A bioinspired scaffold for rapid oxygenation of cell encapsulation systems

Abstract

Inadequate oxygenation is a major challenge in cell encapsulation, a therapy which holds potential to treat many diseases including type I diabetes. In such systems, cellular oxygen (O₂) delivery is limited to slow passive diffusion from transplantation sites through the poorly O₂-soluble encapsulating matrix, usually a hydrogel. This constrains the maximum permitted distance between the encapsulated cells and host site to within a few hundred micrometers to ensure cellular function. Inspired by the natural gas-phase tracheal O₂ delivery system of insects, we present herein the design of a biomimetic scaffold featuring internal continuous air channels endowed with 10,000-fold higher O₂ diffusivity than hydrogels. We incorporate the scaffold into a bulk hydrogel containing cells, which facilitates rapid O₂ transport through the whole system to cells several millimeters away from the device-host boundary. A computational model, validated by in vitro analysis, predicts that cells and islets maintain high viability even in a thick (6.6 mm) device. Finally, the therapeutic potential of the device is demonstrated through the correction of diabetes in immunocompetent mice using rat islets for over 6 months.

Fig. 1. Design and fabrication of the biomimetic SONIC scaffold.

a A digital image of a larva of the mealworm beetle (Tenebrio molitor). b 3D reconstruction of Nano-CT images of the tail of a mealworm (left) and the segmented gas-phase tracheal system (right) inside the body. c Schematic illustrating the tracheal system in a mealworm and O₂ delivery to the surrounding cellular tissue through the tracheae. d A schematic illustrating O₂ delivery from the transplantation site into the cell encapsulation system through a tracheal ladder network-like SONIC scaffold. e Chemical structure of the fluoropolymer PVDF-HFP. f Fabrication of the ladder-like SONIC scaffold. g A digital image of the SONIC scaffold. h Schematics representing the macro- and microarchitecture structure of the SONIC scaffold. i A Nano-CT image of the porous scaffold (the asterisks indicate the pore regions). j 3D reconstruction of Nano-CT images of a selected region (26 × 34.84 × 5.36 µm) inside the SONIC scaffold showing the bicontinuous microstructure (the diffuse red coloring indicates the air phase in the porous skeleton). k Skeletal networks for the polymeric (blue) and the porous (red) regions of the SONIC scaffold.

Fig. 2. Fabrication and characterizations of the SONIC device.

a Schematic illustration of the device fabrication (side view). b, c False-colored SEM images of the SONIC scaffold (b) and polydopamine-coated SONIC scaffold (c). One representative of 3 independent experiments is shown. d, e Digital images of water droplets (colored with food dye) and contact angle goniometer-captured images of a water droplet on a rectangular prism SONIC scaffold before (d) and after (e) polydopamine modification. f A false-colored SEM image of the polydopamine-coated SONIC scaffold with deposited CaSO₄ crystals. One representative of 3 independent experiments is shown. g Stereo microscope image of the SONIC device (top view). h SEM/EDS elemental mapping of F, N, and S on a polydopamine-coated SONIC scaffold with deposited CaSO₄ crystals. The red arrows indicate the lack of polydopamine at a coating crack location. One representative of 3 independent experiments is shown. i SEM image of the polydopamine-coated SONIC scaffold and the corresponding element N distribution profile across a polydopamine coating crack. One representative of 3 independent experiments is shown. j SEM image of the cross-sectional polydopamine-modified SONIC scaffold, showing no polydopamine inside the scaffold. One representative of 3 independent experiments is shown. k An H&E staining slice of an islet encapsulation device showing the polydopamine located at the interface between the SONIC scaffold and alginate hydrogel. One representative of 10 replicates is shown. l Captured images during the perfusion test using a cylindrical SONIC scaffold. A pump-connected needle was inserted into one end of a cylindrical SONIC scaffold, and the other end of the SONIC scaffold was immersed into a vial containing water phase (top, colored with green food dye) and chloroform phase (bottom, colored by Nile Red dye). m Schematics representing the distribution of water and chloroform during the perfusion test.

Fig. 3. Characterizations of rapid O₂ transport through the SONIC scaffold.

a A digital image (left) and schematic (right) showing the sample in a container for the EPR test. b Chemical structure of the EPR spin probe OX063-d24. c Calibration curve of the OX063-d24 relaxation rate versus pO₂. d Average pO₂ of gelatin in the container with a PLA control insert versus time. The initial gray period indicates the deoxygenation of the system. e pO₂ distributions on the tangential plane and transverse plane of a sample with the PLA control insert at different time points (indicated by the arrows in Fig. 3d). f Average pO₂ of gelatin in the container with a SONIC scaffold versus time showing much faster equilibration. g pO₂ distributions on a tangential plane and a transverse plane of the sample with the SONIC scaffold at different time points (indicated by the arrows in Fig. 3f). h Simulation data showing spatial pO₂ profiles over time in the system with the PLA control insert and SONIC scaffold. i Simulated the average pO₂ of gelatin in the container versus time.

