibrils was observed at 1% w/v (Fig. 2, D), while nucleation was not observed at this concentration.

3.4. ThT assay

The ThT assay was used to confirm the presence of β-sheet fibrils in the gel formed from capFFV and observed with TEM. This assay is widely used as an indicator of β-sheet [[28], [29], [30]]. Many peptides form such structures when they self-associate [14,31,32]. ThT fluorescence was measured over time starting at T_0min after preparation of the peptide solution (time lag of ∼ 15 min). Measurements were taken every half hour over 2 h, and after 48 h of incubation of the sample. Contrary to the usual sigmoidal pattern observed for the fibrillation phenomenon [27], Fig. 3, A shows high fluorescence intensity at the start of the assay, T_0min (time lag of ∼ 15 min due to peptide solution preparation). A slight increase in the fluorescence intensity was observed at 2 h, i.e. the time point at which a self-supporting gel was observed (Fig. 3, A). However, it was not clear whether this increase was due to the heterogenicity of the sample or a phenomenon occurring at the molecular level, so it was not possible to conclude based on the results from the ThT assay only.

Fig. 3.

The secondary structure of the peptide gel was evidenced with the ThT assay and FTIR spectroscopy (amide I region). (A) ThT fluorescence over time of capFFV at 2% w/v in buffer A and +1 charge; (B) FTIR spectra of the gel formed from capFFV in buffer A at 2% w/v at T_0min (orange curve) and T_2h (purple curve); (C) FTIR spectra of fresh solutions of capFFV in buffer A at 0.6% (red curve) and 1% w/v (blue curve). All FTIR spectra were background subtracted and the absorbance of the two spectra was normalised in regard to the intensity of the highest absorption band.

3.5. ATR-FTIR

Attenuated total reflectance Fourier-transform infrared spectroscopy (ATR-FTIR) was used to further characterise the peptide fibril nanostructures by focusing on the vibration of the amide I band (∼1600–1700 cm⁻¹) resulting from the C = O stretching vibration [33,34]. A peak attributed to β-sheet secondary structure component was evident at 1625 cm⁻¹ (Fig. 3, B) and indicated the peptide gel (2% w/v) formed from capFFV contained β-sheets [34]. The overlapping of the spectra at T_0min and T_2h showed no significant changes in the β-sheet structure of the peptide over the gelation time frame, indicating the fibrils did not go under significant rearrangement of the secondary structure over the course of two hours and remained under β-sheet conformation. Fresh solutions of capFFV at 0.6% w/v and 1% w/v were also analysed and a peak at 1625 cm⁻¹ attributed to β-sheet [34] was evident (Fig. 3, C) and confirmed the fibrils observed with TEM were β-sheet. The shoulder peak around 1595 cm⁻¹ could be ascribed to glutamic acid side chain group [35].

3.6. Atomic force microscopy

Atomic Force Microscopy (AFM) is a technique used to measure the topography of samples by measuring the height across a surface and is widely used for fibril characterisation [36,37]. AFM analysis was conducted to characterise the structure and morphology of fibrils formed from capFFV at different incubation times and corroborate the TEM observations. AFM images of capFFV at T_0min and T_2h revealed bundles of rather short (the size of the fibrils could not be measured precisely due to their entangled organisation but were overall < 1 μm), coiled and entangled fibrils (Fig. 4, A-B), while fibrils obtained after 7 days of incubation were overall long (1014.7 nm ± 652.6 nm) and straight (Fig. 4, C) supporting the TEM observations.

Fig. 4.

The evolution of the fibril network formed from CapFFV peptide was followed at different times of incubation with AFM. (A) T_0min, (B) T_2h, and (C) T_7days. Scale bars are 500 nm (A), 400 nm (B), and 2 μm (C).

