An Atmosphere-Breathing Refillable Biphasic Device for Cell Replacement Therapy

Abstract

Cell replacement therapy is emerging as a promising treatment platform for many endocrine disorders and hormone deficiency diseases. The survival of cells within delivery devices is, however, often limited due to low oxygen levels in common transplantation sites. Additionally, replacing implanted devices at the end of the graft lifetime is often infeasible and, if possible, generally requires invasive surgical procedures. This report presents the design and testing of a modular transcutaneous biphasic cell delivery device which provides enhanced and unlimited oxygen supply by direct contact with the atmosphere. Critically, the cell delivery unit was demountable from the fixed components of the device, allowing for surgery-free refilling of the therapeutic cells. Mass transfer studies showed significantly improved performance of the biphasic device in comparison to subcutaneous controls. The device was also tested for islet encapsulation in an immunocompetent diabetes rodent model. Robust cell survival and diabetes correction was observed following a rat-to-mouse xenograft. Lastly, non-surgical cell refilling was demonstrated in dogs. These studies show the feasibility of this novel device for cell replacement therapies

Cell replacement therapies take advantage of the dynamic responsiveness and activity of cells and promise to improve treatment for a number of pathologies including endocrine disorders and hormone deficiency diseases.^[1] The encapsulation and protection of the transplanted cells from the host immune system via a semipermeable material or device is required, in many cases, to localize the cells and prevent immune rejection.^[2] While cell encapsulation can overcome several problems of cell replacement therapies such as obviating the coadministration of immunosuppressive drugs, critical limitations barring clinical translation remain.

Primary among these limitations is the lack of adequate oxygen supply to the encapsulated cells.^[3] Immunoisolation by polymer encapsulation necessarily dissociates the cells from the host vasculature and thus the cells rely on oxygen delivery by passive diffusion over distances of hundreds of microns (depending on device design). Moreover, dissolved oxygen levels are considerably lower in common transplantation sites (e.g. the subcutaneous space) between 8 – 35 mmHg,^[4] as compared to arterial oxygen tensions near 100 mmHg.^[5] Graft oxygenation is further impaired by the inevitable formation of a fibrotic capsule around the implant, which creates an additional barrier to oxygen transport.^[6] Of all cell requirements, oxygen is at the lowest concentration with respect to its rate of consumption and is therefore often the most critically limited species in cell encapsulation.^[7]

Strategies to address inadequate oxygenation have included the induction of graft vascularization,^[8] device geometry and materials optimization,^[9] and exogenous oxygen supply.^[10] Oxygen delivery from an external source is often preferred as it can produce supraphysiological oxygen levels. For example, the decomposition of metal peroxides by hydrolysis has been shown to significantly improve graft oxygenation;^[11] however, with these technologies, the window of oxygen production is finite. Oxygen production by electrolysis has also been explored with success,^[12] though this strategy is more complicated from an engineering perspective. Alternatively, the β-Air device (Beta-O2 Technologies) supports injectable oxygen into a gas-permeable chamber;^[13] however, daily oxygen injections are required for graft survival. There is thus a need for developing a device which can support long-term high oxygenation without patient intervention.

The atmosphere is a virtually unlimited source of highly concentrated oxygen. At sea level, the partial pressure of oxygen (pO₂) in the atmosphere is ~160 mmHg—roughly 4 times higher than in common transplantation sites. In fact, the human cornea, which is avascular, is oxygenated by direct contact with the air.^[14] Notably, the aqueous and cellular components of the cornea are protected from evaporation and environmental harm by a lipid/oil-containing layer of the tear film at the atmospheric interface.^[15]

