Charge Directed Selective Co‐Assembly of Ionic Complementary Peptide Binary Mixtures

ABSTRACT

Multicomponent peptide nanostructures offer a powerful platform for designing functional materials, yet controlling their co‐assembly remains a key challenge. Here, we harness electrostatic molecular recognition to drive the selective co‐assembly of five amphiphilic ionic peptide binary mixtures (M₁–M₅). Our results revealed that charge distribution governs β‐sheet strand alignment (parallel vs. antiparallel), assembly kinetics, and hydrogel viscoelasticity. Mixing stoichiometry and pH significantly influences co‐assembly behavior, nanofiber morphology, and network structure (self‐sorted vs. hetero‐aggregated). At pH 7, equimolar mixtures undergo nucleation‐driven co‐assembly into hetero‐aggregates, immediately forming well‐defined nanofibers, while non‐equimolar ratios yield altered morphologies. At a slightly acidic pH of 5–7, both E and K side chains are charged, enabling complementary ionic interactions that promote co‐assembly and gelation. Outside this pH range, co‐assembly is impaired. Notably, M₁ forms β‐sheets and hydrogels at acidic pH (≤4) via independent self‐assembly of its components, suggesting self‐sorted fibers. Overall, we demonstrate that tuning charge complementarity, ionization state, and stoichiometry enables precise control over the molecular, nanoscale, and mechanical properties of multicomponent peptide assemblies, providing a framework for the rational design of advanced peptide‐based materials.

Multi‐component peptide systems enable the development of multi‐functional and tunable materials, yet, controlling the co‐assembly of such complex systems remains challenging. This study demonstrates that rational design of peptide charge distribution, together with controlling parameters (pH, stoichiometry, and concentration), can guide selective co‐assembly, allowing for governing β‐sheet alignment, co‐assembly kinetics, fiber nanomorphology, network structure, and mechanical properties of the resulting hydrogels.

1. Introduction

Supramolecular peptide hydrogels represent an important class of soft biomaterials that have proved to be useful for a wide variety of biomedical and pharmaceutical applications, ranging from tissue engineering [1, 2, 3, 4], drug delivery and antimicrobial materials [5, 6, 7, 8], to disease or inflammation modelling [9, 10] and biosensing [11]. Development of these hydrogels with defined and tailored physicochemical and mechanical properties is crucial for designing these materials to meet the different application needs.

This can be achieved by molecular self‐assembly of the carefully designed primary peptide sequence into higher nanostructures, mainly through non‐covalent interactions. Rod‐like nanofibers are among the most common nanostructures that are formed by the assembly of different peptide sequence designs and often physically entangle into networks, forming hydrogels [12]. Assembly happens in response to various external stimuli, which is depending on the system's design, could be photo [13, 14], enzymatic [15, 16], or thermal [17, 18], as well as a change in pH [19, 20, 21] or ionic strength [22, 23].

Supramolecular nanofibrous hydrogels are commonly formed through the self‐assembly of a single peptide component. In contrast, multicomponent systems, created by mixing two or more peptides, enable the development of functional materials with highly tunable and emergent properties that cannot be achieved by the individual components alone [24, 25, 26]. In principle, the material properties of multicomponent peptide hydrogels could be fine‐tuned by selective modulation of the individual components, giving further flexibility for the system design with tailored physical properties and biologically relevant functions [27].

For a simple binary system, there are different possibilities of assembly patterns for the two components. These are mainly self‐sorting via self‐assembly of the individual components, leading to two different types of nanofibers [28, 29]. The other type may result from the direct molecular interactions between both components, forming either ordered fibers through co‐operative assembly (i.e. selective co‐assembly) or randomly assembled fibers containing random composition of both components [30, 31, 32, 33].

Selective co‐assembly can be engineered by the rational design of the primary sequences of both components to facilitate controlled molecular recognition between the counterparts, enabling the cooperative transition of both peptides from unstructured conformations into ordered secondary structures such as α‐helices [34, 35] or β‐sheets [36, 37, 38, 39, 40, 41, 42], inducing nanofibers formation. This approach can be advantageous over the single peptide component system for encapsulating sensitive drugs and biological cargoes such as cells [37, 38] and proteins [39]. This is because co‐assembly into nanofibers occurs through the non‐covalent molecular recognition between the system components under physiological conditions, thus avoiding the exposure to the drastic external stimuli, such as non‐physiological pH, ionic strength, and temperature, which are commonly needed to trigger self‐assembly of the single peptide component system.

It has been established that β‐sheet‐forming peptides have exceptional tendency to undergo high level of association into extended long nanofibrous structures, controlled and stabilized by various non‐covalent interactions such as H‐bonding, hydrophobic [43, 44, 45], aromatic [20, 44, 46, 47], and/or electrostatic interactions [23, 48, 49, 50], as well as chirality [51, 52, 53, 54]. Generally, most β‐sheet co‐assembling peptides have been designed by introducing charged residues to impart molecular electrostatic recognition sites. This in turn will trigger selective co‐assembly via mutual countercharge electrostatic attraction, while avoiding self‐sorting thanks to self‐repulsion.

For instance, mixing of the countercharge oligopeptide dimer KVW10 (Ac‐WKVKVKVKVK‐Am) and EVW10 (Ac‐EWEVEVEVEV‐Am) at pH6, which are analogues of the β‐hairpin forming MAX1 [55], led to the co‐assembly into β‐sheet‐rich nanofibers forming viscoelastic hydrogels [36]. Similarly, the P11‐13 (Ac‐EQQFEWEFEQQ‐NH₂) and P11‐14 (Ac‐QQOFOWOFOQQ‐NH₂) peptide pair, which were designed based on the P11‐2 peptide [56], co‐assembled via electrostatic attraction at physiological pH upon equimolar mixing in cell culture medium and have shown to support 3D cell proliferation of human dermal fibroblasts [37]. Likewise, mixing the biosynthetic countercharge CATCH (±) peptides that were designed based on Q11 (Ac‐QQKFQFQFEQQ‐Am) [57, 58], showed preference for co‐assembly over self‐association [39, 40, 41]. Increasing the number of charged residues steered the binary system toward co‐assembly, rather than self‐assembly, as well as accelerating the co‐assembly rate between CATCH (±) pairs [40]. The co‐assembly of countercharge N‐palmitoyl peptide amphiphiles has also been reported to be affected by the number of countercharge residues, which govern the cohesiveness of co‐assembly and consequently the nanoscopic morphology of the formed fibers [42]. The bioinspired β‐amyloid derived peptide pairs, KLVFWAK and ELVFWAE, have also showed prevalence of hetero‐assembly when combined, forming assemblies with molecular, structural, and morphological similarities to those obtained from the self‐assembly of a single Aβ polypeptide [59]. Another study reported the decapeptide pairs EEFKWKFKEE (p1) and KKFEWEFEKK (p2) that were designed to complement each other, featuring a central E or K residue and flanked with the opposite charge residues [38]. This design promoted heterotypic interactions over homotypic ones, forming β‐sheet nanofibrous hydrogels over a wide pH range [38].

In essence, selective co‐assembly of a countercharge peptide pair into β‐sheet nanofibers is a complicated process, which could be influenced by a plethora of physicochemical parameters. Most importantly, charge status, distribution, and co‐complementarity of the mixture components, as well as the stoichiometry of the electrostatic interactions, are among the governing factors that could significantly affect the selectivity and magnitude of molecular recognition.

To investigate this, herein we have developed a library of five different ionic co‐complementary binary mixtures, the M₁–M₅, with various charged residues distribution patterns (Table 1). The sequence design for mixture components was based on our previously reported self‐assembling Ultrashort Ionic‐complementary Constrained Peptides (UICPs) [20, 60]. These are self‐assembling β‐sheet forming peptides, composed of 4–5 amino acids with alternating aromatic (Phg; phenylglycine) and charged (E or K) residues, inspired by the Zuotin protein derived EAK16 [48, 61, 62]. The multi‐component system was designed such that co‐operative nanofiber formation and hydrogelation were attained upon mixing the countercharge peptide solutions at pH 7, excluding the need for any non‐physiological external triggers. The binary mixtures were studied for the effect of both charge distribution and co‐complementarity on the mechanism and kinetics of co‐assembly, morphology, and size of nanofibrous structures formed, as well as the mechanical properties of the resulting hydrogels. Furthermore, the effect of components' molar ratio and the medium pH has been studied to explore how these could be manipulated to control the system properties over the length scale.

TABLE 1.

Showing amino acid sequences, abbreviations, codes and characterisation of mixtures M₁–M_5.

