Self‐Assembly Pathway Influence on Dehydropeptide‐Based Gel Properties and Drug Release

Abstract

Low‐molecular‐weight peptide‐based hydrogels formed through self‐assembly have emerged as promising candidates for biomedical applications. While the self‐assembly process is known to affect the network morphology, its impact on mechanical properties and drug delivery remains poorly understood. In this work, it is explored how different gelation conditions influence the morphology, properties, and drug release profiles of dehydropeptide‐based gels. Additionally, it is presented and analyzed, for the first time, the crystal structure of a naphthalene N‐capped dehydropeptide (2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH), which reveals a maximum pore diameter of ≈4.08 Å. By changing the preparation conditions, it is found that the stiffness of the hydrogels can vary by nearly three orders of magnitude. Employing spectroscopic and imaging techniques, the relationship between the gelation methods and the resulting mechanical properties is investigated. These findings suggest that the assembly structure, morphology, and non‐covalent interactions significantly influence the release profile of model drugs such as doxorubicin, methotrexate, and curcumin. These results provide valuable insights into how preparation conditions can impact the properties of peptide‐based hydrogels and their drug release profiles.

In this work, it is explored how the gelation conditions affect the morphology, mechanical properties, and drug release of dehydropeptide‐based gels. The naphthalene‐capped dehydropeptide forms a crystal structure with a pore diameter of ≈4.08 Å. Gelation conditions can influence gel stiffness by nearly three orders of magnitude, in which the assembly structure and non‐covalent interactions critically affect the release of model drugs (doxorubicin, methotrexate, curcumin).

1. Introduction

Self‐assembled hydrogels made from short peptides have attracted significant interest across several technological fields. These gelators exhibit promising potential in biotechnology and biomedicine, as effective platforms for drug delivery, tissue engineering, wound healing and as scaffolds for 3D cell culture media.^{[ 1 ]} Their advantages include simplicity, expedited synthesis, biocompatibility, and cost‐effectiveness, making them highly suitable for diverse applications. These supramolecular systems form a 3D network that entraps water molecules through the anisotropic self‐assembly of the peptides,^{[ 2 ]} driven by cooperative noncovalent interactions such as electrostatic, aromatic and hydrophobic forces, as well as hydrogen bonding. Self‐assembly can be triggered by various stimuli, including pH, temperature, light, solvent switch, ultrasound, enzymatic action, ionic strength alterations, or simply by dispersing the gelator molecules.^{[ 3} , ^{4 ]}

Despite over a decade of exploration, the relationship between molecular structure and gelation ability remains poorly understood.^{[ 5} , ⁶ , ⁷ , ^{8 ]} Small changes in the chemical structure can significantly alter the resulting properties.^{[ 9} , ^{10 ]} For example, modifying a single residue can cause the storage moduli (G’) to vary by over four orders of magnitude.^{[ 11} , ^{12 ]} Given the difficulty in precisely tuning single‐component systems, alternative strategies such as multi‐component approaches,^{[ 13} , ^{14 ]} and combination with nanocomposites,^{[ 2} , ¹⁵ , ¹⁶ , ^{17 ]} have been proposed to improve the control of the gel properties. However, the influence of factors like pH, temperature, ionic strength, buffer composition,^{[ 18} , ¹⁹ , ²⁰ , ²¹ , ²² , ^{23 ]} aging,^{[ 24 ]} and counter ions,^{[ 25 ]} on gel properties is still not fully understood. Additionally, the relationship between G’ and peptide concentration shows wide variability, with exponents ranging from 1.40 to 3.90.^{[ 12} , ²⁶ , ^{27 ]} While this variability allows for tunability, it can also hinder the reproducibility, which is crucial for biomedical applications. As a result, achieving precise control over supramolecular gel properties remains a central focus of intense research and discussion.^{[ 28 ]}

Furthermore, few studies have compared how different pathways affect the assembly and properties of a given gelator.^{[ 18} , ^{29 ]} For instance, factors such as the pH,^{[ 30 ]} peptide concentration,^{[ 31 ]} drug‐peptide interactions,^{[ 32 ]} and molecular size of the drug or network dynamics,^{[ 33 ]} have been shown to influence drug release profiles in peptide‐based gels. However, the impact of network morphologies and properties resulting from different pathways on drug release remains largely unexplored. Understanding how the self‐assembly pathway influences drug release is critical for optimizing peptide‐based gels for biomedical applications and facilitating their clinical translation.

