Modulating refolding and collagen binding of collagen hybridizing peptides through peptoid incorporation

Abstract

Collagen hybridizing peptides (CHPs) selectively bind denatured collagens (dColl), offering a powerful platform for diagnostics and targeted drug delivery. However, the rapid self-trimerization of GPO-derived CHPs (O = hydroxyproline) limits their applications. Here, we present peptoid-based strategies to reduce the triple helix stability and refolding speed of (GPO)₉ by introducing derivatives of Nasp (N-carboxymethyl glycine). Systematic substitution studies revealed that N^α–C^β branched Nsce (N-(S)-carboxyethyl glycine) destabilized the triple helix primarily through steric hindrance rather than electrostatic repulsion. Triple Nsce substitutions not only drastically reduced refolding propensity but also compromised dColl binding. In contrast, the non-branched Nasp residue, under the same conditions, slowed refolding while enhancing gelatin binding compared to (GPO)₉. The CHP incorporating Nasp stained dColl associated with skin melanoma in histological sections as effectively as (GPO)₉. Our findings demonstrate that CHPs incorporating peptoid residues may enable fine-tuning of CHP folding and binding kinetics for the development of next-generation CHPs.

INTRODUCTION

Collagens are the predominant proteins found in the extracellular matrices (ECMs) of vertebrates.¹ A collagen molecule is composed of three polyproline-II (PP-II) polypeptide chains that intertwine to form a distinctive supersecondary protein structure known as the triple helix. The primary structure of collagens necessary for triple helical folding consists of a repetitive Gly-Xaa-Yaa (GXY) tripeptide sequence, where X and Y are commonly proline (Pro, P) and hydroxyproline (Hyp, O), respectively. The triple helix is stabilized by extensive inter-strand hydrogen bonding between the N-H group of the glycine (Gly, G) residue and the C=O group of the X residue in an adjacent strand, as well as intra-strand n $\to$ π* interactions facilitated by the pyrrolidine rings of Pro and Hyp.² Collagens are widely distributed throughout the body, playing pivotal roles in providing not only the structural integrity of tissues but also cell functions, which are crucial for tissue maintenance and regeneration.^{3, 4} In a healthy state, collagen synthesis and degradation are tightly regulated to maintain homeostasis. In contrast, dysregulation of collagen remodeling is observed in a wide range of diseases such as cancers and fibrosis.^{5, 6} Consequently, many diseased tissues contain higher levels of denatured collagens (dColl) compared to healthy tissues. In addition, mechanical damage to load-bearing tissues can denature collagen.^{7, 8} Denatured collagen strands in tissues have proven to be a useful theranostic targets for a variety of pathological features associated with disease and injury.^{9, 10}

Collagen mimetic peptides (CMPs) have been extensively used for understanding the triple helical structure and biological functions of collagens. However, our laboratory conceived an innovative approach to exploit the triple-helical propensity of single-stranded CMPs for probing collagen denaturation.¹¹ In this application, the CMPs are dissociated into monomers and allowed to bind dColl in tissues by triple-helical folding. To more precisely define the property and function of this peptide, we introduced the term Collagen Hybridizing Peptides (CHPs). We and other labs showed that CHPs labeled with imaging agents (e.g., fluorescent molecules and radioisotopes) can i) highlight collagen remodeling activity in tissue sections, ii) detect mechanically damaged collagens in ex vivo models^{7, 8, 12–15}, and iii) selectively accumulate in regions of high collagen degradation in various preclinical disease models.^16–28 This selective binding is attributed to their strong triple-helical propensity and neutrally hydrophilic nature, which together minimize nonspecific interactions. While CHPs offer a promising approach to specifically deliver therapeutics and diagnostic agents to sites with damaged collagens, their clinical translation is limited by their strong tendency to self-trimerize into homotrimers. GPO-based CHPs readily form homotrimers²⁹, which lack the dColl binding property. To dissociate CHP homotrimers into single strands and restore affinity for dColl, the peptides must be heated above their melting temperatures (T_ms). However, upon cooling before or during use, the single-stranded CHPs can reform homotrimers, reducing the concentration of active monomeric strands, which leads to variability between uses.

Several approaches were employed to reduce the homo-trimerization propensity of CHPs. One approach involved substituting the central Gly with N-nitrobenzyl Gly (Nnbz), resulting in a caged CHP that can be UV-activated to fold into the triple helix.¹⁶ However, this approach had drawbacks, including the release of a potentially toxic by-product, o-nitrosobenzaldehyde³⁰, and a rapid self-trimerization, particularly at high CHP concentrations after UV irradiation. Alternatively, we substituted all Pro residues in the (GPO)₉ CHP with (2S,4S)-fluoroproline (fPro, fP), which introduced inter-strand steric hindrance between the (4S)-fluorine of fPro and the (4R)-hydroxyl group of Hyp. Although this modification reduced self-trimerization similar to the work reported previously³¹, it also lowered dColl binding affinity compared to the GPO-based CHP.²¹ The Koide group incorporated charged amino acids into a cyclized CMP, resulting in a drastic reduction in the self-trimerization propensity;³² however, the cyclization also appeared to lower the CHP’s selective binding to dColl.³³ Therefore, a new strategy is needed to mitigate CHP’s self-trimerization while preserving the binding affinity and selectivity for dColl.