Complementary Mixtures (M)		M1		M2		M3		M4		M5
		A1 −	C1 +	A2 −	C2 +	A3 −	C3 +	A4 −	C4 +	A5 −	C5 +
Sequence		EPhgKPhgE	KPhgEPhgK	PhgKPhgEE	PhgEPhgKK	PhgEPhgKE	PhgKPhgEK	PhgEPhgEK	PhgKPhgKE	PhgEPhgE	PhgKPhgK
Abbreviation [60]		E(Phg4)Rev	KPhg4	(Phg4)RevE	Phg4K	Phg4E	(Phg4)RevK	(PhgE)2K	(PhgK)2E	(PhgE)2	(PhgK)2
Code [60] *		UICP10	UICP5	UICP9	UICP4	UICP2	UICP7	UICP13	UICP15	UICP11	UICP12
Charge a		−1	+1	−1	+1	−1	+1	−1	+1	−2	+2
β‐sheets alignment		Antiparallel		Parallel		Parallel		Parallel		Antiparallel
% β‐sheet b		76.27 ± 0.55		56.29 ± 0.90		57.58 ± 1.10		69.36 ± 1.64		56.85 ± 1.61
β‐sheet/random c		15.5 ± 0.28		1.5 ± 0.44		1.7 ± 0.30		3.3 ± 0.34		1.7 ± 0.15
CGC (mM) d		15		30		30		60		75
Hydrogelation pH		3‐8		5‐7		5‐7		5‐7		7.2
Stiffness at 45 mM (kPa) e		1.4 ± 0.39		37.9 ± 0.61		6.45 ± 1.35		N/A (<CGC)		N/A (<CGC)
Fiber morphology		Long, flexible, very thin, entangled fibrils		Rigid, straight, branched, striated belts		Rigid, straight, striated fibers		Extended, thin, straight nanofibers		Extended, thick, curved, flexible fiber bundles
Diameter (nm)	TEM f	5.8 ± 0.2		18.5 ± 0.7		36.3 ± 1.6		13 ± 1.4		32.6 ± 1.6
	SAXS g	4.2 ± 0.2 nm		3.9 ± 0.2		4.05 ± 0.2		9.6 ± 0.3		3.8 ± 0.2

^a Calculated from Equation (1).

^b Calculated from the ratio of β‐sheet peak area in the Amide I band relative to overall area of the Amide I band in FTIR spectra of equimolar mixtures at 75 mM total peptide concentration and pH 7 (n = 3, mean + SD).

^c β‐sheet/random coil ratio was calculated by dividing the β‐sheet peak area in the Amide I band by that of the random coil peak of equimolar mixtures at 75 mM total peptide concentration and pH 7 (n = 3, mean + SD).

^d Minimum total peptide concentration of equimolar mixtures at pH 7 that showed the formation of self‐supported hydrogels in the inverted vial test.

^e Mean storage modulus (G′) values of equimolar mixtures at pH7, measured at frequency 1 Hz.

^f Mean fiber diameter calculated by fitting normal distribution of nanofibers widths (using Origin Pro 2016 software) using frequency counts in TEM micrographs of the 10X diluted samples prepared from equimolar mixtures of 75 mM total peptide concentration at pH 7.

^g Calculations done by fitting rod‐like objects at the lowest tested concentration (<CGC) for each Mixture within the linear region only.

^* Characteristics of individual nanofibers can be found in Soliman et al. 2025 [60].

2. Results and Discussion

In this work, we have employed five pairs of ionic co‐complementary countercharge UICPs, following from previously developed and individually characterized sequences by us [20, 60], to construct new multicomponent nanofibrous systems. The amino acid sequences were designed with distinct charge distribution patterns to achieve N‐to‐C‐terminal ionic co‐complementarity between each pair of countercharge peptides, resulting in five mixtures, M₁–M₅ (Table 1). The molecular co‐assembly driven by electrostatic interactions, as well as the hydrogelation propensity at physiological pH (7–7.5), were systematically investigated. This approach provides insights into the role of ionic co‐complementarity and charge distribution in mediating electrostatic recognition between co‐assembling peptide chains and directing their molecular packing into either parallel or antiparallel β‐sheet structures. Additionally, the kinetics and stoichiometry of co‐assembly were examined, along with the influence of total peptide mixture concentration and pH on secondary structure formation, nanofiber morphology, and hydrogelation behavior, all of which are presented in specific sections.

2.1. Charge Co‐Complementarity

The arrangement of charged residues within the amino acid sequence of the counterpart peptides could affect the mechanism and kinetics of molecular co‐assembly and packing into higher‐order structures [29, 41, 63, 64]. In light of this, the anionic A₁ ⁻ (UICP10) and cationic C₁ ⁺ (UICP5) peptides were designed to have N‐to‐C co‐complementary charge distribution, where A₁ ⁻ possesses E₁ K₃ E₅ (i.e. − + −) while C₁ ⁺ have K₁ E₃ K₅ (+ − —+) residues (Table 1 and Figure 1A). The self‐assembly behavior of the individual peptides was thoroughly investigated in our previous work [60], where both peptides showed to self‐assemble into β‐sheet structures and form hydrogels at slightly acidic pH values (Figure S1A,B). While at pH ∼7–7.5, both peptides neither self‐assembled into β‐sheet nanofibers nor formed hydrogels, with predomination of random coil over β‐sheet structure (Figure 1B,H, Figure S1A,B); thanks to the electrostatic repulsion between the charged peptide chains (overall net charge z = ‐1e⁻ for A₁ ⁻ and +1e⁻ for C₁ ⁺) (Figure S1C).

FIGURE 1.

(A) Schematic showing a top view of the proposed anti‐parallel co‐assembly of M₁ countercharge peptides, revealing the possible electrostatic interactions and lateral association at the hydrophilic and hydrophobic faces, respectively. (B) ATR‐FTIR spectra of A₁ ⁻ (UICP10), C₁ ⁺ (UICP5), and their equimolar mixture; M₁ at pH7 and 75 mM. The dashed line indicates the β‐sheet peak, while the dotted line indicates for random coil. (C) Thioflavin T fluorescence assay spectra (λ_excitation = 440 nm, λ_emission = 460–600 nm) of A₁ ⁻ (UICP10), C₁ ⁺ (UICP5), and their equimolar mixture; M₁ at pH7 and 75 mM. The CM₁ is a control sample of 75 mM M₁ at pH 7 without ThT. (D) TEM micrograph (left panel) and histogram for size distribution (right panel) for nanofibers of M₁ measured from TEM micrographs, with an average size of 5.8 ± 0.2 nm (n = 100 ± SD). The TEM micrograph of 75 mM M₁ sample (10× diluted) shows thin entangled nanofibers (Scale bar = 200 nm). (E) SAXS characterization of M₁ prepared at different concentrations (7.5 mM < CGC and 75 mM > CGC) in log I_N (q) versus q representation. The straight lines depict types of slopes for easier visualization (q⁻¹ and q⁻³). F) AFM Peak Force Error micrograph of a 75 mM M₁ hydrogel (20× diluted) showing entangled nanofibers (Scale bar = 1 µm). (G) Shear moduli (G′) (in log scale) of A₁ ⁻ (UICP10), C₁ ⁺ (UICP5), and their equimolar mixture; M₁ at pH7 and 75 mM. G′ values at 6 rad s⁻¹ obtained from frequency sweep at 0.1% strain were used (n = 3, mean ± SD). (H) Inverted vial test at pH 7 showing CGC of M₁ as 15 mM, while A₁ ⁻ and C₁ ⁺ existed in the solution form at all the tested concentrations.

In contrary to the individual components, equimolar mixing of A₁ ⁻ and C₁ ⁺, that is, mixture M₁, at pH 7 led to co‐assembly into β‐sheet structures mediated by the electrostatic attraction between the co‐complementary counterpart residues E₁‐K₅, K₃‐E_3, and E₅‐K₁ of the A₁ ⁻‐C₁ ⁺ peptide pairs, respectively (Figure 1A,B). The complementary counter charge electrostatic attraction masked the peptide charges (M₁ net charge z = 0), stabilizing co‐assembly into β‐sheet ladders alongside the π–π stacking of the aromatic Phg residues and the inter‐chain hydrogen bonding between the backbone amides (Figure 1A, Figure S1C) [20, 60]. M₁ secondary structure was confirmed by analyzing ATR‐FTIR amide I band, where a prominent sharp peak at 1619 cm⁻¹ (C═O stretching) was detected, confirming the formation of an extended β‐sheet structure (76%), with the 1692 cm⁻¹ peak implying anti‐parallel arrangement (Figure 1B, with deconvolution presented in Figure S2A) [65, 66]. A random coil small peak at 1651 cm⁻¹ was also detected, indicating minor unstructured population of unassembled monomers (5.5%). Overall, the β‐sheet structure is 15‐fold more abundant than the unstructured population (Table 1). Unlike individual components, the amide II band of M₁ showed prominent broad overlapping peaks at 1520 cm⁻¹ (β‐sheet) and 1555 cm⁻¹ (random coil) (Figure 1B).