In this work, we assess the self‐assembly of a naphthalene N‐capped dehydropeptide (2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH) under different pathways and investigate its effect on the gel properties using spectroscopic techniques. Dehydropeptides are highly advantageous for biomedical applications owing to the improved proteolytic stability.^{[ 34 ]} Besides, the dehydroamino acid dehydrophenylalanine (ΔPhe) introduces conformational constraints that promote self‐assembly. Strong supramolecular interactions and their arrangement in the solid state were confirmed through single‐crystal X‐ray diffraction. Additionally, we explore how self‐assembly influences drug release profiles using model drugs with varying net charges at physiological pH.

2. Results and Discussion

2.1. Single‐Crystal X‐ray Diffraction Analysis

To gain a deeper understanding of the higher‐order crystal packing and the intricate patterns of extensive directional supramolecular hydrogen‐bonded (H‐bonded) interactions at the atomic level, a thorough examination of individual crystals of 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH was conducted. The 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH crystals were formed through a vapor diffusion method in an acetonitrile/methanol solvent mixture at room temperature. The formation of colorless needle‐like crystals was observed within a span of five days, reaching their maximum size after 15 days. The detailed single‐crystal data collection and refinement parameters for the 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH crystal is succinctly presented in Tables S1–S5 (Supporting Information). The crystals of 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH exhibited a needle‐like morphology with a length ranging from ≈0.03 to 0.15 mm.

The crystal structure displayed a triclinic symmetry within a unit cell, characterized by the centrosymmetric space group P‐1. The lattice parameters were as follows: a = 4.85542(12) Å, b = 11.4027(4) Å, c = 24.1192(7) Å, α = 85.976(2)°, β = 87.281(2)°, γ = 89.430(2)°, and the volume V = 1330.54(7) ų. The ORTEP (Oak Ridge Thermal Ellipsoid Plot) program demonstrated the thermal characteristics of 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH with ellipsoids of similar sizes, shapes and coherent orientations (Figure S1, Supporting Information). Notably, the asymmetric unit comprised a monomeric 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH molecule, with no inclusion of any solvent molecules, showing a distinction from its unprotected analogues (Figure 1B).

Figure 1.

Single‐crystal X‐ray diffraction study. A) Molecular structure and B) asymmetric unit of 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH. C) Dimeric structure formation of 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH. D) Higher‐order crystal packing diagram showing the face‐to‐face orientation of 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH. E) Demonstration of the crystal packing revealing the formation of the tubular structure resulting from the assembly of four dehydropeptide molecules of 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH (left). The enlarged view of this tubular structure (right) reveals a void space of 13.8% within the aggregation of four dehydropeptide molecules. Aliphatic and aromatic hydrogen atoms are omitted for clarity.

