Following the physical crosslinking protocols adapted from our earlier work^[23], physical crosslinking of the MDP-DOPA network was achieved through repeated freeze–thaw thermal cycling. Briefly, bovine casein (5% w/v; C5890, Sigma-Aldrich, St. Louis, MO, USA) was dissolved in 1 M NaOH, whereas PVA (4% w/v) was solubilized in distilled water (DW) at 90 °C. Before blending, the hot PVA solution was cooled to RT to prevent the thermal denaturation of the protein. The two solutions were mixed thoroughly overnight under continuous stirring at room temperature. The mixed solutions were poured into various casting molds (flat and conduit forms). For in vitro bioevaluations and cellular assays, planar substrates were fabricated using a mold consisting of dual parallel polystyrene sheets separated by a 1.5-mm spacer. The conduit mold consisted of an outer conduit and an inner glass tube to analyze its mechanical properties and conduct in vivo animal studies. The final hydrated conduit had an inner diameter of 1.11 ± 0.08 mm and a wall thickness of 0.80 ± 0.04 mm [mean ± standard deviation (SD), n = 5]. After casting the mixed solution, the molds underwent freeze–thaw cycles. To induce hydrogel crosslinking, the loaded molds underwent five iterative freeze–thaw steps, comprising deep-freezing at -80 °C for 4 h and subsequent ambient thawing for 2 h. The fabricated hydrogel was then washed to remove undesired reactants and byproducts. MDP was modified with DOPA (D9628; Sigma-Aldrich, St. Louis, MO, USA) to immobilize biomolecules and improve cell adhesion. Surface modification was performed by incubating the MDP hydrogels in a 10 mM Tris–HCl solution (pH 8.5; T1503, Sigma-Aldrich, St. Louis, MO, USA) supplemented with 2 mg·mL^-1 DOPA for 16 h at RT. Under these alkaline conditions, the DOPA monomers underwent self-oxidation and polymerization to form a stable poly-DOPA-like coating on the hydrogel surface. After coating, the specimens were thoroughly sonicated in DW to dislodge and remove loosely bound poly-DOPA aggregates or nonspecifically adsorbed particulates. This rigorous cleaning step ensured that only firmly immobilized and chemically cross-linked polymeric layers remained on the substrate.

The external and internal surfaces of the MDP and MDP-DOPA NGCs were mounted on an aluminum stub. To prepare for the scanning electron microscopy (SEM) observations, the NGC samples were sputter-coated with platinum using an automated system (Sputter Coater 108 Auto, Cressington Scientific Instruments, Watford, UK) under vacuum conditions, operating at a 15 mA plasma current for 2 min. The coated samples were characterized by SEM (MERLIN Gemini 2, Carl Zeiss, Oberkochen, Germany; facility equipment registration no. NFEC-2021-05-270500) at the Center for Biomedical Engineering Core Facility (Dankook University, Cheonan, Republic of Korea), at an accelerating voltage of 10 kV.

Hydrogel permeability was evaluated by monitoring the diffusion of toluidine blue (TB; 100 μM; T3260, Sigma-Aldrich, St. Louis, MO, USA) across planar hydrogel disks (n = 5) with a multimode microplate reader (Spark 20M, Tecan, Männedorf, Switzerland). We prepared flat disc-type hydrogels (thickness = 1.5 mm) as previously described, and then fixed the hydrogels with double-sided tape, connecting two glass vials, a bottle containing TB solution, and a bottle containing DW to start the diffusion process. The diffused solution was sampled (1 mL) from the DW bottle at intervals of 30, 60, 120, and 180 min and transferred to a 96-well plate. Diffusion rates were quantified by measuring the optical density of diffused TB at 660 nm. The tensile properties and compressive moduli were measured using a tensile strength machine (Model 5966, Instron, Norwood, MA, USA). For the lateral compressive modulus testing, cylindrical specimens (10 mm in length, n = 5 per group) of each NGC were pressed at a crosshead speed of 1 mm/min until 60% deformation was achieved. Probe displacement and applied force were recorded. Compressive stress (σ) was determined by dividing the applied force by the cross-sectional contact area of the conduit specimen, and compressive strain (ε) was calculated as the displacement normalized to the initial outer diameter (2.71 ± 0.12 mm). The compressive modulus (kPa) was obtained from the slope of the linear region in the stress-strain curve. For the tensile properties testing, flat rectangular specimens (15 × 30 mm², thickness = 1.5 mm; n = 5 per group) were evaluated at a continuous crosshead speed (strain rate) of 1 mm·min^-1 with a load cell force of 100 N.

