Oxygen transporter for the hypoxic transplantation site

Abstract

Cell transplantation is a promising treatment for complementing lost function by replacing new cells with a desired function, e.g., pancreatic islet transplantation for diabetics. To prevent cell obliteration, oxygen supply is critical after transplantation, especially until the graft is sufficiently re-vascularized. To supply oxygen during this period, we developed a chemical-/electrical-free implantable oxygen transporter that delivers oxygen to the hypoxic graft site from ambient air by diffusion potential. This device is simply structured using a biocompatible silicone-based body that holds islets, connected to a tube that opens outside the body. In computational simulations, the oxygen transporter increased the oxygen level to >120 mmHg within grafts; in contrast, a control device that did not transport oxygen showed < 6.5 mmHg. In vitro experiments demonstrated similar results. To test the effectiveness of the oxygen transporter in vivo, we transplanted pancreatic islets, which are susceptible to hypoxia, subcutaneously into diabetic rats. Islets transplanted using the oxygen transporter showed improved graft viability and cellular function over the control device. These results indicate that our oxygen transporter, which is safe and easily fabricated, effectively supplies oxygen locally. Such a device would be suitable for multiple clinical applications, including cell transplantations that require changing a hypoxic microenvironment into an oxygen-rich site.

Introduction

Oxygen (O₂) plays an important role in the body to maintain physiological homeostasis. Under physiological conditions, O₂ is efficiently transported via hemoglobin throughout the body; O₂ molecules bound to hemoglobin are released into tissues exhibiting low O₂ concentration pursuant to the oxyhemoglobin dissociation curve. Adequate O₂ supply is also crucial for maintaining viability and function under non-physiological conditions, such as organs and cells removed from donors for transplantation. In organ transplantation, the arteries and veins of donor organs are surgically anastomosed to recipient vessels to re-establish blood flow through which hemoglobin can then transport O₂ to the graft. In contrast, in cell transplantation, a treatment used to supplement a specific cellular function that has been lost or damaged, the transplant cannot be oxygenated by hemoglobin until revascularization takes place [1]. Therefore, in the early phase following transplantation of cells, O₂ diffusion from tissues surrounding the graft provides the only O₂ source. This period is critical for transplanted cells, because they are exposed to a non-physiological, hypoxic microenvironment that can leads to loss of function and graft death.

Recent advancements in stem cell research have shed light on cell replacement therapy. Generating cell aggregates as a minimum functional unit is a promising method for achieving functional recovery [2–4], and is easier than generating or replacing a whole organ. This concept is illustrated by pancreatic islet transplantation, in which small endocrine units (islets) isolated from the pancreas of a donor are used as cell replacement therapy for diabetic patients with deteriorated function of native insulin-producing beta cells. The transplanted islets supply insulin to the recipient for improved blood glucose control. This therapy is effective for the treatment of Type 1 diabetes [5, 6], and has been used for more than a decade. Currently, islets isolated from a donor’s pancreas are transplanted into the recipient’s liver through the portal vein. However, this often leads to acute and/or gradual loss of a large portion of the islets after transplantation, due to conditions in the liver microenvironment including direct contact with the blood and its immune system [7, 8]. To overcome this challenge, alternative islet transplantation sites have been explored. The subcutaneous (SC) site is an attractive candidate for cell replacement therapy. The SC site provides extensive space for a large number of cells and allows transplantation via minimally invasive procedures. Furthermore, the SC site provides opportunities for frequent graft monitoring and early detection of any potential anomalies for prompt removal. This is an important issue for cell replacement therapy, especially when using a stem cell-derived graft, which has the potential to form malignancies [9–16]. However, despite these advantages, lack of O₂ in SC tissue has hindered the establishment of islet transplantation in this site, as islets are quite susceptible to hypoxia [17–20] and require a sufficient O₂ supply until the graft is re-vascularized. Thus, there is a critical need to develop the means to supply O₂ to the SC site, to maintain transplanted cell viability and ensure transplantation success.

Several methods have been introduced to supply O₂ to a hypoxic site. O₂ inhalation is the simplest method to increase tissue O₂ [21, 22]; we used a rat model to show that the partial pressure of O₂ (pO₂) increases in the SC site from 45 mm Hg under 21% O₂ inhalation to 140 mm Hg under 50% O₂ inhalation, which enhances engraftment of transplanted islets in the SC site. Co-transplantation of cells with O₂-generating materials that directly oxygenate the graft site has been widely explored, leading to improved engraftment and cell survival [23–30]. Administration of O₂ carriers such as perfluorocarbons and oxygenated hemoglobin has also been introduced [31–34]. Of particular relevance, devices to supply O₂ directly to the graft site have been developed to support SC islet transplantation [35–37]. Taken together, these various methods demonstrate the effectiveness of oxygenation, but present shortcomings for clinical applications; e.g., the duration of high O₂ inhalation is limited due to O₂ toxicity, and O₂-generating materials typically require chemical or electrical reactions [38–40]. In the present study, we aimed to fabricate a safe, simply structured, and sustainable O₂ transporting device that supports an improved O₂ microenvironment for SC islet cell transplantations.

