The Design of a Participatory Peptide Nucleic Acid Duplex Crosslinker to Enhance the Stiffness of Self-Assembled Peptide Gels

Abstract

Herein, peptide nucleic acids (PNAs) are employed in the design of a participatory duplex PNA-peptide crosslinking agent. Biophysical and mechanical studies show that crosslinkers present during peptide assembly leading to hydrogelation participate in the formation of fibrils while simultaneously installing crosslinks into the higher-order network that constitutes the peptide gel. The addition of 2 mol% crosslinker into the assembling system results in a ~100% increase in mechanical stiffness without affecting the rate of peptide assembly or the local morphology of fibrils within the gel network. Stiffness enhancement is realized by only affecting change in the elastic component of the viscoelastic gel. A synthesis of the PNA-peptide duplex crosslinkers is provided that allows facile variation in peptide composition and addresses the notorious hydrophobic content of PNAs. This crosslinking system represents a new tool for modulating the mechanical properties of peptide-based hydrogels.

Graphical Abstract

The development of crosslinking agents to modulate the mechanical properties of peptide-based gels increases their scope of utility. We report the design, synthesis and utilization of a peptide nucleic acid duplex crosslinker (DCL) that takes part in peptide assembly leading to gel formation while simultaneously installing non-covalent crosslinks into the higher-order fibrillar network. DCL represents a new modality in the crosslinker arsenal.

Introduction

Peptide hydrogels are an important class of soft materials demonstrating utility in drug delivery,^[1] wound healing,^[2] and immunotherapy.^[3] Many gels prepared by peptide assembly are characterized by moderate mechanical rigidity, which while excellent for many biomedical applications, limit their use in scenarios demanding stiffer materials. Our group has developed a family of amphiphilic peptides, which self-assemble into β-sheet rich fibrillar gel networks under triggered conditions.^[4] Assembly of unstructured monomeric peptide is initiated by adjusting solution pH and/or ionic strength to mitigate charge repulsion. Assembly can be accelerated by increasing the temperature which drives the hydrophobic effect. Peptides assemble into fibrils composed of a bi-layered cross-β structure with individual peptides adopting β-hairpin conformations in their self-assembled state.^[5] The ultimate formation of a percolated network of physically crosslinked fibrils constitutes the formation of hydrogel. Resulting gels display shear-thin/recovery rheological properties that allow their delivery by syringe or spray. Utilizing this property, we have developed gels to deliver a variety of payloads including small molecule drugs,^[6] proteins,^[7] nucleic acids,^[8] nanoparticles,^{[6b, 8]} and cells.^[9] Although these gels are useful for many applications, increasing their stiffness would broaden their potential utility.

The mechanical rigidity of semi-flexible fibrillar peptide gels is dependent on both the mesh size of the fibril network and the stiffness of the fibrils themselves.^[10] Accordingly, G’ ~ κ²/kTαξ⁵, where changes in fibril bending modulus/stiffness (κ) and network mesh size (ξ) effect the storage modulus (G’). We and others have shown that heterochiral assembly can be used to enhance fibril stifness and ultimate gel rigidity.^[11] Mesh size can also be altered by controlling the number of crosslinks between fibrils, where more crosslinks lead to a smaller mesh size and a more rigid gel. For self-assembling systems, simply increasing the wt% of peptide used for gelation can result in the formation of more physical crosslinks. However, peptide solubility can limit the amount of peptide that can be used. As such, alternate strategies have been employed to install crosslinks such as photo-crosslinking,^[12] metal ion complexation,^[13] native chemical^[14] and enzymatic ligations,^[15] disulfide bridging,^[16] chemical crosslinking,^[17] and employing dual networks.^[18] Although these methods can be useful, we were interested in developing a new strategy where noncovalent crosslinks could be installed during the evolution of the gel network, and are incorporated as integral structural units of local fibril morphology.

