Magnetic Retrieval of Encapsulated Beta Cell Transplants from Diabetic Mice Using Dual‐Function MRI Visible and Retrievable Microcapsules

Encapsulated beta cell transplantation offers a potential cure for a subset of diabetic patients. Once transplanted, beta cell grafts can help to restore glycemic control; however, locating and retrieving cells in the event of graft failure may pose a surgical challenge. Here, a dual‐function nanoparticle‐loaded hydrogel microcapsule is developed that enables graft retrieval under an applied magnetic field. Additionally, this system facilitates graft localization via magnetic resonance imaging (MRI), and graft isolation from the immune system. Iron oxide nanoparticles encapsulated within alginate hydrogel capsules containing viable islets are transplanted and the in vitro and in vivo retrieval of capsules containing nanoparticles functionalized with various ligands are compared. Capsules containing islets co‐encapsulated with COOH‐coated nanoparticles restore normal glycemia in immunocompetent diabetic mice for at least 6 weeks, can be visualized using MRI, and are retrievable in a magnetic field. Application of a magnetic field for 90 s via a magnetically assisted retrieval device facilitates rapid retrieval of up to 94% (±3.1%) of the transplant volume 24 h after surgical implantation. This strategy aids monitoring of cell‐capsule locations in vivo, facilitates graft removal at the end of the transplant lifetime, and may be applicable to many encapsulated cell transplant systems.


DOI: 10.1002/adma.201904502
For patients with diabetes, islet transplantation can help to restore insulin secretion and long-term normoglycemia, [1,2] offering a potential alternative to daily insulin injections. In this procedure, islets are isolated from a donor pancreas and are typically transplanted into the liver via hepatic portal vein infusion. [3,4] This often requires patients to take systemic immunosuppressants to prevent graft rejection. Additionally, retrieval of these grafts from the liver in the event of graft failure remains surgically challenging and risks injury to the host. [5] In an alternative approach, cell and organoid grafts can be encapsulated within hydrogel materials prior to transplantation. [6][7][8][9] The hydrogels help to physically isolate the graft from the host immune system, reducing the need for systemic immunosuppression, and limit cellular rejection following transplantation. [10,11] Recently, chemically modified hydrogels have been developed that can also reduce Encapsulated beta cell transplantation offers a potential cure for a subset of diabetic patients. Once transplanted, beta cell grafts can help to restore glycemic control; however, locating and retrieving cells in the event of graft failure may pose a surgical challenge. Here, a dual-function nanoparticleloaded hydrogel microcapsule is developed that enables graft retrieval under an applied magnetic field. Additionally, this system facilitates graft localization via magnetic resonance imaging (MRI), and graft isolation from the immune system. Iron oxide nanoparticles encapsulated within alginate hydrogel capsules containing viable islets are transplanted and the in vitro and in vivo retrieval of capsules containing nanoparticles functionalized with various ligands are compared. Capsules containing islets co-encapsulated with COOH-coated nanoparticles restore normal glycemia in immunocompetent diabetic mice for at least 6 weeks, can be visualized using MRI, and are retrievable in a magnetic field. Application of a magnetic field for 90 s via a magnetically assisted retrieval device facilitates rapid retrieval of up to 94% (±3.1%) of the transplant volume 24 h after surgical implantation. This strategy aids monitoring of cell-capsule locations in vivo, facilitates graft removal at the end of the transplant lifetime, and may be applicable to many encapsulated cell transplant systems.
foreign body responses and associated material fibrosis. [8,12] This has enabled the transplant of cell and organoid therapies to a wider range of accessible extra-hepatic transplant sites in small animal models and non-human primates. [8,9] As these cell and organoid transplantation systems are translated to humans, it is increasingly important to develop strategies to monitor the location of transplanted and encapsulated cell systems. There may also be a need for methods that facilitate removal of transplanted grafts in the event of graft failure post-transplantation. This is of particular interest for capsulebased micro-encapsulation systems, where a curative transplant would require many capsules containing encapsulated islets. [13,14] To address this challenge, we sought to develop methods that enable both monitoring and retrieval of encapsulated cell and organoid therapies.