Peptoids are peptidomimetics composed of N-substituted glycines (N-glys), also referred to as peptoid residues. Previous studies by our group showed that the GXY sequence of collagen can accommodate various types of N-glys at the X and Y positions and that these building blocks are conducive to triple-helical folding similar to Pro.^{34, 35} We showed that a CMP with triple helix stabilizing Nnbz residues at the X position can hybridize to dColl after heat activation. Recently, the Li group demonstrated that substituting three Pro residues with N-(2-picolyl) Gly (N2pic) resulted in a CHP that can be triggered by changes in pH to fold and bind dColl. This is because acidic conditions disfavor the trans isomer of N2pic, while neutral conditions favor it, leading to PP-II folding.³⁶ These studies suggest that N-glys offer an innovative approach to modulate the CHP’s triple helical propensity and dColl binding.

Our lab previously focused on N-gly residues for their role in stabilizing the triple-helical structure of collagen mimetics that assemble as homotrimers.^{34, 35} In this study, we investigated how specific N-gly residues with chiral N^α–C^β branching (Figure 1) disrupt triple-helix folding when incorporated into CHPs, focusing on the effects of substitution position on thermal stability and refolding kinetics, as well as the dColl hybridization efficiency of CHPs with multiple N-gly residues. Through this work, we identified a new CHP sequence that binds dColl more effectively than the state-of-the-art (GPO)₉ sequence and demonstrated its use in histological staining of dColl associated with skin melanoma.

Figure 1.

Peptoid residue design. (A) Substitution of central Pro with Nval or Nasp reduces the T_m of a triple-helical peptide (GPO)₇ by 19 and 5 °C, respectively.³⁴ Relying solely on Nval to reduce the high homotrimeric propensity of CHP may result in CHPs with nonspecific binding due to the substantial hydrophobicity of the Val side chain. Combining structural features of Nval and Nasp results in two N-carboxyethyl glycine enantiomers, Nrce and Nsce, which are less hydrophobic than Nval. (B) Molecular modeling shows potential steric interactions of Nrce (orange), Nsce (lime green), and fluorine (light green) within the triple helix. The N^α–C^β methyl group of Nrce in the X position (N^xrce) clashes with the carbonyl O of Gly in a nearby strand. For Nsce in the X position (N^xsce), its methyl branch has a steric clash with the OH group of Hyp from another peptide chain, which is similar to the steric clash induced by the fluorine atom of fluorinated Pro (fP) in the (GfPO)₉.²¹

EXPERIMENTAL SECTION

Chemicals and materials used in the synthesis and purification of peptides are listed in Table S1 of the Supporting Information. They were purchased from commercial suppliers and used as received, without further purification.

Peptide synthesis and carboxyfluorescein labeling

All peptides were synthesized manually using Fmoc-mediated solid-phase peptide synthesis on Fmoc-Rink amide OctaGel Resin (0.35 mmol/g, 100–200 mesh), with dimethylformamide (DMF) as the solvent. Peptoid residues were incorporated into the growing peptide chains using the bromoacetic acid-mediated submonomer method.^{34, 37} L-Alanine tert-butyl ester hydrochloride, glycine tert-butyl ester hydrochloride, and L-alaninamide (each at 10 equivalents) were used as amine submonomers to introduce N-(S)-carboxyethyl glycine (Nsce), N-carboxymethyl glycine (Nasp), and N-(S)-propanamide (Nspam) residues, respectively, to the CHPs. L-Alanine tert-butyl ester hydrochloride and glycine tert-butyl ester hydrochloride were each premixed with equimolar amounts of N,N-diisopropylethylamine (DIEA) prior to their addition to the reaction mixtures.

Amino acid coupling was performed using Fmoc-amino acids, hexafluorophosphate benzotriazole tetramethyl uronium (HBTU), and 1-hydroxy-7-azabenzotriazole (HOAt) in equimolar amounts. The coupling of Fmoc-Gly-OH to Nsce and Nspam residues was mediated by (7-azabenzotriazol-1-yloxy)tripyrrolidinophosphonium hexafluorophosphate (PyAOP) with a 1.5-hr reaction time, repeated twice. Fmoc removal was carried out using 20% piperidine in DMF. Fmoc-amino acids and coupling reagents were added at 5 equivalents relative to the growing peptide chains.

For non-labeled sequences, the N-termini of the peptides were acetylated with 30 equivalents of acetic anhydride in the presence of an equimolar amount of DIEA. After acetylation, the CHPs were globally deprotected and cleaved from the resin by treatment with a cleavage solution composed of trifluoroacetic acid, water, and triisopropylsilane (95:2.5:2.5, v/v/v). The crude peptides were precipitated from the cleavage solution using cold diethyl ether and separated by centrifugation and decantation of the supernatant. This process was repeated twice. The residual ether was allowed to evaporate overnight. The resulting crude peptides were then solubilized in deionized water and purified by reverse-phase HPLC. Purified peptides were lyophilized and redissolved in deionized water to prepare stock solutions. Concentrations of the resolubilized peptides were measured by SpectraMax M2e microplate reader (Molecular Devices, San Jose, CA, USA).