In the crystal packing, significant intermolecular interactions between the carboxylic acids of the dehydropeptide were observed, resulting in the formation of robust face‐to‐face dimers that incorporate the macrocycles, as indicated by the emergence of the smallest graph‐set value of ${\mathrm{R}}_2^2$(8)‐type rings (Figures 1C,D; S2, Supporting Information). The notation “R,” commonly used to represent macrocycles formed by intermolecular H‐bonds, signifies the number of donors (in this case, 2) and acceptors (also 2), along with the total number of atoms (8) participating in the macrocycle formation. The formation of the tubular structure resulted from the collective aggregation of four dipeptide molecules (Figures 1E; S3, Supporting Information). Each monomeric unit of 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH contributed to the formation of two rectangular channels, forming a void space that occupies 13.8% of the unit‐cell volume and featuring a maximum pore diameter of 4.08 Å (Figures 1E; S3, Supporting Information). In comparison, the unprotected analogue, Phe–ΔPhe, exhibited very similar rectangular channels formed by the translation of four peptide molecules.^{[ 35 ]} However, in this case, each monomeric unit contributed to the formation of four tubular structures, showcasing a solvent void space of 20.6% of the unit‐cell volume and a pore diameter of 4.03 Å. Further, these two structures, Phe–ΔPhe and 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH, exhibited notable differences in their molecular conformations. In the case of Phe–ΔPhe, the torsion angle was calculated as 149.70°, indicating that the side chains were positioned on both sides of the peptide bond plane. Moreover, 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH, featuring an aromatic naphthalene protecting group on the N‐terminus side of the peptide, showcased a slightly different arrangement of amino acid side chains. This structural dissimilarity is evident in the larger torsion angle value of 176.53°, highlighting the linear orientation of the side chains. In addition, the stability of the assembled structure of the dehydropeptide was facilitated by intermolecular C─H···π interactions, specifically involving the stacked Phe‐ΔPhe and naphthalene aromatic rings. Moreover, the observed aromatic‐aromatic interactions were found to be stronger in 2‐Naph‐L‐Phe‐Z‐ΔPhe‐OH (4.85 Å) compared to the unprotected Phe–ΔPhe dipeptide (5.58 Å) (Figure S4, Supporting Information). In addition, the inter‐amide segments also revealed an interesting H‐bonded network, with a short bond distance of 1.94 Å (N─H⋯O) and bond angle of 170.79°(N─H⋯O) (Figure S5, Supporting Information).

2.2. Self‐Assembly Pathways

Gels were prepared through different methods: 1) slow pH decrease achieved by adding glucono‐δ‐lactone (GdL method); 2) mixing a basic pH gelator solution with pH = 7.4 phosphate buffer (buffer switch method), with a heat‐cooling cycle of the peptide solution before (HC‐BS) and after (BS‐HC) the mixing; 3) heat‐cooling method (HC); and 4) solvent‐switch method (SS) assisted with ultrasounds.

The final buffer concentration was 100 mM for the heat‐cooling and solvent‐switch methods, and 50 mM for the buffer switch, HC‐BS, and BS‐HC methods. Gelation was confirmed via the vial inversion test (Figure 2). Moreover, all gels could recover their gel state after disruption, except for those prepared through the GdL method (Figure S6, Supporting Information), which showed phase separation after a day. Hereafter, for clarity, the gels are referred to by their respective preparation pathway.

Figure 2.

Photographs and TEM images of the hydrogels (0.5 wt.%) were obtained using the different preparation methods. A) Slow pH decrease achieved by adding glucono‐δ‐lactone. B) Buffer switch. C) Heating‐cooling before buffer switch. D) Heating‐cooling after buffer switch. E) Heating‐cooling. F) Solvent‐switch. The scale bar is 1 µm.

2.3. Fibres Morphology

Although drying the gels can introduce artifacts,^{[ 36 ]} TEM images (Figure 2 ) revealed distinct changes in the fibre morphology depending on the preparation method. The GdL method (Figure 2A) produced thin, straight fibres, while other pathways resulted in more flexible and entangled fibres. Notably, the solvent‐switch method (Figure 2F) exhibited junctions and nucleation points, consistent with other observations in gels prepared by solvent‐switching from DMSO to water.^{[ 18} , ²⁰ , ^{37 ]} Interestingly, solvent switching initially yielded a precipitate, which could be solubilized through sonication. While ultrasound is known to assist hydrogel formation,^{[ 38} , ^{39 ]} this approach was ineffective when the peptide solid was directly dispersed in a buffer solution.

The HC method led to fibres with a helical‐ or twisted‐like morphology, as described elsewhere.^{[ 16 ]} Similar structures have been observed in other peptides assembled through different pathways,^{[ 24} , ^{40 ]} or co‐assembly.^{[ 14 ]} In the buffer switch methods, a turbid, viscous solution was initially formed, which became translucent upon heating and returned to a turbid state upon cooling. Although the fibres exhibited similar morphologies, the heating‐cooling cycle produced thinner fibres (Figure S7, Supporting Information), consistent with findings in other peptide‐based gels.^{[ 41} , ^{42 ]} The thinnest fibres, with a cross‐section of ∼14 nm, were obtained using the GdL method.