Inspired by this clever natural oxygen delivery strategy, we designed a novel modular biphasic (BP) system. Cells were encapsulated in a hydrogel (liquid phase) and oxygen supply was provided by contact with the atmosphere (gas phase). In mimicry of the cornea, the device was implanted in a transcutaneous position, thereby exposing one face to the air and the other to the subcutaneous space. Further, environmental protection was provided by a perfluorinated carbon (PFC) oil-infused film at the atmospheric interface, such as the role of the surface layer of the tear film. (Figure 1). The BP device consisted of four fundamental components (Figure 1a): (1) the PFC cover to provide environmental protection and prevent dehydration, (2) PFC channels within the cell encapsulation domain for improved oxygen transport and mechanical reinforcement, (3) a hydrogel for cell encapsulation and immunoisolation, and (4) a frame to fasten the device in a transcutaneous configuration. Importantly, the cell encapsulation module was attached to a demountable cap, which allowed for the replacement of the therapeutic cells within a few minutes non-surgically. Eventual graft decline may necessitate additional transplantations for sustained therapeutic activity. The modularity of the BP device provided a platform to circumvent complicated and invasive surgical procedures traditionally required for graft replacement.

Figure 1. The BP device: design components and function.

(a) Annotated schematic illustrating the fundamental components and functions of the BP device. (b) Digital images of the empty device for mouse implantation from the top view (left), side view (center), and bottom view (right). (c) Annotated schematic illustrating device components and dimensions; schematic and digital image of the cell encapsulation module (left panel); schematic of each component, including the (top to bottom) titanium frame cap, cell encapsulation module, alginate-impregnated nylon mesh, and titanium frame base (center panel); schematic of the cell encapsulation module, featuring the PDMS sealing O-ring (right panel). (d) SEM image of the PTFE nanomembrane. (e) A digital image of a water droplet on the PTFE nanomembrane. (f) A contact angle goniometer-captured image of a water droplet sliding on the PFC-infused PTFE nanomembrane after tilting to ~5°. (g) SEM images of the PVDF-HFP scaffold displaying its spiral configuration and porous structure. (h) Chemical structure of the PVDF-HFP copolymer. (i) SEM image of the nylon mesh. (j) Confocal image of the alginate-impregnated nylon mesh.

In this report we present the characterization and testing of the BP device. Simulation and in vitro investigations show improved graft oxygenation in comparison to subcutaneous transplantation. Encapsulated pancreatic islet transplantation—a promising application of cell replacement therapy—offers to improve type 1 diabetes treatment by eliminating or reducing the need for exogeneous insulin injections.^[16] Islets (cell clusters between tens and hundreds of microns in diameter containing hundreds to thousands of insulin-secreting β cells and other secretory cell types) are particularly vulnerable to hypoxia.^[17] Stripped of their native microvasculature during isolation,^[18] islets maintain a steep oxygen consumption rate due to the high metabolic demand of insulin secretion,^[19] and their low capacity for anaerobic respiration.^[20] Moreover, the capacity of islets to secrete insulin in response to glucose is significantly reduced at even moderately low oxygen levels,^[17] and after acute exposure to hypoxia.^[21] We therefore tested rat islet-encapsulating BP devices in immunocompetent streptozotocin (STZ)-induced diabetic mice and confirmed that the device was able to maintain islet health and provide diabetes correction in vivo. Finally, robust cell survival and a proof of concept of the cell refilling procedure was shown in dogs.

First, a prototype was designed for implantation in a mouse (Figure 1b, c). The frame of this device, comprising both a base and a cap, was composed of titanium for its well-documented biocompatibility.^[22] Several portals (1 mm diameter) were fabricated into the cap to allow gas exchange with the atmosphere, whereas the bottom face of the base was open as to totally expose the graft to the subcutaneous tissue to ensure nutrient exchange between the host and the encapsulated cells. A PFC (Krytox®, GPL103)-infused polytetrafluoroethylene (PTFE) nanomembrane was applied below the titanium cap at the device-atmosphere interface (Figure 1d). The low surface energy and nanoporosity (~200 nm pore size) of the PTFE mesh created a strong capillary force that enabled PFC infiltration and retention. Prior investigation demonstrated that bacterial adhesion to this membrane was severely limited.^[23] Moreover, the mesh pore size corresponded to that of standard bacterial filtration membranes, therefore the membrane was thus expected to prevent both the adhesion and passage of bacteria through this interface. In addition, this composite material was non-wettable, non-volatile, and omniphobic,^[24] and was accordingly an optimal material for both barring environmental stressors from graft interference and preventing hydrogel dehydration (Figure 1e, f). Cell and bacterial culture on the PFC-infused PTFE membrane also showed significantly impaired adhesion, though cell viability was preserved, suggesting that this material would both reject bacterial infiltration while providing no harm to the encapsulated cells (Figure S1, Figure S2a, b, Supporting Information). Furthermore, an in vitro test suggested that bacterial migration through the membrane was prohibited (Figure S2c – e, Supporting Information).