3.7. Oscillatory rheology

The gel formation and mechanical properties of capFFV were studied with oscillatory rheology measurements for gels of varying incubation times (2 h, 7 days) and concentrations (0.6, 1 and 2% w/v). First, DTS experiments were conducted to assess the gelation and stiffness of the gels by monitoring the storage modulus (G’) and loss modulus (G’’) over time, at constant shear strain and frequency, and within the linear viscoelastic region (LVE) (Fig. 5, top row). Then, amplitude sweep measurements were carried out to study the viscoelastic properties of the gels (Fig. 5, bottom row). The results from the oscillatory rheology experiments presented in Table 1 showed that all samples were hydrogels characterised by elastic-dominated behaviour (G’>G’’) over the range of concentrations and periods of incubation studied. Variations in stiffness distinguished the hydrogels, with increased concentration resulting in increased stiffness. In particular, the sample incubated for 7 days was the stiffest gel with a G’ value of ∼ 20 kPa after 1 hour of DTS experiment (Table 1). The amplitude sweep measurements highlighted the real-time formed gel, the 7-day incubated gel, and the 1% sample had similar limiting strain amplitude, γ_L, of 0.0148 strains which marks the transition between the linear and nonlinear regions. The real-time formed gel and the 7-day incubated gel samples also shared the same flow point γ_y value of 0.47 at which G’=G’’. However, the LVE region was not easily observed for the 0.6% w/v sample. Interestingly, the flow point γ_y (G’=G’’) was determined at 1 shear strains for the 0.6% w/v and the 1% w/v samples, which is higher than the γ_y value of the gel samples.

Fig. 5.

The mechanical properties of the gels were estimated with oscillatory rheology. Top row: DTS measurements: (A) 2% w/v 7-day incubated gel, (B) 0.6% and 1% w/v samples, (C) 2% w/v real time-forming gel. Bottom row: amplitude sweep measurements: (D) 2% w/v 7-day incubated gel, (E) 0.6% and 1% w/v samples, (F) 2% w/v real time-forming gel.

Table 1.

G’ values of the different samples at different time points of the DTS experiment.

Concentration (% w/v)	Incubation time	G’ at the start (Pa)	G’ after 1 hour (Pa)	G’ after 2 h (Pa)
2	7 days	∼ 6.5.103	∼ 20.103	n/a
2	2 h	∼ 1.4	∼ 13.8	∼ 30
1	24 h	∼ 9.6	∼ 16.5	n/a
0.6	24 h	∼ 0.8	∼ 0.9	n/a

n/a = not applicable.

3.8. UncapFFV

In the same way as capFFV, the self-assembling and gelation properties of uncapFFV were studied as a function of the pH. While capFFV comprised two protective groups at both termini, the charge of the amine and carboxylic groups of uncapFFV varied with the pH. As a result, the peptide net charge was distributed differently across the pH scale. Contrary to capFFV that self-assembled at charge +1, uncapFFV was insoluble at an equivalent charge in buffer J (pH 4.5), and it was also insoluble in buffer B (+2 charge, pH 3.5). However, the peptide was soluble in the other buffers for charges $\le$0. Nevertheless, no self-supporting gel was observed at any of the pHs and charges tried. This was unexpected as the peptide was selected for its gel-forming potential during the in silico approach described above. At neutral charge, the peptide was found to form amorphous aggregates (Fig. 6, A). Some fibrillar structures and nuclei were seen in buffers L and G (Fig. 6, B-C). For both buffers, the addition of lyophilised peptide resulted in a decrease in the final pH of around 1 unit. At pH 7.5 in buffer L, the calculated charge was close to 0 and −1 at pH 9.5 in buffer G [38,39]. A strategy to accentuate self-assembly observed in buffer G was to increase the peptide concentration to favour peptide-peptide interaction [20,40]. For this, the pH of uncapFFV was adjusted to pH 9.5 and pH 10.5 to obtain a peptide net charge between −1 and −2. At pH 10.5 and −1.5 charge, uncapFFV formed amorphous aggregates reminiscent of the ones obtained at pH 6.5 at a neutral charge (Fig. 6, D). TEM images of the peptide at pH 9.5 and −1 charge showed peptide aggregates but also short fibrils along with nucleation suggesting the peptide was still in a lag or nucleation phase (Fig. 6, E).

Fig. 6.

TEM images of uncapFFV (A) at neutral charge, (B) in buffer L, (C) in buffer M, (D) at pH 10.5, (E) at pH 9.5. Scale bars are: 0.2 µM (A), 100 nm (B, C), and 200 nm (D, E).

