Structural Elucidation and Quantitative Characterization of BDDE-Crosslinked Recombinant Human Type III Collagen by NMR Spectroscopy: Crosslinking Sites, Product Distribution, and Enhanced Properties

1. Introduction

Collagen encapsulates a whole family of extracellular matrix proteins that are characterized by the amino acid repeating sequence [Gly-X-Y]n and a complex supramolecular structure, which leads to a dominant role in maintaining the biological and structural integrity of various tissues and organs, including skin, bone, cartilage, tendons, and ligaments.[1,2] Until now, 29 different types of collagen in the human body have been classified, with type I, III and IV being the most abundant and functionally significant.[3] Notably, type I and type III collagen are main ingredients of adult skin tissue, constituting 80-90% and 8-12% of the total content, respectively.[4] Type I collagen endows the fibril matrix with high tensile strength and rigidity, while type III collagen is often found in close association with type I to form hybrid fibrils to increase the tissues’ flexibility and distension.[5,6,7] Moreover, type III collagen can induce the transcription of growth factors for fibroblasts, thus playing an important role in dermal collagen fibrillogenesis and skin regeneration.[8,9] Especially, it has been reported that fetal skin has a higher proportion of type III collagen than that in childhood and adult skin, accounting for about 34-65% of the total collagen.[10,11,12] The decrease of type III collagen is directly associated with the aging of the skin, leading to the lack of elasticity and renewability.[13,14] Naturally, the supplemention of exogenous type III collagen becomes an important way to improve skin conditions, and type III collagen-based materials show a broad spectrum of applications in beauty cosmetics, tissue filling, cutaneous wound healing and other fields.[15,16]

To date, animal-derived type III collagen is the primary source of commercially available collagen, predominantly prepared via acid and alkaline hydrolysis and enzymatic digestion of animal connective tissue.[17] However, there are many inherent concerns related to the use of animal-derived materials in implantable devices, including elevated costs, low purity, potential epidemic transmission risks, and the tendency to elicit immunogenic reactions.[18,19] With the development of genetic engineering and biosynthesis technology, recombinant human type III collagen offers an alternative approach to traditional animal-derived collagen with enhanced hydrophilicity, low variability and immunogenicity.[20,21] Yan et al. prepared a novel recombinant collagen (rhCLA) by repeating the segment of Gly300-Asp329 of type III collagen for 18 times. RhCLA was successfully expressed in E.coli, and a systematic separation and purification procedure was established through alkaline precipitation, acid precipitation and cation exchange chromatography, yielding rhCLA purity surpassing 98% and an overall recovery of approximately 120 mg/L culture.[22] Moreover, Wang et al. demonstrated that recombinant humanized collagen type III (rhCol III), which was prepared by 16 tandem repetitions of a 30 amino acid triple-helix region from human collagen type III (hCOL3A1, Gly483-Pro512) showed superior biocompatibility and capability to facilitate the repair and regeneration of damaged skin.[23]

The breakthrough of recombinant humanizaed type III collagen marks a pivotal advancement in the field of collagen-based biomaterials. However, the potential drawbacks of collagen, including low mechanical strength and poor resistance to enzymatic degradation, restrict its application in contemporary aesthetic and medical fields to some degree.[24,25,26] Thus, it is paramount to develop innovative and efficient strategies to augment the durability and stability of collagen-based biomaterials. Among them, chemical cross-linking has attracted tremendous interest due to its simplicity and heightened cross-linking density.[27,28] Glutaraldehyde (GTA) cross-linking, which involves the formation of a Schiff base type compound between the aldehyde groups of GTA and the ε-amino groups of collagen, has been found to increase the viscosity and resistance to deformation and flow of the collagen dispersion.[29,30,31] Nonetheless, GTA is associated with potential cytotoxicity and calcification risks.[32,33] While the cross-linking strategy mediated by 1-(3-Dimethylaminopropyl)-3-carbodiimide hydrochloride/N-hydroxysuccinimide (EDC/NHS) enhances the tensile strength and mechanical elasticity of collagen hydrogels via the covalent amide bonds formation between the free amine and carboxylate groups, it also poses risks of impairing cell viability, migration and proliferation.[34,35,36] Similarly, genipin(GP)-crosslinked collagen strikes a balance between mechanical properties and biological functions through spontaneous reaction with the amine groups.[37,38] However, the accompanying generation of dark blue pigments during cross-linking limits its clinical biomedical applications.[39,40] In regard to the various concerns about the above common chemical crosslinkers, developing highly biocompatible and durable collagen fillers remains a considerable challenge in biomaterials research.

1,4-Butanediol diglycidyl ether (BDDE) has been widely employed as a biofunctional cross-linking agent to improve the structural stability and functional performance of collagen-based biomaterials[41,42,43]. The crosslinking efficiency and resulting material properties are critically dependent on the extent and nature of BDDE–collagen interactions. BDDE crosslinking can produce different structural outcomes, including double-arm crosslinks bridging two collagen strands, single-arm modifications with only one epoxide group reacted, and hydrolyzed/inactivated BDDE molecules[42]. Among these, only double-arm crosslinks contribute meaningfully to the mechanical stability and three-dimensional network structure of the hydrogel. Therefore, comprehensive characterization of both the total degree of modification and the proportion of effective crosslinks is essential for quality control and product optimization.