Herein, we design, synthesize and utilize a bifunctional duplex crosslinker (DCL) that can take part in peptide assembly leading to fibril formation while installing a non-covalent crosslink into the higher-order fibrillar network. The duplex crosslinker (DCL) comprises a central peptide nucleic acid (PNA) duplex that displays two opposing peptides, each capable of engaging in fibril formation via assembly, Figure 1A. PNAs are synthetic oligonucleotide analogs that contain pseudo-peptide backbones that displays adenine, thymine, cytosine, and guanine nucleobases, Figure 1B.^[19] PNAs can bind to RNA, DNA and other PNAs with sequence specificity, and can act as antisense oligonucleotide probes to target genes,^[20] as in vivo gene editing tools,^[21] and as biosensors.^[22] PNAs have also been used in the preparation of self-assembled nanostuctures^[23] and incorporated into material networks to enable DNA crosslinking.^[24] Our crosslinking strategy takes advantage of strong and specific PNA-PNA duplex formation to extend attached peptides in an antiparallel fashion from the central portion of the crosslinking unit. We employed PNA:PNA as opposed to PNA:DNA duplexes since they are reported to be more stable.^[25] Importantly, the duplex-conjugated peptides can take part in fibril formation by assembling with free peptides in solution. Thus, although the attached peptides are initially unstructured, they are designed to adopt a folded β-hairpin conformation when assembled into a fibril.^[20c] Duplex crosslinkers (DCL) are prepared by simply mixing PNA-peptide monomers. Although crosslinkers prepared in this study contain the same peptide sequence, this PNA-based system is versatile and allows the display of two different peptides within a single croslinking unit by simply altering the sequence at the monomer level. As will be shown, addition of small amouts of duplex crosslinker to solutions of self-assembling peptide substantially rigidifies resultant hydrogel networks while preserving the native local morphology of fibrils within the network.

Figure 1.

(A) PNA duplex crosslinker (DCL) formed by the hybridization of individual PNA-peptide monomers. When present during peptide assembly, the DCL participates in fibril formation while simultaneously crosslinking the evolving hydrogel network. (B) Structure of PNA monomer and nucleobases.

Results and Discussion

The duplex crosslinker is comprised of two individual PNA-peptide monomers (Pep-PNA1 and Pep-PNA2), where identical peptides are covalently attached to two distinct self-complimentary PNAs. The PNA sequences were chosen to form a tight and thermally stable duplex using a minimal number of nucleobases to limit their potential for non-specific aggregation. We selected two previously reported 10-mer self-complementary PNA sequences (PNA1 and PNA2 in Table 1)^[26] and determined their binding affinity to be in the nanomolar range. We targeted this range of affinity since we plan to use the crosslinker at micromolar concentrations during gel formation. Ultimately, we limited the number of bases to ten and chose a composition to restrict the number of purine bases (adenine and guanine) to be less than 70% to decrease overall hydrophobicity. Importantly, the duplex melting temperature must be higher than the temperatures used to accelerate peptide assembly leading to gel formation. As will be shown, the complimentary PNA sequences in Table 1, provide a thermally stable duplex that meets these criteria. The N-terminus of each PNA appends an amphiphilic peptide connected through a PEG-20 unit that aids solubility and distances the peptide from the PNA duplex once formed. The sequence of the appended peptides was chosen from a family of b-hairpin forming peptides developed by our group that are known to assemble into fibrillar gels,^{[4b, 8, 27]} and is the same sequence as the free peptide used for gel formation in this study.

Table 1.