2.4. Hydrogels Mechanical Properties

Oscillatory time sweeps (Figure 3 ) suggested a fast gelation for the buffer/solvent switch and heating‐cooling pathways. Gels prepared via the buffer switch, HC‐BS and solvent switch methods exhibited elastic‐dominated behavior (G’>G’’) from the earliest measurement time. In contrast, the sol‐gel transition for the heating‐cooling or BS‐HC methods began near 30 °C. Gels formed through the slow pH decrease using GdL showed a lag time of ≈100 min, as reported elsewhere.^{[ 16 ]}

Figure 3.

Shear storage G’ (filled symbols) and loss G’’ (empty symbols) modulus during the A–C) kinetic process of gelation, D–F) frequency sweep, and G–I) strain sweep. Gels were prepared through the (A,D,G) GdL method, (B,E,H) heating‐cooling (HC) and BS‐HC methods and (C,F,I) buffer switch (BS), HC‐BS, and solvent‐switch (SS) methods.

The frequency sweep profile revealed that most gels displayed a weak dependence of G′ on the frequency, a characteristic of strong gels. However, the buffer switch pathway resulted in gels with stronger frequency dependence, suggesting a less permanent network.

Strain sweep assays further highlighted differences in network structures. The buffer switch gels exhibited a low storage modulus and sustained high strain, indicative of entangled or sparsely cross‐linked networks of flexible chains, consistent with the morphology observed in TEM images.^{[ 43 ]} In contrast, gels prepared via the solvent switch or heating‐cooling cycle withstood significant deformation, indicating stiffer and more permanent networks, as also supported by the frequency sweep. According to elasticity theory, this behaviour may result from a reduced average distance between entanglements or cross‐linking points.^{[ 44 ]}

Considering the gels have the same peptide concentration, gels prepared by the solvent‐switch and heating‐cooling methods likely formed more fibrils and/or entanglements compared to those prepared by the buffer switch. Specifically, the solvent switch method produced gels with a G’ and strain sweep profile comparable to previously‐reported gels prepared via solvent switch from DMSO, which commonly display highly entangled fibres.^{[ 26} , ^{45 ]} Regarding the temperature effect, Draper et al.^{[ 42 ]} reported that similar N‐capped dipeptides assemble into worm‐like micelles at basic pH, wherein a heating‐cooling cycle induced thinner filaments through dehydration of the fibrils core, leading to stiffer gels as observed in this work. Subtle differences were also noted between the HC‐BS and the BS‐HC methods: the latter yielded slightly larger G’ and coincided with thinner fibres. This suggests that the dehydrated core might re‐swell after buffer addition, whereas it remains dehydrated and more entangled if the heating‐cooling cycle occurs after the buffer switch. The heating‐cooling gel, while less stiff, withstood large deformation, similar to other dipeptide gels with twisted fibres.^{[ 14 ]} These mechanical profiles contrast with those of gels prepared via the GdL method,^{[ 16 ]} which are stiffer and more brittle, characteristic of densely cross‐linked/entangled networks of rather rigid links.^{[ 43} , ^{46 ]}

2.5. Influence of Assembly Method on the Hydrogels

Fluorescence measurements (Figure 4A) provided further insights into the peptide assemblies, revealing emission bands near 350 and 450 nm, corresponding to monomers and aggregates, respectively.^{[ 11} , ^{16 ]} The GdL method produced a large aggregates band and a strongly quenched, red‐shifted monomers band, indicating significant aromatic ring stacking.^{[ 47 ]} For other gels, similar maximum emission wavelengths and slight variations in band broadness suggested subtle differences in naphthalene stacking, as supported by the excitation spectra (Figure S8A–C Supporting Information). UV–vis spectra also confirmed that the preparation methods led to different assemblies (Figure S8D Supporting Information). A red‐shift in the aggregates band with longer excitation wavelength pointed out to emission from distinct species, which may result from the red‐edge excitation shift (REES) effect. This phenomenon, described for other supramolecular gels,^{[ 48} , ^{49 ]} is associated with slow rates of solvent relaxation (reorientation) around the fluorophore excited state relative to the fluorescence lifetime.^{[ 50 ]} The decrease of the Stokes shift with longer excitation wavelengths (Figure S8C, Supporting Information) is consistent with other studies,^{[ 48} , ^{49 ]} further supporting this analysis. Additionally, the peptide matrix could also act as a relaxing “solvent”, behaving like a dense fluid due to substantial side chain fluctuations.^{[ 51 ]} The results suggest that the stacked aromatic rings are located in a motionally‐restricted polar environment. The stronger REES observed for the GdL and BS‐HC gels align with their more rigid fibres and thinner aqueous core, respectively, while the weaker REES in the solvent switch gel suggests a more flexible environment.