While NMR spectroscopy has been widely applied to characterize BDDE-crosslinked hyaluronic acid (HA) fillers[44,45], its application to BDDE-crosslinked collagen systems has not been reported. In HA systems, the basic structural unit is a simple repeating disaccharide composed of D-glucuronic acid and N-acetyl-D-glucosamine, and BDDE reacts predominantly with the well-defined hydroxyl groups on these two sugar residues through base-catalyzed ring-opening etherification reactions. This structural simplicity enables straightforward NMR interpretation: the N-acetyl methyl signal of GlcNAc at δ 1.90 ppm (¹H) or δ 21.9–22.6 ppm (¹³C) serves as a convenient internal reference for quantifying the degree of modification (MoD) and cross-linker ratio (CrR) via integration against BDDE methylene signals at δ 1.60 ppm[45]. By contrast, collagen is a heterotrimeric protein comprising over 20 types of amino acid residues with highly overlapping NMR signals, and the crosslinking reaction sites are far more diverse and condition-dependent. Consequently, there is no single well-established internal reference signal as compared to the N-acetyl methyl in HA, and the severe spectral congestion arising from the different amino acid types precludes the straightforward integration-based quantification methods, which have been well resolved for HA-BDDE systems.

All in all, the primary objective of this study was to establish and validate a comprehensive NMR spectroscopy methodology for the structural characterization of BDDE-crosslinked rhCol III. Specifically, we aimed to definitively identify the crosslinking sites in BDDE-crosslinked rhCol III using 1D and 2D NMR spectroscopy firstly and to develop quantitative NMR methods for determining key crosslinking parameters including MoD, mono-crosslink ratio (RMCR), and CrD, while evaluating the impact of crosslinking on physicochemical properties.

2. Materials and Methods

Materials

Recombinant humanizaed collagen type III (rhCol III, Mw=55.0 kDa) was purchased from Jiangsu JLand Biotech Co., Ltd., China. In brief, rhCol III was prepared by taking the gene sequence of human type III collagen as a template, selecting highly hydrophilic gene fragments, transforming the target gene into Pichia pastoris for fermentation via genetic recombination, and finally conducting multi-stage isolation and purification.

1,4-butanediol diglycidyl ether (BDDE) was sourced from Shanghai Macklin Biochemical Co., Ltd., China. Normal saline (0.9% NaCl) was supplied by Jiangsu Qinfen Pharmaceutical Co., Ltd., China. Dulbecco’s modified eagle medium (DMEM) and fetal bovine serum (FBS) were bought from Gibco, USA. The cell counting kit-8 (CCK-8) was acquired from Shanghai Yeasen Biotechnology Co., Ltd., China.

RhCol III fibers cross-linked with BDDE (rhCol III^C) were prepared and provided by Harbin Fuerjia Technology Co., Ltd., China. All samples were stored at 4°C prior to characterization studies.

BDDE residue content

Gas chromatography (GC) was used for the quantitative analysis of BDDE residues in rhCol III^C fibers.[46] Test solution preparation: Approximately 0.1g of the freeze-dried samples was dissolved completely in 3mL of distilled water. Afterwards, 1.0 mL of ethyl acetate was added to the above solution, followed by vortexing for 3min and centrifuging at 14000 rpm for 5min to collect the first supernatant. The remaining aqueous phase was processed following the same procedure to collect the second supernatant. Then the two supernatants were pooled and used as the final test solution. Reference solution preparation: BDDE was dissolved in ethyl acetate at a concentration of 2.0 μg/mL in accordance with the Chinese Pharmacopoeia.

The reference and test solutions were transferred to an autosampler for analysis by gas chromatography (Agilent 8890). HP-5 (Agilent) was used as a column, with a constant flow rate of 1.0 mL/min and the oven temperature ranging from 130 to 230 °C. Record the chromatogram and calculate the residual BDDE content X (μg/g) of the sample using the following formula:

X (μg/g) = C_S × (A₁ + A₂)/(A_S × W_i)

where CS represents the concentration of the BDDE reference solution (2.0 μg/mL); A1 and A2 are the BDDE peak areas of the first and second ethyl acetate supernatants from the test sample solution, respectively; AS is the peak area of BDDE in the reference solution; Wi denotes the weight of the test sample (g).