The propagation of co‐assembly into extended β‐sheet ‘amyloid‐like’ nanofibers was revealed by Thioflavin T (ThT) fluorescence measurement [40, 67, 68, 69]. ThT is a molecular rotor, which is in low viscosity solvents and in the unbound state exhibits torsional motion of the benzthiazole ring relative to the aminobenzene ring, accounting for the low fluorescence intensity. On the other hand, binding of ThT to β‐sheet fibers prevents this rotation leading to increased fluorescence intensity [70]. Indeed, a significant enhancement of the ThT fluorescence intensity was observed in the case of M₁ as compared to the individual components A₁ ⁻ and C₁ ⁺ (Figure 1C), with a linear increase in fluorescence intensity as a function of total peptide mixture concentration (Figure S4A,B). No fluorescence was detected for the control ThT solution and the peptide hydrogel without ThT, thus ruling out the possibility of autofluorescence of the unbound ThT and the peptide molecules, respectively (Figure 1C).

Nanofiber morphology was examined by Transmission Electron Microscopy (TEM), where micrographs clearly showed the ability of M₁ to form thin curved entangled nanofibers with an average diameter of d ∼5.8 ± 0.2 nm (Figure 1D and Table 1). This morphology was confirmed by Atomic Force Microscopy (AFM), where an entangled network of curved fibers was also observed (Figure 1F).

Entanglement of M₁ nanofibers formed a stable continuous network (Figure 1D,F), resulting in self‐supportive hydrogel formation at a low critical gelation concentration (CGC) of 15 mM, unlike the individual mixture components, A₁ ⁻ and C₁ ⁺, which both failed to form hydrogels at the tested concentration range (7.5–75 mM) (Figure 1H, Figure S4B). SAXS was then used to confirm the presence of fibers and the structure features as a function of concentration (Figure 1E). The system q^{− 1} behavior, typical for rods, confirmed that at both concentrations, 7.5 and 75 mM, the samples consist of fiber‐like structures. At lower q values, the q^{− 3} behavior suggests sharp interfaces or large‐scale compact structures (possibly due to aggregation or bundle formation), consistent with the observed elongation and bundling in TEM (Figure 1D) and AFM (Figure 1F). The transition from q^{− 1} to q^{− 3} across the whole q range implies a hierarchical structure, where individually building fibers are aggregated into larger compact bundles or domains. At 75 mM (>CGC), the scattering intensity is higher and shows a more pronounced q⁻³ regime at low q, indicating greater aggregation or compact domain formation compared to the more dispersed, rod‐like structures at 7.5 mM (<CGC). It has been previously shown that for infinitely long rod‐like objects, in the dilute regime and for qR_σ < 1.3, R_σ being the cross‐section radius of gyration of the rod‐like object, a linear behavior should be observed in a ln qI_N(q) versus q² plot [71, 72]. We found this regime valid in each of the lowest concentrations measured in SAXS (across the analysis for each of the mixtures presented across the whole manuscript), allowing for an estimation of the R_σ, and hence the associated diameter of the smallest building fibers (Table 1). The smallest building fibers for M₁ were estimated to have diameter of 4.2 ± 0.2 nm, coinciding well with the lower end of size distribution from TEM (Figure 1D).

Next, hydrogelation of M₁ was verified by rheological characterization, where mechanical stiffness was assessed using frequency sweep experiments, which established the formation of viscoelastic materials with a concentration‐dependent linear increase in the storage shear moduli (G′), from ∼0.2 to 5 kPa at the concentration range 15–75 mM (Figure 1G, Figure S4B). This can be directly correlated to the concentration‐dependent linear increment in nanofiber formation, as indicated from ThT fluorescence (Figure S4A,B). Interestingly, M₁ demonstrated phenomenal thixotropic properties, with ∼70% recovery of the initial G′ was achieved after exposure to high shear strain (1000% for 1 min) within 5 min of relaxation in the first cycle, and ∼100% recovery in the following two cycles (Figure S1D). This could be attributed to the high stability of the fiber–fiber lateral association via electrostatic attraction and π‐π stacking (Figure 1A) and the entanglement into a stable network (Figure 1D,F). This could have led to hydrogel fragmentation at high shear, rather than network disentanglement and fiber alignment, with the subsequent fast dynamic re‐assembly upon relaxation [72]. Indeed, these rheological properties explain the remarkable injectability and sprayability of M₁ (Videos S1 and S2), which are desired properties for pharmaceutical and biomedical applications.

Finally, we investigated the effect of mixing order on the hydrogel properties, that is, adding A₁ ⁻ to C₁ ⁺ versus C₁ ⁺ to A₁ ⁻, and found out it did not affect the resulting mixtures, as indicated from the secondary structure population and mechanical stiffness profiles for the formed hydrogels (Figure S3).

2.2. Charge Distribution

Next, we have then investigated the effect of varying the positions of charged residues on the mechanism, thermodynamic stability, and kinetics of co‐assembly. Moving the N‐terminal charged residue in both A₁ ⁻ and C₁ ⁺ components to the C‐terminus led to a different design of another charge complementary peptide pair; A₂ ⁻ (UICP9) with K₂ E₄ E₅ (+ − −) and C₂ ⁺ (UICP4) with E₂ K₄ K₅ (− + +) charged residues (Figure 2A and Table 1), that is, M₂. We have previously reported that A₂ ⁻ self‐assembled into β‐sheet nanofibers with the formation of hydrogels at the pH range of ∼3–6, but not at pH ≥ 7 (Figure S5A,D), while the cationic C₂ ⁺ counterpart failed to self‐assemble into any ordered secondary structure at a wide range of pH values (including pH 7) and molar concentrations (data not shown) [60].

FIGURE 2.

(A) Schematic showing a top view of the proposed co‐assembly mechanism of the anionic A⁻ and the cationic C⁺ peptides of M₂‐M₄ into parallel‐packed β‐sheet dimers, revealing the possible electrostatic interactions and lateral association. (B) ATR‐FTIR spectra with peak deconvolution of M₂‐M₄ at pH 7 and 75 mM (deconvolution data reported in Table S1). (C) ThT fluorescence of the equimolar peptide mixtures, M₁–M₅ (at pH 7), shows significant enhancement of intensity as a function of concentration, indicating the high population of β‐sheet nanofibers. (D) Inverted vial test at pH 7 showing CGC of mixtures M₂, M₃, and M₄ as 30, 30, and 60 mM, respectively, while the corresponding individual components A⁻ and C⁺ existed as solutions at all the tested concentrations. (E) Log–log plot of shear moduli (G′) versus molar concentrations obtained for mixtures M₁–M₅. G′ values at 6 rad s⁻¹ obtained from frequency sweep at 0.1% strain were used (n = 3, mean ± SD).

Unlike M₁, the charged residue distribution of both A₂ ⁻ and C₂ ⁺ led to the preferential co‐assembly of M₂ into parallel β‐sheet ladders, guided by the co‐complementary electrostatic lateral association (Figure 2A). Antiparallel co‐assembly is thermodynamically unfavored, in this case, due to the possible electrostatic repulsion between the K₂–K₄ and E₄–E₂ of the A₂ ⁻–C₂ ⁺ peptide pairs, respectively (Figure 2A).