To evaluate enzymatic degradation profiles, MDP-DOPA hydrogel disks (20 mm in diameter, n = 5) were placed in 5 mL of a 1 mg·mL^-1 collagenase B solution (11088815001, Roche Diagnostics, Basel, Switzerland) prepared in phosphate-buffered saline (PBS, pH 7.4; P3813, Sigma-Aldrich, St. Louis, MO, USA) and maintained at 37 °C with gentle agitation at 30 rpm^[24,25]. Unmodified MDP hydrogel disks were processed alongside MDP-DOPA samples as comparative controls under identical conditions, with n = 5 per experimental group. Each hydrogel was weighed at different time points (1, 2, 4, 6, and 8 weeks) after washing with DW and complete lyophilization to remove residual water. Day 1 was included solely to show the initial appearance of the samples after fabrication and before initiation of the degradation test. The weight loss at each time point (t) was calculated using the following equation: weight loss (%) = (W_i - W_t)/W_i × 100, where W_i represents the initial weight of the hydrogel at day 0, and W_t represents the weight of the hydrogel at each time point.

RAW 264.7 macrophages (RRID: CVCL_0493; ATCC, Manassas, VA, USA, as certified by the vendor) were cultured on planar MDP and MDP-DOPA hydrogel substrates measuring 100 mm in diameter (1 × 10⁶ cells hydrogel^-1, n = 4) with lipopolysaccharide (LPS; 250 ng·mL^-1, L2630, Sigma-Aldrich, St. Louis, MO, USA)-treated media [PM; 1% penicillin/streptomycin (PS; 30-002-CI, Corning, Corning, NY, USA) and 10% fetal bovine serum (FBS; 35-010-CV, Corning, Corning, NY, USA) in Dulbecco’s modified eagle medium (DMEM; 10-013-CV, Corning, Corning, NY, USA)] for 24 h to evaluate the ratio of M1/M2 macrophage polarization. The cells were then washed three times with PBS and treated with the proliferation medium. The CM were collected from RAW 264.7 cells incubated with tissue culture polystyrene (TCPS, 430167, Corning, Corning, NY, USA), C-CM, MDP (M-CM), and MDP-DOPA (MD-CM), centrifuged for 20 min at 300 g, and filtered with a 0.45-μm syringe filter (431220, Corning, Corning, NY, USA). RAW 264.7 cells were detached from each hydrogel, and M1/M2-related genes were evaluated. Total cellular RNA was isolated using an AccuPrep® Universal RNA Extraction Kit (K-3140, Bioneer, Daejeon, Republic of Korea) according to the manufacturer’s protocol. RNA concentration and optical purity ratios (A₂₆₀/A₂₈₀) were assessed spectrophotometrically using a NanoDrop One instrument (Thermo Fisher Scientific, Waltham, MA, USA). For cDNA synthesis, 1 μg of the isolated RNA template was reverse-transcribed into complementary DNA utilizing AccuPower® CycleScript RT PreMix (K-2044, Bioneer). The resulting cDNA served as a template to quantify M1/M2 macrophage-specific gene target levels via real-time polymerase chain reaction (PCR) using SYBR Green. Target transcription was determined using the comparative 2^-ΔΔCT approach, with results expressed as relative fold-change compared to RAW 264.7 cells cultivated on standard TCPS. Other surface markers of RAW 264.7 cells were analyzed using the intraclass correlation coefficient. To fix the cultured RAW 264.7 cells, specimens were incubated in 4% paraformaldehyde (PFA; 58127, Sigma-Aldrich, St. Louis, MO, USA) for 15 min and subsequently washed three times with PBS. To prevent non-specific binding, the washed cells were incubated in a blocking solution consisting of 5% goat serum (G9023, Sigma-Aldrich, St. Louis, MO, USA) and 0.1% Triton™ X-100 (T8787, Sigma-Aldrich) in PBS for 1 h at ambient temperature. The samples were then probed with primary antibodies against CD206 (PA5-101657, Invitrogen, Thermo Fisher Scientific, Carlsbad, CA, USA) and CD86 (ab220188, Abcam, Cambridge, UK) overnight at 4 °C. For signal visualization, the secondary antibodies were conjugated to fluorescein isothiocyanate (FITC; A-11012, Invitrogen) and rhodamine (A-11008, Invitrogen), followed by nuclear counterstaining with 4′,6-diamidino-2-phenylindole (DAPI; Vectashield, H-1200, Vector Laboratories, Burlingame, CA, USA). Fluorescence imaging was performed using a fluorescence microscope (Eclipse Ts2R; Nikon, Tokyo, Japan), and target-positive signals were quantified using ImageJ software (NIH, Bethesda, MD, USA).