Sequences of peptide, PNAs and peptide-PNA conjugates used in this study

S. No.	Name	Sequence
1	Pep-PNA1	VLTKVKTKVDPPTKVQVKVFV-PEG10-PEG10-TGTTACGACT
2	Pep-PNA2	VLTKVKTKVDPPTKVQVKVFV-PEG10-PEG10-AGTCGTAACA
3	PNA1	TGTTACGACT
4	PNA2	AGTCGTAACA
5	Pep	VLTKVKTKVDPPTKVQVKVFV

PNAs are notoriously hydrophobic making the synthesis of their derivatives difficult. To synthesize Pep-PNA1 and Pep-PNA2 monomers, we developed a robust strategy wherein the peptide and PNA portions were first prepared separately by Fmoc solid-phase synthesis and then ligated together using strain promoted azide-alkyne cycloaddition. Figure 2 shows the synthesis of Pep-PNA1; an identical strategy was used to prepare Pep-PNA2. The key was to include a PEG-10 unit on both the PNA and peptide portions and then ligate the two PEG units to ultimately form the Pep-PNA product containing a PEG-20 linker. Including PEG₁₀ on both partners of the ligation reaction increased their solubility and enabled the final reaction. Attempts to synthesize either Pep-PNA1 or Pep-PNA2 monomers by a single linear solid phase synthesis required very long coupling times, ultimately providing a low yield of material containing multiple truncated sequences. With respect to the ligation reaction, we first attempted a copper catalyzed reaction employing a propargyl glycine-modified PNA. This failed due to copper-mediated precipitation of the lysine-rich peptide. However, the approach outlined in Figure 2 afforded highly pure intermediates and final products at milligram scale, Figures S1–S7. Lastly, the length of the PEG unit proved critical for preparing well-behaved Pep-PNA monomers. Using shorter PEG units as linkers, for example a PEG₅, resulted in Pep-PNA monomers prone to non-specific aggregation.

Figure 2.

Synthesis of Pep-PNA1. (A) Solid phase preparation of Pep-PEG₁₀-azide fragment (1). (a) Coupling: Fmoc-AA-OH (5 eq) or Fmoc-PEG₁₀-OH (5 eq), Oxyma (5 eq), DIC (5 eq), DMF, microwave, 90 °C, 4 min. For amino acids marked by arrows (V,K,T), Fmoc-AA-OH (5 eq), HCTU (5 eq), DIPEA (10 eq), DMF, r.t., 30 min. (b) Deprotection: 20% Piperidine/DMF, 90 °C, 2 min. (c) Cleavage conditions: 95:2.5:2.5; TFA:H₂O:TIS, r.t., 3 hr. (B) Solid phase preparation of DBCO-PEG₁₀-PNA1 fragment (2). (d) Coupling: Fmoc-protected PNA monomers (5 eq), HATU (4.9 eq), DIPEA (5 eq), NMP, r.t., 3 h. (e) Deprotection: 20% Piperidine/DMF, r.t., 7 min. (2X). (f) Cleavage: 95:2.5:2.5;TFA:H₂O:TIS, r.t., 3 h. (g) DBCO conjugation: DBCO-NHS (4 eq), DIPEA (5 eq), DMF/water (9/1), 40 °C, 24 h. (C) SPAAC: (h) Fragment 1 (1.2 eq), fragment 2 (1 eq), H₂O/CH₃CN (9/1), 40 °C, 24 h.

Gelation of free peptide will ultimately be performed at 37 °C to accelerate self-assembly by driving the hydrophobic effect. Thus, the duplex portion of the crosslinker should possess a higher melting temperature to ensure it stays intact during gelation. We first characterized the melting temperature of the duplex portion of the crosslinker alone without ligated peptide, namely (PNA1:PNA2). Individual PNAs used to form the PNA1:PNA2 duplex were prepared on solid support and include a C-terminal lysine residue to improve their water solubility, Figures S8–9. Temperature-dependent UV absorption measurements were performed at 260 nm to monitor the thermally induced denaturation of the duplex. UV spectra showed a characteristic hyperchromic transition due to unstacking of aromatic nucleobases upon duplex melting.^[28] The melting temperature (T_m) of the duplex is 68.8 ± 3.7 °C, which is over 30 °C higher than the temperature used for gel formation, Figure 3A. Isothermal titration calorimetry (ITC) indicates 1:1 binding (N = 0.84) that is tight (K_d = 15 nM), and that duplex formation is largely enthalpically driven ( ΔH ~ −72 kcal/mol), Figure 3B, Table 2.