Transplanted islets have previously been tracked in vivo through cellular labeling with fluorescent dyes, nanoparticlebased contrast agents, or radiolabels. [15][16][17][18][19][20][21][22][23] Co-encapsulation of contrast agents within hydrogels capsules has also been used to locate encapsulated islets via magnetic resonance imaging (MRI) in vitro [21,24] and recently, to track the movement of unconstrained capsules implanted in vivo. [25] Although these strategies facilitate localization of implanted capsules, they do not directly facilitate the retrieval of such implanted grafts. We have developed a dual-function hydrogel capsule, which can facilitate graft retrieval under a directed magnetic field and aids graft localization via MRI.
Here, we show the first example of an encapsulated cell therapy that can be magnetically retrieved following transplantation. Our system provides a single integrated approach for both in vivo capsule localization and retrieval. We demonstrate the therapeutic use of these materials to transplant encapsulated rat islets into immunocompetent diabetic mice. These technologies provide a tool for the encapsulation of a range of functional cells and cell organoids with dual imaging/retrieval capabilities and facilitate graft monitoring and graft removal at the end of the transplant lifetime.
Alginate hydrogels can be used for cell encapsulation and transplant. Hydrogel capsules can be formed through electrostatic droplet generation, followed by divalent cation crosslinking of guluronic acid residues on the alginate polymer backbone, shown schematically in Figure 1A. [26] To develop hydrogel capsules that can be magnetically retrieved, we incorporated iron oxide nanoparticles coated with different functional groups into the hydrogel aqueous phase before droplet generation (detailed in Experimental Section). We hypothesized that interactions between functional groups on the surface of the nanoparticles, carboxylic acid groups in the guluronate block within alginate, and divalent cations used during alginate crosslinking may stabilize nanoparticle-hydrogel interactions. [27] This could enable production of hydrogel capsules that respond to magnetic fields and are suitable for co-encapsulation with viable cells for long-term implantation and retrieval in vivo.
Hydrogel capsules with iron oxide concentrations between 0 and 5 mg mL -1 were made at two clinically relevant capsule sizes (0.5 and 1.5 mm) [9,14] and the mobility of these capsules in a magnetic field was tested. To explore capsule response to magnetic fields, we measured the distance at which capsules containing nanoparticles were able to move through saline (against gravity) toward a magnet at a fixed distance ( Figure 1E-H; Video S1, Supporting Information). Briefly, capsules were placed in the bottom of a syringe filled with saline, allowed to settle, and then a magnetic plunger was slowly lowered in the syringe. The distance from the magnet at which 1 × 1.5 mm capsule or 2 × 0.5 mm capsules moved toward the magnet was recorded. All three nanoparticle-loaded systems were magnetically responsive. As expected, the distance that capsules could cross toward a magnet increased as nanoparticle concentration increased, demonstrating that the retrieval distance is proportional to nanoparticle concentration at loadings between 0.1 and 5 mg mL -1 , with responsiveness ranked as NP-COOH > NP-PEG > unfunctionalized NP. In all cases, at loading densities above 1 mg mL -1 , capsules could be collected on magnets positioned 1-2 cm away in vitro, which suggested a dynamic retrieval range suitable for magnetically assisted retrieval of capsules surgically implanted in mice. At this concentration, the NP-COOH capsules are significantly more responsive (can be retrieved at greater distances) than the NP or NP-PEG systems. There was no statistical difference in retrieval distance between large and small capsules ( Figure 1H; Figure S1, Supporting Information).
