After the spinal cord is fully severed, the adult central nervous system virtually stops growing, and secondary inflammation at the injury site seals off the road to regeneration. In 2026, a team at Wenzhou Medical University published a study in Materials Today Bio: taking GelMA — a classic bio 3D printing ink — as the substrate, doping in N-acetylcysteine (NAC)-modified silver nanowires, and casting a conductive, spinal-cord-soft hydrogel with axial microchannels.

In a 2 mm full-transection rat spinal cord model, six weeks after implantation, axons on both sides of the injury re-crossed and formed synapses, with significant recovery of hindlimb motor function. The paper further proposes and validates the molecular switch behind it — the FGF13/Ca²⁺/CaMK2A/CREB pathway.

*All values quoted from the original paper: microchannels from Figure 1G; conductivity from Section 3.1 and Figure 2G; spinal cord reference range from Introduction; behavioral data from Section 3.4.

PART 01 The Spinal Cord “Cable”: Why It Cannot Reconnect After Severing

The spinal cord is the main signal conduit between the brain and the body’s nerves — why can’t it reconnect after severing? The paper’s introduction offers two layers of reasons. First, adult central nervous system regeneration is extremely limited; axons on either side of the injury do not regrow like peripheral nerves. Second, and more troublesome — mechanical transection is only the first hit; the secondary pathology is long-term: excessive inflammation, demyelination, and failed axonal reconnection unfold one after another, turning the injury site into a “no-go zone” highly unfriendly to regeneration.

More critically, the spinal cord is not a homogeneous “jelly.” The paper cites diffusion tensor imaging (DTI) evidence: white matter nerve fiber tracts are highly aligned, and regenerating axons require axially oriented microchannels to grow in an ordered manner; meanwhile, the spinal cord is an electrically active organ — conductivity ~0.02–0.60 S/m, elastic modulus ~100–3000 Pa. An ideal scaffold must satisfy three requirements at once: provide direction, conduct electricity, and match softness.

Existing hydrogel scaffolds mostly provide only passive support; carbon nanotubes, graphene, PEDOT, and other conductive materials each have shortcomings — poor dispersion, insufficient mechanical compliance, or single functionality.

PART 02 Meet the Two Protagonists: GelMA Ink and “Retrofitted” Silver Nanowires

The scaffold substrate, GelMA (gelatin methacryloyl), is a classic bio 3D printing ink raw material — retaining cell adhesion sites, tunable stiffness, and good biocompatibility, naturally close in softness and hydrophilicity to the spinal cord extracellular matrix. This study did not use a printhead but adopted a forming process better suited to tubular scaffolds (detailed in the next section). GelMA’s inherent weakness is electrical insulation — it cannot support the conductive microenvironment the spinal cord requires.

To supply conductivity, silver nanowires (AgNWs) were used: 90 nm in diameter, 20–60 μm in length, one-dimensional filaments with high aspect ratio, forming a continuous conductive network at low loading without sacrificing softness. But bare silver nanowires have two old problems — high contact resistance and unstable dispersion. The team used NAC for surface modification: thiols form robust Ag–S coordination bonds with silver atoms, stabilizing the nanowire interface and lowering electron transfer resistance; carboxyl and amino groups introduce zwitterionic character, making the nanowires more compliant. NAC itself is a clinically common antioxidant, so loading it into the scaffold is like embedding a sustained-release antioxidant “medicine cabinet.”

PART 03 The “Track” Structure

How to create straight axial channels inside a gel? Let ice do the work. A Teflon mold filled with precursor solution was placed vertically on a −40 °C cold stage, allowing unidirectional heat dissipation so that ice crystals grow in an ordered manner against the temperature gradient — wherever the ice goes, it leaves a channel behind. After 20 minutes of directional freezing, it was transferred to −20 °C for 24 hours, crosslinked and set via the APS/TEMED redox system. After the ice crystals melt, what remains is a hydrogel column perforated with parallel microchannels. The control group (R-GelMA) was placed directly in a −40 °C freezer for random freezing; SEM comparison: the aligned group showed parallel, straight grooves, while the random group was a tangle.