Figure 3.

(A) Thermal denaturation of PNA1:PNA2 duplex (5 μM in pure water). Data shown as average ± standard deviation (n=3). (B) ITC thermogram for duplex formation in HEPES buffer (125 mM, 150 mM NaCl) at 25 °C. Upper panel shows raw data for 19 sequential injections of a solution of PNA1 into PNA2. Lower panel shows normalized heat as a function of molar ratio.

Table 2.

Thermodynamic parameters associated with the duplex formation

Parameters	Values
N	0.84 ± 0.007
ΔH	−71.6 ± 1 kcal/mole
TΔS	−5.1 kcal/mole
K	(6.7 ± 1.4)*107 M−1
Kd	(1.5 ± 0.3)*10−8 M

We next studied the effect of the ligated peptides on duplex formation. Both complimentary PNAs as well as the peptide are quite hydrophobic. Thus, it’s possible that Pep-PNA1 and Pep-PNA2 could aggerate on their own, negatively impacting duplex formation. Gratifyingly, sedimentation equilibrium analytical ultracentrifugation (sed-AUC), which measures apparent molecular weights (M_app) in solution, shows that each Pep-PNA construct behaves as a single ideal monomer when dissolved in water, Figures 4A and B. Sed-AUC also shows that when added together, the constructs cleanly form the desired Pep-PNA1:Pep-PNA2 duplex, Figure 4C. For each case, the values of M_app are slightly lower than the calculated molecular weights. This is common for highly charged molecules where their sedimentation is affected by the primary charge effect.^[29] Thermal denaturation of the duplex crosslinker provided a T_m of 64.2 ± 2.5 °C, Figure 4D. This is 4 °C lower than the PNA1:PNA2 duplex, suggesting that the added peptide minimally influences the thermal stability of the duplex portion of the crosslinker. In control experiments, no thermal transitions were observed for Pep-PNA1 nor Pep-PNA2 (Figure S10A). Further control experiments showed that the peptide portion does not contribute to the absorbance signal at 260 nm, the wavelength used to monitor the dissociate of the duplex, Figure S10B. Lastly, CD spectroscopy shows that the attached peptides are unstructured when the duplex crosslinker is dissolved in water, Figure S11A. However, if buffer is added, the peptides begin to form β-sheet structure (Figure S11B), indicating that they are capable of assembly under the permissive conditions of gel formation. Control experiments show that the duplex portion of the crosslinker does not contribute to the far UV-CD signal, which would complicate spectra interpretation, Figure S11C. Taken together, the data show that the designed duplex crosslinker is thermally stable existing as a single species in water, but poised to participate in assembly events under solution conditions that trigger gelation.

Figure 4.

Sedimentation analytical ultracentrifugation of aqueous solutions of (A) Pep-PNA, (B) Pep-PNA2, and (C) Pep-PNA1:Pep-PNA2 duplex crosslinker. Data collected at 26,000 (blue); 41,000 (yellow); and 51,000 rpm (red) for each sample and fit globally to a single ideal species model. Apparent and calculated molecular weights are shown along with the residuals of fit. (D) Thermal denaturation of Pep-PNA1:Pep-PNA2 duplex crosslinker. Data shown as average ± standard deviation (n=3).