NMR Spectroscopy

All NMR experiments were performed on a Bruker AVANCE NEO 600 MHz spectrometer (Germany) equipped with a 5 mm triple-resonance probe operating at 298 K (25 °C) or a Bruker Ascend Evo 800 MHz spectrometer (Germany) equipped with a cryogenically cooled probe cooled by liquid helium. For the 600 MHz NMR spectrometer, the ¹H resonance frequency was 600.13 MHz and the ¹³C resonance frequency was 150.90 MHz. During quantitative ¹³C spectrum acquisition on the 800 MHz NMR spectrometer, the ¹³C resonance frequency was 201.24 MHz. For protein NMR analysis, lyophilized collagen samples (30 mg, native or crosslinked) were dissolved in 500 μL of D₂O at a concentration of approximately 60 mg/mL. Complete dissolution was achieved by gentle vortexing for 15 min at room temperature. Samples were transferred to 5 mm NMR tubes and allowed to equilibrate at 298 K for at least 10 min before data acquisition. All NMR spectra were processed and analyzed using MestReNova software (version 14.2.0, Mestrelab Research, Santiago de Compostela, Spain). Peak assignments were based on previously published NMR data for collagen[47], related peptides[48] and BDDE-modified polysaccharides[49].

Figure 1. Schematic representation of the rhCol III^C fibers preparation process and NMR analysis, including the followings: A. pH adjustment of the reaction system; B. Crosslinking in the thermostatic reactor; C. Ultrafiltration of the crude product to removed trace amount of BDDE; D. Lyophilization following freezing; E. The freeze-dried sample is dissolved in 500 μL of D₂O solution, and then transferred to a 5 mm NMR tube; F. Sealed tube is then inserted into an NMR instrument for data acquisition.

Figure 1. Schematic representation of the rhCol III^C fibers preparation process and NMR analysis, including the followings: A. pH adjustment of the reaction system; B. Crosslinking in the thermostatic reactor; C. Ultrafiltration of the crude product to removed trace amount of BDDE; D. Lyophilization following freezing; E. The freeze-dried sample is dissolved in 500 μL of D₂O solution, and then transferred to a 5 mm NMR tube; F. Sealed tube is then inserted into an NMR instrument for data acquisition.

Fourier transforms-infrared (FT-IR)

RhCol III and rhCol III^C fibers were characterized by attenuated total reflection Fourier transform infrared spectroscopy (Nicolet 6700, Thermo Electron Corporation, USA). FT-IR spectra were recorded between 4000 and 400 cm^-1 with a resolution of 4 cm^-1. The data were processed using OMNIC software.

Scanning electron microscope (SEM)

The SEM (Hitachi S-800, Japan) was utilized to observe the morphological characteristics of rhCol III^C fibers, operating at an accelerating voltage of 5.0 kV. Prior to imaging, the freeze-dried samples were carefully mounted onto copper stubs, positioned in an ion sputterer and subjected to gold spraying.

Viscosity

The viscosity testing of samples was measured using a rheometer (MCR92, Anton Paar). RhCol III and rhCol III^C fibers were sequentially loaded onto the center of a 25mm diameter parallel plate, and subjected to rotational mode testing to evaluate the viscosity variations over shear rates ranging from 0.001 to 1000 s^-1.

Differential scanning calorimetry (DSC)

The thermal stability of rhCol III^C fibers was investigated using a DSC 3500 instrument (NETZSCH, Germany) over the temperature range of 20 °C to 200 °C. The scan rate was set at 5 °C min^-1.

Cell proliferation

The cytotoxicity of the rhCol III^C fibers was assessed using the cell proliferation experiment.[50] Initially, 100 μL sterile sample solutions at concentrations of 2.5, 5.0, and 10.0 mg/mL, as well as the national standard (porcine Col I) at 0.5 mg/mL, were added into wells of a 96-well plate and incubated at 37.0 °C for 24h. Each experimental group was represented by 3 replicate wells. After blocking with 1% BSA and rinsing with PBS, human skin fibroblasts (HSF) were seeded into individual wells at a density of 0.5×104 cells per well, and cultured at 37 °C in a humidified incubator supplemented with 5% CO2 for 72h. Following this, 10 μL of CCK-8 solution was introduced to each well and incubated at 37 °C for an additional 2h. The absorbance of the solution was measured at 450 nm using a Tecan Infinite E Plex multifunctional microplate reader (Tecan, Switzerland). The cell viability was assessed as:

$$\mathrm{Cell\; viability}={\displaystyle \frac{{\mathrm{OD}}_{\mathrm{s}}-{\mathrm{OD}}_{\mathrm{b}}}{{\mathrm{OD}}_{\mathrm{c}}-{\mathrm{OD}}_{\mathrm{b}}}}\times 100\%$$

where ODs, ODc and ODb denote the absorbance values of sample, control and blank groups, respectively.

Subcutaneous implantation

In vivo biocompatibility of the rhCol III^C fibers was evaluated via subcutaneous implantation assay.[51] The animal study was carried out in accordance with the requirements of GB/T 16886.6-2022, and with ethics approval from Laboratory Animal Management Committee of Sichuan Province (SYXK 2023-0017). Before treatment, six SD rats (203.1-220.0 g, Charles River) were acclimatized for more than 5 days under standard housing and feeding conditions in compliance with laboratory animal care and welfare guidelines. All surgery procedures were performed under sterile conditions. The Rats were anaesthetized by intraperitoneal injection of pentobarbital sodium (40 mg/kg body weight), and the operative areas on both sides of the rat spine were exposed and conventionally prepared. Along the spine’s midline of each rat, the left area approximately 2cm away from the spine was taken as the experimental group, and the symmetrical right area was seen as the control group correspondingly. The control group received normal saline injections at 4 separate points, while the experimental group received rhCol III^C injections in the same way. RhCol III^C fibers were dispersed in normal saline at a concentration of 5 mg/mL for experimental group preparation. 0.5 mL of the corresponding solutions was injected subcutaneously at each implantation point. After the surgery, the rats were kept in individual cages for free movement at a controlled temperature of 20-25 °C, a relative humidity of 44-68% and on a 12 h light/dark cycle. Meanwhile, standard chow and water were available for freely intake.