The ATR‐FTIR spectrum of M₂ demonstrated a prominent β‐sheet peak at 1625 cm⁻¹ (∼56% relative abundance), representing a blue shift to the M₁ antiparallel β‐sheet peak at 1619 cm⁻¹, along with the relatively smaller intensity and splitting of the higher wavenumber band at 1683 cm⁻¹ (Figures 1B and 2B, Figure S6A). These observations are consistent with the abundance of parallel β‐sheet orientation for M₂ co‐assembly [66, 73, 74, 75]. In addition, a prominent peak at 1648 cm⁻¹ denoting the presence of a significant population of random coil was observed (∼37%), indicating partial disassembly of/unassembled M₂ mixture components (β‐sheet/random ratio is 1.5). (Figure 2B and Table 1). In addition, two broad overlapped peaks were observed at 1525 cm⁻¹ (β‐sheet) and 1550 cm⁻¹ (random coil) within the amide II band region, confirming the mixed population (Figure 1B). M₂ also showed lower nanofiber content as revealed from the lower ThT fluorescence intensities compared to M₁ at the same mixture molar concentrations (Figure 2C). Moreover, the kinetics of M₂ nanofiber formation is significantly slower than M₁, where M₂ nanofibers were detected after 24 h of mixing A₂ ⁻ and C₂ ⁺, while an instant nanofiber formation with significantly higher content was observed for M₁ after 1 min of mixing of the countercharge mixture components (Figure S1E). Overall, these findings indicate a relatively lower thermodynamic stability of the M₂ parallel co‐assembly in comparison to the M₁ antiparallel arrangement.

Despite the slow co‐assembly kinetics, M₂ successfully formed self‐supportive hydrogels after 24 h of equilibration at 4°C, with a CGC of 30 mM, which is 2‐fold higher than that of M₁ (Figures 1, 2, and Table 1). The mechanical stiffness of M₂ hydrogels was significantly higher than M₁ at the same molar concentrations. A concentration‐dependent increase in G′ value was observed starting at ∼6 kPa at the 30 mM CGC and gradually increasing up to ∼300 kPa at 75 mM (Figure 2E, Figure S4C).

The higher G′ values of M₂ hydrogels are a result of the slow co‐assembly kinetics that led to the formation of rigid, straight, thicker, and occasionally branched nanofiber bundles (average d ∼18.5 ± 0.7 nm) (Figure 3A). The M₂ bundles are striated, showing morphological similarity to those formed by the co‐assembly of the previously reported ionic co‐complementary decapeptides, EEFKWKFKEE and KKFEWEFEKK, with the latter mixture exhibiting well‐defined striations along the tape's long axis (Figure 3A) [38]. These striations indicate that electrostatic attraction plays a key role besides nucleation, which is the propagation of lateral association along the nanofiber axis [60]. The observed striations in TEM could also be a result of artefactual lateral association from sample dehydration due to the collapse of the solvent network. While the observed striations in TEM could indeed arise from drying‐induced bundling that is not present in the hydrated SAXS samples, it is also possible that such features are partially embedded within the observed SAXS signal (Figure 3D). The low‐q q^{− 3} slope combined with the high‐q peaks is consistent with hierarchical structures in which lateral features contribute to the overall scattering profile, particularly at the concentrations used. In SAXS, such striations often influence the apparent fractal dimensions, which commonly fall between slopes of −2 and −3.

FIGURE 3.

(A–C) TEM micrographs (top panel) (Scale bar = 200 nm) and corresponding histograms (bottom panel) for nanofibers size distribution of (A) M₂, (B) M_3, and (C) M₄ samples prepared at 75 mM and pH 7 (10× diluted) with average sizes of 18.5 ± 0.7, 36.3 ± 1.6, and 13 ± 1.47 nm (n = 100 ± SD), respectively. (D–F) SAXS characterization of (D) M₂, (E) M_3, and (F) M₄ mixtures and their single‐component measurements in a log I_N (q) versus q representation. The straight lines depict types of slopes for easier visualization (q⁻¹ and q⁻³).

The apparent initial lower thermodynamic stability of the M₂ mixture implies that the un/disassembled peptides could have re/assembled into more electrostatically stable and thicker bundles later within the 24 h of incubation. On the other hand, the significantly faster co‐assembly kinetics of M₁ hindered the formation of superstructures, ending up with thermodynamically stable, but much thinner, nanofibers (Figures 1D and 3A, Figure S1E). In line with this, Liu et al. showed that CATCH(6K+/6D‐) pair, of sequences KQKFKFKFKQK (6K+) and DQDFDFDFDQD (6D‐) respectively, formed stiffer hydrogels and thicker bundled nanofibers, thanks to the considerably slower kinetic profile than any of the other CATCH(X+/Y‐) pairs reported [76]. Also, Stupp and co‐workers have reported the co‐assembly of countercharge N‐palmitoyl peptide amphiphiles (PAs), where they demonstrated that shorter PAs assemblies with weak cohesive interactions formed superstructures of thick bundle filaments, due to the possible molecular escape followed by reassembly into more stable bundles, while the more cohesive longer PAs formed thinner stable β‐sheet structures [42].

Replacing the M₂ residues at positions P2 and P4 with the opposite charge amino acid (i.e., E to K and vice versa), generated the A₃ ⁻ (UICP2) with E₂ K₄ E₅ (‐ + ‐) and C₃ ⁺ (UICP7) with K₂ E₄ K₅ (+—+), forming the components of M₃ mixture (Figure 2A and Table 1). Like M₂, both M₃ counterparts have two exposed countercharge residues at the hydrophilic face of the peptide chain and one exposed C‐terminus charged residue at the hydrophobic side. These structural similarities are thought to drive co‐assembly of M₃ components the same way as M₂, forming parallel β‐sheets (∼57%) with the presence of unstructured un/disassembled chains (random coil peak at 1650 cm⁻¹, with ∼34% relative abundance) (Figures 2A,B, Figure S6B). Nanofiber morphology and content of M₃ are akin to M₂, where rigid straight fibers are associated into slightly thicker bundles of d ∼36.3 ± 1.6 nm (Figures 2C and 3B). M₃ has also formed self‐supported hydrogels at the same CGC as M₂ (30 mM) with significantly higher mechanical stiffness than M₁ at all tested concentrations (Figure 2D,E).

By reversing the charges of the C‐terminus P4 and P5 residues of M₃ peptide components, we created A₄ ⁻ (UICP13) with E₂ E₄ K₅ (− − +) and C₄ ⁺ (UICP15) with K₂ K₄ E₅ (+ + −) (Figure 2A and Table 1), components of mixture M4. M₄ hydrogels were successfully formed by mixing equimolar quantities of both peptides at pH 7, with a CGC of 60 mM, twice that of M₂ and M₃ mixtures (Figure 2D). Despite the structural similarity of M₄ to both M₂ and M₃, all co‐assembling into parallel β‐sheet structures (Figure 2B, Figure S6C), M₄ has strikingly adopted significantly different nanoscopic assemblies. Interestingly, unique monodispersed interwoven extended nanofibers were formed, with a thinner uniform size (average d ∼13 ± 1.4 nm) and very little to no observed association/aggregation into thicker bundles (Figure 3C). This could be attributed to the limited tendency of M₄ β‐sheets toward propagation of lateral growth through the hydrophilic sides, hindered by the expected repulsion between the projecting either all anionic (E₂ E₄) or all cationic (K₂ K₄) residues (Figure 2A) [60].

As can be seen from SAXS curves obtained for M₂, M₃ and M₄ mixtures, in each case, the mixture was characterized by the strong well‐defined structural homogeneity with q⁻³ behavior across the whole q scale and peaks occurring in the region 0.6–1.4 nm⁻¹ (Figure 3D–F). This behavior corresponds well to the obtained TEM characteristics for homogeneous fibers with repeated inter‐structuring that can correspond to the peaks obtained in SAXS. In d‐space, this regime corresponds to 10.5–4.5 nm, matching the sizing in the TEM (Figure 3A–C).

The M₅ mixture based on all anionic A₅ ⁻ (UICP11; E₂ E₄) and all cationic C₄ ⁺ (UICP12; K₂ K₄) tetrapeptides, has conversely formed significantly thicker and polydisperse nanofiber bundles (d ∼ 32.6 ± 1.6 nm) at pH 7 (Table 1 and Figure 4D). This is attributed to the different co‐assembly mechanism into the more stable anti‐parallel β‐sheet structures, as revealed from ATR‐FTIR (amide I at 1617 and 1690 cm⁻¹ bands, and amide II at 1520 cm⁻¹) (Figure 4A,B,D, Figure S2Band Table S1). Besides those, a prominent random coil peak at 1648 cm⁻¹ was also observed in the amide I region (∼33.7% relative abundance), as for the 1555 cm⁻¹ peak (amide II), both indicating the presence of unassembled peptide monomers (β‐sheet/random ratio is 1.7) (Table 1, Figure S2B). This is ascribed to the repulsion between the highly charged peptide chains (overall net charge z = ‐2e⁻ for A₅ ⁻ and +2e⁻ for C₅ ⁺) (Table 1). Like the anti‐parallel β‐sheet forming M₁, M₅ exhibited a significantly faster nanofiber formation kinetics (<1 h) compared to the parallel β‐sheet forming mixtures (Figure S1E). M₅ mixture had the least hydrogelation tendency with the highest CGC ∼ 75 mM (Figure 4C). Furthermore, the hydrogels exhibited the lowest storage modulus G′ values among all studied mixtures at the same molar concentration (Figures 2E and 4F, Figure S4C).