We investigated the stability and mechanical properties of hydrogels containing various nanoparticles to determine suitability for cell encapsulation and potential transplantation. We compared the hydrogel morphology ( Figure 1B), electrostatic interactions ( Figure 1C,D), mechanical properties ( Figure S2, Supporting Information), and long-term nanoparticle retention ( Figure S3, Supporting Information). In general, nanoparticle-loaded alginate capsules can be formed up to a nanoparticle loading density of 1-2.5 mg mL -1 . Above this threshold, there is some disruption to the spherical morphology of the capsules, inducing "tail" formation ( Figure 1B, indicated by the white circles). We evaluated storage (Gʹ) and loss modulus (Gʺ) of different gels in response to strain using a rheometer across the frequency range of 0.2-150 rad s -1 . This allows us to evaluate whether inclusion of nanoparticles significantly alters the mechanical properties of the hydrogels. In all systems, viscoelastic hydrogels were formed, with Gʹ an order of magnitude greater than Gʺ ( Figure S2, Supporting Information).
We used a spectroscopic approach to assess nanoparticle leaching and the long-term stability of these nanoparticleloaded hydrogel systems. Iron oxide nanoparticles strongly www.advmat.de www.advancedsciencenews.com absorb in the 300-400 nm wavelength range; we assessed UVvis absorbance of supernatants taken from gels incubated in calcium-supplemented saline for up to 6 months ( Figure S2, Supporting Information). NP and NP-COOH capsules were stable, with no detectable nanoparticle leaching over a 6 month timeframe. In contrast, leaching of NP-PEG from the NP-PEG hydrogel capsules occurred over a 6 month timeframe (<0.2 mg mL -1 , Figure S3, Supporting Information). These data suggest that the NP-PEG alginate hydrogel system is not stable, and that NP-PEG are able to move within and out of the hydrogel matrix over time.
As capsule formation occurs via an electrostatic droplet generation system; it is possible that changes in the electrostatic interactions of the alginate/nanoparticle mixture may affect capsule integrity. Zeta potential of nanoparticles in alginate solutions ( Figure 1D) and conductivity studies ( Figure 1C Figure 1. Properties of iron oxide nanoparticle and nanoparticle-loaded hydrogels. A1) Schematic of nanoparticle-loaded hydrogel capsule formation and A2) application in diabetes transplants with MRI imaging and magnetic retrieval capabilities. B) TEM images of the different nanoparticle systems used and microscopy images of alginate hydrogel capsules containing nanoparticles at loading densities from 0.25 to 5 mg mL -1 . C) Conductivity and D) zeta potential of nanoparticles in saline or a saline-alginate aqueous phase. Statistical analysis performed using one-way ANOVA with multiple comparisons, statistics represent comparison to nanoparticles suspended in alginate. E-H) Magnetic retrieval of nanoparticle-loaded hydrogels in saline against gravity. E) Video stills of the magnetic retrieval process (stills time = 0, time = 16 s, and time = 22 s). F) Comparison between retrieval distances (the distance at which the first capsule moves against gravity toward the magnet) for NP, NP-COOH, and NP-PEG 1.5 mm alginate hydrogel capsules at concentrations between 0.25 and 5 mg mL -1 . G,H) Average retrieval distance for 1.5 mm capsules at 1 mg mL -1 iron oxide loading, and a comparison between capsules of different sizes. Statistical analysis performed using one-way ANOVA with multiple comparisons. All graphs show mean values ± SEM with *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. www.advmat.de www.advancedsciencenews.com NP-COOH systems that did not alter alginate conductivity. COOH-functionalized nanoparticles may also interact electrostatically via complexation between the divalent cation and COOH groups on the nanoparticle and alginate backbone to form a more stable hydrogel system better suited for cell encapsulation ( Figure 1A, schematic).