To investigate the synergistic relationship between the retrieved CM and neurogenic cues, each CM was thoroughly mixed with neuronal differentiation media (NDM; DMEM supplemented with 10% FBS, 5% horse serum (H1138, Sigma-Aldrich, St. Louis, MO, USA), 1% PS, and NGF at 100 ng·mL^-1 at a volumetric ratio of 6.5:3.5 (CM:NDM), yielding a sub-optimal final NGF concentration of 35 ng·mL^-1 within the culture microenvironment. After culturing 5 × 10³ PC12 [RRID: CVCL_0481, American Type Culture Collection (ATCC), Manassas, VA, USA] cells in each well of a 96-well plate (3599, Corning, Corning, NY, USA) in various CMs, the cells were observed under a light microscope (AZ100, Nikon, Tokyo, Japan) at each time point (biological n = 4). Morphometric quantification, including neurite length and the proportion of neurite-positive cells (defined as cells exhibiting processes exceeding 10 μm), was performed on captured images using ImageJ.

Adult male SD rats (8 weeks old, weight: 260 ± 20 g at the time of surgery; Dayun, Gyeonggi-do, Republic of Korea) were used to make the sciatic nerve defect model, as previously described^[26]. All experimental procedures involving animals were approved by the Institutional Animal Care and Use Committee of Dankook University (approval no. DKU-22-001) and complied strictly with the NIH Guide for the Care and Use of Laboratory Animals. Animals were maintained in individual cages under controlled environmental conditions (a temperature of 23-25 °C and 45%-50% relative humidity) with ad libitum access to food and water. After 1 week of housing, surgical interventions were performed under general anesthesia induced with isoflurane (Forane®, Choongwae Pharma, Seoul, Republic of Korea). After hair removal, an incision was made through the skin and underlying muscle of the left gluteal region to expose the sciatic nerve. A 10-mm nerve defect was created by completely transecting and excising the nerve 5 mm distal to the hip joint. Both the proximal and distal nerve stumps were subsequently inserted into a 14-mm-long NGC (n = 4 per group) and coapted into the epineurium using 10-0 nylon sutures (NK 1013; AILEE Co., Busan, Republic of Korea). At each evaluation time point (1, 4, and 8 weeks post-implantation), 4 animals per group were assigned. Finally, the muscular, subcutaneous, and cutaneous layers were sequentially closed using 6-0 (SK 617; AILEE Co.) and 3-0 (SK 312; AILEE Co.) silk sutures. All rats were euthanized using CO₂ at weeks 1, 4, and 8, and their hind limbs were retrieved. All in vitro experiments were conducted between February and December 2021, and all in vivo experiments were conducted between February 2022 and 2023.