At 1 and 4 weeks post-operation, the rats were euthanized and tissue specimens from the injection sites were collected for further study. Following the fixation in 10% neutral-buffered formalin, tissues were further dehydrated through a graded ethanol series, embedded in paraffin, and sectioned into 5-micron slices. Theses sections were then stained with hematoxylin and eosin (H&E) for histopathological analysis. Finally, histological images were captured using an upright fluorescence microscope.

Statistical analysis

The data were expressed as mean ± standard deviation (SD) and statistically analyzed by One-way ANOVA. Statistical significance was defined as *p<0.05.

3. Results

Cross-Linked rhCol III Fibers Design

RhCol III applied in this study was encoded by a gene fragment within the full-length sequence of human type III collagen (Gly519-Ala617), and further expressed by Pichia pastoris with 6 tandem repeats. The theoretical molecular weight of rhCol III was 55,027 Da based on the amino acid sequence. Previous studies have demonstrated that rhCol III possesses excellent hydrophilicity and biocompatibility; however, its susceptibility to collagenase-mediated degradation restricts its practical applications. Notably, benefiting from its high crosslinking efficiency and low cytotoxicity, BDDE has been widely used as a chemical crosslinker for collagen to enhance the degradation resistance, particularly for animal-derived collagen. At present, very few studies focus on the modification of recombinant human collagen, let alone rhCol III. In this study, cross-linked rhCol III fibers (rhCol III^C) were successfully fabricated through reaction between the amine groups of the collagen and the epoxide groups of BDDE, which follows the widely recognized mechanism (Figure 2).[42]

Crosslinking Analysis

As previously reported, the excessive amount of BDDE in crosslinked products can be toxic and have harmful effects on human health[52]. Accordingly, we quantified the BDDE levels in our rhCol III^C fibers, based on a standard testing method called gas chromatography. The results showed that the residual BDDE concentration in rhCol III^C was 1.7 μg/g, which complies with the industrial specification limit of ≤ 2.0 μg/g. This negligible BDDE contamination indicates rhCol III^C fibers are safe biomaterials.

Chemical Characterization

The differences in chemical structures between rhCol III and rhCol III^C determined by FTIR are shown in Figure 3(A). The FTIR spectra of rhCol III and rhCol III^C exhibited identical characteristic vibration peaks, including amide A (3287 cm-1, N-H stretching), amide B (3070 cm-1, C-H stretching), amide I (1621 cm-1, C=O stretching), amide II (1529 cm-1, N-H bending) and amide III (1228 cm-1, C-N stretching), which were typical absorbance peaks of collagen as reported in previous literature.[56] In the FTIR spectra of cross-linked rhCol III^C, absorption band shifts at 968 cm-1 and 864 cm-1 were detected relative to native rhCol III, indicative of structural alterations induced by BDDE treatment. When combined with reduced free-amino content and supplementary characterization data, these observations support successful crosslinking of rhCol III.

As depicted in Figure 3(B), SEM was employed to further evaluate the morphological changes of rhCol III^C. All images were obtained at identical magnification conditions (×500, ×1000, SEI and 5.0 kV). RhCol III displayed a conventional lamellar structure (Figure 3(A,C)), while rhCol III^C revealed an intricate porous network structure (Figure 3(B,D)). Meanwhile, Image-Pro Plus 6.0 analysis showed that the porosity of rhCol III^C fibers was as high as 36.03%-41.46%. These findings suggested that BDDE crosslinking optimized the arrangement of collagen fibers, which provided favorable mechanical support for elevating its anti-enzymatic degradation performance.

Figure 3. Chemical characterization of rhCol III^C. (A) FTIR spectra, (B) SEM images, (C) viscosity and (D) DSC curves of rhCol III and rhCol III^C.

Figure 3. Chemical characterization of rhCol III^C. (A) FTIR spectra, (B) SEM images, (C) viscosity and (D) DSC curves of rhCol III and rhCol III^C.

Additionally, the impact of BDDE cross-linking on mechanical properties of rhCol III^C fibers was evaluated through rheological viscosity measurements (Figure 3(C)). At a low shear rate of 0.1s^-1, the apparent viscosity of rhCol III was 122.82 mPa·s, while the viscosity of rhCol III^C was 12148 mPa·s. BDDE-crosslinked rhCol III exhibited a marked increase in apparent viscosity, reaching approximately 100-fold value of the native material, which indicated that rhCol III^C has superior mechanical strength and structural stability.