FIGURE 4.

(A) Schematic showing a top view of the proposed anti‐parallel co‐assembly of M₅ countercharge peptides, revealing the possible electrostatic interactions and lateral association at the hydrophilic and hydrophobic faces, respectively. (B) ATR‐FTIR spectra of A₅ ⁻ (UICP11), C₅ ⁺ (UICP12), and their equimolar mixture; M₅ at pH7 and 75 mM. (C) Inverted vial test at pH 7 showing CGC of M₅ as 75 mM, while A₅ ⁻ and C₅ ⁺ existed in the solution form at all tested concentrations. (D) Histogram for size distribution for nanofibers of M₅ samples at 75 mM (10× diluted) measured from TEM micrographs (inset, scale bar = 200 nm), with average size of 32.6 ± 1.6 nm (n = 100 ± SD). (E) ATR‐FTIR spectra of scrambled mixtures SM₁, SM₂, SM₃, and SM₄ at pH 7 and 75 mM. (F) Shear moduli (G′; in log scale) of the charge co‐complementary mixtures M₁–M₅ and scrambled mixtures SM₁–SM₄ prepared at pH7 and 75 mM. G′ values at 6 rad s⁻¹ obtained from frequency sweep at 0.1% strain were used (n = 3, mean ± SD). Inset: Inverted vial test of scrambled mixtures (SM₁‐SM₄) at pH 7 and 75 mM, showing failure to form self‐supported hydrogels apart from SM_3, which showed the formation of a viscous liquid.

Finally, scrambled mixtures (SMs) with random distribution of charged residues were examined to assess the importance of ionic co‐complementarity as an essential structural pre‐requisite for co‐assembly into β‐sheet structures (Table 2). Each SM has at least one pair of core same charge residues at the hydrophilic side of the binary peptide components, which could potentially interfere with the co‐assembly into β‐sheet structures via electrostatic repulsion (Table 2). With the exception of SM₃, none of the tested SMs neither co‐assembled into β‐sheet structures nor formed hydrogels (Figure 4E,F). ATR‐FTIR analysis of SM₃ showed a very weak β‐sheet peak at 1624 cm⁻¹ (∼5% of the amide I band area), accounting for a low population of β‐sheet structures (Figure 4E). Rheological characterization of SMs revealed failure to form viscoelastic self‐supportive hydrogels, except SM_3, which showed a storage modulus G′ slightly >100 Pa, indicating the formation of a viscous liquid rather than a hydrogel (Figure 4F). These findings confirm the crucial role played by the positions of charged residues and ionic co‐complementarity for the successful selective intermolecular recognition between the mixture components to trigger co‐assembly.

TABLE 2.

Showing amino acid sequences and abbreviations of scrambled mixtures SM₁‐SM_4.

Scrambled Mixtures (SM)	SM1		SM2		SM3		SM4
	A2 −	C5 +	A1 −	C5 +	A5 −	C2 +	A5 −	C1 +
Sequence	PhgKPhgEE	PhgKPhgK	EPhgKPhgE	PhgKPhgK	PhgEPhgE	PhgEPhgKK	PhgEPhgE	KPhgEPhgK

2.3. Stoichiometry of Co‐Assembly

We demonstrated that equimolar mixing of ionic co‐complementary UICPs at pH 7 enabled masking peptide charges and triggered co‐operative assembly and lateral growth into packed β‐sheet ladders. Indeed, the stoichiometry of this countercharge molecular recognition could impact the thermodynamic stability and kinetics of co‐assembly and nanofiber formation, as well as hydrogelation.

To study these effects, M₁ was prepared at a wide range of A₁ ⁻:C₁ ⁺ molar ratios (from 1:9 up to 9:1), allowing for fine‐tuning the mixture's net charge (Figure 5A, bottom panel). Remarkably, M₁ formed self‐supported hydrogels for the whole range of molar ratios examined by the inverted vial test (Figure 5A, top panel). Relative β‐sheet abundance as a function of molar ratios showed a ‘bell‐shaped’ curve with the highest β‐sheet contents (∼78%–81%) observed at and around the equimolar ratio (i.e., 3:7–7:3), with slight reduction in relative β‐sheet content (∼61%–65%) for samples with the lowest (1:9) and highest (9:1) A₁ ⁻:C₁ ⁺ ratios (Figure 5B). Rheological properties for the formed hydrogels can be correlated to their β‐sheet content. Mixtures prepared below and above the A₁ ⁻:C₁ ⁺ equimolar ratio (2:8–8:2) exhibited similar to slightly lower G′ values (range of ∼1.7–5.8 kPa; where G′ for equimolar M₁ was ∼5.7 kPa) (Figure 5B). While both mixtures prepared at the lowest (1:9) and highest (9:1) ratios were of significantly lower stiffness (∼250 Pa) (Figure 5B) and possessed poor thixotropy compared to other ratios, with <20% recovery from high shear strain (Figure S4E).

FIGURE 5.

(A) Inverted vial test of M₁ samples at different A₁ ⁻:C₁ ⁺ molar ratios, pH7, and 75 mM (Top panel) and theoretical net charge state of M₁ as a function of pH and the A₁ ⁻:C₁ ⁺ molar ratio (Bottom panel). Theoretical net charge was calculated using Equation (1) (reported in methods). The dashed line indicates zero net charge. (B) Relative β‐sheet peak area (left y‐axis; closed symbol data points) and Shear moduli (G′) in log scale (right y‐axis; open symbol data points) for M₁ samples at different A₁ ⁻:C₁ ⁺ molar ratios, pH7, and 75 mM. G′ values at 6 rad s⁻¹ obtained from frequency sweep at 0.1% strain were used (n = 3, mean ± SD). (C) Co‐assembly kinetics of M₁ at different A₁ ⁻:C₁ ⁺ molar ratios, pH 7 and 15 mM (= CGC), using ThT fluorescence assay at 4°C (n = 3, mean ± SD). (D) Histograms for size distribution for nanofibers of M₁ samples prepared at 1⁻:9⁺ (left panel) and 9⁻:1⁺ (right panel) molar ratios at 75 mM and pH 7 (10× diluted) measured from TEM micrographs (insets, scale bar = 200 nm), with average sizes of 15.3 ± 0.6 and 8.4 ± 0.3 nm, respectively (n = 100 ± SD). White arrows pointing to fiber bundle twisting. (E) SAXS characterization of M1 mixture at different stoichiometric ratios in a log I_N (q) versus q representation. The straight lines depict types of slopes for easier visualization (q⁻¹ and q⁻³). (F) AFM Peak Force Error micrograph of representative fibers in a 75 mM M₁ hydrogel showing nanofibers with A₁ ⁻:C₁ ⁺ molar ratios of 1:9 (left; 20× diluted) and 9:1 (right; undiluted) (Scale bar = 1 µm).

Nanofiber formation kinetics were also governed by the stoichiometry of countercharge interactions. For instance, in M_1, both the rate and extent of nanofiber formation were clearly affected by the A₁ ⁻:C₁ ⁺ molar ratio, where both the slowest formation and lowest nanofiber content were noticed for the 1:9 and 9:1 mixtures, followed by the 2:8 and 8:2 ones (Figure 5C). However, there were insignificant differences in nanofiber formation kinetics and content for the other M₁ ratios (i.e., 3:7–7:3) (Figure 5C).

Both M₁ mixtures with the lowest (1:9) and highest (9:1) A₁ ⁻:C₁ ⁺ molar ratio formed nanofibers with different morphologies, which are also distinct from that of the equimolar mixture (Figures 1D and 5D). The 1:9 mixture formed twisted fiber bundles with an average diameter d ∼ 15.3 ± 0.6 nm, which is ∼3‐fold thicker than the nanofibers formed by the equimolar M₁ (d ∼5.8 ± 0.2 nm) (Figure 5D,F). Although the nanofiber size of the 9:1 mixture (d ∼8.4 ± 0.3 nm) was close to that of the equimolar mixture, a clearly different nanofiber shape was observed, where both TEM and AFM showed straight rigid structures (Figure 5D,F) as opposed to the apparently more flexible curved entangled fibers of the 1:1 mixture (Figure 1D,F). The differences in SAXS between stoichiometric mixtures were minimal at the level of nanofibers (similar to the original 1:1 M₁ mixture) (Figure 5E). There wer,e however noticeable differences in the low q region corresponding to difference in aggregation between different ratios, with 1⁻:9⁺ slightly deviating from q⁻³, whereas 9⁻:1⁺ simply displayed slightly lower intensities. SEM confirmed that the three mixtures formed a continuous network of entangled nanofibers, which is a characteristic feature of viscoelastic hydrogels (Figure 6).