Additionally, a spiral poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) scaffold (Figure 1g, h) was fabricated directly on the PTFE membrane. This fluorinated polymer scaffolding was porous, which allowed for PFC infiltration and thus provided both structural reinforcement and improved oxygen delivery. Around the scaffolding, a suspension of cells within ultrapure sodium alginate (Pronova SLG100) was added via pipet and crosslinked by submersion in a 95 mM CaCl₂ and 5 mM BaCl₂ buffer. Alginate was selected as the cell encapsulation hydrogel for its biocompatibility and common application in cell encapsulation.^[25] A cell-free alginate-impregnated nylon mesh (70 μm thickness) was applied for mechanical reinforcement at the device-host interface (Figure 1i, j). The above components constituted the cell encapsulation module (Figure S3, Supporting Information). Finally, a PDMS O-ring was included between the cell encapsulation module and the frame base to ensure a tight seal.

Subsequently, the oxygen transfer advantages of the transcutaneous concept were investigated by theoretical and in vitro analyses (Figure 2). Computational models were developed to compare the oxygenation of randomly seeded islets in the “transcutaneous” BP device, the “transcutaneous” device without the spiral scaffold, and a “subcutaneous” control (see the Mass Transfer section and Figure S4 in the Supporting Information for model development details). The difference between the subcutaneous and transcutaneous configurations was implemented by applying a top boundary condition of oxygen tension of 24 mmHg (3% oxygen) or 160 mmHg (21% oxygen), corresponding with subcutaneous and atmospheric levels respectively (the boundary condition at the device-subcutaneous space interface was 24 mmHg for both conditions).

Figure 2. Mass transfer.

(a) Simulation-predicted oxygen concentration profiles in the BP device, the BP device without the spiral scaffold, and a fully implanted subcutaneous control encapsulating islets. (b) Quantification of spatially averaged islet oxygen concentration (islet number labelled #1 through #8 from left to right). (c) Surface plot of simulation-predicted oxygen profiles in the BP device and a subcutaneously transplanted control encapsulating dispersed (INS-1) cells. (d) Simulation-predicted oxygen concentration in the BP device and a subcutaneously transplanted control along a horizontal cross section (labelled A-A) and a vertical cross section (labelled B-B) from surface plots shown in (c). (e) Schematic of experimental design (top) and microscope images (bottom) of live/dead-stained INS-1 cells in BP device (left) exposed to the atmosphere while partially submerged in media at a pO₂ of 24 mmHg and a control alginate slab (right) fully submerged in media at a pO₂ of 24 mmHg.

Simulation revealed that the BP device provided significantly higher predicted oxygenation in the alginate-cell domain in comparison to the subcutaneous control. The concentration of oxygen in the PFC film was also predicted to be higher than that within the hydrogel due to its superior oxygen solubility. Most importantly, robust islet survival was predicted for the BP device, whereas a high degree of islet necrosis was predicted in the subcutaneously implanted control (Figure 2a; Figure S5, Supporting Information). Further, quantification of spatially averaged oxygen concentration within individual islets suggested that the inclusion of the spiral scaffold modestly to significantly improved islet oxygenation, depending on the proximity of the cell cluster to the scaffold (Figure 2b). This result, in addition to the qualitatively noted improved mechanical strength, encouraged the incorporation of the spiral scaffold in all ensuing testing.