Carboxyfluorescein (CF) was coupled to the N-terminal amino group of the Gly₃ linker on the CHPs. Briefly, each 5 equivalents of CF and PyAOP were mixed with 10 equivalents of DIEA in DMF to activate CF for 1 min. The activated CF solution was added to the resin-bound peptides, and the coupling reaction was allowed to proceed for 24 hr. CF-conjugated peptides were cleaved, purified, and characterized as described above. Peptide characterization was performed using a Thermo Scientific Exploris 120 Orbitrap mass spectrometer and a Waters Xevo G2-S QTOF [for Ac-(GPO)₉] at the Mass Spectrometry Facility of Department of Chemistry of University of Utah. Analytical HPLC was also performed to assess peptide purity. The HPLC conditions and peptide characterization data are provided in Tables S2 to S5 in the Supporting Information.

Circular dichroism spectroscopy

Prior to circular dichroism (CD) spectroscopy, all stock solutions were heated at 90 °C for 10 min and stored at 4 °C for ≥ 48 hr. Each stock solution was aliquoted to prepare working solutions at 150 μM (total single strands) in a cold 1x PBS (pH 7.4) or 0.1 M HCl. Working solutions were prepared in a cold room maintained at 4 °C. All CD experiments were performed using a JASCO J-1500 CD spectrometer (JASCO, Easton, MD, USA) with a 1-mm quartz cuvette. Data processing and conversion of recorded ellipticity (θ) to molar ellipticity ([θ], 10³ deg cm² dmol⁻¹) were done using JASCO Spectra Manager Version 2 software (Version 2.15.01). CD spectra of each sequence were recorded every 0.1 nm, ranging from 200 to 250 nm at 4 °C. The original CD spectra were smoothed using the means-movement method available in the JASCO Spectra Manager software. Changes in the CD signal at 225 nm were monitored over a temperature range of 5–90 °C (heating rate: 0.5 °C/min) to obtain CD thermal melting profiles of the CHPs. First derivative plots were generated from the melting profiles using the Savitzky-Golay algorithm. The temperature corresponding to the minimum of each first derivative plot was defined as the T_m. The T_m values were determined in either the 1× PBS (pH 7.4) and 0.1 M HCl (pH 1.3) and were accurate within ±0.5 °C between multiple measurements. The CD spectra and CD thermal melting data are provided in Figures S1 and S2 of the Supporting Information.

For refolding kinetics experiments, the working solutions were transferred into a 1-mm quartz cuvette and heated at 90 °C for 10 min in a water bath to fully dissociate CHP homotrimers. The cuvette was then quickly placed in the CD sample holder preset to 4 °C and equilibrated for 1 min. Then, the change in θ at 225 nm was monitored over a 24-hr period. The θ value prior to heating was defined as 100% folded (native state). The fraction folded (F) at each time point was calculated using the following equation: F = (θ_t – θ_in)/(θ_nat – θ_in), where θ_t is the θ at time t, θ_in is the θ at the initial time point (after 1 min of cooling at 4 °C), and θ_nat is the θ signal at the fully folded native state. The time at which F equals 0.5 was defined as the t_1/2(refold) when 50% of monomeric CHP has refolded into CHP homotrimers.

Gelatin binding assays

Gelatin powder from porcine skin (Sigma-Aldrich, St. Louis, MO) was mixed with 1x PBS (pH 7.4) and heated at 80 °C to prepare a porcine gelatin solution at a concentration of 10 %w/v. The gelatin solution (15 μL) was added to each well in a 96-well plate (black wall, clear bottom; Corning, Kennebunk, ME) to fully cover the well bottom. Excess gelatin solution was then removed by pipetting. After the coating, the plate was stored in a refrigerator for at least 15 min. Gelatin in each well was subsequently crosslinked with 100 μL of a solution containing 2 mM NHS and 10 mM EDC HCl in MES buffer (pH 4.7). The plate was agitated overnight at room temperature. After crosslinking, the solution was removed, and the gelatin films were washed four times with 100 μL of 1x PBS (pH 7.4).

To assess gelatin binding, 50 μL of each CF-labeled CHP (prepared at 10 μM) was added to designated wells and incubated for 24 hr to allow binding to the gelatin films. CF-labeled (GPO)₉ was used as a positive control, while scrambled sequences of the peptoid-containing CHPs were used as negative controls. For the “+Heat” group, the CF-labeled peptide solutions were heated at 90 °C for 5 min and cooled in an ice bath for 1 min prior to addition to the wells. Peptide solutions in the “−Heat” group were not heated prior to addition. After incubation, the peptide solutions were removed, and each well was washed four times with 100 μL of cold 1x PBS (pH 7.4). Fluorescent signals from all treatments were measured by the SpectraMax M2e microplate reader (λ_ex = 492 nm, λ_em = 537 nm).

