Engineering Vascularized Islet Macroencapsulation Devices: An in vitro Platform to Study Oxygen Transport in Perfused Immobilized Pancreatic Beta Cell Cultures

Abstract

Islet encapsulation devices serve to deliver pancreatic beta cells to type 1 diabetic patients without the need for chronic immunosuppression. However, clinical translation is hampered by mass transport limitations causing graft hypoxia. This is exacerbated in devices relying only on passive diffusion for oxygenation. Here, we describe the application of a cylindrical in vitro perfusion system to study oxygen effects on islet-like clusters immobilized in alginate hydrogel. Mouse insulinoma 6 islet-like clusters were generated using microwell plates and characterized with respect to size distribution, viability, and oxygen consumption rate to determine an appropriate seeding density for perfusion studies. Immobilized clusters were perfused through a central channel at different oxygen tensions. Analysis of histological staining indicated the distribution of viable clusters was severely limited to near the perfusion channel at low oxygen tensions, while the distribution was broadest at normoxia. The results agreed with a 3D computational model designed to simulate the oxygen distribution within the perfusion device. Further simulations were generated to predict device performance with human islets under in vitro and in vivo conditions. The combination of experimental and computational findings suggest that a multichannel perfusion strategy could support in vivo viability and function of a therapeutic islet dose.

Keywords: artificial vascularization; beta cells; encapsulation; immobilized culture; oxygen mass transport; oxygen model; type 1 diabetes.

FIGURE 1

Mouse insulinoma 6 (MIN6) aggregate size distribution and viability. Bar graphs show the size distribution and volume mean moment (De Brouckere mean diameter) D [4,3] with standard deviation SD of MIN6 islet-like clusters (ILCs) generated using AggreWell™400 microwell plates. Fluorescence images show representative live/dead staining results using calcein-AM (green) and ethidium homodimer-1 (red), from which the mean percentage of live cells is determined. (A) Seeding density = 3,000 cells/well, D [4,3] = 289 (SD 65) µm, viability = 93.9% (SD 0.4%); (B) Seeding density = 2,000 cells/well, D [4,3] = 306 (SD 60) µm, viability = 93.9% (SD 2.6%); (C) Seeding density = 1,000 cells/well, D [4,3] = 269 (SD 34) µm, viability = 85.0% (SD 2.2%); (D) Seeding density = 500 cells/well, D [4,3] = 241 (SD 24) µm, viability = 84.1% (SD 0.3%); and (E) Seeding density = 200 cells/well, D [4,3] = 208 (SD 17) µm, viability = 96.6% (SD 1.2%). N = 3 biological replicates with 30 images (>400 ILCs) analyzed per condition. Error bars represent the standard deviation. Viability results were analyzed using a one-way analysis of variance and Tukey multiple comparisons post-hoc test. Conditions (C,D) were significantly different from Conditions (A,B,E) (p < 0.001). All other combinations were not significantly different (p > 0.05). Scale bar = 200 µm.

FIGURE 2

Mouse insulinoma 6 (MIN6) oxygen consumption rate (OCR) normalized to DNA content. OCR was measured for MIN6 islet-like clusters (ILCs) generated using different seeding densities in AggreWell™400 microwell plates. Conditions are SC = single-cell suspension (control); A = 3,000 cells/ILC; B = 2,000 cells/ILC; C = 1,000 cells/ILC; D = 500 cells/ILC; E = 200 cells/ILC. Using a one-way analysis of variance, no significance difference was observed between the means of conditions (p = 0.5041). N = 3 biological replicates. Error bars represent the standard deviation.