4. Discussion

In this study, the self-assembling and gelation properties of a food-derived bioactive peptide were investigated using synthetic peptides. Self-assembling bioactive peptides (SABP) are of special interest as they represent a relatively simple way of delivering bioactives compared to conventional treatments and delivery routes. Furthermore, food sources such as co-products or secondary processing streams hold promise as viable and economical raw materials for engineering peptide gels while circumventing the need for de novo synthesis [7]. This study aimed to explore and inform on the assembly of synthetic peptides derived from bulk proteins down the track, and in our case for oral delivery applications. For this purpose, we used a combination of in silico tools to help us identify candidate food-derived SABP, and we selected FFVAPFPEVFGK peptide, a bioactive peptide derived from bovine casein. Prior to selecting the peptide, its gelation potential was estimated by a computational approach [11]. In this paper, we studied the gelation potential of the peptide under its natural form (FFVAPFPEVFGK, uncapFFV), and we also investigated the influence of protecting groups on the self-assembling properties using a capped peptide (Ac-FFVAPFPEVFGK-NH₂, capFFV), using synthetic peptides in both cases. Capping groups have been shown to influence the self-assembling and gelation properties of peptides [41,42]. Therefore, this study helps better understand the chemistry behind the self-assembling mechanism of FFVAPFPEVFGK peptide and will, in turn, help design peptides with the right self-assembling properties for specific applications. The first part of the study consisted of triggering the self-assembly of capFFV and uncapFFV peptides by varying the pH and, by extension, the peptide's net charge. This method was chosen as it is a simple and effective way to trigger self-assembly through protonation/deprotonation of the functional groups [43]. Next, the morphology and secondary structure of the resulting nanostructures as well as the mechanical properties of the hydrogels were characterised with microscopic, spectroscopic and rheology measurements. Finally, the peptide concentration and incubation time influence on the morphology, secondary structure, and mechanical properties were investigated. This discussion will first present the results obtained for the capped peptide, capFFV, and then compare with the uncapped peptide, uncapFFV.

4.1. Triggering self-assembly and gelation through variations in pH

A buffer was prepared at each pH corresponding to a protonation/deprotonation of a functional group, i.e. a change in the peptide net charge, which is, in the case of capFFV, from glutamic acid (E) and lysine (K) residues. CapFFV was insoluble when the overall charge was < +1. Despite the presence of charged amino acid residues (K, E) that could favour the peptide solubility, capFFV contains four F, two V residues and one A residue, all hydrophobic residues [44,45]. Common solvents used for dissolving peptides include dimethyl sulfoxide (DMSO), acetonitrile, and dimethylformamide (DMF) [44]. However, these solvents were not considered here as the end goal application chosen for the food-derived SABP involved human consumption. Interestingly, capFFV was soluble in buffer A and formed a self-supporting gel (no flow was observed upon inversion of the vial) within 2 h. At pH 3.5, capFFV carries a positive charge from K protonated side-chain group, indicating like-charge repulsion of K [46]. If repulsion forces only were present or were the dominating forces, no assembly would be observed and indicates that the contribution from electrostatic interactions is not the only factor driving the self-assembly of capFFV. Instead, it is suspected that other interactions such as hydrophobic interactions and aromatic π-π stacking (from the F residues) along with hydrogen bonding played a role in the mechanism of self-assembly of capFFV and were sufficient to overcome electrostatic repulsion forces to lead to optimum self-assembling conditions [46,47]. Moreover, the distribution of the charges along the peptide sequence is also suspected to play a role in the self-assembling mechanism of capFFV. It is suspected that the presence of a positive charge from K at the C-terminus allows for other interactions to occur along the peptide sequence under these conditions. Furthermore, aromatic π-π stacking is known to contribute to the self-assembly and gelation of peptides containing F residues [19,48]. For these reasons, our hypothesis behind the self-assembly mechanism of capFFV is based on a combination of hydrophobic interactions, aromatic π-π stacking, hydrogen bonding, electrostatic forces, and the distribution of the charges in the peptide sequence . On the contrary, uncapFFV was insoluble at charge $\ge$+0.5 and soluble at charge $\le$0, but it did not form a gel at any of the pHs tried. These observations showed the presence of protecting groups clearly influenced the capacity to self-assemble as this is the only parameter that differed between the two peptides. Strategies employed to trigger the self-assembly of the peptide will be further discussed in Section 4.4. The TEM and ThT results indicated that capFFV self-assembles instantaneously at low pH and an overall charge of +1. The results also highlighted that the fibrillation of the peptide occurs on a shorter time scale (instantaneous) than gelation (2 h) and agree with previous findings in the literature [49]. A decrease in fluorescence over time was also observed. This could indicate all monomers have been incorporated into fibrils and/or the formation of mature fibrils. In the presence of fibrils, the surface area for ThT to bind decreases as a result of β-sheets already entangled with one another, forming the fibrils. The fluorescence decrease could also result from the sedimentation of the fibrils observed at the bottom of the microplate, after reaching a specific size [50]. In addition, FTIR analyses and the ThT assay revealed that the peptide nanostructure was made of β-sheet fibrils that formed spontaneously and did not undergo any significant changes over the gelation timeframe of 2 h. On the 3D level (or macromolecular level), rheological measurements followed the real-time formation of the gel. They indicated that the gel structure formed rapidly (within the first 15 s of the test despite a lag phase of 15 min) and continued forming overtime and became stiffer as observed with the self-supporting behaviour/no flow was observed upon inversion of the vial.