We next examined the influence of the duplex crosslinker (DCL) on the mechanical properties of the hydrogel. The DCL was designed to install crosslinks during the evolution of the gel network while simultaneously participating in peptide assembly and fibril formation. We first studied the effect of added crosslinker on the evolution of mechanical stiffness observed during the self-assembly of a 1 wt% solution of free peptide (Pep, Table 1). Here, self-assembly leading to gelation is initiated by adjusting the pH and ionic strength of a cold aqueous peptide solution by the addition of buffer and then raising the temperature to 37 °C. Time-sweep oscillatory rheology shows that in the absence of duplex crosslinker, Pep forms a moderately stiff gel with a storage modulus (G’) of 800 Pa, Figure 5A and S12A (orange data). Gelation is biphasic with an initial fast phase followed by a slow increase in G’ over extended times. The gel displays shear-thin/recovery behavior, recovering its mechanical stiffness after the application of 1000% strain. When 2 mol% duplex crosslinker is added and assembly triggered, the resulting gel network is nearly two-fold stiffer (1,500 Pa) and the initial rate of gel formation is significantly faster, Figures 5A/D and S12B (blue data). The crosslinked material also maintains its ability to recover after shear thinning (Figures 5A and S13), recovering greater than 90% of its original stiffness. Figures 5B and C show that the time-sweep data was collected in the linear regime of both frequency (6 rad/sec) and strain (0.2%). Interestingly, although the crosslinker increased the rigidity of the gel (Figure 5D), there was little observable change of the loss modulus (G”), Figure 5E. The loss factor (tanδ = G”/G’,) which corresponds to the damping character of the material, decreased two-fold upon crosslinking; 0.044 ± 0.029 versus 0.023 ± 0.008, respectively (Figure 5F). Further, a greater than 50% decrease in the crossover point is observed, Figure 5G. The cross-over point corresponds to the intersection of G’ and G” as a function of strain. When material formation is triggered at 50 °C, a temperature approaching the DCL T_m, there is no difference in G’ between 0% and 2% DCL-containing gels, Figure S14. Thus, when using temperature to accelerate gel formation for this system, a balance must be struck between driving peptide assembly via the hydrophobic effect whilst minimizing the breaking of H-bonds that stabilize the PNA duplex. Taken together, the rheological data indicated that the duplex crosslinker rigidifies the viscoelastic gel mainly by modulating its elastic component while allowing shear-thin/recovery behavior.

Figure 5.

(A) Dynamic time sweeps measuring the storage modulus (G’) of 1 wt% pep hydrogels with 0% (orange data) and 2 mol% (blue data) duplex crosslinker (DCL) at constant strain (0.2 %) and frequency (6 rad/s). At 180 min gels are shear-thinned at 1000% strain and allowed to recovery after reduction of strain to 0.2%. (B) Dynamic frequency sweeps at 0.2% strain for 0% and 2 mol% duplex crosslinked Pep hydrogels. (C) Dynamic strain sweeps of Pep hydrogels with 0% and 2 mol % duplex crosslinking at a constant frequency of 6 rad/sec. The bar graphs show the difference in (D) storage moduli (G’), (E) loss modulus (G”), (F) loss factor (tan δ), and (G) crossover point between 0% and 2 mol% duplex crosslinked Pep hydrogels that are collected from the dynamic time sweep at 180 min Data shown as average ± standard deviation (n=3).

Finally, we examined the influence of DCL on fibril formation and morphology. The DCL is designed to form network level crosslinks, but as an integral structural unit of the fibrils constituting the gel. Thus, the secondary structure content of the fibrils, their rate of formation, and local morphology should be minimally perturbed by the duplex crosslinker if it is joining in self-assembly as a benign participant and not altering the intrinsic assembly mechanism. Figure 6A shows CD spectra of fibers formed by Pep alone and Pep containing 2 mol% duplex crosslinker (DCL). Spectra show characteristic minima at 216 nm, typical of β-sheet secondary structure. Importantly the magnitude of the mean residue ellipticity values are nearly identical. Figure 6B shows that the rates of β-sheet evolution, which reports on fibril formation, are also very similar. Data fitting provided first order rate constants of 0.017 and 0.019 min⁻¹ for fibrils formed without and with crosslinker, respectively. Finally, TEM shows that fibrils isolated from hydrogels formed by Pep alone or with crosslinker have very similar local morphologies, characterized by measured widths of 4.6 and 4.2 nm, respectively (Figures 6C/D). Taken together, the data indicate that the crosslinker takes part in peptide assembly leading to fibril formation without affecting large changes in assembly mechanism nor morphological outcome.