To design an implantable micromechanical O₂ transporter to augment O₂ supply, we combined two parts: a highly O₂-permeable silicone “diffuser,” on which islets are placed in a gel matrix; and a cannula made of stainless steel tubing, which traverses the skin to provide a direct path for the diffusion of atmospheric O₂ into the device (Fig.1A). This device is driven by diffusion potential; thus, O₂ is transported from the ambient air (higher O₂) to the SC site (lower O₂) without chemical or electrical reactions, which ensures safety in clinical applications (Fig.1B). Further, because the O₂ source for this device is ambient air, O₂ never depletes. Our data indicate that the device effectively supplies oxygen locally to support islet transplantation.

Figure 1: Device design and fabrication of the O₂ transporter.

(A) Overview of the O₂ transporter, consisting of diffuser and cannula. (B) Schematic of the O₂ transporting system; O₂ is transported from ambient air (higher O₂) to the subcutaneous site (lower O₂) by diffusion potential. (C) Dimensions of the device. Cross sectional view indicated by red color; whole device (left) and enlarged section of diffuser (right). (D) Microfabricated molds for photoresist casting of the diffuser parts. Scale bar: 5 mm. (E) Assembly of the silicone parts and stainless tubing. (F) Photograph of a complete device. A filleted neck prevents stress concentration between cannula and diffuser (arrowhead). (G) Fabrication of the O₂ blockage control device and comparison to the O₂ transporter. The stainless tubing was replaced by a stainless rod to block O₂ transportation. (H) Schematic of the loading of the islet graft; islets were suspended in the gel for fixation on the transporter. (I) Photograph of the islet graft on the device.

Materials and Methods

Device Design and Fabrication

The O₂ transporter device is composed of a medical-grade PDMS (polydimethylsiloxane) hollow body diffuser connected to a stainless steel cannula. Simulations of O₂ transport were used to guide the relative dimensions of the diffuser and tubing to ensure O₂ transport was limited by diffusion into the gel/islets, but not by diffusion into and through the device. A 15-mm stainless steel tube (304 stainless steel tube, outer diameter=0.91 mm, inner diameter = 0.71 mm, McMaster-Carr, Santa Fe Springs, CA, USA) is connected to the diffuser (10 mm outer diameter), which has a 120-μm-thick top surface to facilitate high O₂ permeation and a 240-μm-thick bottom surface (Fig.1C). The layers are separated by an array of micropillars (300 μm in diameter) to produce an air gap of 120 μm, sufficient for transport of O₂ evenly throughout the device, despite the stainless tubing connecting on only one side. A 480-μm-tall rim and quadrant dividers provide a retaining reservoir for the islet graft.

The diffuser portion of the O₂ transporter was built in two parts by casting medical-grade PDMS (MED4–4210, NuSil Technology LLC, Carpentaria, USA; mixed at 10:1 ratio by weight and degassed,) into microfabricated molds by photoresist and partially curing for 10 min at 100°C (Fig.1D). To form the transporter, the two halves were removed from the molds, glued together with a thin layer of silicone, and fully cured overnight at 100°C (Fig.1E). Then the stainless steel tube was inserted into the diffuser and glued with silicone. The connection of the diffuser to the stainless steel tube features a filleted neck to prevent stress concentration (Fig.1F). The assembled device was cured overnight at 100°C, followed by 3× acetone extraction (at 56°C, minimum for 6 h) to remove unreacted polymer or contaminants. Devices were leak-tested by submerging the diffuser in water and blowing air into the stainless tube. An O₂ blockage device (Neg-CTL device) was fabricated similarly, except the stainless tubing was replaced by a stainless rod, which blocks O₂ transportation from the ambient air to the O₂ diffuser (Fig.1G).