4.2. Influence of peptide concentration

Previous studies have characterised the effect of peptide concentration on self-assembly and gelation properties [31,40,51,52]. In particular, high peptide concentrations can favour peptide-peptide interactions and the formation of supramolecular networks [40]. The influence of peptide concentration on the ability of capFFV to form fibrils was assessed by preparing fresh peptide solutions at 0.6% and 1% w/v, and by diluting a 2% w/v self-supporting gel to the same concentrations. The hypothesis was that the formation of fibrils and gelation was concentration-dependant. The results obtained with the fresh peptide solutions at 0.6% and 1% w/v confirmed the two-step gelation process of capFFV, with fibrillation and gelation occurring on a different time scale. Furthermore, the concentration-dependency of fibrillation and gelation was validated as fewer fibrils along with nucleation were observed at lower concentrations, and no gel was obtained after more than 10 days of incubation. Moreover, the TEM images from the fresh solutions and diluted ones showed remarkably similar nanofibril networks, implying fibrillation was very fast. In addition, FTIR analyses carried out on the fresh peptide solutions demonstrated that the concentration did not affect the secondary structure as the fibrils were formed from β-sheets. Interestingly, results from oscillatory rheology demonstrated that although no gel was observed for both concentrations, an elastic-dominated behaviour was evident (G’ > G’’), while the lowest concentrated sample (0.6% w/v) was characterised as a weaker gel of higher fluidity. An interesting finding here was obtained with the amplitude sweep measurements. Indeed, it was found that the low concentrated samples could resist slightly higher strains before they started to flow. Such behaviour is characteristic of a sparsely cross-linked network of flexibles and potentially entangled network of flexible chains, unlike the highly concentrated gel that resisted lower strains and is characterised by a densely cross-linked – entangled network of rather rigid links [53]. Zhao et al. also observed a greater resistance to deformation from gels of lower concentration [54]. Concentration seemed to influence the gels' mechanical properties, especially the stiffness and rigidity of the gels with lower concentrations resulting in less stiff and more deformable gels and higher concentrations in stiffer and more rigid gels, as observed elsewhere [20]. Therefore, our results demonstrate that it is possible to manipulate parameters like the concentration to obtain distinct assemblies (fibrils, gels) useful for specific purposes.

4.3. Influence of incubation time

Incubation time seemed to influence the fibril network between T_0min (preparation of the peptide solutions), T_2h (formation of a self-supporting gel) and T_7days (gel incubated for 7 days) with a more defined organisation of the fibrils over time. In addition, the 7-day incubated fibrils appeared longer, straighter, and showed a high degree of entanglement. This implies the fibrils continued forming, elongating, and re-arranging over time even though a self-supporting gel had already formed. Similar behaviour was studied by Bouchard et al. with bovine insulin fibrils [15]. The authors showed the evolution of the fibril network over time, from aggregated protein in the first few hours of incubation through to well-organised, elongated, and twisted fibrils after several weeks. The hypothesis of a higher degree of entanglement of the long-time incubated gel was further supported by rheology measurements where the storage modulus (G’) was the highest of all the samples right from the start of the test and after 1 hour of test. The massive increase in the gel strength could result from an arrangement into long and straight fibrils from increased inter- and intramolecular interactions [55], as observed with TEM at different time points. This suggests the stiffness of the gels formed from capFFV peptide can be controlled by varying the incubation period and thus be modulated depending on the desired application.

