Biopatch

Chronic wounds fail for two reasons at once, and conventional dressings address neither.

A wound that will not close is usually held in an unfavourable biochemical state. Sustained oxidative stress keeps the tissue in a pro-inflammatory loop, and excess reactive oxygen species have been linked experimentally to the onset and persistence of chronic wounds. A passive dressing covers that environment without changing it.

The second failure mode is mechanical. Repeated deformation at the wound margin is associated with delayed closure and heightened inflammation, and it is invisible to the clinician between dressing changes. Biopatch was specified to address both: a matrix that carries ROS-reducing chemistry, and a sensing layer that makes local mechanical load a measurable quantity.

One feedstock class, two functions: structural matrix and candidate bioactive.

Biopatch is built from aquatic biomass. Seaweed-derived alginate serves twice over, as the hydrogel matrix and as the substrate for enzymatic production of alginate oligosaccharides. Water hyacinth, an invasive freshwater macrophyte whose dense mats depress dissolved oxygen and degrade habitat, supplies the second candidate ingredient through hydroethanolic extraction.

Both candidates enter the same cell-based screen. The selected ingredient is then carried into matrix fabrication, and the sensing layer is deposited on the finished patch. Each stage is specified as an independently testable unit, so a result at one stage does not have to be inferred from the behaviour of the assembled device.

Three alginate lyases screened for soluble expression, two carried forward.

Alginate depolymerisation is run with recombinant lyases expressed in E. coli BL21(DE3). Three constructs were evaluated against an empty-vector background: FlAlyA (PDB 5Y33) from Flavobacterium sp., Alyw203 from Vibrio sp., and AlgL7 from Microbulbifer sp. Expression was screened across three temperatures and two inducer concentrations in parallel, so favourable conditions were identified without iterating the design cycle three times over.

5Y33 and Alyw203 both yielded soluble protein at 18 °C and 26 °C; AlgL7 expressed poorly under every condition tested and was dropped. The two surviving candidates were run at 26 °C and 0.2 mM IPTG for functional work. Alyw203 showed a temperature-dependent shift in apparent mass, consistent with proteolytic cleavage near the fusion junction at lower temperatures, which remains an open question rather than a settled mechanism.

Depolymerisation was confirmed by dinitrosalicylic acid assay. Reactions containing either enzyme generated significantly higher reducing-sugar equivalents than the empty-vector control, which establishes that alginate was cleaved. Reducing-sugar equivalents are a measure of cleavage activity and are not a yield figure for isolated oligosaccharide.

The screen selected the extract and rejected the oligosaccharide route.

Both candidate ingredients were assayed for intracellular ROS in LPS-stimulated RAW 264.7 macrophages using a dichlorodihydrofluorescein diacetate probe, read at 485/528 nm across five replicates. The oligosaccharide preparations produced no significant reduction at either tested loading. The botanical extract produced a concentration-dependent reduction, falling below the unstimulated control at the two higher concentrations.

That result set the formulation. Extract at 100 µg/mL was selected for matrix loading, the concentration at which the LPS-induced signal is suppressed to approximately the level of untreated cells. The oligosaccharide route is not closed, but it requires different preparation conditions and concentrations before it can be assessed again.

Deformation becomes a resistance trace, in real time.

A flexible silver nanowire layer is deposited on the patch as a strain gauge. Bending, stretching and repeated deformation change its resistance; the patch is wired as the variable leg of a voltage divider against a 1 kΩ fixed reference, and the divider output is digitised by the analogue-to-digital converter of an nRF52840 microcontroller and streamed over USB serial at 115200 baud.

Resistance is recovered from the measured voltage by rearranging the divider relation, and plotted continuously in a ten-second moving window refreshed every 50 ms. The conversion path is deliberately transparent: the converter measures a voltage, the software computes a resistance, and the resistance trace is the quantity used to compare deformation events. Raw voltage is never treated as the sensor output.

Once an unloaded baseline R₀ is recorded, the same path yields relative change ΔR/R₀ for comparison between deformation trials. The computation assumes a stable supply, a constant reference resistor and reliable contacts, and it reports absolute resistance rather than a calibrated strain figure.

Three models, coupled only where the physics couples.

The platform carries an electrical model, a release model and a ROS model. They share parameters only where a shared parameter is physically justified: matrix thickness governs molecular transport, while the nanowire resistance model describes an electrically independent layer. Keeping them separate prevents a convenient result in one domain from propagating into another.

Release is approximated as one-dimensional Fickian diffusion through the matrix thickness, with an effective diffusion coefficient for the double network and release from both faces of the sheet. Stronger ionic crosslinking enters the model as a lower effective diffusivity where release data support it. The ROS model is a reaction–diffusion balance in which a source term represents continued generation at an inflamed site and a scavenging term scales with the locally available agent.

The coupling is one-directional and physically ordered: the matrix must release the agent before the agent can act, so the diffusion model supplies the availability term A(t) to the ROS model. Faster release brings the effect forward in time; slower release spreads it out. The parameter set below is illustrative, used to interrogate design choices such as thickness and crosslink density, and every value requires wet-lab calibration before it describes a real wound.