Fig. 4. The SONIC scaffold improves cell survival under hypoxic conditions.

a Simulation-predicted pO₂ distributions of INS-1 cells encapsulated control device in three tangential cross-sections (labeled A-A, B-B, and C-C) and a transverse cross-section (labeled D-D, a corresponding ${\mathrm{c}}_{{\mathrm{O}}_2}$ distribution labeled as D’-D’). b Quantitative pO₂ and ${\mathrm{c}}_{{\mathrm{O}}_2}$ distributions along a radial line in the transverse cross-section of the control device showing rapid O₂ dropping from 40 mmHg (0.05 mM) at the surface to ~3 mmHg (0.004 mM) at the center. c Simulation-predicted pO₂ distributions of INS-1 cells encapsulated SONIC device in three tangential cross-sections and a transverse cross-section. d Quantitative pO₂ and ${\mathrm{c}}_{{\mathrm{O}}_2}$ distributions along a radial line in the transverse cross-section of the SONIC device showing high pO₂ over 35 mmHg in the whole device, corresponding with high ${\mathrm{c}}_{{\mathrm{O}}_2}$ over 0.044 mM in the cell/hydrogel phase and a substantially higher ${\mathrm{c}}_{{\mathrm{O}}_2}$ of ~1.85 mM in the SONIC scaffold due to the preferential partitioning of O₂ into the gas-containing SONIC scaffold. e, f Microscope images of live/dead staining of INS-1 cells encapsulated in the control device (e) and the SONIC device (f). One representative of 2 independent experiments is shown. g Simulation-predicted pO₂ distributions of rat islets (size-distributed) in a control device in three tangential cross-sections (labeled A–A, B-B, and C-C) and two transverse cross-sections (labeled D-D and E-E) showing massive hypoxic regions in the center of the device with necrosis observed in many islets (white regions in the islets represent necrosis). h Simulation-predicted pO₂ distributions of rat islets (size-distributed) in a SONIC device in three tangential cross-sections (labeled A-A, B-B, and C-C) and two transverse cross-sections (labeled D–D and E-E) showing well-oxygenated islets in the entire device, with negligible necrosis observed in rare large islets (arrow in A-A cross-section). i, j Average pO₂ (i) and fraction of necrosis (j) of the islet populations in control devices (n = 20), and SONIC devices (n = 20). Mean ± SD, ****p < 0.0001 (unpaired two-sided students t-test). k, l Scatter plots of islet location versus average pO₂ in simulated islets (all 150 µm in diameter) in control devices (k) and SONIC devices (l).

Fig. 5. The SONIC device enables 6-month diabetes correction in mice.

a, b Schematic (a) and microscope image (b) of rat islets encapsulated in a SONIC device (~4 mm in diameter). One representative of 11 replicates is shown. c BG measurements of diabetic C57BL6/J mice following IP transplantation of SONIC devices (pink, n = 5, retrieved on day 60; red, n = 6, one device was retrieved on day 145 after BG rising and the other five were retrieved on day 181), or control devices (black, n = 5, two mice were sacrificed on day 96 and day 128 due to poor health, the other three devices were retrieved on day 181); ****p < 0.0001 (one-way analysis of covariance (ANCOVA)). d IPGTT on day 58; mean ± SD; ****p < 0.0001 (diabetic mice versus SONIC device-treated mice, diabetic mice versus healthy mice), n.s. (p = 0.5591; SONIC device-treated mice versus healthy mice). e IPGTT on day 180; mean ± SD; ****p < 0.0001 (diabetic mice versus SONIC device-treated mice, diabetic mice versus healthy mice, control device-treated mice versus SONIC device-treated mice, and control device-treated mice versus healthy mice), n.s. (p = 0.6667; diabetic mice versus control device-treated mice), n.s. (p = 0.9966; SONIC device-treated mice versus healthy mice). Statistical tests in d and e were analyzed via a two-way analysis of variance (ANOVA) followed by Sidak’s post hoc p-value adjustment for multiple comparisons. f Static GSIS test of devices (n = 3) retrieved on day 60; mean ± SD, ***p = 0.0006 (paired two-sided students t-test). g Live/dead staining of islets from a retrieved SONIC device on day 60. One representative of 2 replicates is shown. h, i H&E and immunohistochemical staining of tangential cross-sections of retrieved devices on day 60 showing intact morphology and insulin/glucagon-positive islets in both peripheral regions (h) and central regions (i) of the device. One representative of 3 replicates is shown. j H&E staining of the transverse section of a retrieved device on day 181 showing healthy islets in both peripheral regions and central regions of the device. The asterisk indicates the host side of the device-host interface. One representative of 2 replicates is shown. k A selected transverse surface plot collected from the simulation device showing well-oxygenated islets in both peripheral regions and central regions. l H&E staining of a retrieved device on day 181 showing healthy islets in the device even with some deposited fibrosis on the device. The asterisk indicates the host side of the device-host interface. One representative of 4 replicates is shown. m Selected surface plots collected from the simulation of devices with fibrosis on one half of the device (implemented by a no-flux condition on the bottom half). The PLA control device (left) showed a significantly lower pO₂ at the blocked bottom side in comparison to the unblocked top side, and a necrotic islet (yellow arrow) was observed near the blocked face. The SONIC device (right) showed a slighter lower pO₂ at the blocked bottom side in comparison to the unblocked top side, but the islets (blue arrows) at both sides were sufficiently oxygenated.