DSC measurements were further conducted to assess the thermal stability of rhCol III and rhCol III^C, as shown in Figure 3(D). The thermal denaturation temperature (Tm), marked as a peak on the DSC curves, represents the transition point where the collagen undergoes structural loosening due to the cleavage of stabilizing hydrogen and covalent bonds.[57] In our study, the Tm value of rhCol III was 55.94 °C, whereas that of rhCol III^C was 68.18 °C, indicating BDDE crosslinking improved rhCol III stabilization and contributed to maintaining its structural integrity under physiological conditions.

The ¹H NMR spectra of native rhCol III and BDDE-crosslinked rhCol III^C are compared in Figure 7, Figures S1 and S2. Native rhCol III exhibited characteristic collagen proton signals, including the α-CH protons at 3.5–4.7 ppm, side chain CH₂ and CH₃ protons at 1.3–2.5 ppm, and the well-resolved lysine ε-CH₂ signal (HK5) at approximately δ 2.95–3.05 ppm, which appears as a triplet due to coupling with the adjacent HK4 methylene protons. Following BDDE crosslinking, several significant spectral changes were observed after comparison of the two spectra. The first one is the decreased intensity of lysine HK5 signal. The triplet signal at δ 2.95–3.05 ppm (HK5), corresponding to the ε-methylene protons of unmodified lysine residues, was significantly reduced in intensity after crosslinking. This provides direct evidence that lysine residues are involved in the crosslinking reaction. The second one is the appearance of new BDDE-derived signals. As shown in Figure S2, new signals appeared in the δ 1.5–1.7 ppm region corresponding to the central methylene protons of the BDDE butanediol spacer chain (H5′/H5″), confirming covalent incorporation of BDDE into the collagen structure. The third one is the outcome of new signals in the δ 3.2–3.6 ppm region. Following crosslinking, new proton signals emerged in the δ 3.2–3.6 ppm region, which can be assigned to BDDE methylene protons adjacent to ether linkages (Figure 6, H3′/H3″) and to the modified lysine ε-methylene protons shifted downfield due to the change from primary amine to secondary amine after reaction with BDDE. These new signals partially overlap with serine side chain protons and other collagen signals but are clearly discernible in difference spectra.

The ¹³C NMR spectra provided more structural information and enabled clearer differentiation between different BDDE product types (Figures S3 and S4). Key observations from the ¹³C NMR spectra of rhCol III^C include the disappearance of unmodified lysine Cε signal (CK5) and the presence of BDDE backbone carbon signals. In the native collagen ¹³C NMR spectrum, a clear signal was observed at δ 40.8 ppm (Figure 8), assigned to the ε-carbon of unmodified lysine residues (CK5). In the crosslinked sample, this signal was significantly reduced in intensity, consistent with the ¹H NMR findings and confirming that lysine residues are the primary sites of modification. Additionally, a prominent signal at δ 26.89 ppm was assigned to the central methylene carbons of the BDDE spacer chain (C5′/C5″), which are equivalent in both mono-linked and double-crosslinked BDDE molecules (Figure 9). This signal serves as a measure of total BDDE incorporation.

Figure 6. molecular structural diagram of BDDE-crosslinked rhCol III^C (A and B stand for double-crosslinked and mono-linked BDDE molecules, respectively).

Figure 6. molecular structural diagram of BDDE-crosslinked rhCol III^C (A and B stand for double-crosslinked and mono-linked BDDE molecules, respectively).

For the crosslinked sample, a distinct signal at δ 64.37 ppm was observed and it was assigned to the terminal hydroxymethyl carbon (C10″) of mono-crosslinked BDDE, where the second epoxide group has undergone hydrolysis to form a primary alcohol. This signal is only present in mono-linked BDDE modifications and serves as a quantitative marker for ineffective crosslinks. The assignment was confirmed by comparison with the ¹³C NMR spectrum of glycerol, where the primary hydroxymethyl carbons resonate at δ 64.29 ppm, and with BDDE hydrolysis product (BDPE) simulation data predicting a chemical shift of approximately δ 63.8 ppm for the corresponding carbon (Figure 10).

Figure 7. ¹H NMR spectrum of BDDE, rhCol III and rhCol III^C.

Figure 7. ¹H NMR spectrum of BDDE, rhCol III and rhCol III^C.

Figure 8. Expanded ¹³C NMR spectrum of BDDE, rhCol III and rhCol III^C.

Figure 8. Expanded ¹³C NMR spectrum of BDDE, rhCol III and rhCol III^C.

Figure 9. Expanded ¹³C NMR spectrum of rhCol III^c (, δ 24-35 ppm).

Figure 9. Expanded ¹³C NMR spectrum of rhCol III^c (, δ 24-35 ppm).

Figure 10. Expanded ¹³C NMR spectrum of rhCol III^c (, δ 62-73 ppm).

Figure 10. Expanded ¹³C NMR spectrum of rhCol III^c (, δ 62-73 ppm).