FIGURE 6.

SEM micrographs of 75 mM M₁ samples (20× diluted) showing the continuous network of entangled nanofibers for the mixtures of A₁ ⁻:C₁ ⁺ molar ratios (A) 1:9, (B) 1:1, and (C) 9:1, respectively (Scale bar = 100 µm).

A very similar co‐assembly and hydrogelation trends were also observed for M₂–M₅ mixtures (Figure S7A–H). With the exception of mixtures of 1:9 A⁻:C⁺ molar ratio, both β‐sheet formation and hydrogelation of M₂, M_3, and M₅ were successfully achieved at all the other studied molar ratios (Figure S7A,B,D). The highest relative β‐sheet content and G′ values were recorded for the equimolar mixtures, with a gradual reduction of both β‐sheet content and mechanical stiffness by either increasing or decreasing the A⁻:C⁺ molar ratio (Figure S7E–H). However, at a 1:9 ratio, mixtures M₂, M_3, and M₅ failed to form hydrogels at the tested concentration, with a significantly very low relative β‐sheet content in M₂ and M₃, and absence of β‐sheet structure in M₅ (Figure S7E,F,H). This is a result of the expected electrostatic repulsion between the C‐terminus cationic K₅ residues, in the case of C₂ ⁺ and C₃ ⁺, as for K₂ and K₄ in C₅ ⁺ (Figures 2A and 4A, and Table 1). In addition, K residues are highly soluble as their sidechain has strong H‐bonding affinity to water, which competes with the C⁺ peptides ability to assemble into β‐sheet structures [72, 77]. β‐sheet formation and hydrogelation of M₄, on the other hand, were achieved at a narrower range of A⁻:C⁺ molar ratios (4:6–7:3) (Figure S7C,G), due to the high possibility of electrostatic repulsion between similar charges exposed at the hydrophilic sides of both A₄ ⁻ and C₄ ⁺, if the molar proportion of one of each exceeded the other (Figure 2A and Table 1).

These structural differences were less evident in SAXS measurements. All studied ratios displayed similar profiles for the M₂ mixture, indicating similarly built nanofibrillar objects (Figure S7I). On the other hand, mixture 1:9 deviated from the scattering profiles for other ratios in M₃ and M₄ mixtures, indicating lower propensity for bundling and larger structure formation (Figure S7J–K). Finally, all scattering profiles were similar for M₅ mixtures with a ratio of 9:1, depicting lower intensities and hence indicating concentration‐driven effects might be key in deciphering the exact structural details (Figure S7L). Altogether, SAXS effectively confirmed formation of nanofibers and their bundling across most of the cases; however, this solely does not appear to be a sufficient factor leading to hydrogelation, which is, as noted above, a concentration‐driven process.

Based on the above results, we could conclude that the nanofiber formation mechanism for non‐equimolar mixtures is mediated by a nucleation process, where equimolar co‐assembly of the countercharge peptides occurs first till depletion of the limiting component. Aggregates of the multi‐component dimers could then act as nuclei to trigger the propagation of assembly of the major component into β‐sheet structures that are further stabilized by both π‐π stacking of Phg residues and inter‐chain H‐bonding [20, 60]. These interactions are thought to overcome the electrostatic repulsion that could happen between same‐charge sidechains of the major component. This nucleation process will ultimately lead to the assembly of nanofibers forming hetero‐aggregates (Scheme 1A).

SCHEME 1.

Depiction of the proposed mechanisms of (A) co‐operative assembly between countercharge peptides at a non‐equimolar ratio and pH 7 through nucleation mediated by electrostatic attraction, followed by assembly of the major component into nanofibrous hetero‐aggregates, (B) self‐assembly of individual components of M₁ at pH 4, which, upon mixing lead to the formation of self‐sorted aggregates.

Overall, the stoichiometry of co‐assembling components can be manipulated to develop hydrogels with tunable mechanical and physicochemical properties, including hydrogel viscoelasticity and overall net charge, thus giving the flexibility of tailoring the materials for various biomedical applications.

2.4. Effect of pH on Co‐Assembly and Hydrogelation

Co‐assembly through electrostatic attraction is highly sensitive to the medium pH, as this would affect the protonation/deprotonation status of both the E and K residues' side chains. Considering this, equimolar mixtures were studied at a pH range of ∼2–10 and investigated for propensity to co‐assembly and hydrogelation.

Over a wide pH range ∼2–8, M₁ mixture successfully formed self‐supported hydrogels. M₁ hydrogels formed at pH <6 were translucent, while formation of cloudy to opaque hydrogels was observed at pH range 6–8, indicating highly aggregated denser nanofiber networks (Figure 7A,D). Nanofiber formation was confirmed by the ThT fluorescence enhancement detected at pH 7 immediately after equimolar mixing of the A₁ ⁻ and C₁ ⁺ peptides, which was not observed for mixtures prepared at pH 2 and 10 (Figure 7B, Figure S8A). Increasing the pH from 2 to 7 was associated with enhancement in molecular co‐assembly, with the highest β‐sheet content, and hence hydrogel stiffness, seen at pH 5–7 (Figure 7C).

FIGURE 7.

(A) Inverted vial test of 75 mM M₁–M₅ samples prepared at the equimolar A⁻:C⁺ ratio at a range of pH values (∼2–10). (B) Thioflavin T fluorescence assay spectra (λ_excitation = 440 nm, λ_emission = 460–600 nm) of equimolar A₁ ⁻:C₁ ⁺ M₁ samples (75 mM) at pH 2, 7, and 10. The CM₁ is a control sample of 75 mM M₁ at pH 7 without ThT. (C) Relative β‐sheet peak area (left y‐axis; closed symbol data points) and Shear moduli (G′; in log scale) (right y‐axis; open symbol data points) for 75 mM M₁ samples at the equimolar A₁ ⁻:C₁ ⁺ ratio, as a function of pH. G′ values at 6 rad s⁻¹ obtained from frequency sweep at 0.1% strain were used (n = 3, mean ± SD). (D) SEM micrographs of 75 mM M₁ samples (20× diluted) at different pH values showing the continuous network of entangled nanofibers at pH7, compared to a less dense network at pH2 and ‘flake‐like’ aggregates at pH 10 (Scale bar = 100 µm).

At this pH range, both E and K side chains are charged (pKa 4.25 and 10.53, respectively), therefore, mixing M₁ components could lead to molecular recognition through countercharge electrostatic attraction, masking the side chains’ charge and leading to overall neutralization (i.e., M₁ net charge z = 0) (Figure 5A, bottom panel). On the contrary, in strong acidic (pH ∼ ≤4) and basic (pH >8) media, the E is protonated, and the K is deprotonated, respectively. This could interfere with the molecular recognition via countercharge attraction, consequently lowering the chances for co‐assembly of the mixture components into β‐sheet ladders (Figure 7C). Despite this, M₁ showed hydrogel formation at acidic pH values ∼≤4, which can be attributed to the self‐assembly of the individual A₁ ⁻ (UICP10) and C₁ ⁺ (UICP5) components rather than co‐assembly, suggesting the possibility of self‐sorted fiber formation (Figures S1A,B and S8B, Scheme 1B) [30, 32, 33, 60]. However, because of the net positive charge of the resulting mixture (Figure 5B, bottom panel), M₁ demonstrated lower β‐sheet content, as well as G′ moduli values at acidic pH values ≤4, compared to the mixtures formed at pH 5–7 (Figure 7C). M₁ nanofibrous network formation and structure were characterized by SEM, where micrographs revealed a less dense nanofibrous network for the pH 2 hydrogel and ‘flake‐like’ aggregates for the non‐gelling pH 10 mixture, compared to the continuous dense network of nanofibers of the pH 7 mixture (Figure 7D).

Unlike M₁, hydrogelation of the parallel β‐sheet forming mixtures M₂‐M₄ was only possible at the slightly acidic to neutral pH range of 5–7 (Figure 7A), over which these mixtures showed the highest β‐sheet content and hydrogel stiffness (Figure S8C). SEM micrographs showed the characteristic continuous nanofibrous network of typical hydrogels for all three mixtures at pH7, whereas amorphous aggregates have been observed for mixtures prepared in strong acidic (pH 2) and basic (pH 10) media (Figure S9). Finally, M₅ formed a nanofibrous hydrogel only at pH 7, but not in basic media (Figure 7A, Figure S9). It was not possible to prepare and investigate the M₅ mixture in acidic media due to the poor aqueous solubility of A₅ ⁻ at the tested concentrations.