We evaluated the effect of various iron oxide nanoparticles on primary rat islets in vitro. Islets were co-cultured with various nanoparticle systems (no NP, unfunctionalized NP, NP-COOH, or NP-PEG) for 48 h. Islet viability was then evaluated using Calcein AM staining and metabolic activity using a luciferase ATP assay. We found that the percentage of viable cells per islet remains consistent across these conditions ( Figure 2B; median 94.0% in untreated cells; and 92.0%, 94.5%, and 95.5% for NP, NP-COOH, and NP-PEG, respectively) compared to islets treated with ethanol for 1.5 h (28.0% viable). We also compared the metabolic activity of islets following treatment with nanoparticles, using a luciferase assay to quantify ATP activity within cells. Figure 2C illustrates that there is a slight (nonsignificant) decrease in mean ATP activity within islets (19.5%, 25.1%, and 24.0% decrease in ATP activity for NP, NP-COOH, and NP-PEG, respectively) following incubation with nanoparticles.
Next, islets were encapsulated in alginate systems containing the three different iron oxide nanoparticles at 1 mg mL -1 ,  Islets were encapsulated in nanoparticle-loaded alginate hydrogels and exposed to low (2 mM) then high (20 mM) glucose. D) Insulin secretion 24 h post-encapsulation was measured and E) the resulting stimulation index was analyzed to determine the fold change in insulin secretion on exposure to low and high glucose concentrations. Statistical analysis was performed using one-way ANOVA with multiple comparisons; statistics in (B,C) represent comparison to untreated islets using paired t-tests (GSIS). D,E) Comparisons between low and high insulin, compared to islets encapsulated in alginate alone. All graphs show mean values ± SEM with *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 and represent mean values from three to five independent experiments.
www.advmat.de www.advancedsciencenews.com and the secretion of insulin in response to glucose stimulation was monitored. Encapsulated islets were exposed to basal low (2 mM) and high glucose (20 mM) solutions and the stimulation index was calculated at 24 h post-encapsulation. In the absence of nanoparticles, islets demonstrated a 5.5-fold increase in insulin secretion following exposure to low and high glucose conditions post-encapsulation. Islets coencapsulated with NP and NP-COOH showed a similar trend (5.1-fold and 5.2-fold increase, respectively). Islets co-encapsulated with NP-PEG showed a decreased stimulation index at 24 h (stimulation index 2.6), suggesting impaired cellular function. Earlier, we found that NP-PEG leached from capsules ( Figure S2, Supporting Information). PEG ligands are often included in nanoparticle delivery systems due to their ability to improve nanoparticle circulation times and facilitate penetration of specific biological barriers. [28][29][30][31][32] It is possible that the disruption to beta cell function may be related to NP-PEG leaching or cellular penetration; however, further studies on the interaction between islets and nanoparticles would be required to understand the precise biological mechanism driving this effect. In all cases, islets encapsulated with nanoparticles demonstrated suppressed insulin secretion and ATP activity compared to islets alone, although this was not statistically significant.
The inclusion of iron oxide nanoparticles within the alginate capsule renders the capsules responsive to magnetic fields. We, therefore, evaluated whether hydrogel capsules containing iron oxide nanoparticles could also be visualized post-implantation using MRI, allowing for localization of transplanted capsules (Figure 3A, schematic). First, we fabricated 0.5 mm alginate capsules containing iron oxide nanoparticles with nanoparticle concentrations between 0.05 and 1.00 mg mL -1 ( Figure 3B) and embedded these capsules in agarose gels, to stabilize capsule location, before MRI imaging. Capsules were visible via MRI at all concentrations tested.
Next, we tested whether capsules loaded with 1.00 mg mL -1 iron oxide in two capsule diameters (0.5 and 1.5 mm) were visible once implanted into the intraperitoneal (IP) space of mice. Figure 3B illustrates example MRI images, with several 0.5 and 1.5 mm capsules highlighted using insets; capsules of both sizes were visible using MRI as hypointense regions. We also compared visibility of large capsules containing iron oxide at lower concentrations of 0.25, 0.50, and 1.00 mg mL -1 ( Figure 3C). Results show that in contrast to capsules embedded in agarose gel systems, capsules implanted in vivo in the IP space are not easily identifiable at lower iron oxide concentrations of 0.25 mg mL -1 . In contrast, 1.5 mm capsules containing iron oxide nanoparticles at concentrations above 0.50 mg mL -1 can be located in vivo using this method. The stronger magnetic susceptibility artifacts present at higher iron oxide concentrations make the capsules appear larger in size, thus further facilitating their identification. [25] The identification of these capsules using MRI requires radiographic expertise, and we expect iron oxide concentrations may need to be carefully selected to optimize imaging with other MRI machines with different magnet strengths.