The ¹H-¹H COSY spectrum of native rhCol III showed a characteristic strong coupling signal between HK5 (δ 3.01 ppm) and HK4 (δ 1.66 ppm) of lysine residues (Figure S12). In the crosslinked sample, the intensity of this HK5–HK4 cross peak was significantly reduced, consistent with the modification of lysine ε-amino groups. Additionally, new cross peaks appeared in the COSY spectrum of the crosslinked sample corresponding to the BDDE spin system: coupling between the spacer methylene protons (δ 1.65–1.70 ppm) and the adjacent ether-linked methylene protons (δ 3.55–3.65 ppm), confirming the intact BDDE alkyl chain structure in crosslinked products.

The HSQC spectrum of native collagen revealed a characteristic cross peak at δ_H 3.02 ppm / δ_C 40.8 ppm, corresponding to the lysine ε-CH₂ group (HK5/CK5). In the crosslinked sample, this cross peak was significantly reduced in intensity, while new cross peaks appeared at δ_H 3.2–3.4 ppm / δ_C 50–55 ppm, consistent with the downfield shift of lysine Cε after substitution by the BDDE moiety through secondary amine formation. The magnitude of this shift (approximately +10–15 ppm in ¹³C) is characteristic of nitrogen substitution on a methylene carbon and provides definitive confirmation that the crosslinking occurs at the lysine ε-amino position. The HSQC spectrum also confirmed the presence of BDDE-derived cross peaks, including the spacer CH₂ at δ_H 1.65–1.70 ppm / δ_C 26.89 ppm and the ether-linked CH₂ at δ_H 3.5–3.6 ppm / δ_C 70–72 ppm.

An important issue in BDDE crosslinking of proteins is whether serine hydroxyl groups participate in the reaction, particularly under alkaline conditions. The NMR data provide strong evidence that serine hydroxyl groups do not react with BDDE under the employed conditions. The serine β-methylene proton signal (HS2, δ 3.88 ppm) showed no significant change in chemical shift or peak shape before and after crosslinking. In the H-H COSY spectrum, the coupling signal between serine HS1 and HS2 (blue box in spectra) showed no reduction in intensity compared to the raw collagen spectrum. Using ethanol and diethyl ether as model compounds, we confirmed that etherification of hydroxyl groups results in an upfield shift of the adjacent methylene protons in the ¹H NMR spectrum. The absence of such a shift for serine side chain protons in the crosslinked collagen samples provides additional evidence that serine hydroxyl groups remain unmodified. The pKa of serine hydroxyl groups is approximately 13–14, meaning that under the reaction conditions (pH 10.0–8.4), serine hydroxyl groups exist almost entirely in the protonated form and are not sufficiently nucleophilic to attack the epoxide rings of BDDE. In contrast, the pKa of lysine ε-amino groups is approximately 10.5, so a significant fraction exists in the deprotonated, nucleophilic form at the reaction pH.

Quantitative Analysis: MoD, RMCR, and CrD

Degree of Modification (MoD)

In the quantitative analysis, the degree of modification (MoD) represents the percentage of total lysine residues modified by BDDE (including both double-crosslinking and single-crosslinking modifications). MoD was determined by ¹H NMR based on the relative decrease in intensity of the lysine ε-methylene proton signal (HK5, δ 2.95–3.05 ppm) after crosslinking, using the alanine methyl proton signal (HA2, δ 1.3–1.6 ppm) as an internal reference that does not participate in the crosslinking reaction. The formula was performed as follows:

R₀ = I₀(HK5) / I₀(HA2) (native rhCol III)

Rₓ = Iₓ(HK5) / Iₓ(HA2) (crosslinked sample)

MoD (%) = [(R₀ − Rₓ) / R₀] × 100 (I)

where I₀(HK5) and Iₓ(HK5) are the integrated peak areas of lysine ε-methylene protons in uncrosslinked and crosslinked samples, respectively, and I₀(HA2) and Iₓ(HA2) are the integrated peak areas of alanine methyl protons in the corresponding samples.

The MoD values for the standard reaction condition were determined by ¹H NMR using the relative intensity decrease of the lysine HK5 signal referenced to alanine HA2. The results are summarized as follows:

Native rhCol III: R₀ = I₀(HK5) / I₀(HA2) = 36.00 / 24.67 = 1.46

BDDE-crosslinked rhCol IIIc: Rₓ = Iₓ(HK5) / Iₓ(HA2) = 3.26 / 24.00 = 0.136

MoD = [(1.46 − 0.136) / 1.46] × 100% = 90.7%

This high degree of modification indicates that the majority of lysine residues in rhCol III have reacted with BDDE under the optimized reaction conditions. The 90.7% MoD is consistent with the accessibility of most lysine residues on the collagen surface and the high reactivity of BDDE under alkaline conditions.

Ratio of Mono-Crosslinks (RMCR)

The RMCR represents the proportion of BDDE molecules that form mono-crosslink modifications versus double-crosslinks. This parameter was determined by quantitative ¹³C NMR based on the characteristic signal of the terminal hydroxymethyl carbon (C10″) of mono-crosslinked BDDE at δ 64.37 ppm (Figure 10), relative to the BDDE backbone methylene carbon signal at δ 26.89 ppm (C5′/C6″) (Figure 9).