Figure 4.

A) Fluorescence emission spectra of gels prepared using different methods (λ_exc = 280 nm). Inset: Normalized fluorescence emission spectra. B) Dependence of the maximum fluorescence emission wavelength on the excitation wavelength. Inset: Comparison of the total change in maximum fluorescence emission wavelength for the assessed range of excitation wavelengths (280 to 500 nm). C) FTIR spectra in the Amide I region. Grey dashed lines highlight the bands near 1640 cm⁻¹ and 1680 cm⁻¹. D,E) Raman spectra in the (D) 900–1500 cm⁻¹ and (E) Amide I region. Dashed black lines highlight the phenylalanine characteristic Raman peaks. F) Powder X‐ray diffraction of GdL, buffer switch, heating‐cooling and solvent switch gels. The black dashed lines highlight the reflections near 2θ = 7.6°, 18.6°, 22° and 26.6°.

Fourier‐transform infrared (FTIR) spectroscopy (Figure 4C) further supported the similar packing for the buffer switch, HC‐BS, and BS‐HC gels. The amide I absorption peak near 1640 cm⁻¹, characteristic of β‐sheets, is consistent with other small peptide‐based gels.^{[ 11} , ⁵² , ^{53 ]} Diphenylalanine molecules were suggested to adopt a supramolecular β‐sheet secondary structure with the J‐aggregate nature of aromatic residues.^{[ 11} , ⁵⁴ , ^{55 ]} The broader amide I band in gels prepared by heating‐cooling or buffer/solvent switch suggests a greater conformational freedom compared to GdL gels, spanning the characteristic wavelength range of random coils (1640–1650 cm⁻¹) and various turn structures (1660–1685 cm⁻¹).^{[ 14} , ^{52 ]} The bands near 1690 cm⁻¹ have been associated with anti‐parallel β‐sheets,^{[ 14 ]} but it remains a topic of debate.^{[ 56} , ^{57 ]} Moreover, dehydrophenylalanine‐containing peptides are known to adopt a β‐turn conformation,^{[ 58} , ^{59 ]} or a mixture of β‐like and extended structures,^{[ 60 ]} suggesting the presence of different types of β‐turn.

Raman spectroscopy was carried out to obtain further details on the assemblies. The spectra (Figure 4D) displayed the characteristic peaks of phenylalanine (1002 cm⁻¹, 1031 cm⁻¹, 1207 cm⁻¹, 1576 cm⁻¹, 1597 cm⁻¹),^{[ 61 ]} including a peak near 1383 cm⁻¹, previously reported for other dehydropeptides.^{[ 11} , ⁶² , ^{63 ]} The amide I region (1600–1700 cm⁻¹, Figure 4E) indicated major differences between GdL gels and those prepared by other methods. The buffer switch, HC‐BS, and BS‐HC gels displayed similar profiles, while slight changes were observed for the heating‐cooling and solvent switch gels. The narrow bandwidth and downshift of the amide I band in the GdL gel suggest reduced conformational freedom, and a larger number of strands and/or stronger hydrogen bonding,^{[ 64 ]} consistent with the larger REES and fluorescence aggregate band. Regarding the secondary structure, the bands near 1675 cm⁻¹ and 1690 cm⁻¹ indicate β‐sheet, β‐turn and random structures, while the peaks between 1630–1650 cm⁻¹ can be associated with β‐sheet and random coil,^{[ 65} , ^{66 ]} or the stretching mode of ΔPhe C_α = C_β bond.^{[ 63 ]}