Another model was developed to simulate oxygen transport in a dispersed cell encapsulation system. Again, significantly higher graft oxygenation was predicted for the BP device in comparison to the subcutaneous control (Figure 2c). Along a horizontal cross section, predicted graft oxygenation was nearly an order of magnitude higher in the BP device compared to the control; along a vertical cross section, the highest oxygen concentration of the control device represented the lowest oxygen concentration of the BP device (Figure 2d). These simulated predictions were validated by an in vitro analysis. INS-1 cells were encapsulated at a density of 3 million cells mL⁻¹ in a BP device and an alginate slab and placed in media which had been previously reduced to a pO₂ of 24 mmHg (3% oxygen) in a hypoxia chamber. A barrier layer (MitoXpress oil) was applied at the media-air interface to impede oxygen transport into the media from the atmosphere. The top face of the BP device was exposed to the atmosphere while the remainder of the device was submerged in media, whereas the alginate slab was positioned beneath the oil barrier and completely submerged in the media. Robust cell survival was observed in the transcutaneously-positioned BP device, whereas only an outer layer of viable cells remained in the subcutaneous control (Figure 2e). These studies and analyses corroborated the modeling results and reaffirmed our hypothesis that exposing the device to the atmosphere would significantly improve graft oxygenation.

We next tested the therapeutic capability of the BP device in a rat-to-mouse xenotransplantation model (Figure 3). Mouse devices encapsulating isolated rat islets (500 islet equivalents; IEQ) within ~50 μL of alginate, were fabricated for transcutaneous transplantation in STZ-induced diabetic C57BL/6J mice (Figure 3a–d). Hyperglycemia reversal (blood glucose, BG < 200 mg dL⁻¹) was observed after 1 day and for the duration of the study (15 days) in animals treated with the BP device. Brief BG lowering was observed in subcutaneously transplanted alginate slab controls encapsulating 500 IEQ rat islets (see Supplementary Figure S4b for dimensions), though the mice returned to a hyperglycemic state (BG > 450 mg dL⁻¹) within 1 week following transplantation. Diabetic controls were hyperglycemic at all readings over the course of the study.

Figure 3. Diabetes correction in STZ-induced diabetic mice.

(a) Schematic illustration of the BP device featuring a titanium frame for mouse implantation. (b) Schematic illustration of the islet encapsulation module. (c) Microscope image of islets encapsulated within the spiral alginate hydrogel. (d) Digital images showing the surgical procedure for fastening the device in the transcutaneous position: (left) a circular section of skin was excised and a purse-string suture pattern was placed in the surrounding skin; (center) the device was placed in the space of the excised skin; (right) the sutures were pulled tight and the device was fixed in the transcutaneous position. (e) BG readings of mice receiving BP devices (n = 5), subcutaneous transplantation controls (n = 3), and nontreated diabetic mice (n = 5); mean ± SD; ***P < 0.001 (BP device versus subcutaneous control), ***P < 0.001 (BP device versus diabetic control). (f) IPGTT at day 7; n = 5 for BP devices, n = 3 for subcutaneous controls, n = 5 for nontreated diabetic controls, n = 5 for non-diabetic controls; mean ± SD; ***P < 0.001 (BP device versus subcutaneous control), ***P < 0.001 (BP device versus diabetic control), n.s. (P > 0.05; BP device versus non-diabetic control). (g) Live/dead staining of islets from one retrieved BP device at day 15. (h) Static GSIS test of retrieved BP devices (n = 3) and subcutaneous controls (n = 3); mean ± SEM; ***P < 0.001 (2.8 mM versus 16.7 mM conditions for retrieved BP devices), n.s. (P > 0.05; 2.8 mM versus 16.7 mM conditions for retrieved subcutaneous controls), ***P < 0.001 (retrieved BP device versus retrieved subcutaneous control for both 2.8 mM and 16.7 mM conditions). (i) H&E staining of islets in one retrieved BP device at day 15. (j) Immunohistochemical staining of islets in a retrieved BP device at day 15.