RMCR (%) = [I(δ 64.37) / (I(δ 26.89) / 2)] × 100 (II)

where I(δ 64.37) is the integrated peak area of the terminal hydroxymethyl carbon of mono-linked BDDE (C10″) , and I(δ 26.89) is the integrated peak area of the BDDE backbone methylene carbons (C5′ and C6′ in mono-crosslinked BDDE, or C5″ and C6″ in double-crosslinked BDDE as two equivalent carbons). The division by 2 accounts for the two equivalent backbone carbons per BDDE molecule.

Accordingly, the proportion of effective double-arm crosslinks (CrR) is given by the following formula:

CrR = 1 − RMCR/100 (III)

The RMCR value for BDDE-crosslinked rhCol III^C was determined by corresponding quantitative ¹³C NMR analysis based on the ratio of the terminal hydroxymethyl carbon signal (δ 64.37 ppm) to the BDDE backbone carbon signal (δ 26.89 ppm):

I(δ 64.37) = 0.39 (relative integration)

I(δ 26.89) = 1.00 (relative integration, normalized)

RMCR = [0.39 / (1.00 / 2)] × 100% = 0.39 / 0.50 × 100% = 78%

This result indicates that 78% of BDDE molecules incorporated into collagen form single-arm (pendant) modifications, while only 22% form effective double-arm interchain crosslinks. The predominance of mono-arm products is consistent with the kinetic competition between the second crosslinking reaction and hydrolytic inactivation of the second epoxide group. Given the high initial pH (10.03) and the relatively high BDDE: lysine ratio, base-catalyzed hydrolysis of the second epoxide group competes effectively with intermolecular crosslink formation, leading to a high proportion of single-arm modifications.

Degree of Crosslinking (CrD)

The overall degree of crosslinking represents the number of lysine residues per collagen molecule that participate in effective double-crosslinks. The calculation accounts for the stoichiometry that each double-crosslink involves two lysine residues (one from each collagen chain), while each mono-arm crosslink involves only one lysine residue. Based on the established formula for BDDE crosslinking of collagen[42], the CrD is calculated according to the following formula:

CrD (Lys residues per molecule) = N × MoD × (1 − RMCR/100) / (2 − RMCR/100) (IV)

where N is the total number of lysine residues per rhCol III molecule (N = 18), MoD is the degree of modification expressed as a fraction, and RMCR is the mono-arm crosslink ratio expressed as a value in percentage. This formula is derived from the stoichiometric relationship that total modified lysines = mono-crosslinked BDDE × 1 + double-crosslinked BDDE × 2, and total BDDE = mono-crosslinked BDDE + double-crosslinked BDDE.

The CrD value for BDDE-crosslinked rhCol III^C was calculated using the well-established formula (IV) with 18 lysine residues per molecule (N = 18), MoD = 90.7%, and RMCR = 78%:CrD = 18 × 0.907 × (1 − 0.78) / (2 − 0.78) = 18 × 0.907 × 0.22 / 1.22 = 18 × 0.1803 / 1.22 = 18 × 0.1635 = 2.94

This means that on average, approximately 2.94 out of 18 lysine residues per rhCol III molecule participate in effective double-arm crosslinks, corresponding to approximately 1.47 crosslinking bridges per molecule. This level of crosslinking is sufficient to form a stable three-dimensional network while preserving the biological activity of the collagen matrix.

Cell Proliferation In Vitro

Cell proliferation assay was conducted to assess the cellular compatibility of rhCol III^C. Incubation of HSF cells with extracts of rhCol III^C fibers and porcine Col I for 72h resulted in significant differences. As illustrated in Figure 11, the cell viability of the national standard group was 106%. In contrast, treatment with rhCol III^C at concentrations of 2.5, 5.0, and 10.0 mg/mL displayed higher cell viabilities of 107%, 113% and 110%, respectively. The results demonstrated that rhCol III^C possessed superior biocompatibility and moderate cell proliferation-promoting ability, supporting its prospective biomedical applications.

In Vivo Performance

Next, in vivo subcutaneous implantation experiments in SD rats were undertaken to further evaluate the biocompatibility of rhCol III^C. The back of each rat was divided into two groups along the midline of the spine: the experimental group on the left, and the control group on the right (Figure 12(A)). The experimental and control groups received injections of 0.5 mL of 5 mg/mL rhCol III^C and normal saline, respectively. The skin status at rats implantation sites was monitored continuously. As presented in Figure 12(B), no overt edema, erythema or vascular proliferation symptoms appeared at all injection sites within 4 weeks postoperatively, which implied the favorable histocompatibility of rhCol III^C.