Tissue staining and fluorescence imaging

A melanoma skin tissue was isolated from a syngeneic mouse generated by the previously described procedure³⁸. The tissue was embedded in OCT, frozen, and sectioned into 15 μm-thick slices. Prior to staining, the tissue slides were equilibrated to room temperature. The OCT-embedded sections were then rinsed with 1x PBS (pH 7.4) for 5 min. CHP staining was performed as follows: working solutions of CF-(GPO)₉ and CF-(N^xasp)₃-CHP were prepared at 20 μM in 1x PBS (pH 7.4). The peptide solutions were heated at 90 °C for 5 min in a water bath, then immediately quenched in an ice bath for 1 min. A 100 μL aliquot of the 20 μM CF-(GPO)₉ or CF-(N^xasp)₃-CHP solution was promptly pipetted onto each tissue section. The slides were incubated overnight in a humidity chamber at 4 °C. After incubation, the peptide solutions were removed, and the slides were washed three times with 1x PBS (pH 7.4). The slides were mounted with Fluoroshield. For total collagen staining, tissue sections were incubated in a tissue steamer (IHC World, LLC.) for 5 min, followed by CHP staining as described above. A separate tissue section was treated with hematoxylin and eosin (H&E) staining. Fluorescent imaging was performed using EVOS FL Auto Cell Imaging System (Thermo Fisher Scientific) with a 4x objective lens, while H&E images was taken with an AMG EVOS XL Cell Imaging System (Thermo Fisher Scientific) at 4x and 20x magnifications. Fluorescence from the CF-labeled CHPs was detected using the GFP channel. The fluorescent signals corresponding to CHP binding were measured as mean intensity after background subtraction using ImageJ software. Tissue staining was performed in triplicate using separate sections from the same OCT-embedded melanoma tissue block.

RESULTS AND DISCUSSION

Peptoid residue design

Conventional CHPs have strong triple-helical propensity with a triple helix melting temperature above 70 °C. Therefore, they need to be briefly heated (to melt the homotrimeric CHPs) before they can be used to target dColl. Because CHPs can trimerize during cooling, this process can lead to inconsistencies in the amount of active single-stranded CHPs available for dColl binding. Under concentrated conditions (e.g. 40 μM for in vivo imaging), CHPs with high T_m may refold into homotrimers rapidly enough to deter dColl binding. Our previous study showed that substituting the central Pro of (GPO)₇ with the peptoid analog of Val (Nval) caused significant triple helix destabilization due to steric effect, resulting in a 19 °C decrease in T_m, while the peptoid analog of Asp (Nasp) reduced the T_m by 5 °C at pH 7.4—an effect not observed at pH 1.8—indicating that its destabilizing influence is primarily due to charge interactions.³⁴ Therefore, we thought that these two peptoid residues, which destabilize the triple helical homotrimers via different mechanisms, could be used to retard the homo-trimerization of CHPs and improve their dColl binding for various applications.

We investigated a new peptoid residue, N-carboxyethyl glycine (Nr/sce) (Figure 1A), which combines the chemical features of both Nval and Nasp. This design was expected to reduce the homo-trimerization propensity of CHP through a combination of steric hindrance and electrostatic repulsion without making the peptide too hydrophobic, since Val’s hydrophobic side chain³⁹ can cause non-specific binding, which is undesirable for both ex vivo tissue staining and in vivo imaging applications. The design was also aimed to promote dColl binding because the peptoid residues with similar side chains favor the trans conformation, thereby stabilizing the PP-II backbone.^{40, 41} Since Nr/sce is chiral, molecular modeling was performed to determine the enantiomer best suited for CHP development.

We built 3D molecular models of triple helical CMPs that contain Nrce (N^xrce-CMP) or Nsce (N^xsce-CMP) in the X position of (GPO)₃GXO(GPO)₃ using UCSF ChimeraX software.^42–44 These models were modified from the N^xphe-CMP crystal structure (PDB entry: 7JX5). N^α–C^β bonds of Nrce and Nsce were rotated to find conformations with a minimal number of clashes, resulting in rotamers shown in Figure 1B. The N^α–C^β methyl group of the N^xrce residue clashed with the carbonyl oxygen of the peptide backbone on an adjacent strand, which could significantly impair its ability to make the triple helix. In contrast, the N^α–C^β methyl group of N^xsce residue showed a steric clash with the hydroxyl group of Hyp on a nearby strand. This steric interaction appeared similar to that previously observed in CHPs containing fluorinated Pro, where an interchain steric clash between the fluorine atom and the hydroxyl group of Hyp was shown to lower the T_m while preserving the peptide’s capacity to bind dColl.²¹ Therefore, we proceeded to explore the Nsce monomer for our study.

Effect of Nsce residue’s position on the triple helix stability

We investigated how the position of the Nsce monomer in the CHP affects T_m reduction. Substituting a single Pro residue with an Nsce residue at different triplet locations of (GPO)₉ resulted in a decrease in T_m by 9~16 °C (Figure 2). The destabilizing effect was the least when the substitution was at the two termini of the peptide, particularly in the first triplet from the N-terminus. This outcome was expected, since fraying is typically observed at the ends of helices,⁴⁵ and the influence of the N-terminal residue on triple-helical stability is reported to be generally lower than that of the C-terminal residue.^{46, 47} Interestingly, the Nsce residue had the strongest destabilization effect when it was placed at the penultimate positions, with its destabilization effect diminishing as substitution was moved toward the center of the peptide. This finding contrasts with previous studies, which demonstrated that substitutions at the center of triple helices produced the greatest effect (ΔT_m)—for example, the substitution of the central Gly with Nnbz or α-aza-Gly.^{16, 48, 49}

Figure 2.

Schematic sequences of CHPs containing Nsce residue at different triplet position (N^xsce-CHPs) with melting temperatures (T_m) and changes in T_m (ΔT_m) in comparison to (GPO)₉. The T_m values were determined in 1× PBS (pH 7.4) and 0.1 M HCl (pH 1.3) and were accurate within ±0.5 °C between multiple measurements.