In summary, pH has proven to play an essential role in switching on/off co‐assembly by control of the ionization degree of the mixture components, hence, the charge magnitude available for electrostatic attraction and formation of stable supramolecular structures.

3. Conclusion

Multicomponent peptide‐based nanostructures have recently emerged as a promising platform for the development of composite materials with highly tunable and desirable properties. Nevertheless, guiding the co‐assembly process remains as a major hurdle against the development of rationally designed functional materials. Here, we have investigated the control over physicochemical parameters for the co‐assembly of ultrashort ionic‐complementary peptides (UICPs), mainly focusing on electrostatic interactions. A library of five N‐to‐C‐terminus amphiphilic charge complementary binary mixtures, M₁–M₅, were designed to study the effect of: (i) charge distribution of the peptide components on β‐sheet strand directionality (parallel versus anti‐parallel); (ii) mixing stoichiometry and medium pH on the nanofiber morphology and network structure (self‐sorted versus hetero‐aggregates); and (iii) in correlation to the mechanical properties of the formed hydrogels.

Our results revealed that binary peptides with a charge distribution pattern of X^+/− ArX^+/− ArX^+/− (M₁) and ArX^+/− ArX^+/− (M₅) (where Ar is aromatic phenylglycine and X^+/− is either lysine or glutamate) showed preference toward co‐assembly into anti‐parallel β‐sheet ladders at pH 7, triggered and stabilized by the countercharge electrostatic attraction along the hydrophilic face of the sequence. On the other hand, the co‐assembly of M₂–M₄ mixtures of peptide components flanked with two C‐terminus charged residues (i.e. ArX^+/− ArX^+/−X^+/− ) adopted parallel β‐sheet formation. This directionality was guided by the additional electrostatic attraction between position P5 countercharge residues at the hydrophobic face of the binary components. Parallel co‐assembly was also driven by the ionic co‐complementarity of core countercharge residues at the hydrophilic side (P2 and P4) for the three mixtures and avoided core charge repulsion of M₂ and M₃ components. The anti‐parallel β‐sheet forming mixtures (M₁ and M₅) showed a significantly faster nanofiber formation kinetics (∼1 h) compared to the parallel β‐sheet forming ones (∼24 h) (e.g., M₂ and M₃). The mechanical stiffness of M₂–M₄ hydrogels (storage modulus G′ ∼1–10 kPa) was found to be significantly higher than M₁ and M₅ (G′ ∼0.1–1 kPa) at the same molar concentrations.

The molar ratios of the anionic (A⁻) to cationic (C⁺) components (A⁻:C⁺) affected the stoichiometry of countercharge molecular recognition, hence the thermodynamic stability and kinetics of co‐assembly and nanofiber formation. As a general trend, the highest relative β‐sheet formation and G′ values were recorded for the equimolar mixtures, with a gradual reduction of both β‐sheet content and mechanical stiffness by either increasing or decreasing the A⁻:C⁺ molar ratio. Although individual components showed a lower tendency toward self‐assembly and an inability of hydrogelation at pH7, both β‐sheet formation and hydrogelation were successfully achieved for all mixtures at the studied range of molar ratios (1:9–9:1). Nanofiber formation for non‐equimolar mixtures is thus mediated by an electrostatic nucleation process leading to the formation of hetero‐aggregates resulting in different nanofiber morphology compared to equimolar mixtures.

The rate and extent of co‐assembly was found to be significantly affected by the medium pH for equimolar mixtures. At slightly acidic to neutral pH (∼5–7), both E and K side chains are charged facilitating the countercharge molecular recognition of the mixture counterparts, enabling co‐assembly and hydrogelation. With exception of M₁, deviation from this pH range impaired the co‐assembly process. M₁ though, showed the ability to form β‐sheet structures and hydrogels at acidic pH (∼≤4), which was mediated by the self‐assembly of the individual components A₁ ⁻ and C₁ ⁺, suggesting self‐sorted fiber formation. Network structure was also affected by medium pH, where in neutral medium continuous nanofibrous network forming hydrogel was detected, while amorphous peptide aggregates were formed in acidic (pH 2) and basic (pH 10) media.

Overall, this study showed that strategic control over charge distribution and the molar ratios of ionic complementary peptide counterparts, as well as controlling ionization level by adjusting the medium pH, could be manipulated to tailor the molecular, nanoscopic, and mechanical properties of multicomponent peptide nanomaterials. Such tunability establishes a versatile framework for designing extracellular matrix‐mimetic nanofibers with defined molecular and nano‐structural properties, enabling adaptation to accommodate diverse cell types with distinct biomechanical requirements for regenerative applications, in the future.

4. Materials and Methods

4.1. Materials

All peptides (purity >95%) were purchased from Biomatik Corporation (Wilmington, DE, Canada) and checked in‐house for chemical identity and purity using mass spectroscopy (MS) and reverse‐phase high‐performance liquid chromatography (RP‐HPLC). The purity profiles (HPLC chromatograms) of the peptides are presented in the Supplementary Information – Peptides Purity Characterization Data, while the full chemical characterization was published in our previous work in [60]. Thioflavin T (ThT) was supplied by Alfa Aesar (Heysham, UK). All other reagents and solvents (analytical grade) were sourced from Sigma‐Aldrich (Gillingham, UK) and used as received.

4.2. Methods

4.2.1. Preparation of Hydrogels

Powders of both anionic (A⁻) and cationic (C⁺) peptides were separately dissolved in HPLC water by mixing at 2500 rpm for 1 min using a vortex mixer. The pH of each peptide solution was then adjusted to the desired value, separately, through stepwise addition of 0.5 M NaOH solution in increments of 5–10 µL, while vortex mixing at 2500 rpm for 30–60 s after each addition at room temperature. The final volume of each peptide solution was adjusted by HPLC water to obtain the required final peptide concentration. To trigger co‐assembly, countercharge peptide solutions were then mixed at a range of A⁻:C⁺ molar ratios (1:9 to 9:1) by vortex mixing at 2500 rpm for 1 min, at room temperature. Peptide mixtures were then kept in the fridge at 4°C for 24 h to equilibrate before being evaluated the following day. Inverted vial test was performed to visually inspect the formation of self‐supported hydrogels and to determine the critical gelation concentration (CGC), which is the minimal total molar peptide concentration of the binary mixture at which the sample did not flow upon inversion.

4.2.2. Net Charge Calculation

Equation (1) was used to calculate the theoretical net charge of different peptide sequences at each pH value [78]:

$$Z=\sum _i{N}_i\frac{{10}^{pK{a}_i}}{{10}^{pH}+{10}^{pK{a}_i}}-\sum _j{N}_j\frac{{10}^{pH}}{{10}^{pH}+{10}^{pK{a}_j}}$$ (1)

where N_i/j are the numbers and pKa_i/j are the pKa values of the basic (i – pKa > 7) and acidic (j – pKa < 7) groups present on the peptide. The overall theoretical net charge of binary mixtures prepared at different molar ratios of A⁻ and C⁺ peptides was calculated using Equation (1). The total numbers of basic and acidic groups were determined by summing the contributions from each peptide, calculated as the number of respective groups in each peptide sequence multiplied by its molar fraction in the mixture (i.e., N_i/j = [N_i/j in A⁻ peptide × molar fraction of A⁻] + [N_i/j in C⁺ peptide × molar fraction of C⁺]).

4.2.3. Attenuated Total Reflectance–Fourier Transform Infrared Spectroscopy (ATR‐FTIR)

Peptides' co‐assembly into secondary structures was evaluated using a Bruker Alpha FTIR spectrometer equipped with a diamond multibounce ATR plate. The peptide hydrogel specimen was applied onto the crystal, and transmittance readings were recorded within the wavenumber range of 4000–400 cm⁻¹ at 128 scans and a resolution of 2 cm⁻¹. HPLC water was used as background, automatically subtracted from the sample spectrum by OPUS version 8.1 software of the instrument. Deconvolution of amide I peaks (1600–1700 cm⁻¹) was performed using OriginPro 2016 software for peak separation and accurate calculation of areas. Quantification of β‐sheet content in each sample was carried out by calculating the ratio of β‐sheet peak area, which was defined as the prominent peak at the wavenumber range 1616–1630 cm⁻¹, to the overall area of the amide I band. The β‐sheet/random coil ratio was calculated by dividing the deconvoluted β‐sheet peak area by that of the random coil peak.