Moreover, histological inflammatory responses to injection of the rhCol III^C at 1 and 4 weeks post-surgery was analyzed using H&E staining. It is clear from the Figure 12(C) that the surrounding subcutaneous tissues remained normal at 1 week after injection of normal saline and rhCol III^C. There were thimbleful polymorphonuclear leukocytes, macrophages and lymphocytes distributed in the subcutaneous tissues of the two groups. At the 4th week post-injection, the inflammatory response was markedly alleviated in all SD rats. In addition, there was obvious formation of granulation tissue and epithelial layers around the implantation sites, comparable to native skin. These findings suggested that BDDE cross-linking did not induce the inflammation response and endowed the rhCol III^C with excellent histocompatibility.

Figure 12. In vivo assessment of rhCol III^C. (A) Schematic illustration of the experimental procedure. (B) Gross appearance at 1 week and 4 weeks after surgery. (C) H&E staining of animal samples at 1 week and 4 weeks after surgery.

Figure 12. In vivo assessment of rhCol III^C. (A) Schematic illustration of the experimental procedure. (B) Gross appearance at 1 week and 4 weeks after surgery. (C) H&E staining of animal samples at 1 week and 4 weeks after surgery.

4. Discussion

The NMR results presented in this study provide clear evidence that BDDE crosslinks rhCol III primarily through covalent modification of lysine ε-amino groups, consistent with the established reaction chemistry of bifunctional epoxides with proteins[42,43]. Under the alkaline reaction conditions employed, the ε-amino groups of lysine residues (pKa ~10.5) are partially deprotonated and act as nucleophiles, attacking the electrophilic epoxide rings of BDDE to form stable secondary amine linkages via ring-opening reactions. The appearance of new NMR signals corresponding to the BDDE alkyl backbone in both ¹H and ¹³C spectra, combined with the absence of unreacted BDDE signals after thorough dialysis, confirms the covalent nature of the BDDE–collagen conjugation. The significant reduction in intensity of the lysine ε-CH₂ signals (HK5 at δ 2.85–3.0 ppm in ¹H NMR and CK5 at δ 40.8 ppm in ¹³C NMR), along with the appearance of new downfield-shifted signals corresponding to modified lysine residues, provides definitive evidence for lysine as the primary crosslinking site.

The selectivity for lysine amino groups over serine hydroxyl groups can be rationalized on the basis of both pKa considerations and inherent nucleophilicity. The pKa of lysine ε-amino groups (~10.5) means that at pH 9–10, a significant fraction (~10–30%) exists in the deprotonated, nucleophilic form. In contrast, serine hydroxyl groups have a much higher pKa (~13–14), so they are essentially fully protonated and non-nucleophilic at the reaction pH. Additionally, amine nitrogen is inherently more nucleophilic than hydroxyl oxygen towards electrophilic centers such as epoxide rings. These factors combine to make lysine ε-amino groups the reaction sites for BDDE crosslinking in rhCol III under alkaline conditions.

Traditional quality control methods for crosslinked collagen, such as the TNBS assay for free amine determination or mechanical testing for gel strength, provide only indirect or bulk measurements of crosslinking status. The TNBS assay measures the total number of free amino groups but cannot distinguish between double-crosslinking and single-crosslinking modifications[42]. Mechanical testing provides functional information but is influenced by multiple factors including water content, collagen concentration, and network topology, making it difficult to isolate the specific contribution of crosslinking chemistry.

Benefiting from multiple superiorities including low immunogenicity, higher water solubility, simple component structure and controllable production preparation process, rhCol III emerges as a promising alternative of animal-derived type III collagen, showing broad application prospects in cosmetics, regenerative medicine, wound dressings, and even artificial organs. However, the issue of intrinsically rapid degradation is an inherent bottleneck restricting the development of rhCol III. Herein, BDDE- a clinically prevalent chemical crosslinker- was utilized to strengthen intermolecular crosslinking of rhCol III and build a more compacted microstructure. Based on the modification, rhCol III^C was expected to provide enhanced structural stability, and mechanical integrity. Indeed, rhCol III^C was still in the early stages of development. Subsequent research will concentrate on not only the product performance both in vitro and in vivo, but also more comprehensive and in-depth structural characterization. In the future, we believe rhCol III^C with outstanding performances can serve as various physical forms, such as fibers, membranes, sponges or hydrogels tailored to specific requirements for extensive biomedicine applications.

5. Conclusions

In summary, a novel crosslinked recombinant humanized collagen fiber (rhCol III^C) was successfully fabricated by BDDE chemical crosslinking. In addition, a comprehensive NMR spectroscopy methodology for the structural characterization of BDDE-crosslinked recombinant human type III collagen was established and also validated. By combining 1D ¹H and ¹³C NMR with 2D COSY and HSQC experiments, we demonstrated that BDDE crosslinks rhCol III exclusively through the ε-amino groups of lysine residue. Meanwhile, rhCol III^C demonstrated prominent biocompatibility, promoting the proliferation of HSF cells. The in vivo subcutaneous implantation study on SD rats, with observation durations of 1 and 4 weeks, further verified that rhCol III^C was safe with favorable histocompatibility. Physicochemical characterization confirmed that BDDE crosslinking significantly enhances the structural stability, and the thermal stability of rhCol III.