Next, we investigated whether the charge–charge repulsion contributed to the destabilization of CHP homotrimers. The N^xsce-CHPs were prepared in 0.1 M HCl (pH 1.3) to suppress the ionization of the Nsce residues, thus eliminating the side chain’s negative charges. The T_m of each neutral N^xsce-CHP was measured and summarized in Figure 2. Interestingly, the ΔT_m values remained similar between neutral (measured at pH 1.3) and anionic CHPs (measured at pH 7.4) (ΔΔT_m: 0~3 °C), which suggested that the Nsce exerted minimal electrostatic repulsion, and the destabilizing effect was primarily due to inter-strand steric hindrance.

We believe that the steric hindrance exerted by Nsce causes the greatest structural destabilization of the triple helix in regions of the CHP where local stability is intermediate—the locations between the middle and the two termini. The middle region is less susceptible to destabilization by Nsce due to the presence of flanking GPO repeats, which provide high stability, whereas the terminal regions are already unstructured for further destabilization by Nsce. The relative magnitude between the destabilization induced by the Nsce side chain and the stabilization provided by adjacent GPO triplets appears to be critical. If the destabilizing effect of Nsce were sufficiently strong to overcome the stabilizing influence of the flanking GPO repeats, such as the case of substitution of Gly with Nnbz¹⁶, it would likely result in greater destabilization at the peptide center rather than other locations. Altogether, the results demonstrate that the Nsce can reduce the stability of the CHP homotrimers (as represented by the decrease in T_m) without increasing hydrophobicity; however, in contrast to what we had expected, there was no synergistic destabilizing effect of combining the side chain features of Nval and Nasp.

Effect of Nsce residue’s position on the triple helix refolding rate

We further investigated the impact of Pro → Nsce substitutions on the refolding rate of the CHP homotrimers using temperature-jump CD experiments. The number of CHP monomers available for dColl binding after heat activation depends on the rate of homotrimer refolding—a faster refolding rate results in fewer monomeric strands available for binding. Information from these refolding kinetics studies, combined with homotrimer stability data, may help design CHPs with tailored folding and dColl binding properties.

Representative refolding profiles of N^xsce-CHPs in 1× PBS (pH 7.4) and 0.1 M HCl (pH 1.3) are shown in Figures 3A and S3, respectively. The refolding halftime, t_1/2(refold) values are summarized in Table S6. A single Pro → Nsce substitution increased the t_1/2(refold) by 2~15 min, depending on the triplet position. As shown in Figure 3B, substitution at the N-terminal triplet had no significant effect on the refolding rate (p > 0.05), while substitution at the C-terminal triplet moderately retarded the refolding. This is consistent with prior kinetic studies, which showed the C-terminal modification having slightly greater effect than the N-terminal modification.^{50, 51} The substitution at the 2^nd triplet from the N-terminus increased the t_1/2(refold) to a similar extent as the C-terminal substitution, although both had smaller effects than substitutions at the internal triplets of the CHP [t_1/2(refold) 31 – 34 min]. Although we cannot rule out peptoid residues’ effect on nucleation step, since GPO segments are highly effective in nucleation and nucleation can occur at any segments of the peptide^{50, 52, 54}, it seems that the difference in folding rates observed here can be attributed partly to the bulky branched side chain creating additional energy barrier for the propagation of triple helical folding particularly at the internal positions of the peptide during the propagation of the triple helix. Ionization (pH 1.3 vs pH 7.4) had minimal effect on the t_1/2(refold) values (Figure 3C), suggesting that this energy barrier originates mainly from the steric effect rather than the charge effect.

Figure 3.

Refolding kinetics of homotrimeric N^xsce-CHPs (150 μM) determined at 4 °C. (A) Representative CD refolding profiles of N^xsce-CHPs in 1× PBS (pH 7.4). (B) The refolding rate as represented by the refolding half time, t_1/2(refold) varies with the substitution position. The data were analyzed by one-way analysis of variance (ANOVA) with Tukey’s test. The peptide sequences sharing same letters have the calculated p-values for mean comparison ≥ 0.05. (C) The t_1/2(refold) values between charged (at pH 7.4) and neutral (at pH 1.3) N^xsce-CHP were not statistically different. Two-way ANOVA with Šídák’s multiple comparison test was performed for data analysis. The data presented in (B) and (C) represent the mean ± SD of three independent measurements and are ordered according to the substitutions from the N- to C-terminus.

Stability and refolding rate of CHPs with three Nsce substitutions

In order to lower the T_m of the CHP further down, three Pro → Nsce substitutions were made—two at the penultimate triplet positions and one at the middle triplet positions [(N^xsce)₃-CHP, Figure 4A]. We thought that this CHP would bind dColl at a level similar to (GPO)₉ because Nsce favors the trans conformation⁴¹, thereby promoting the PP-II and triple helical folding. CD thermal melting temperature of (N^xsce)₃-CHP was 31 °C (Figure 4A), which is close to the additive destabilization (ΔT_m) from each of the three N^xsce positions. Under the acidic condition (pH 1.3), where the Nsce residues are neutral, the T_m of (N^xsce)₃-CHP increased by only 5 °C, further supporting that steric hindrance mediated by the carboxy ethyl group, not the electrostatic repulsion, played a dominant role in destabilizing the triple helix. The CF conjugation increased the T_m of (N^xsce)₃-CHP (Figure 4A) by 2 °C. Compared to fPro, Nsce exhibited a stronger destabilizing effect: three Nsce substitutions were sufficient to reduce the T_m below body temperature (37 °C), which is useful for in vivo application, while nine Pro → fPro substitutions were required to achieve a similar outcome.²¹

Figure 4.

