Researchers have shown that a flat, cell-compatible hydrogel pattern can be programmed to become several three-dimensional forms, including a dome-like spherical cap, a saddle and a stingray-inspired structure that flapped as temperature was cycled. Fibroblast viability was high in the tested transformed constructs during culture. The findings suggest that shape programming and cell-laden fabrication can be combined in the tested material system.
The approach separated two jobs inside one printed architecture: cell support and actuation. Multimaterial digital light lithography patterned temperature-unresponsive GelMA domains within a temperature-responsive pNIPAm matrix. GelMA served as the cell-supportive part of the design, while the pNIPAm matrix was used for temperature-responsive morphing. The resulting architecture linked a planar material pattern to a prescribed three-dimensional shape.
For cell experiments, the bioink contained 5 wt.% GelMA and 3T3 fibroblasts at 2 million cells per milliliter. The GelMA precursor was equilibrated to 37°C before printing. The metabolic measurements also included non-morphing planar disk constructs as controls.
Programming the bend
To tune the response, the team measured how the patterned material changed area at different temperatures. At 37°C, the area-based shrinkage ratio increased from 0.3 at a GelMA area fraction of zero to 1.05 at a fraction of 0.45. At 25°C, the area-based swelling ratio decreased as the GelMA area fraction increased. The calibration showed that the material's measured response changed with the proportion of GelMA in the pattern.
That calibration was used to set target curvature, or the way a surface bends. Gaussian curvature captures bending in two directions, making it useful for describing both dome-like and saddle-shaped forms. It was estimated from two perpendicular side views by fitting central cross-sectional profiles to circular arcs, then calculating two principal curvatures from the fitted radii and multiplying them. For the spherical-cap construct, the measured value was 0.056 per square millimeter, compared with a programmed value of 0.060 per square millimeter. For the saddle, it was −0.092 per square millimeter, compared with a programmed −0.090 per square millimeter. The study also tested a range of programmed curvature values. The close matches in these examples indicate that the printed structures followed the intended curvature.
The same design logic was extended to more complex assemblies. The framework combined target metrics through integration, transformation and modular assembly, broadening the range of programmable three-dimensional constructs.
A structure that moved
One demonstration used a stingray-inspired construct. During cyclic temperature modulation from 28.5°C to 34°C, it exhibited synchronized oscillatory flapping. The motion was observed under the tested temperature cycle, but high-cycle lifetime, fatigue and possible interfacial degradation were not characterized. That leaves unanswered how the material would perform under prolonged repeated actuation.
Cells inside the forms
The cell experiments measured fibroblast viability in the transformed shapes during culture. In spherical-cap constructs, viability rose from 87.1% on day 1 to 91.3% on day 14. On day 14, viability reached 96.8% in saddle constructs and 94.4% in stingray-inspired constructs. The transformed cell-laden constructs maintained structural integrity for up to 15 days in culture.
Metabolic activity, the study's measure of cell activity, increased over 14 days and was largely comparable among spherical-cap, saddle and non-morphing planar disk constructs. No significant differences were reported on days 3, 5 and 10, while spherical-cap constructs showed a greater increase than disk controls on days 7 and 14. Quantitative results were reported as means with standard deviations. The comparisons used one- or two-way ANOVA with Tukey's multiple-comparisons test, with results below 0.05 treated as statistically significant.
The cells also changed shape as culture continued. They were rounded and uniformly distributed on days 1 and 3, formed multicellular assemblies by day 7, and were predominantly elongated by day 14. Prominent aligned actin stress fibers were reported by day 14. The observations are consistent with fibroblast spreading, assembly and organization during culture.
What the tests leave open
The results describe a cell-compatible morphing platform in the tested fibroblast constructs, but they do not establish that curvature itself caused the observed cellular behavior. The study did not isolate curvature-dependent cellular functional responses from baseline behavior in GelMA using non-morphing controls or constructs with prescribed curvatures. The reported cell responses should therefore be read as observations in transformed constructs, not as evidence that a particular curvature caused cells to spread, assemble or align.
A separate domain-size analysis also came with a statistical caution. Viability showed no statistically significant effect of domain size and no day-by-size interaction. The reported p-values were 0.07 and 0.70, respectively, based on four independent constructs at each time point. Because no equivalence test or power calculation was reported, the result means that no difference was detected in the tested comparison, not that all domain sizes were demonstrably equivalent.
The experiments report programmable curved and moving hydrogel constructs alongside viability and cell-behavior measurements, while leaving open whether geometry can be tied to a specific cellular function and how the interfaces will hold up under repeated actuation.
Paper data and sources
Original title: Discrete 2D Material Programming for 3D Shaping and Morphogenesis-Inspired 4D Bioprinting.
Authors: Fereshteh Family, Aneela Davuluri, Athulya Martin, Kyungsuk Yum
Journal/Repository: Advanced science (Weinheim, Baden-Wurttemberg, Germany)
Status: Peer-reviewed
First online: 2026-08-21
DOI: 10.1002/advs.77228
Original paper