Fig. 2: Multi-gradient engineering.

a Responsive hydrogel system. The hydrogel pillars are formed by curing a mixture of precursor molecules (monomers and DNA crosslinkers), MNPs and initiators under UV irradiation. Upon DNA target incubation, the hydrogel network is disrupted, increasing its porosity and releasing the encapsulated MNPs. b Characterization of pillar formation and MNP release efficiency. We optimized various hydrogel conditions to achieve > 95% pillar formation efficiency, and further optimized these conditions for specific and efficient MNP release (right). c MR responsiveness gradient. Magnetized MNPs were simulated at various vertical distances (z) above the GMR sensor surface, confirming that the sensor is the most responsive at its surface. d MNP gradient. An external magnetic field was applied before pillar gelation to form a complementary MNP gradient. We measured the relative abundance of ATTO590-conjugated MNPs at different heights (h) of the formed hydrogel pillars and confirmed MNP enrichment near the pillar tips and the GMR sensor. e Analyte gradient. We adjusted the gap spacing between the hydrogel pillars and the sensor surface to shape the analyte gradient and concentrate the analyte molecules on the pillar tips. f MNP release profile. By embedding ATTO590-labeled MNPs, either in gradient or uniform distribution, in the hydrogel pillars, we showed through both magnetic and fluorescence measurement that the gradient-distributed system showed a faster release kinetics. g Titration analysis of the triple-matched, dual-matched and unmatched assays. We compared the performance of the four architectures. The triple-matched system (analyte–MNP–MR) showed the best LOD, while the other configurations (analyte–MR, MNP–MR, no matching) showed less sensitive detection. The LOD is defined as 3 × s.d. above the signal of the no-target sample. Data (n = 9) were collected from different pillars in d–e. All measurements were performed in triplicate (n = 3 independent experiments) and the data are displayed as mean ± s.d. in f–g, and as mean in b. Source data are provided as a Source Data file.