Effects of three Pro → Nsce substitutions in the CHP homotrimer formation and dColl hybridization. (A) Peptide sequences and T_m of the Ac-(N^xsce)₃-CHP and CF-(N^xsce)₃-CHP. The T_ms were measured by CD in 1× PBS (pH 7.4). (B) CD refolding kinetic experiments performed in 1× PBS (pH 7.4) showed that (N^xsce)₃-CHP (at 150 μM) refolded substantially slower than (GPO)₉ (see Figure 5B). Neutralization of Nsce side chain minimally changed the refolding rate of (N^xsce)₃-CHP. (N^xsce)₃-CHP showed no sign of refolding at 37 °C. (C) N-terminal carboxyfluorescein conjugation increased the refolding rate of (N^xsce)₃-CHP. (D) Comparative gelatin binding assay with and without a heat-activation step (90 °C for 5 min) of (N^xsce)₃-CHP, (GPO)₉ (positive control), and the scrambled peptide (N^xsce)₃-CHP^Scr (negative control). The peptide solution (10 μM in 1× PBS, pH 7.4) was applied to thin gelatin films and incubated at 4 °C for 24 hr before fluorescent measurement. Graphs represent average fluorescent intensity, which is directly proportional to the extent of binding (mean ± SD values, n = 4). Two-way ANOVA with Šídák’s multiple comparison test was used for statistical analysis (* p ≤ 0.05; *** p ≤ 0.001; **** p ≤ 0.0001).

We also examined the triple-helical refolding rate of (N^xsce)₃-CHP at 150 μM in 1× PBS (pH 7.4). The t_1/2(refold) was found to be ~1,300 min, which was a dramatic increase from N^xsce-CHP5’s 31 min and (GPO)₉’s 20 min. Neutralization of Nsce’s negative charge did not alter the refolding rate (Figure 4B). Unlike T_m, the refolding speed of (N^xsce)₃-CHP was not a simple sum of each Nsce residue. The dramatic slowing of the refolding suggests that the rate-limiting step could be the nucleation of the triple helix, a process requiring three PP-II domains from each chain interacting with each other. If this is the case, the PP-II propensity of the CHP backbone could be considered the effective concentration of reactants in such trimeric nucleation reaction and the reduction in PP-II propensity can drastically slow the triple helix nucleation reaction. Faster refolding with N-terminal addition of the hydrophobic CF, which facilitates nucleation through hydrophobic interactions⁵³ (Figure 4C), suggests that the nucleation might be the rate-determining step for triple-helical folding of this peptide. Additional experiments are needed to gain full understanding of the reaction kinetics including reaction order and rate constants. Nonetheless, incorporation of three N^xsce residues, significantly delayed the homotrimeric refolding, which could be desirable for the application of CHPs as long as high dColl affinity (i.e., hetero-trimerization) is retained.

Gelatin binding of (N^xsce)₃-CHP and (N^xasp)₃-CHP

The designed (N^xsce)₃-CHP exhibited desirable homotrimeric folding properties as a CHP; it remained monomeric at a relatively low temperature (37 °C) and refolded into triple helices at a slow rate, even at 4 °C (Figure 4B). We next investigated this peptide’s capacity to hybridize to dColl strands using gelatin as a model for dColl. In this assay, the scrambled peptide (N^xsce)₃-CHP^Scr was used as a negative control, while (GPO)₉ was used as a positive control. Given the relatively low T_m of the (N^xsce)₃-CHP, we also tested the gelatin binding without heating. Although (N^xsce)₃-CHP exhibited greater binding to gelatin than the scrambled sequence, its “+Heat” group showed significantly lower gelatin binding compared to (GPO)₉ (p ≤ 0.0001; Figure 4D). Even when unheated samples were compared, the (N^xsce)₃-CHP’s binding was no more statistically significant than (GPO)₉’s (p = 0.08). The low gelatin binding suggested that Nsce may not only destabilize homotrimers but also impair heterotrimer formation with dColl. The poor binding performance of (N^xsce)₃-CHP compared to (GfPO)₉ (Figure S4) implies that the Nsce side chain causes not only steric clashes against the hydroxy group of the Hyp but also other major interactions that disfavor the formation of the triple helix with dColl strands.

We investigated whether the charge–charge repulsion could be the primary factor negatively affecting the dColl binding of (N^xsce)₃-CHP. Acidic conditions would not be suitable for such study as acids can not only neutralize the Nsce but also influence the charge state of the gelatin and can dissolve gelatin substrate during the binding study. Therefore, we tested (N^xspam)₃-CHP, which is structurally similar to (N^xsce)₃-CHP but neutral in 1× PBS (pH 7.4). (N^xspam)₃-CHP’s T_m was 37 °C, which is nearly identical to that of (N^xsce)₃-CHP in acidic condition, confirming that charges on Nsce had negligible effect on homotrimeric stability of the CHP. Gelatin binding assay showed that (N^xspam)₃-CHP also had low affinity to dColl, indicating that Nsce’s anionic charges did not contribute to low dColl binding (Figure 5A).

Figure 5.