Two rounds of “screening” nailed the formulation. Among four GelMA concentrations from 2.5% to 5%, 2.5% had a Young’s modulus of about 1.9 kPa, exactly in the spinal cord’s soft range, and was selected. NAC-AgNW loadings of 4/8/12/16 mg/mL gave conductivities of 0.126/0.247/0.267/0.185 S/m respectively — increasing further actually decreased conductivity (nanowire aggregation disrupted the network), and live/dead staining indicated noticeably more dead cells from 12 mg/mL upward. 8 mg/mL emerged as the balance point between conductivity and viability. Final formulation: 25 mg/mL GelMA + 8 mg/mL NAC-AgNW, with microchannel diameters uniformly landing in the 150–200 μm range.

PART 04 Health Report: Soft Like the Spinal Cord, Conductive Like Nerve, with a “Sustained-Release Medicine Cabinet”

Three scaffold types (pure GelMA, bare silver nanowires, NAC-AgNW) were put through a “health check,” and the differences were marked. All groups had storage modulus G′ consistently exceeding loss modulus G″, indicating stable elastic gel behavior; the NAC-modified group showed stable modulus across time, frequency, and strain sweeps, without internal perturbations from bare silver nanowire aggregation. It withstood 80% deformation, with Young’s modulus about 3 kPa — almost matching spinal cord softness; swelling rate exceeded 90%, enabling gentle filling of defect cavities after implantation.

Conductivity was the core goal of this retrofit: the three groups gave 0.03/0.14/0.24 S/m — once the dispersion problem was solved, the network became continuously interconnected, and 0.24 S/m falls right in the spinal cord’s physiological range of 0.02–0.60 S/m. Cyclic voltammetry and impedance spectroscopy agreed: anodic peak as low as 0.29 V, smallest impedance arc and charge-transfer resistance, and the brightest in the lightbulb demonstration. On degradation, pure GelMA nearly disappeared within 12 hours in a collagenase environment, while the two silver-containing groups slowed markedly — enzymes cannot chew through a denser network, so the scaffold stands longer in vivo. The release curve was equally critical: free thiols released continuously, reaching 1.8 mM cumulatively by day 8 — the NAC “medicine cabinet” covers exactly the early post-injury window when inflammation is most intense.

PART 05 Cell-Level Validation: Live Longer, Migrate Farther, Align Straighter, Resist Oxidation

Materials passed the exam — now do cells buy it? PC12 neural cells were cultured on all three scaffolds: days 1 and 3 showed mostly live cells in all groups, with no obvious toxicity from silver nanowires; by days 7 and 14 viability still exceeded 90%. On alignment, the random-channel group (R-GelMA) showed disorganized cell long-axis orientation, while cells on aligned scaffolds lined up along microchannels — proportion with long-axis within ±10°: A-GelMA 20%, A-AgNW-GM 21%, A-NAC-AgNW-GM 27%, with the NAC-modified group (uniform conductive network) marshalling the “ranks” most neatly.

Migration and antioxidant tests carried more weight: PC12 cells on the NAC-modified group migrated significantly farther and covered more area; the antioxidant test first created a “fire scene” — 6 hours of 0.1 mM tert-butyl hydroperoxide (TBHP) challenge — then visualized reactive oxygen species (ROS) with the DCFH-DA probe: the injury control and pure GelMA groups showed blinding fluorescence, the bare silver nanowire group lowered it somewhat via trace silver redox chemistry, and the NAC-modified group was at the bottom — dual antioxidant defense pushed the “fire” to a low level.

PART 06 Immune Dispatch: Persuading “Wrecking Crew” Inflammatory Cells into a “Road-Building Crew”

The first responders after spinal cord injury are microglia/macrophages: M1 phenotype releases inflammatory factors that fan the flames, while M2 phenotype secretes repair signals that clean up — the direction of injury depends heavily on the balance between the two. The paper used lipopolysaccharide (1 μg/mL) to force BV2 microglia into an M1 state, then had the scaffold “mediate” across a Transwell. The result was clear: pro-inflammatory genes iNOS, IL-6, CD86, TNF-α, IL-1β were all downregulated, while anti-inflammatory genes IL-10, Arg-1, CD206 were upregulated; immunofluorescence and immunoblotting at the protein level gave consistent conclusions — the balance tipped toward M2. Mechanistically, NAC scavenging ROS broke the cycle of “oxidative stress → inflammation → more M1,” while silver nanowires actively steered immune cells toward the repair phenotype; the two are complementary and synergistic.