An intraperitoneal glucose tolerance test (IPGTT) was performed on the mice at day 7 to further test device function. The BG of the BP device-treated group returned to a lowered state after 60 minutes, which was similar to healthy controls, whereas the blood glucose of the subcutaneously implanted controls did not lower over the 120 minutes investigated, similar to the diabetic controls (Figure 3f). Live/dead staining of retrieved islets showed that the encapsulated cells in BP devices were largely viable following retrieval (Figure 3g). Furthermore, a static GSIS performed on the retrieved cell encapsulation modules and subcutaneous controls showed glucose responsiveness of the BP devices, whereas insulin secretion was significantly impaired in the retrieved subcutaneous controls (Figure 3h). Maintained islet function in the BP device was further corroborated by healthy islets found in hematoxylin and eosin (H&E) stained slides and the robust presence of insulin following immunostaining (Figure 3i, j). In contrast, islet health was significantly impaired in retrieved subcutaneous samples (Figure S5, Supporting Information).

The engineering of a BP device for large animal transplantation and cell refilling was subsequently investigated (Figure 4). A series of design iterations were pursued to overcome translational hurdles (the evolution of the design is illustrated in Figure 4a and Figure S6, Supporting Information). While the fundamental components of the BP device were preserved, the final design featured several new functionalities inspired by iterative analysis (Figure 4b, c). Instead of titanium, the frame was fabricated by 3-dimensional (3D) printing (Form2 3D printer) with the Class IIa biocompatible Dental LT resin as this material provided greater flexibility over design modifications. On the frame base, six anchor rings were incorporated to fasten the device within the subcutaneous tissue via suturing; this was motivated by the successful application of this technique in a dog in the first design (Figure S7, Supporting Information). Testing of the first design in dogs also showed that poor device fixation led to infection (Figure S6d, Figure S8, Supporting Information).

Figure 4. BP device transplantation and cell refilling in a dog.

(a) Schematics illustrating the design evolution of the resin-based BP device; the first design (left) was upgraded to include a porous structure for tissue integration and was elongated to expose the cell encapsulation module to the deeper subcutaneous space; the second design (center) was upgraded to the final design (right) by including a hexagon depression for frame cap removal by a hex wrench and a trimmed rim diameter of the top of the frame base to reduce skin coverage. (b) Digital image of the final BP device design. (c) Schematic and dimensions of the BP device components including a (top to bottom) PFC-infused PTFE nanomembrane, frame cap with a PDMS washer, cell encapsulation module, frame base with a porous exterior, anchor rings, and threading, and the alginate-impregnated nylon mesh. (d) Digital image of the device in a dog at 1-month post-implantation. (e) Digital image during the non-surgical refilling procedure: a hex wrench was placed in hexagon depression and twisted as the base was stabilized with forceps. (f) Digital image of the device in a transcutaneous position after the frame cap (including the cell encapsulation module) has been removed. (g) Digital image of the frame cap with the cell encapsulation module following retrieval at 1-month. (h) Digital image of the BP device with the replaced cap containing encapsulated rat islets. (i) Digital image of the BP device at 1-month post cell refilling. (j) Digital image of the retrieved cell encapsulation module at 1-month post cell replacement. (k) H&E staining of rat islets from the retrieved BP device. (l) Immunohistochemical staining of islets in the retrieved BP device. (m) Schematic highlighting the porous structure on the exterior of the frame base. (n) H&E staining of the device and surrounding subcutaneous tissue showing tissue integration into the porous structure.

We therefore incorporated a macroporous structure on the frame base, which resulted in robust tissue ingrowth and the transcutaneous fixation of the new device design in mice, rats, and dogs (Figure S9–S12, Supporting Information) for over 1 month. This was consistent with another finding in the literature that demonstrated that porous implants improved tissue integration in the subcutis and cutis, which was further hypothesized to lower the risk of bacterial infection.^[26] Furthermore, the frame height was increased such that the bottom face of the cell encapsulation module was exposed to the deep subcutaneous tissue following the excision of the cutis and some subcutaneous adipose tissue (Figure S12f, g, Supporting Information). Implantation in this region was desirable as it has been suggested that oxygen levels are higher in the deep subcutaneous space in comparison to superficial regions of the tissue.^{[4, 27]} The foreign body response at the interface of the alginate-impregnated nylon mesh and the host subcutaneous tissue in both mice and rats was characterized by immunohistochemical staining, revealing a modestly vascularized collagenous layer (Figure S10, Figure S11f – h, Supporting Information).