X‐ray diffraction (Figure 4F) suggested the presence of polymorphic states, consistent with other observations for peptide‐based gels.^{[ 20 ]} Distinct structures were observed for the GdL gel, and slight differences among the buffer switch, heating‐cooling and solvent switch gels. Peaks with d‐spacing near 11.6 Å (2θ = 7.6°), 4.8 Å (2θ = 18.6°), and between 4.0 Å (2θ = 22°) and 3.4 Å (2θ = 26.6°) were detected, which are commonly associated with inter‐sheet, inter‐strand, and aromatic rings π‐π stacking distances, respectively.^{[ 20} , ⁶⁷ , ⁶⁸ , ^{69 ]} Further comparison of crystals formed in methanol/acetonitrile mixtures with PXRD analyses of the gels showed matching XRD patterns, indicating structural similarities across the samples (Figures S9, Supporting Information). The complete FTIR and Raman spectra, along with XRD diffractograms, are included in the Supporting Information (Figures S10–S12, Supporting Information).

Overall, the assembly structure varied with the preparation method, influencing the aromatic stacking, solvation, and dehydropeptide conformation. The GdL gel exhibited the strongest packing, while other gels showed minor differences. The similar assembly of the buffer switch, HC‐BS, and BS‐HC gels suggests that differences in mechanical properties may arise from the heat‐cooling cycle, which induces thinner fibres through core dehydration, leading to more cross‐links/entanglements. This may also result from the formation of more fibres requiring fewer monomer units while maintaining the dehydropeptide conformation. In contrast, the differing mechanical properties of the heating‐cooling and solvent switch gels likely result from variations in network morphology, including subtle changes in assemblies, solvation and peptide conformation.

2.6. Influence of the Assembly Method on Drug Release

Co‐assembly with drugs,^{[ 70} , ⁷¹ , ^{72 ]} and other molecules can modify peptide assemblies,^{[ 14} , ^{73 ]} thereby affecting the morphology and gel properties. To explore this, we first studied the effect of the encapsulated drugs on the secondary structure of the gels. Figure 5A–C displays the amide I region of gels loaded with doxorubicin, curcumin and methotrexate. The complete Raman spectra are included in Supporting Information (Figure S13, Supporting Information). Compared to the neat gels (Figure 4E), no major changes in the peak wavenumber associated with the secondary structure were observed, suggesting that the drugs did not interfere with the peptide assemblies. Even at higher drug concentrations (Figure S14, Supporting Information), where the spectra were dominated by the drug signals, the frequencies of gel‐related peaks remained similar. However, changes in the characteristic frequencies of protonated (≈1700 cm⁻¹) and deprotonated (≈1620 cm⁻¹) carboxylic acid were noted.^{[ 74} , ^{75 ]} The disappearance of the latter suggests that the terminal carboxylic acid preferentially adopted a protonated state, likely due to hydrogen bonding interactions with the drug molecules. Besides, slight differences in the amide I peak intensity ratio were observed, possibly reflecting slight modifications in the microenvironment.

Figure 5.

A–C) Raman spectra of gels prepared using different methods loaded with (A) doxorubicin, (B) curcumin, and (C) methotrexate in the Amide I region. Dashed black lines highlight the carboxylic acid characteristic Raman peaks. D–F) Release profiles of (D) doxorubicin, (E) curcumin, and (F) methotrexate in pH = 7.4 phosphate buffer (10 mM) from gels prepared using different methods.

Notably, while all gels could be loaded with a drug concentration of 0.1 mM, increasing it to 1 mM led to different outcomes. Here, we compare GdL and BS systems considering their contrasting morphology and mechanical properties. In the case of doxorubicin, gel formation was completely inhibited in both GdL and BS systems at this higher concentration. However, while GdL‐based gels could still be obtained with curcumin and methotrexate at 1 mM, BS gels failed to form with any of the drugs at this higher concentration.