Axonal functional recovery post-implantation was determined using gastrocnemius electromyography (EMG) recordings and SFI measurements. For EMG evaluation, the rats were anesthetized using isoflurane, and electrical stimulation (1.5 mA) and signal recording were performed using the Medelec Synergy system (Oxford Instruments Medical Inc., Surrey, UK). To determine SFI values, footprints were collected biweekly by having the rats traverse a custom walking track lined with white paper. The print length, toe spread, and intermediate toe spread were measured and applied using the following formula: SFI = -38.3 × [(EPL - NPL) NPL^-1] + 109.5 × [(ETS - NTS) NTS^-1] - 13.3 × [(EIT - NIT) NIT^-1] - 8.8, where EPL is the experimental print length, NPL is the normal print length, ETS is the experimental toe spread, NTS is the normal toe spread, EIT is the experimental intermediary toe spread, and NIT is the normal intermediary toe spread.

To assess nerve injury-induced muscle atrophy, the gastrocnemius muscles were harvested at 4 and 8 weeks post-implantation. After rinsing with PBS, wet weights of the harvested muscles were immediately recorded. For tissue preparation, samples were fixed overnight in 4% PFA, dehydrated through a graded sucrose gradient (10%, 15%, and 30%) for 2 h each at 4 °C, and embedded in OCT compound (4583, Sakura Finetek USA Inc., Torrance, CA, USA) prior to cryosectioning. The sections were stained with hematoxylin (GHS232, Sigma-Aldrich, St. Louis, MO, USA) and eosin (HT110232, Sigma-Aldrich, St. Louis, MO, USA) [hematoxylin and eosin (H&E)] and observed by light microscopy. Muscle fiber diameters were quantified from the captured histological images using the ImageJ software (National Institutes of Health, Bethesda, MD, USA).

The sciatic nerve of each group was harvested 1 week after implantation. The harvested nerves were snap-frozen and divided into the proximal region and remaining tissue. The proximal regions were used for IHC analysis, and the remaining regions were used for quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis. The RNA was isolated according to the manufacturer’s instructions. Other processes were the same as those used in the in vitro qRT-PCR analysis, with the autograft group serving as the reference calibrator (relative expression level = 1) for the 2^-ΔΔCt calculation. Four and eight weeks after implantation, the harvested nerves were subjected to IHC and osmic acid staining. For IHC analysis, the harvested nerve tissues were fixed in 4% PFA and cryo-sectioned via a Leica CM3050 S cryostat (Leica Microsystems, Wetzlar, Germany). Permeabilization and blocking of nonspecific binding were performed using 0.3% Triton X-100 supplemented with 5% goat serum for 1 h at room temperature. The prepared sections were then incubated with the following target primary antibodies for 1-week specimens: rabbit anti-CD86 (1:100; Invitrogen, Thermo Fisher Scientific, Carlsbad, CA, USA) and rabbit anti-CD206 (1:100; Bioss, Woburn, MA, USA). For 4- and 8-week specimens, rabbit anti-neurofilament 200 monoclonal (1:300; Abcam, Cambridge, UK) and rabbit anti-S-100 polyclonal antibodies (1:1,000; Abcam) were used. Following three sequential PBS washes, tissue sections were incubated for 1 h at room temperature with fluorophore-conjugated secondary antibodies, either FITC- or rhodamine-labeled goat anti-rabbit IgG (1:200; Jackson ImmunoResearch Laboratories, West Grove, PA, USA), diluted in PBS containing 1% goat serum. Nuclear counterstaining was performed using DAPI. Immuno-stained specimens were photographed using a fluorescence microscope. All positive signals for each antibody were evaluated using the ImageJ software. For osmic acid staining, the samples were fixed with 4% PFA and stained with osmic acid (1% w/v) (75632, Sigma-Aldrich, St. Louis, MO, USA). After washing with excess DW for 6 h, paraffin sections (3 µm) were cut and observed by light microscopy. The thickness of the myelin sheath and axon diameter were calculated using the ImageJ software.