Effect of electrostatic repulsions and inter-strand steric hindrance on dColl binding. (A) Nasp, Nasn, and Nspam residues were incorporated into the (GPO)₉ sequence at the same positions as those of (N^xsce)₃-CHP. The resulting (N^xasp)₃-CHP, (N^xasn)₃-CHP, and (N^xspam)₃-CHP were tested for gelatin binding at 4 °C, the results of which were compared between test CHPs and their scrambled sequences (negative controls). Each peptide was prepared at 10 μM in 1× PBS (pH 7.4). Graphs show the average measured fluorescent intensity, which is directly proportional to the binding extent [mean ± SD values, n = 4 for all sequences except (N^xasp)₃-CHP where n = 3 due to an elimination of one outlier]. (B) Refolding profiles of CF-(GPO)₉ and CF-(N^xasp)₃-CHP in 1× PBS (pH 7.4) at 4 °C were determined at low (30 μM) and high (150 μM) concentrations. (C) Gelatin binding of (GPO)₉ and (N^xasp)₃-CHP was compared over different incubation times. The peptides at 10 μM in 1× PBS (pH 7.4) were incubated with gelatin film at 4 °C before washing and fluorescent measurement at the indicated time points. Graphs represent average measured fluorescent intensity [mean ± SD values, n = 4 except for (N^xasp)₃-CHP at 6 and 24 h where n = 3 due to the removal of one outlier]. One-way ANOVA with Šídák’s multiple comparison test was used for statistical analysis in (A), and two-way ANOVA with Šídák’s test was used for statistical analysis in (C) (* p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001; **** p ≤ 0.0001).

Finally, upon recognizing that N^α–C^β branching destabilizes both CHP homo-trimerization and dColl binding—reducing the refolding rate only when accompanied by a similar loss in dColl affinity—we returned to the drawing board and synthesized (N^xasp)₃-CHP, which lacks the N^α–C^β methyl branch of the (N^xsce)₃-CHP. The T_m of (N^xasp)₃-CHP was 55 °C, which was higher than (N^xsce)₃-CHP’ T_m; however, we expected it to have high dColl binding while still offering slow refolding rate, which may benefit its use in histology and ex vivo tissue staining. The gelatin binding assay showed that (N^xasp)₃-CHP hybridized to dColl significantly more than (N^xsce)₃-CHP, (N^xspam)₃-CHP, and their corresponding scrambled sequences (Figure 5A). These findings showed that steric hindrance caused by the N^α–C^β branching indeed was responsible for low dColl binding as well as the low T_m.

Interestingly, in our assay, (N^xasp)₃-CHP exhibited the gelatin binding level even higher than those of (GPO)₉ and its neutral analogue, (N^xasn)₃-CHP (T_m = 63 °C), despite having a lower T_m. This contradicts the common understanding that CHPs with higher T_m exhibit stronger binding to dColl.^{34, 36} We suspected that this could be because CHPs with slower refolding rates can remain in the single-stranded form for a longer period of time during the gelatin binding assay, increasing their availability to hybridize with dColl. As we had expected, our refolding study (Figure 5B) showed that (N^xasp)₃-CHP refolded substantially slower than (GPO)₉ at both low (30 μM) and high (150 μM) concentrations. Unlike (GPO)₉, which showed similar refolding rate between the two concentrations, (N^xasp)₃-CHP’s refolding rate was markedly reduced by concentration reduction from 150 μM to 30 μM. It is known that (GPP)₁₀ nucleates triple helix formation approximately 1,000 times slower than (GPO)₁₀.⁵⁴ Because of such slow nucleation rate, the triple-helical folding of (GPP)₁₀ is highly concentration-dependent: t_1/2(refold) of less than 100 min at 1 mM, 500 min at 180 μM, and over 1,400 min at 60 μM. It seems that the triple helical folding of (N^xasp)₃-CHP is also concentration dependent, at least in the concentration range we tested (30–150 μM), suggesting that it also undergoes slow nucleation. This slower refolding rate can be attributed in part to the electrostatic repulsion, since ionized (N^xasp)₃-CHP refolded much more slowly than (N^xasn)₃-CHP at pH 7.4 (Figure S5). As a result of this slow triple-helical nucleation, a high concentration of monomeric species remained during the gelatin binding assay for (N^xasp)₃-CHP, whereas for (GPO)₉, rapid nucleation and trimerization reduced the effective concentration of monomeric species, resulting in lower overall gelatin binding compared to the (N^xasp)₃-CHP. To validate this, we performed a gelatin binding experiment at different incubation time (Figure 5C). Both peptides showed a similar level of gelatin binding at the first 4 h of incubation; however, at later time points (6, 8, and 24 h), (N^xasp)₃-CHP exhibited significantly higher binding than (GPO)₉. Despite its lower thermal triple-helical stability, (N^xasp)₃-CHP’s slower self-trimerization allowed more monomeric strands to remain available for dColl hybridization after the 4 h of incubation time. This kinetic delay and overall higher binding level for (N^xasp)₃-CHP could be useful for CHP binding conditions that require extended incubation time, such as ex vivo imaging of dense, collagen-rich tissues that require prolonged time for CHP diffusion and penetration. For CHPs with very low refolding rate of more than a day of t_1/2(refolding), such as (N^xdxn)₅-CHP or (N^xsce)₃-CHP, gelatin binding was also very low under similar gelatin binding conditions, because the dColl hybridization is also hindered by the slow refolding rate.¹¹