A simple approach was implemented to allow cell refilling. Threading was included on the frame cap and base, and therefore the cap could be removed and replaced by counterclockwise and clockwise rotation respectively. The cell encapsulation module was attached to the frame cap by the inclusion of a PDMS O-ring, as in the previous design; thus, cell refilling was performed by unscrewing the current cap and replacing it with a new one. In the final design, a hexagon depression was integrated into the frame cap such that this process could be performed more easily using a hex wrench (i.e. Allen wrench). In addition, robust attachment of the alginate-impregnated nylon mesh was achieved by situating it within a small depression in the bottom of the frame base and allowing the infiltration of some alginate into adjacent macropores prior to gelation (Figure S13, Video S1, Supporting Information). Lastly, an additional PFC-impregnated PTFE nanomembrane was placed on top of the frame cap for increased environmental protection, and a PDMS washer was included between the frame base and cap to ensure sealing (Figure S14, Supporting Information).

The BP device was tested in a healthy dog. A cell-free BP device was transcutaneously implanted by the method described above. While the surgical procedures were performed under sterile conditions, animals were kept in AAALAC-approved non-sterile enclosures and allowed daily outdoor access, thus the graft was exposed to potential environmental stressors. At 1 month following transplantation, device integration was robust, and no adverse reaction was observed (Figure 4d). Using a hex wrench, the frame cap was unscrewed, and the (cell-free) cell encapsulation module was removed (Figure 4e, f; Video S2, Supporting Information). The retrieved cell encapsulation module was not infected or noticeably affected by either the environment or the immune system according to qualitative observation (Figure 4g). Next, a new cell encapsulation module containing encapsulated rat islets (2000 IEQ in ~75 μL alginate) was attached to the frame cap and twisted manually (i.e. non-surgically) into the frame base at the transplantation site (Figure 4h; Video S3, Supporting Information). The removal-and-replacement procedure was performed in a few minutes.

At 1 month following cell refilling, the device maintained tissue integration and structural integrity (Figure 4i). The cell encapsulation module was removed by the same mechanism described above (Figure 4j, Video S4, Supporting Information). Retrieved encapsulated islets were found to be mostly healthy and insulin positive as confirmed by H&E and immunohistochemical staining (Figure 4k, l). H&E-stained slides of the excised frame base and surrounding tissue revealed robust tissue ingrowth into the negative space of the macroporous structure (Figure 4m, n). This study demonstrated the feasibility of replacing the transplanted cells when necessary without surgical intervention.

While several groups have reported improved graft oxygenation, there are a few salient advantages of the biphasic system worth reiterating. Foremost, continuous contact with the atmosphere does not require patient intervention nor the introduction of oxygen generating technologies. Nonetheless, as atmospheric contact exposes the device to environmental harm and bacterial contamination, two key materials strategies were employed to overcome this challenge. First, the application of the omniphobic PFC-infused PTFE nanomembrane was critical for avoiding infection through the portals of the frame cap. It has also been hypothesized that tissue ingrowth facilitates the introduction of immune components into the device-tissue interface, which act to resist bacterial passage.^[26] Therefore, the incorporation of the porous structure on the exterior of the frame base, which was demonstrated to encourage tissue ingrowth, may have played an equally important role in preventing infection.

Cell replacement therapies have the potential to shift the paradigm of chronic disease treatment, although this technology is often constrained by limited oxygen supply and the difficulty of refilling the therapeutic cells after graft decline. In this report we present the design, engineering, and testing of a highly oxygenated biphasic cell encapsulation platform, which enabled gas exchange by contact with the atmosphere and supported cell refilling without requiring surgery. These benefits were realized by rational design, where immune protection and cell survival were accomplished by hydrogel encapsulation and environmental protection was facilitated by a PFC oil-infused nanomembrane interface. High oxygenation in comparison to subcutaneously implanted grafts was confirmed by theoretical analysis and in vitro studies. Moreover, the therapeutic efficacy of this device was shown in a rat-to-mouse xenograft. Finally, non-surgical cell refilling was shown in a transcutaneous canine implantation. The continued investigation of practical solutions for persisting problems in cell encapsulation will contribute to translation of such devices to the clinic.