At the molecular level, the Raman spectra overlapped significantly with drugs‘ peaks at higher drug concentrations, which led us to use FTIR for further insights into how the drugs influenced the self‐assembled structures (Figure S15, Supporting Information). In GdL gels, the drugs induced an enhancement of the peak at ≈1640 cm⁻¹, which was tentatively assigned to β‐sheet structures. However, at 1 mM of doxorubicin, this peak was suppressed, whereas it remained present for gels containing curcumin or methotrexate – consistent with the gel formation. In contrast, the BS gels at 1 mM of the drug showed significant spectral broadening in the amide I region and loss of the characteristic band structure observed in the neat gels, suggesting the loss of the self‐assembled structure.

Hence, given that the assembly structure of drug‐loaded gels remained similar to neat gels at a drug concentration of 0.1 mM, any structural modification is expected to have a negligible impact on drug release. Instead, drug release is anticipated to be mainly influenced by other parameters, such as the gel network morphology and dynamics, drug molecular size and charge, and drug‐fibre interactions.

Drug‐fibre interactions can significantly alter gel network morphology and properties, which could not be capture from FTIR and Raman spectroscopy alone. Instead, to assess these changes, the mechanical properties of drug‐loaded BS gels were determined (Figure S16, Supporting Information). Doxorubicin induced a 100× increase of the storage modulus, likely due to additional inter‐fibre cross‐linking. The larger density of cross‐linking was also evidenced by a strain overshoot, which is commonly associated with the rearrangement of network microstructures.^{[ 76 ]} These results are in line with the findings by Xue et al.^{[ 72 ]} who reported that at comparable conditions (≈0.1 eq. DOX to gelator), doxorubicin forms positively charged nanospheres. These particles interact electrostatically with the negatively charged fibres, acting as cross‐linker agents that enhance the gels‘ inter‐fibre interactions, leading to an enhancement of the storage modulus. In contrast, curcumin only slightly increased G', implying a minimal impact on the fibre cross‐linking. On the other hand, methotrexate, introduced additional negative charges, increasing inter‐fibre repulsion and reducing cross‐linking density. This led to a pronounced decrease in G' and a less solid‐like behaviour, as suggested by the lower G'/G″.

The drug release profiles were generally biphasic (Figure 5D–F), characterized by an initial faster release stage followed by a slower release during the longer second phase. These profiles were fitted to several mathematical models (Table S6, Supporting Information), with the Korsmeyer‐Peppas and Gompertz models providing the best fit.^{[ 77} , ^{78 ]} The parameter n of the Korsmeyer‐Peppas model, which is associated with the diffusion mechanism (Table S7, Supporting Information), yielded values < 0.5 during the first 6 h of release, consistent with findings in other dehydropeptide‐based gels.^{[ 11} , ^{79 ]} This indicates a diffusion‐controlled release mechanism that remained similar across all gels, despite differences in drug release kinetics. The cumulative release amounts varied by drug, following the order: anionic (methotrexate) > neutral (curcumin) > cationic (doxorubicin). This trend aligns with the effect of the drug on the mechanical properties, and previous observations that a slower drug release occurs from hydrogels with complementary charge to the drug.^{[ 79} , ^{80 ]} Additionally, the large planar aromatic surface of doxorubicin can favour its intercalation into peptide fibres, further slowing its release.^{[ 81 ]} In this context, the negligible differences in doxorubicin release profiles suggest that the electrostatic interactions dominate over variations in network morphology.

For neutral curcumin, which does not interact via electrostatics with the gel network, the cumulative release was larger than the cationic drug. However, only minor differences were observed in the release profiles, with the heating‐cooling gel showing the slowest release. Considering the variations in storage modulus (spanning several orders of magnitude), the similar release profiles suggest that the network morphology, electrostatic interactions, and mechanical properties alone cannot fully explain the drug release profiles. Instead, specific non‐electrostatic interactions, such as π‐π interactions, might play a key role as described elsewhere.^{[ 80 ]} For instance, while electrostatic repulsion between methotrexate and the anionic gel fibres would be expected to induce a faster release, the similar cumulative release of methotrexate and curcumin in some gels implies the involvement of non‐electrostatic interactions. Interestingly, the methotrexate release also appeared to be influenced by the viscoelasticity, as the release profiles followed a trend similar to the G″ and G″ values (buffer switch < solvent switch < heating‐cooling < HC‐BS ∼ BS‐HC < GdL). However, the GdL gel displayed an unexpectedly similar release profile to the buffer switch gel, suggesting that the stronger peptide packing hinders methotrexate intercalation/adsorption into/onto the fibres.