Figure 6.

(A) Circular dichroism spectra show no difference in the β-sheet content of Pep fibrils containing either 0% (orange data) or 2 mol % (blue data) Pep-PNA1:Pep-PNA2 duplex crosslinker (DCL). Data collected at 37 °C in 25 mM HEPES buffer containing 150 mM NaCl, pH 7.4. (B) Rates of β-sheet formation during fibrillogenesis for samples containing 0% or 2 mol% Pep-PNA1:Pep-PNA2 duplex crosslinker. Data collected at 50°C in 1X HBS buffer (25 mM HEPES, 150 mM NaCl), pH 7.4. Inset shows kinetic parameters derived from first order best fit, R² = 0.96 and 0.95 for 0% and 2 mol% crosslinked systems, respectively. Data shown as average ± standard deviation (n=3). (C,D) TEM images show no difference in fibril morphology for fibrils form with (C) 0% and (D) 2% duplex crosslinker; (inset) width measurements of fibrils formed in presence of 0% and 2% duplex crosslinker. N= 101 and 111, respectively. Scale bar = 100 nm

Lastly, we studied the release rate of two drugs, BAY 2416969 and Amonafide from crosslinked and non-crosslinked 1wt% peptide gels, Figure 7. BAY 2416964 inhibits aryl hydrocarbon receptor (AhR), an immune metabolomic target in cancer.^[30] Amonafide is a known DNA intercalator and topoisomerase II inhibitor.^[31] After loading the same concentration of each drug into the peptide gels, cumulative release was followed over a week. Amonafide showed burst release from the gel network within hours (70% release within 24 hr), Figure 7D. The burst release of protonated Amonafide from the electropositve gel (each peptide comprising the gel carries a +5 charge at pH 7.4) is most likely due to electrostatic repulsion. In contrast, BAY 2416964 showed slower and more sustained release (70% release over 7 days), Figure 7C. Crosslinking had little influence on the release rates. This is expected given that the mesh size of the gels ranged from 25 to 28 nm, for crosslinked and non-crosslinked gels, respectively.

Figure 7:

Structures of anticancer drugs, (A) BAY 2416964 and (B) Amonafide. Cumulative bulk release of (C) BAY 2416964 (D) Amonafide from 1 wt% Pep hydrogels containing 0% (orange data) and 2 mol% (blue data) duplex crosslinker (DCL) at 37 °C. Hydrogels formed at pH 7.4 (25 mM HEPES, 150 mM NaCl) at 37 °C.

Conclusion

Although PNAs are widely used as antisense oligonucleotides, biosensors, and gene-editing tools, their application as gel crosslinkers is less explored. We show that the complexation of PNA-peptide hybrids can afford duplex crosslinkers that are capable of rigidifying peptide gels during the evolution of the fibrillar network. Duplex crosslinkers take part in peptide assembly without significantly influencing fibrillogenesis while installing network-level non-covalent linkages. A synthesis of the starting monomeric PNA-peptide hybrids is provided that circumvents problems due to high hydrophobic content and should allow the facile incorporation of difference peptide sequences by click ligation. Subsequent complexing of PNA-peptide monomers represents a generalized platform for constructing duplex crosslinkers of varying peptide composition. Although, identical sequences were used in this study, one can envision preparing duplexes that display different peptide sequences that could be used to modulate the rigidity of mixed peptide gels, for example heterochiral systems. Finally, it would be interesting to explore PNA crosslinking for peptide gels prepared by phase separation.^[32]

Data Availability Statement:

The data that support the findings of this study are available in the supporting information of this article.