Our previous results indicated alginate capsules loaded with iron oxide nanoparticles at 1 mg mL -1 could be retrieved on a magnet placed 1-2 cm away ( Figure 1F), located in vivo using MRI ( Figure 3C), and were generally well tolerated by islets in vitro ( Figure 2B-D). To evaluate the clinical utility of this system, we encapsulated rat islets in these nanoparticle alginate systems followed by transplantation into an immune competent diabetic transplant model (Figure 4A, schematic). We then monitored graft function and animal glycemic control, before determining graft location using MRI and subsequent in vivo retrieval of encapsulated rat islets ( Figure 4B-G).
A total of 500 rat islets were encapsulated in nanoparticleloaded alginate hydrogel capsules containing 1 mg mL -1 of NP-PEG or NP-COOH ( Figure 4B). These capsules were spherical and nanoparticles appeared well distributed inside ( Figure S5, Supporting Information). We transplanted these capsules into the IP space in STZ-induced diabetic mice and show these can be easily identified via MRI ( Figure 4B). As expected, capsules containing no nanoparticles, and therefore no contrast agent, have limited visibility using MRI. However, capsules containing islets with both the NP-PEG and NP-COOH systems can be visualized as hypointense regions using MRI, allowing us to monitor the anatomical distribution of capsules following transplant. Although individual capsules can be located within the abdominal cavity, their distinction from other hypointense structures, and indeed clusters of capsules in the peritoneal cavity, is nevertheless challenging and requires radiographic expertise.
Once capsules had been localized to within the IP cavity, we sought to retrieve them through a magnetically assisted surgery. Briefly, we developed a new device for magnetically assisted capsule retrieval in vivo, consisting of a peristaltic flushing system surrounded by an annular magnet to produce a combined mechanical and magnetic retrieval system ( Figure 4C). We surgically implanted 200 nanoparticle-loaded alginate capsules (NP) into the IP space of mice, closed this incision, and in a second surgery (performed within 24 h of initial implantation) used our device to flush and retrieve capsules, which were then counted. Application of the device for as little as 90 s facilitated retrieval of 94% (±3.1% SD) of the iron oxide-loaded magnetic capsules ( Figure 4C).
In a separate experiment, we transplanted capsules containing islets in alginate capsules loaded with unfunctionalized NP, NP-COOH, or NP-PEG into diabetic animals ( Figure 4D-G) and successfully surgically retrieved these capsules ( Figure 4D,E) following application of a magnetic field for up to 90 s. The inclusion of magnetic nanoparticles and the use of our device aid magnetic surgical retrieval; capsules containing magnetic nanoparticles (NP or NP-COOH functionalized) were retrieved at significantly higher yields compared to plain alginate capsules ( Figure 4D). Capsules remain intact post-retrieval ( Figure S5, Supporting Information). To evaluate whether these systems could be magnetically retrieved following longer term implantation, we also compared the retrieval of capsules at the 6-8 week timepoint and at 4 months ( Figure 4E). We find that capsules can be recovered from the abdominal cavity at later timepoints, and that the capsule volume recovered is not significantly different between 6-8 weeks and 4 months using application of a magnetic field for 90 s.
In general, capsules containing iron oxide nanoparticles could be retrieved in greater yields than capsules containing no iron oxide nanoparticles (increase in retrieval yield over plain www.advmat.de www.advancedsciencenews.com alginate NP: 18.7%, NP-COOH: 44.2%, NP-PEG: 50.6% across all conditions, all timepoints). Surgical retrieval of implanted microcapsules may be influenced by the microcapsule material and any foreign body responses, the location and duration of implantation, and the surgical approach and expertise of the surgeon performing the retrieval. With further optimization and surgical training, it may be possible to retrieve an even greater proportion of transplanted capsules in this short time frame, reducing the time required for surgical retrieval of encapsulated cell systems.