Quantitative data are expressed as means ± SDs. The SPSS software (IBM SPSS Statistics) was used to perform Bonferroni and Tukey tests.

In this study, we developed multifunctional NGCs from MDP and PVA and functionalized them with DOPA to enhance peripheral nerve regeneration^[43,44]. The fabrication process utilized a freeze–thaw method to create physical crosslinks, eliminating the need for chemical crosslinking agents that could cause denaturation and functional failure of bioactive proteins within casein.

The highly hydrophilic nature of PVA limits the adsorption of essential cell-binding proteins, thereby posing a challenge in providing cell-adhesion motifs^[21]. To address this, we coated NGCs with DOPA, a mussel-derived adhesive protein^[45]. This DOPA coating was hypothesized to enhance the adsorption of cell-binding proteins through electrostatic interactions, thereby promoting favorable cell responses such as the polarization of M2-lineage macrophages, which is critical for nerve regeneration^[45]. Our characterization results confirmed that the DOPA coating process did not adversely affect the porous morphology or mechanical properties of the hydrogel, thereby ensuring that the conduit possesses the requisite stability to prevent luminal collapse after implantation^[46,47]. Although their mechanical strength was lower than that of some brittle synthetic grafts, the flexibility and elongation capacity of our conduits were similar to those of human nerves, which, combined with the presence of bioactive peptides, may be more effective for nerve regeneration^[30]. Because peripheral nerve repair is generally performed under tension-free coaptation, an extremely high tensile strength is not a prerequisite for successful implantation.

The nerve repair process after injury involves a complex immune response, starting with an inflammatory phase dominated by M1 macrophages, followed by a regenerative phase promoted by M2 macrophages^[41,48]. We hypothesized that the bioactive peptides within the MDP and DOPA coatings would modulate this response to favor regeneration. Our in vitro results strongly supported this finding, demonstrating that the MDP-DOPA surface significantly promoted a shift from the M1 to the M2 macrophage phenotype. This shift can be attributed to the κ-casein fragments in MDP, which induce M2 polarization via TLR stimulation, and the enhanced cell adhesion provided by the DOPA surface, which facilitates interaction with macrophage integrins^[15,16].

Although we acknowledge the inherent complexity and heterogeneity of macrophage subpopulations in vivo beyond this binary framework, our comprehensive cytokine array analysis of the macrophage secretome [Supplementary Figure 5] directly validated this pro-regenerative transition. Secretome profiling showed that the MDP-DOPA interface robustly suppressed a broad spectrum of pro-inflammatory cytokines, while concurrently driving elevated secretion of vital neurotrophic and angiogenic factors, establishing a highly favorable microenvironment for downstream tissue remodeling.

Furthermore, the secretome from M2-polarized macrophages promotes axonal regrowth^[34,36,49]. Our study confirmed this indirectly, as CM from macrophages cultured on MDP-DOPA (MD-CM) significantly enhanced neurite outgrowth in PC12 cells to a level comparable to NDM. This potent neurogenic effect is likely a synergistic result of M2-secreted paracrine factors such as IL-10 and the direct action of bioactive casein peptides such as β-casomorphins and FLPYPY, released from the hydrogel. Importantly, PC12 cells maintained excellent viability and achieved highly accelerated neurite outgrowth under MD-CM conditions, demonstrating the exceptional cytocompatibility and the complete absence of cytotoxicity from our hydrogel platform and its degradation byproducts [Supplementary Figure 6].