In vivo data continued the trend: one week post-surgery, CD68-positive cell density in the injury area was lower in all silver-containing groups than the injury-alone group, with the NAC-modified group lowest; iNOS positivity decreased and Arg-1 positivity increased. The safety ledger was also clean: major organ H&E showed no inflammation or toxicity; ICP-MS showed silver content in all organs remained below 1 μg/g throughout and decreased over time, approaching baseline by day 42, with no liver or kidney accumulation. Even neurogenic bladder, a common complication, showed improvement — Masson staining showed relief of pathological bladder wall thickening in the NAC-modified group.

PART 07 The Rat Test: Across the Injury, Nerves “Reconnect”

The ultimate test was in rats: 60 SD rats (220–230 g) evenly divided into four groups — injury alone, pure GelMA, bare silver nanowires, and NAC-modified; the T9–10 segment was resected for a 2 mm spinal cord to create complete transection, followed by immediate implantation of the corresponding hydrogel, with six weeks of testing. BBB motor scores once per week: the NAC-modified group led throughout, opening a significant gap by week 6; DeepLabCut tracking showed larger hip, knee, and ankle swing angles in this group, with markedly less hindlimb dragging in footprint assays; motor evoked potentials (MEP) showed highest amplitude and shortest latency — better signal quality from brain command to hindlimb.

Histology provided the explanation: on H&E, the injury cavity in the NAC-modified group was significantly smaller than in the injury-alone group; LFB showed markedly reduced demyelination; Nissl staining showed the most Nissl bodies among all four groups — the most surviving neurons. The critical question was answered by BDA tracing: injecting BDA into the motor cortex “lit up” the corticospinal tract, and only the NAC-modified group showed large numbers of BDA-positive axons densely crossing the entire injury region and blending with distal host tissue; regenerating terminals were in close contact with NeuN-positive neurons and clearly co-localized with the postsynaptic marker PSD95 and the vesicle protein Syn — 3D reconstruction showed visible “pre-synapse / post-synapse” paired structures. The new neural circuit from brain to distal injury site did not merely “grow across” — it “plugged in.”

PART 08 The Molecular Switch: FGF13/Ca²⁺/CaMK2A/CREB Axis Lights Up

No matter how clear the phenomenon, it must answer “why.” Transcriptome sequencing showed: compared with the injury-alone group, the NAC-modified group had 1,080 genes upregulated and 1,089 downregulated, with enrichment pointing to synaptic organization, neurotransmitter secretion, calcium ion binding, and other repair processes. The most striking upregulation was an “unconventional” member — fibroblast growth factor 13 (FGF13): it is not secreted extracellularly but acts intracellularly, stabilizing microtubules and regulating voltage-gated sodium channels (Nav) to direct axon growth and neuronal excitability — two skills that fit the scaffold’s two design principles exactly: physical guidance provided the “road,” FGF13 provided the “navigation”; the scaffold provided the “external grid,” FGF13 amplified the “endogenous electrical”.

Pathway validation was in three steps. First, KEGG and GSEA both converged on the calcium signaling pathway (NES 1.56, p<0.001, FDR<0.001): FGF13 enhances sodium channel activity → action potentials trigger calcium channel opening → increased calcium influx, and FGF13 can directly bind the intracellular domain of the L-type calcium channel CaV1.2, stabilizing the pore subunit and promoting its membrane insertion. Second, immunoblotting confirmed: FGF13, p-CaMK2A, and p-CREB, which were suppressed after injury, all rebounded significantly in the NAC-modified group — the cascade was relit. Third, immunofluorescence plus BDA double-labeling showed FGF13, CaMK2A, and CaV1.2 co-localized along regenerating axons; lentiviral overexpression of FGF13 in primary cortical neurons produced significantly elongated neurites, upregulated GAP43, and more elaborate branching in Sholl analysis — the gains were even more pronounced on the NAC-modified scaffold. “The material’s electrical cue, passing through the molecular amplifier FGF13, was plugged into the neuron’s own growth program” — the paper proposes and validates (the authors claim for the first time) this FGF13/Ca²⁺/CaMK2A/CREB axis.

The paper is candid about limitations: observation window only 6 weeks; full-transection model still distant from the clinically more common contusion/compression injuries; in vivo electrical signal propagation was not directly measured; in vivo necessity of FGF13 (e.g., knockdown experiments) remains to be supplemented.

Closing Thoughts

From rat full-transection to clinical application, there is still a long road: the paper offers a validated pathway and a reusable design paradigm, not a product about to land.

📄 Paper Information

· END ·