We evaluated the effect of nanoparticle-loaded alginate capsules on a curative dose of rat islets (500 per mouse) used to treat C57-B6 diabetic mice. Blood glucose levels were evaluated for 6-8 weeks post-transplant to determine whether these transplants could restore insulin secretion and normoglycemia in vivo in mice ( Figure 4F,G). Figure 4G shows blood glucose measurements of animals following transplant of 500 rat islets encapsulated in alginate or nanoparticle alginate capsules. Figure 4F  www.advmat.de www.advancedsciencenews.com following transplant of 500 rat islets encapsulated in alginate or nanoparticle alginate capsules. Three consecutive measurements above 250 mg dL -1 post-transplant surgery were considered "transplant failure"; [33,34] the cure rate, therefore, describes how long animals retained normal glycemia (remained cured) until transplant failure.
For NP and NP-COOH systems, islets encapsulated in nanoparticle-loaded alginate hydrogels were able to restore normal glycemia in immunocompetent diabetic mice without immunosuppression for up to 6-8 weeks, with over 80% of mice remaining normoglycemic at day 42. However, the islets co-encapsulated with NP-PEG demonstrated significantly impaired function compared to the other systems. NP-PEG systems were able to reduce blood glucose levels to approximately 200 mg dL -1 for the first 4-5 weeks, before blood glucose levels begin to rise. After 6 weeks, the capsules were retrieved as described earlier and blood glucose measurements were monitored for an additional 3-5 days before termination of the experiment. Blood glucose levels increased immediately after removal of the capsules containing islets, confirming that Adv. Mater. 2020, 32,1904502  implanted into C57BL6 mice. Capsules are indicated with red lines-only nanoparticle-loaded capsules were visible. C) Schematic of devices used to test magnetic properties of capsules and retrieval results; 200 NP capsules were implanted into the IP space of C57B6 mice. In a separate surgery performed within 24 h, our magnetic device was applied for up to 90 s and retrieval assessed (three to four experiments). Graph represents the proportion of iron oxide-loaded capsules that could be collected within 90 s. D) Average volume of capsules retrieved after 6-8 weeks in unfunctionalized NP, NP-COOH, and NP-PEG systems using magnetic retrieval. E) Aggregated data comparing average retrieval volume of magnetic systems (NP, NP-PEG, and NP-COOH) after implantation for 6-8 weeks, or 4 months, with application of the magnetic retrieval device limited to 90 s. D,E) All graphs show mean values ± SEM with *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 and represent mean values from 4-10 mice per group. Statistics show a one-way ANOVA with multiple comparisons between individual conditions, compared to unmodified alginate capsules. F,G) A total of 500 rat islets were encapsulated in alginate, alginate NP, alginate NP-COOH, or alginate NP-PEG systems and transplanted into the IP space of STZ-induced diabetic C57B6 mice. F) Percentage normal glycemia in various encapsulation systems. A curative euglycemia threshold of 200 mg dL -1 was applied, [12] with failure defined as three consecutive blood glucose measurements above 250 mg mL -1 . [33,34] G) Average blood glucose levels following transplantation of rat islets encapsulated in alginate, or alginate containing nanoparticles. F,G) Experiments were grouped and conducted two to three times and graphs represent average of 8-12 animals. Statistical analysis was performed using a Mantel-Cox survival curve analysis.
For a subset of type I diabetic patients, islet microencapsulation has the potential to enable transplantation of cells and organoids to extra-hepatic transplantation sites, and reduce the need for systemic immune suppression. However, monitoring and retrieving grafts from these sites may prove challenging. Here, we develop a dual-function nanoparticle-loaded hydrogel capsule that facilitates magnetically assisted surgical retrieval at the end of transplant lifetimes and enables graft localization via MRI.