Fluorescence emission measurements of drug‐loaded BS and GdL gels provided further insight (Figure S17, Supporting Information). In BS gels, all drugs induced a strong fluorescence emission quenching and shape change of the monomers band, while the aggregates band disappeared. This effect could result from the drugs disturbing the aromatic stacking, and affecting the microenvironment of the peptide aromatic rings. The same effect was observed for curcumin in GdL gels, while doxorubicin and methotrexate led to distinct behaviours. Doxorubicin appeared to disrupt the aromatic stacking, as observed from the disappearance of the aggregates band and increased monomers band intensity, which aligns with the inhibition of gelation by larger concentrations of doxorubicin. In contrast, methotrexate showed minimal interaction with the GdL gel fibres. It neither significantly quenched the gel‘s fluorescence emission nor strongly affected the aggregates band, thus suggesting a weak association with the GdL gel fibres.

Overall, the gelation pathway led to subtle changes in the release profile of complementary‐charged and neutral molecules, while the release of similarly charged molecules was more strongly influenced by the assembly structure. These results suggest that non‐covalent interactions can have a greater impact on drug diffusion than gel morphology and viscoelastic properties.

Abraham et al.^{[ 80 ]} also found that, in addition to the charge, specific interactions between the drug and gel fibres could determine cargo release, with network viscoelasticity having a minimal impact. Kurbasic et al.^{[ 71 ]} observed that drugs with larger aromatic units (e.g., naphthalene) were more efficient in engaging in non‐covalent interactions with the aromatic units of the peptide compared to those with smaller aromatic rings (e.g., benzene). Similarly, Chan et al.^{[ 81 ]} reported that intercalation had a greater impact on fibril‐drug interaction compared to electrostatic interactions. Hence, in this work, aromatic and electrostatic interactions, along with peptide assembly structure and morphology, were found to influence the release of similarly‐charged molecules, while their impact on neutral and complementary‐charged molecules was minimal.

3. Conclusion

In this work, we explored how the self‐assembly pathway affects the morphology, properties, and drug‐release behaviour of dehydropeptide‐based gels. The crystal structure of the naphthalene N‐capped dehydropeptide, 2‐Naph‐L‐Phe‐Z‐ΔPhe OH displayed a maximum pore diameter of ≈4.08 Å. The self‐assembly pathway significantly influenced the hydrogel network morphology and stiffness, with variations of nearly three orders of magnitude. Specifically, the pH change kinetics and a heating‐cooling cycle could strongy affect the stiffness of gels. The pathway also affected the stacking of aromatic rings and influenced the peptide conformation. This conformation remained unchanged after drug encapsulation. Electrostatic interactions played a key role in drug release rates, with slower release observed for the complementary charge drug (doxorubicin). The assembly structure also influenced the release of similarly charged molecules but had minimal effects on complementary charge and neutral molecules. These results demonstrate the potential of single peptide‐based gels for the sustained release of complementary charge and neutral molecules, with mechanical properties that can be independently adjusted by altering the self‐assembly pathway. Besides, stiffer gels with faster or slower release rates of similarly‐charged molecules could be achieved through a slow pH decrease or a heating‐cooling cycle, respectively. Overall, this work advances our understanding of how the assembly pathway influences peptide‐based gel properties and drug release, providing valuable insights for designing new gelators for biomedical applications.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Supporting Information

Veloso S. R. S., Vijayakanth T., Shankar S., et al. “Self‐Assembly Pathway Influence on Dehydropeptide‐Based Gel Properties and Drug Release.” Macromol. Biosci. 25, no. 9 (2025): 25, 70003. 10.1002/mabi.202400449

Data Availability Statement

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