Fig. 6. Demonstration of the therapeutic potential of a thick SONIC device.

a, b Schematics representing the simulated geometry for a thick control device (a) and thick SONIC device (6.6 × 6.6 × 6.6 mm) (b) containing rat islets (size-distributed). Simulation-predicted pO₂ distributions of the control device in three cross-sections (labeled A-A, B-B, and C-C) showing a massive hypoxic central region, with necrosis observed even in some small islets (arrows in C-C cross-section). However, the islets were well-oxygenated throughout the SONIC device with negligible necrosis. c, d Average pO₂ (c) and the fraction of necrosis (d) in the islet populations in control devices (n = 20) and SONIC devices (n = 20) at the size of 4.2 mm and 6.6 mm, mean ± SD. e, f Scatter plots of islet location versus average pO₂ of islets (generated as all 150 µm in diameter) in the thick control device (e) and thick SONIC device (f). g, h Schematic (g) and microscope image (h) of rat islets encapsulated in a thick cubic SONIC device with a side length of ~6.6 mm. One representative of 5 replicates is shown. i BG measurements of diabetic C57BL6/J mice following IP transplantation of the 6.6 mm SONIC devices (pink, n = 5, retrieved on day 60; red, n = 5, retrieved on day 123). j IPGTT on day 58; mean ± SD; **p = 0.0066 (two-way ANOVA). k Live/dead staining of islets from a retrieved device on day 120. l, m H&E (l) and immunohistochemical staining (m) of retrieved devices on day 120 showing morphology-intact and insulin/glucagon-positive islets. One representative of 5 replicates is shown.

Fig. 7. Computational exploration of a SONIC spiral device for delivering a clinically relevant islet dose.

a Annotated schematic of the central section of a hypothetical SONIC spiral device, including the SONIC scaffold (blue) arranged in an Archimedean spiral (with the distance between turns fixed at 500 μm) and hydrogel-encapsulated human islets (red). A thickness of 1.2 mm ensures a maximum distance of insulin diffusion of 600 μm; a diameter of 4.75 mm was used for simulations, representing the central section of a device scaled radially to achieve a sufficient encapsulated islet payload. b Schematics showing the SONIC spiral device and scaffold-free control device encapsulating 4%, 6%, 8%, and 10% human islets (volume of islets per volume of device), as tested in the simulations. c, d Simulation predictions of the mean islet population pO₂ (c) and fraction of necrosis (d) of human islets encapsulated at variable densities in the SONIC spiral device (n = 3) and the scaffold-free control device (n = 3); mean ± SD; c: ****p < 0.0001 (control device versus SONIC spiral device at all islet densities); d: ***p = 0.0004 (control device versus SONIC spiral device at 4% islet density), **p = 0.0018 (control device versus SONIC spiral device at 6%), ****p < 0.0001 (control device versus SONIC spiral device at 8% and 10% islet densities). Statistical tests in c and d were analyzed via a two-way ANOVA followed by Sidak’s post hoc p-value adjustment for multiple comparisons. e, f Surface plots of pO₂ gradients (right) at selected cross-sections (left; labelled A-A, B-B, and C-C) in the control device (e) and the SONIC spiral device (f) at 8% human islet loading density showing significantly higher pO₂ and negligible necrosis in the SONIC spiral device compared to the control device (white regions in the islets represent necrosis, and yellow arrows in C-C section indicate necrosis even in some small islets).