We have explored capsules loaded with iron oxide nanoparticles functionalized with various ligands (unfunctionalized NP, NP-COOH, or NP-PEG) and compared the properties of these three systems, focusing on islet functionality, capsule tracking via MRI, and retrievability. The inclusion of iron oxide nanoparticles within alginate capsules enables magnetic retrieval of capsules in vitro and in vivo, with the NP-COOH system the most easily retrieved in vitro. When coupled with application of a magnetic field for 90 s using a magnetically assisted retrieval device, nanoparticle-loaded hydrogels facilitate rapid retrieval of up to 94% (±3.1%) of the transplant volume 24 h post-implantation.
Following transplantation, the inclusion of the iron oxide nanoparticles in the capsule structure at concentrations above 1 mg mL -1 facilitated tracking of encapsulated islets in vivo as identifiable, distinct hypointense structures in the IP space using MRI imaging. Islets encapsulated within nanoparticleloaded alginate hydrogels using NP and NP-COOH systems were also able to restore normal glycemia in immunocompetent diabetic mice without immunosuppression for up to 6-8 weeks in vivo. These capsules remained stable for up to 6 months in vitro, with no detectable nanoparticle leaching. In contrast, although PEG-functionalized nanoparticle-loaded gels were also retrievable, nanoparticle leaching was observed from these hydrogels after a 6 month time frame, and islet function was impaired in vitro and in vivo, making them unsuitable for translation. We have, therefore, identified islets encapsulated in alginate hydrogels loaded with 1 mg mL -1 of carboxylated iron oxide nanoparticles (NP-COOH) as an optimal system for magnetic retrieval of microencapsulated islets.
Nanoparticle Characterization: NP, NP-COOH, and NP-PEG were imaged by TEM. Briefly, nanoparticles were loaded onto carbon stubs and imaged in a JEOL 2100 FEG TEM at 120 kV. Dynamic light scattering and zeta potential were measured on a Malvern Zetasizer. Briefly, NP, NP-PEG, or NP-COOH were suspended in saline or a 0.1% alginate in saline mixture and loaded into capillary cuvettes before analytical measurements of zeta potential, conductivity, and particle size.
Microcapsule Fabrication and Testing: Alginate capsules were formed as previously described. [8,35] In brief, electrostatic droplet generation at 0.1-0.2 mL min -1 under a voltage of 5-10 kV was used to form capsules using a PicoPump syringe pump. Capsules were crosslinked in a 20 mM BaCl 2 solution, washed in HEPES or in Krebs buffer, and stored in 0.9% saline containing 2 mM CaCl 2 (Ca 2+ -supplemented saline). To generate iron oxide-loaded capsules, iron oxide nanoparticles coated with different functional groups were incorporated into the hydrogel aqueous phase before droplet generation. Nanoparticle leaching was determined by incubating capsules containing NP, NP-PEG, or NP-COOH iron oxide nanoparticles in 2 mM calcium-supplemented saline at 37 °C for up to 6 months. Supernatants were taken at regular intervals and absorbance was assessed against a nanoparticle standard curve using a Tecan plate reader (300-400 nm). Mechanical properties were assessed in bulk hydrogels formed using the nanoparticle-loaded aqueous phase for NP, NP-PEG, and NP-COOH systems. Briefly, hydrogel disks were crosslinked as described, and parallel plate rheometry was performed at 0.1% strain across the frequency range of 0.2-150 rad s -1 .
Magnetic Distance Testing: A cylindrical, 1/4″ by 3/4″-diameter rareearth metal magnet (K&J Magnetics, Plumsteadville, PA, USA) was glued to a syringe plunger with plugged nozzle. Alginate capsules, in ≈10-15 mL of storage buffer, were layered inside the syringe body. The plunger was submerged in the liquid without disturbing the capsules or trapping air bubbles and gradually lowered until one capsule (for 1.5 mm capsules) or two capsules (for 0.5 mm capsules) rose toward the magnet. The distance from the bottom of the magnet to the top of the capsules was recorded as the ">0%" magnetic response distance. The plunger was then lowered incrementally until all the capsules had risen to the magnet ("100%" distance).