Histological staining of defective collagen in skin melanoma

Melanoma has been reported to alter the collagen remodeling activity of adipose tissue by inducing dedifferentiation of fat cells into fibroblast-like cells with increased expression of metalloproteinases and collagen synthesis.^{55, 56} Therefore, skin with melanoma often show an adipose layer expressing a high level of collagen.⁵⁷ Direct evidence of elevated collagen remodeling in melanoma and nearby tissues, such as adipose tissue, may improve our understanding of the crosstalk between cancer cells and non-cancer cells, as well as its impact on cancer progression. We utilized the newly developed CF-(N^xasp)₃-CHP to visualize dColl in mouse melanoma skin tissue sections, comparing its performance with the conventional CHP, CF-(GPO)₉ (a positive control), and their respective scrambled sequences (negative controls). Histological tissue staining was performed by applying a heat-activated CHP solution (20 μM) to the tissue sections, followed by incubation at 4 °C overnight, which is a common condition for tissue staining with commercially available CHPs. Hematoxylin and eosin (H&E) staining was also performed to visualize the tissue structure.

H&E staining of a tissue section revealed an adipose tissue adjacent to a tumor mass (Figure 6A). A 20x magnified image of this structure showed cells with morphological features typical of white adipocytes, including peripherally compressed nuclei and absence of cytoplasmic staining, indicating the presence of lipid droplets. There was a continuous layer of adipose tissue with a clearly delineated boundary intensely stained by eosin. Fluorescence imaging of sections stained with CF-(N^xasp)₃-CHP (Figure 6B) showed signals concentrated around and inside the layer-like adipose tissue, suggesting the presence of dColl. Interestingly, no fluorescence signals were observed within the tumor mass. Tissue sections stained with scrambled sequences exhibited almost no fluorescence. We also performed total collagen staining by intentionally denaturing the collagen by heating the tissue sections followed by CHP application (Figure 6B). The results demonstrated that CHP signals from the melanoma skin tissue sections (unheated) were from specific dColl hybridization rather than nonspecific binding. The results suggest the presence of abundant dColl within the layer-like adipose tissue. However, the origin of this dColl—whether from matrix metalloproteinase-mediated degradation of secreted collagen or from defective collagen production by adipose cells influenced by adjacent cancer cells—remains to be elucidated.

Figure 6.

Melanoma tissue staining by (N^xasp)₃-CHP and (GPO)₉. (A) H&E staining of mouse melanoma tissue sections at 4x magnification (left, scale bar = 1000 μm) and 20x magnification (right, scale bar = 200 μm) showed layer-like adipose tissue (labeled with AT) adjacent to the tumor mass (labeled with T). (B) Representative fluorescence images (4× magnification, scale bar = 1000 μm) of mouse melanoma skin tissue stained with 20 μM of (GPO)₉, (GPO)₉^scr, (N^xasp)₃-CHP, or (N^xasp)₃-CHP^scr at 4 °C overnight showed that the CHPs primarily localized in and around the layer-like adipose tissues. Total collagen (Total Coll) staining was obtained by intentionally heat-denaturing the tissue sections before the CHP staining. (C) Fluorescent signals from each group were quantified and compared. Graphs show the average measured fluorescent intensity (mean ± SD values, n = 3) One-way ANOVA with Šídák’s multiple comparison test was used for statistical analysis (* p ≤ 0.05).

We also found that the fluorescence intensity and pattern of tissue sections stained with CF-(GPO)₉ were comparable to that of CF-(N^xasp)₃-CHP (Figure 6C), suggesting that (N^xasp)₃-CHP is a promising alternative to (GPO)₉ for CHP hybridization. Its strong dColl binding, despite a significantly slower self-trimerization rate compared to (GPO)₉, may enable more flexible and user-friendly conditions for diagnostic or therapeutic delivery applications.

CONCLUSION

This study presents a new approach to modulate the folding behavior of CHPs by incorporating peptoid residues that introduce steric hindrance during triple-helical folding. We systematically investigated how Nsce substitution and its positional placement affect CHP’s triple helix stability, refolding kinetics, and dColl binding. Our data showed that Nsce, which has chiral N^α–C^β branch, reduced the thermal stability and refolding rate of the triple helix, primarily through steric hindrance instead of charge repulsion. However, multiple Nsce substitutions compromised the peptide’s ability to hybridize with dColl, suggesting that excessive destabilization from N^α–C^β branched peptoid residues also hinders non-self-triple helix formation. In contrast, the unbranched variant (N^xasp)₃-CHP showed a moderately slow self-refolding rate while maintaining high dColl binding. This new CHP outperformed the conventional (GPO)₉ in gelatin binding assays and effectively detected dColl in melanoma tissue histology. Importantly, the reduced self-trimerization rate of (N^xasp)₃-CHP enhanced the consistency and usability of CHPs without sacrificing specificity, potentially enabling broader adoption in diagnostic and therapeutic applications. Overall, our findings demonstrate that peptoid side chains allow modulation of triple-helical refolding properties, providing a design framework for advancing CHP applications.

Supplementary Material

ASSOCIATED CONTENT

Supporting Information

The Supporting Information contains list of chemicals, HPLC conditions, peptide characterization (m/z values, HPLC chromatograms, CD spectra, CD thermal melting profiles and their first derivatives), and additional refolding rate and gelatin binding data.