4.2.4. Thioflavin T (ThT) Fluorescence Assay

ThT fluorescence assay was carried out for the detection of amyloid‐like β‐sheet fibril formation and evaluation of the co‐assembly kinetics of countercharge peptides. Peptide solutions were titrated to the desired pH using 0.5 M NaOH as described above. ThT stock aqueous solution (1 mM) was added to each peptide solution of the binary mix, separately, with a final concentration of 100 µM (10‐fold dilution). The peptide solutions were mixed, and ThT fluorescence was detected by a SpectraMax M5 spectrofluorometer plate reader (Molecular Devices) with excitation wavelength set to λ_ex 440 nm, and the emission was set to either scan range of 460–600 nm (spectrum), or to 490 nm (for kinetic studies at different time intervals).

4.2.5. Oscillatory Rheology

The dynamic rheological properties of the developed materials were tested by a stress‐controlled Anton Paar MCR‐501 Rheometer equipped with a temperature‐controlled Peltier plate and a 25 mm parallel plate geometry. A sample volume of ∼500 µL was gently loaded onto the bottom plate using a plastic spatula and the gap between the sample stage and the upper plate was set to 250 µm. Any excess material was removed by a spatula. Samples were then allowed to equilibrate at 37°C for 2 min before measurement, and a solvent trap was used to reduce evaporation. Storage (G′) and loss (G″) moduli were recorded as a function of amplitude sweep (strains of 0.01%−1%, 1 Hz, and 37°C), to determine the linear viscoelastic region, which was used to maintain the strain at 0.1% for further measurements. Radial frequency sweep tests were performed between 0.1–100 rad s⁻¹ at 0.1% strain and 37°C, to assess the mechanical stiffness of the samples. Time sweeps were carried out to examine the thixotropic properties of the samples. Measurements were recorded at 0.1% strain for 5 min, followed by a cycle of 1000% strain for 1 min, followed by 0.1% strain for 5 min, with the frequency kept at 1 Hz. All measurements were repeated in triplicates.

4.2.6. Scanning Electron Microscopy (SEM)

A Carl Zeiss EVO LS 15 SEM was used to characterize the morphology of the peptide nanofibrous network. Hydrogel samples were diluted 20‐fold using HPLC water, then mounted onto an aluminum stub with a carbon tab. Samples were freeze‐dried overnight using a Lablyo Mini freeze‐dryer and gold sputter‐coated (10 nm) in a Quorum QR150S coater before imaging. Samples were analyzed with an accelerating voltage of 10 kV and a probe current of 100 pA.

4.2.7. Transmission Electron Microscopy (TEM)

The morphology of peptide nanofibers was examined using a JEOL JEM‐1400 TEM outfitted with an Emsis Xarosa digital camera with Radius software at an accelerating voltage of 120 keV. Samples were diluted 10‐fold using HPLC water, and a 10 µL volume was placed onto the carbon‐coated surface of the copper grid (400 mesh) for 10 s. To mitigate sample aggregation, grids were subjected to glow discharge in a Quorum Gluqube plus at 20 mA for 30 s before use. Samples were then blotted away using filter papers and washed with 10 µL of pure water. Subsequently, a negative staining was performed using 10 µL of a 2% (w/w) uranyl acetate solution, and the samples were observed under the microscope after complete drying.

4.2.8. Atomic Force Microscopy (AFM)

Samples were prepared by 20‐fold dilution of a 75 mM hydrogel in deionized water (ddH2O). A total of 50 µL was deposited onto a freshly cleaved mica surface, allowed to stand for 2 min. The excess solution was then removed, followed by a single wash with 1 mL of HPLC grade water. Residual water was gently wicked away using Whatman No. 1 filter paper, and the samples were air‐dried overnight before imaging. AFM imaging was performed in PeakForce Tapping (PFT) mode in air using a Bruker Resolve BioAFM equipped with a Nanoscope V controller operating under Nanoscope v9.7 software. ScanAsyst‐Air probes (silicon nitride tips with an aluminum coating; nominal tip radius ≈2–5 nm; spring constant 0.4 N·m^{− 1}; Bruker AFM Probes, Camarillo, CA, USA) were used for imaging. Height and PFT images with a scan size of 5 µm were acquired at a scan rate of 1.98 Hz. The instrument was routinely calibrated using a grating with 180 nm deep and 10 mm² depressions. Data was first‐order flattened using the Nanoscope Analysis (v3.0) software before image export.

4.2.9. Small Angle X‐Ray Scattering (SAXS)

SAXS experiments were performed on beamlines B21 and I22 at the Diamond Light Source (DLS) facility in Didcot, UK.

For I22 the following setup was used. The energy of the beam was 12.4 keV, corresponding to the X‐ray wavelength of 1 Å. Quartz capillaries (1.5 mm outer diameter, 0.01 mm wall thickness) were supplied from the Capillary Tube Supplies Ltd., UK. All samples were introduced to capillaries manually via syringe. Samples were collected at 21 ± 1°C for 99 consecutive frames at the exposure time of 1 s each to check for radiation damage. Calibration of the SAXS detector (Pilatus P3−2 M, Dectris, Switzerland) was performed using silver behenate powder. The distance between samples and the detector was fixed to 4.2 m to cover a momentum transfer vector range of to an accessible momentum transfer vector range of 0.049 nm⁻¹ < q = (4π/λ) sin(θ/2) < 4.7 nm⁻¹, where θ is the scattering angle and λ is the wavelength of incident photons.

Measurements performed on beamline B21 followed the setup previously described [60]. The energy of the beam was 13.05 keV, corresponding to an X‐ray wavelength of 0.95 Å. An Arinax sample change robot was used to automatically injected the samples into Quartz capillaries (1.5 mm outer diameter, 0.01 mm wall thickness, Capillary Tube Supplies Ltd. UK). The sample‐to‐detector distance was fixed to 3.7 m, corresponding to an accessible momentum transfer vector range of 0.03 nm⁻¹ < q = (4π/λ) sin(θ/2) < 3.4 nm⁻¹, where θ is the scattering angle and λ the wavelength of the incident photons. Calibration of the SAXS detector (Eiger 4 M, Dectris, Switzerland) was performed using silver behenate powder. Samples were collected at 20 ± 1°C for 20 consecutive frames at an exposure time of 1 s each to check for radiation damage.

Empty capillaries were used as background and subtracted from all measurements, while the subtraction mask was created using glassy carbon. Data were reduced using the processing tools at DawnDiamond software suite. The 2D scattering photon patterns were integrated using the azimuthal integration tool to obtain a 1D scattering pattern.

4.2.10. Statistical Reproducibility and Data Analysis

To ensure data reproducibility, experiments were performed using three different batches of peptide hydrogel preparations per mixture and per experimental condition tested. Deconvolution of ATR‐FTIR amide I band peaks was performed for spectra measurements of three different samples to calculate the mean ± standard deviation (n = 3, mean ± SD) of the relative secondary structure percentage and β‐sheet/random coil ratio. Rheological measurements were performed on three different hydrogel sample conditions to calculate the mean ± SD of storage modulus G′ values at 6 rad s⁻¹. TEM measurements of nanofiber diameters were calculated as the average of nanofiber thickness at three different locations along the fiber length in all TEM micrographs obtained from imaging three different batches per sample (total ≈80–100 diameter measurement per sample). Nanofibers' size distribution was estimated from histograms by LogNormal fitting to calculate the average diameter and size range using OriginProTM 2016 software.

Author Contributions

Conceptualization: A.K., M.A.N.S., J.K.W., and M.A.E. Methodology: A.K., M.A.N.S., A.C., T.S., J.T.J., M.H.A., N.A., J.K.W., and M.A.E. Formal analysis: A.K., M.A.N.S., J.T.J., M.H.A., J.K.W., and M.A.E. Data curation: A.K., M.A.N.S., N.A., J.T.J., M.H.A., R.A., J.K.W., and M.A.E. Writing – original draft preparation: A.K., M.A.N.S., J.K.W., and M.A.E. Writing – review and editing: A.K., M.A.N.S., T.S., N.A., Z.A., J.K.W., and M.A.E. Supervision: Z.A. and M.A.E. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: smll72431‐sup‐0001‐SuppMat.pdf.

Supplemental File 2: smll72431‐sup‐0002‐VideoS1.mp4.

Supplemental File 3: smll72431‐sup‐0003‐VideoS2.mp4.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.