Islet Encapsulation and Functional Tests: For islet encapsulation, rat islets were isolated as previously described. [35] Briefly, islets were washed in calcium-free Krebs (4.7 mM KCl, 0.58 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 25 mM HEPES, and 135 mM NaCl), and encapsulated within 12 h postisolation. Capsules containing islets were stored in media in an incubator overnight, and islet function was assessed using glucose-stimulated insulin secretion (GSIS). Briefly, capsules were washed, and exposed to low KR2 (KR0 + 2 mM glucose) for 30 min, washed, and then exposed to high KR20 (KR0 + 20 mM glucose) for 30 min. Insulin production was measured by ELISA. Cell viability of islets was assessed post-isolation, post-encapsulation, and post-retrieval by dual fluorescence staining with the inclusion/exclusion dyes fluorescein diacetate (FDA; Sigma) for live cells and propidium iodide (PI; Sigma) for dead cells. Briefly, naked or encapsulated islets were rinsed twice with 10 mL HBSS (Mediatech) and then mixed with FDA and PI in HBSS. For immunohistological analysis, capsules were fixed in formalin post-retrieval and stained with Newport Green dye (a zinc/insulin dye) and imaged. A fluorescence microscope with bright field view, plus filters for FDA or Newport Green (excitation wavelength 488 nm, emission wavelength 520 nm), and PI (excitation wavelength 534 nm, emission wavelength 617 nm) were used to assess the viability of the islets, and to image encapsulated islets and capsules. Percentages of total viable cells within 25-50 whole islets were estimated by a single operator trained in islet isolation protocols.
Microcapsule Transplantation and Retrieval: Animal procedures were approved by the MIT Committee on Animal Care. STZ-induced diabetic C57BL/6 mice were purchased from Jackson Laboratory. Surgery was performed under isoflurane anesthesia and postoperative buprenorphine. Briefly, capsules were infused into the abdominal cavity through an abdominal incision and closed with surgical sutures, tissue glue, and wound clips. Mice were monitored post-surgery, and blood glucose measurements were recorded using a tail prick and AlphaTRAK commercial glucose meter (Zoetis, Kalamazoo, MI, USA). A euglycemia threshold of 200 mg dL -1 was applied, [12] with failure defined as three consecutive blood glucose measurements above 250 mg mL -1 . [33,34] Both survival and euthanized retrieval surgeries were performed. The incision site was opened and IP space was flushed with Krebs solution to enable www.advmat.de www.advancedsciencenews.com Adv. Mater. 2020, 32,1904502 capsule collection. For nonterminal retrievals, mice were anesthetized as described above. A surgical flushing instrument aided mechanical and magnetic retrieval of capsules. An annular magnet (i.e., 1/8 in., o.d. 1/4 in. × 1/4 in.) was fixed to a pipette hose and peristaltic pump (VWR, Radnor, PA, USA). Sterile saline was used to flush and retrieve capsules. Magnetic capsules loosely attached to the magnet, and could be flushed off into a collection chamber.
MRI Imaging: Capsules were embedded into agarose gels prior to in vitro imaging, or were transplanted into the IP cavity of mice as described. Mice were anesthetized and placed in the supine position within a volume coil of a 7T preclinical MRI system (Varian 7T/310/ ASR, Agilent). Vital signs and temperature were continuously monitored. Respiratory gating was performed for motion artifact correction. Images were collected using 1 mm thick coronal slices of 50 mm × 50 mm field of view. A fast spin echo pulse sequence (TR = 2000 ms, TE = 12.7 ms, data matrix: 256 × 256, 2 averages) was used. Data were stored in DICOM format and was visualized using MATLAB.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.