As a comparison, the storage modulus (G’) of MAX1, a 20 amino acid peptide that adopts a β-hairpin secondary structure, was ∼1600 Pa at the same concentration (2% w/v) [56]. Similarly, Ac-I₃SLGK-NH₂ and Ac-I₃SLKG-NH₂ peptides had a storage modulus of ∼1000 Pa. MAX1 was designed for tissue engineering [2], while Ac-I₃SLGK-NH₂ and Ac-I₃SLKG-NH₂ peptides were designed as drug delivery vehicles [28]. Furthermore, dilution of the capFFV sample did not impact either the fibril formation as seen with TEM analysis or the gel and elastic behaviour of the samples at 0.6 and 1% w/v. Thus, the storage modulus values obtained with the different samples from capFFV are within the range of values used in the literature for oral delivery applications and so they could be further investigated in that perspective.

4.4. Comparison with the natural peptide

In contrast to capFFV that self-assembled into fibrils when the charge was +1, no fibrillar network was observed with uncapFFV for the same charge. The strategies tried for triggering the fibrillation of uncapFFV were unsuccessful. The increase of the peptide concentration and the change in peptide net charge, which are both conventional catalysts of peptide fibrillation, showed little impact, although the peptide formed aggregates and short fibrils. Although we considered that the computational approach (to predict the gelling potential) was developed with dipeptides rather than longer peptides as presented in this study, the predictions differed from the results obtained in the lab. This highlights the importance of the experimental phase and the need for the development of such predictive tools using longer peptides [57]. Nevertheless, because nucleation was observed after 24 h of incubation, suggesting the peptide was still in a lag phase, it is hypothesised that an increase of the incubation time will lead to an exponential phase where a fibrillar network can be observed. Together with the capped peptide observations, these results indicate that the end-capping added to capFFV influenced the peptide solubility and self-assembling properties. An attempt to shorten the nucleation phase and accelerate fibrillation [58,59] was to increase the incubation temperature to 60 °C but was not successful (data not shown). Future work should also investigate the capacity of uncapFFV to form fibrillar species with the ThT assay, and its secondary structure under FTIR analysis. Strategies to further intensify fibrillation of uncapFFV include the addition of salts (composition, ionic strength) or biopolymers, and can both be further investigated [25]. More importantly, it would be useful to find a way to make uncapFFV form the same or a similar fibrillar network to the one observed with capFFV to avoid the need for capping the peptide in the context of food.

5. Conclusion

This paper describes the study of the self-assembling potential of a synthetic bioactive peptide derived from bovine casein: FFVAPFPEVFGK. We studied the influence of peptide concentration, incubation time and the net charge on the self-assembling properties of the peptide in its natural form (uncapFFV) and capped form (capFFV). CapFFV gave promising results as it was able to form a network of β-sheet fibrils and a self-supporting gel as observed with the different techniques used. Investigation of the fibril network structure with oscillatory rheology demonstrated that the degree of entanglement of the fibrils and the concentration impacted the mechanical properties of the gels, namely the stiffness and deformation. This means that slightly altering the self-assembling conditions (peptide concentration and incubation time) of the peptide gels studied herein could result in different possibilities in terms of end goal applications. Future studies could focus on other variables to further develop and exploit the peptide's properties as well as testing the peptide's potential for applications like oral delivery. In the case of uncapFFV, future work should look into other ways of initiating the self-assembly of the peptide, such as the addition of salts (composition, ionic strength) or biopolymers. This study helped unravel the chemistry behind the self-assembly mechanism of a food-derived peptide and opens the field of possibilities for a range of applications, including oral delivery, and the use of bulk proteins as sources of such peptides.

CRediT authorship contribution statement

Noémie Petit: Conceptualization, Methodology, Formal analysis, Investigation, Writing – review & editing. Jolon M. Dyer: Conceptualization, Methodology, Formal analysis, Funding acquisition, Writing – review & editing. Juliet A. Gerrard: Conceptualization, Methodology, Formal analysis, Funding acquisition, Writing – review & editing. Laura J. Domigan: Conceptualization, Methodology, Formal analysis, Funding acquisition, Writing – review & editing. Stefan Clerens: Conceptualization, Methodology, Formal analysis, Funding acquisition, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

• Data will be made available on request.