The superior in vitro performance of MDP-DOPA translated into remarkable in vivo efficacy. In a rat sciatic nerve defect model, the MDP-DOPA conduit led to functional recovery (SFI and electrophysiology) and prevented muscle atrophy, significantly better than unmodified MDP and closely approaching the therapeutic trajectory of the clinical autograft gold standard. Although commercially available or well-established natural polymer scaffolds such as collagen- or chitosan-based conduits are frequently investigated as conventional controls, their clinical efficacy in large nerve gaps remains heavily constrained by poor mechanical stability, leading to luminal collapse and a lack of active immunomodulatory cues, often resulting in regenerative outcomes far inferior to autografts^[3,4]. By contrast, our study used an autograft, the clinical gold standard for peripheral nerve repair, as the primary positive control to establish the most rigorous benchmark for functional integration. The fact that the soft, bioinstructive MDP-DOPA conduit achieved histological and functional recovery comparable to this ultimate clinical standard negates the necessity for an intermediate natural polymer control, further highlighting its superior potential for clinical translation. The MDP-DOPA group showed significantly greater axonal regrowth, Schwann cell migration, and remyelination (thicker myelin sheaths and larger axon diameters) in the distal nerve segment than the MDP group.

A key aspect of our proposed mechanism is early-stage immunomodulation at the injury site^[6,39]. Analysis at one week post-implantation revealed that the MDP-DOPA conduit environment was dominated by M2 macrophages, in stark contrast to the M1-dominant environment in the MDP group. qRT-PCR data further confirmed that the tissue within the MDP-DOPA group showed high expression of M2-related cytokines and growth factors, including TGF-β, VEGF, and NGF. These factors are crucial for recruiting Schwann and endothelial cells, which in turn create the regenerative microenvironment necessary for robust axonal growth and functional recovery^[6-8,41,42]. Thus, MDP-DOPA actively orchestrated the sequential processes of regeneration, beginning with modulation of the initial immune response. The predominant presence of M1 macrophages in the MDP group likely explains their inability to effectively transition to the healing phase, leading to poorer regenerative outcomes^[33].

Despite the encouraging functional, histological, and immunomodulatory outcomes observed in this study, some limitations should be considered. First, the in vivo evaluation used a 10-mm sciatic nerve defect in adult male rats, with the longest observation period being limited to 8 weeks. This model was sufficient to demonstrate the regenerative efficacy of the MDP-DOPA conduit relative to the MDP conduit and autograft control. Longer follow-up periods and validation in more challenging nerve defects or larger animal models will be required to confirm the durability and broader translational applicability of the observed effects.

Second, we primarily evaluated degradation under controlled in vitro conditions using a collagenase-containing medium. The in vivo findings confirmed that the conduit maintained its tubular architecture without significant collapse at 8 weeks; however, the quantitative in vivo degradation kinetics, local tissue response to degradation residues, and long-term safety were not comprehensively assessed. Therefore, future studies should examine the temporal relationships among conduit degradation, tissue remodeling, and long-term biocompatibility.

Finally, although the present results support the involvement of MDP-derived bioactive components and DOPA-mediated cellular interactions in macrophage modulation and nerve regeneration, the exact primary peptide sequences and their specific downstream signaling cascades operating in this hydrogel platform were not individually isolated or blocked in the present study. While the observed M2 polarization and neurite outgrowth are consistent with the known bioactivities of κ-casein derivatives and β-casomorphins, future studies focused on peptide-specific sequence analysis and pathway inhibition will be necessary to further refine the underlying molecular mechanism. The current study characterized cumulative total protein release and changes in macrophage-associated gene and protein profiles, providing evidence for the proposed bioactivity; however, peptide-specific identification, quantitative release analysis, and pathway-blocking experiments are necessary to define the detailed molecular mechanisms. Additional studies would strengthen, rather than alter, the regenerative effects demonstrated in this study.

Peripheral nerve repair remains challenging because the current NGCs lack the ability to actively modulate the immune microenvironment and support robust regeneration. In this study, we developed a milk-derived immunomodulatory hydrogel conduit functionalized with DOPA that promoted macrophage polarization toward a pro-regenerative phenotype and enhanced neurite outgrowth. In a rat sciatic nerve defect model, the conduit achieved functional and histological recovery comparable to autografts without in vivo delivery of exogenous growth factors. This study introduced a growth factor-free biomaterial strategy that integrates intrinsic bioactivity and immunomodulation, thereby providing a new design paradigm for next-generation NGCs and regenerative biomaterials.