FIGURE 3

Parametric analysis for a computational model of a single-channel macroencapsulation device. (A) Schematic of the major device components. Cell clusters are immobilized in a hydrogel encapsulation matrix featuring a central perfusion channel. (B) Radial oxygen profiles at varying cell densities within the device. (C) Sensitivity of the model to cell fraction X. Vertical lines indicate “low” (1,000 ILCs/mL) and “therapeutic” (8,600 ILCs/mL) doses of MIN6 islet-like clusters (ILCs). The therapeutic dose is calculated by scaling a therapeutic dose of human islets to an equivalent number of MIN6 ILCs based on oxygen consumption rate (OCR). (D) Sensitivity of the model to the maximum oxygen consumption rate R _max. Vertical lines indicate the oxygen consumption rate of MIN6 ILCs (obtained experimentally) and human islets (from literature) (Papas et al., 2007a). (E) Effect of flow rate on the concentration profile. The colour legend for all panels and data legend for panels (C–E) are located at the bottom of the figure.

FIGURE 4

Macroencapsulated MIN6 behaviour after 48-h normoxic (140 mmHg O₂), arterial (76 mmHg O₂), and hypoxic (15 mmHg O₂) perfusion culture. (A) Hematoxylin and eosin (H&E) stained sections show the range of islet-like cluster (ILC) morphologies observed: Intact (smooth, rounded), Early Disruption (partial breakage), and Advanced Disruption (pervasive breakage). Scale bars = 50 µm. (B) Cleaved caspase-3 (CC3) and hypoxia inducible factor-1α (HIF-1α) staining was performed to assess apoptosis and response to hypoxia, respectively. The average mean grey value (inversely proportional to positive staining intensity) was quantified for all intact ILCs and normalized to the image background. Error bars represent the standard deviation. (C) MIN6 ILC morphology relative to distance from the perfusion channel. A box-and-whisker plot illustrates the distribution of Intact and Early Disruption ILCs for each oxygen condition. Abbreviations: Int = Intact, ED = Early Disruption, Norm = Normoxia (140 mmHg O₂), Art = Arterial (76 mmHg O₂), Hyp = Hypoxia (15 mmHg O₂). N > 5 ILCs for each category except Early Disruption (Arterial), where n = 2. The whiskers extend from the minimum to the maximum distance measured from the perfusion channel. Total perfusion culture data is collected from N = 3 (arterial), N = 4 (hypoxia), and N = 5 (normoxia) independent biological replicates. The embedded table indicates the position of the theoretical 7 mmHg (blue dashed line) and 0.1 mmHg (red dashed line) O₂ contours determined from computational simulations. (D) Simulations of the 48-h perfusion experiments in (A–C), performed using COMSOL Multiphysics software. Images represent cross-sections taken near the middle of the perfusion device at steady state. The 2-mm perfusion channel is outlined in black at the centre of the cross-section. The colour legend indicates oxygen tension in mmHg. The white contour represents an oxygen tension of 7 mmHg and the red contour represents 0.1 mmHg, corresponding to the positions listed in the table in (C).

FIGURE 5

Anticipated effect of device design parameters on oxygen distribution. All images shown are radial cross-sections taken at the middle of the device at steady state. (A) Mouse insulinoma 6 (MIN6) islet-like clusters (ILCs) encapsulated in 2% alginate at a concentration of 1,000 ILCs/mL gel and perfused with cell culture medium under normoxic conditions (140 mmHg O₂). (B) Same parameters as (A) but substituting MIN6 ILCs with human islets exhibiting an oxygen consumption rate of 202 ± 87 nmol/min/mg (Papas et al., 2007a). (C) Same parameters as (B) but cell density increased to a therapeutic dose of 600,000 IEQ per device and perfused at an arterial oxygen tension (76 mmHg O₂) (Buder et al., 2013). (D) Same parameters as (A) with an arrangement of two perfusion channels, each 2 mm in diameter and spaced 2 mm apart. (E) Same parameters as (B) with an arrangement of two perfusion channels. (F) Same parameters as (C) with an arrangement of two perfusion channels. In each panel, a white contour represents an oxygen tension of 7 mmHg and a red contour represents 0.1 mmHg O₂. The colour legend indicates oxygen tension in mmHg O₂. Islet equivalent, IEQ = islet of diameter 150 µm.