Simulation O₂ microenvironment of the graft with O₂ transporter

The O₂ microenvironment was simulated at the scale of the O₂ transporter and at the scale of an individual islet sitting on the transporter, when transplanted in a rat SC site. The finite-element simulation used COMSOL 5.3 (COMSOL, Los Angeles, CA, USA) based on the steady-state reaction-diffusion equation:

$$H\cdot D\cdot {\nabla }^2\left[P\right]=Q$$ (1)

where ∇² denotes the Laplace operator and Q represents the islet O₂ consumption governed by Michaelis-Menten type kinetics:

$$Q=\rho \cdot k\cdot \frac{P}{P+\frac{K_M}{H}}$$ (2)

Michaelis-Menten kinetics approximate cellular metabolism under various pO₂ conditions. The form of the equation captures the asymptotic behavior of O₂ uptake rate to a maximum rate k when pO₂ is above K_M; when pO₂ is below K_M, the O₂ uptake rate is proportional to the available O₂. The experimentally measured O₂ consumption rate (OCR) of the islets in low glucose and high glucose environments was used in the simulations. Because the K_M for beta cells has not been experimentally determined, we used a value similar to that of the mitochondria, consistent with other work modeling the islet [41].

We used pO₂ = 45 mmHg (0.0592 atm) in the SC site [22] and atmospheric pO₂ = 159 mmHg. Concentrations of O₂ across regions of the system were linked through Henry’s Law with the assumption of equal interfacial pO₂ (i.e., C_1,i/H₁ = P_1,i = P_2,i = C_2,i/H₂, where subscript n,i refers to the interfacial surface in region n). A full set of parameters is presented in Table 1. For scaled simulations of the O₂ transporter, the O₂ consumption of 600 islets was modeled as uniformly distributed throughout a 150-μm thick region on top of the upper PDMS membrane. The stainless steel tube inner surface had a no-flux boundary condition with the open end held at atmospheric pO₂. The Neg-CTL device was simulated by enforcing a no-flux boundary condition at the inner face of the rod/tube region. To model the situation in vitro, a 2-mm-thick fluid compartment was placed around the device to account for hypoxic PBS at 45 mmHg pO₂, mimicking in vivo SC site pO₂.

Table 1.

Parameters for islet O₂ simulation

Description	Values	Note / Reference
Diffusivity O2 in islets, fluid, gel	3 × 10−9 $\left[\frac{m^2}{s}\right]$	[41]
Diffusivity O2 in air	2.2 × 10−5 $\left[\frac{m^2}{s}\right]$	[42]
Diffusivity O2 in PDMS	3.4 × 10−9 $\left[\frac{m^2}{s}\right]$	[43]
Solubility O2 in islet, ECF, gel	165 × 10−6 $\left[\frac{\mathit{\text{mol}}}{L\cdot \mathit{\text{atm}}}\right]$	[41]
Solubility O2 in air	0.04 $\left[\frac{\mathit{\text{mol}}}{L\cdot \mathit{\text{atm}}}\right]$	Ideal Gas Law
Solubility O2 in PDMS	8 × 10−3 $\left[\frac{\mathit{\text{mol}}}{L\cdot \mathit{\text{atm}}}\right]$	[43]
Subcutaneous pO2 in rat	45 [mmHg]	[22]
Atmospheric pO2	159 [mmHg]	Assumed
Oxygen Consumption Rate (High glucose environment)	8.3 × 10−11 $\left[\frac{\mathit{\text{mol}}}{m{m}^3\cdot s}\right]$	Measured
Oxygen Consumption Rate (Low glucose environment)	6.3 × 10−11 $\left[\frac{\mathit{\text{mol}}}{m{m}^3\cdot s}\right]$	Measured
Michaelis O2 constant	1 × 10−3 $\left[\frac{\mathit{\text{mol}}{O}_{\text{2}}}{m^3}\right]$	[44, 45]
Islet diameter	150 × 10−6 [m]	[46]

Isolation of pancreatic islets

Islets were isolated from the pancreata of Lewis (LEW) rats using our standard procedure [47]. Briefly, ice-cold Hanks’ balanced salt solution (HBSS; Sigma-Aldrich, St. Louis, MO, USA) with 0.1% bovine serum albumin (Sigma-Aldrich) and 1 mg/mL of Liberase (Roche Diagnostics GmbH, Mannheim, Germany) was injected into the pancreatic duct. The distended pancreas was dissected out, then digested at 37°C for 30 min. The tissue was further dissociated mechanically by shaking and washing cycles. Islets were purified using gradient centrifugation on Histopaque-1077 (Sigma-Aldrich) and handpicked to ensure high purity. Isolated rat islets were cultured in RPMI 1640 (Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum and 5 mM glucose at 37°C in a tissue culture incubator under 21% O₂ plus 5% CO₂.

Preparation of islets on the O₂ transporter

The number of islets used in experiments was determined using our standard procedures [47]. Briefly, cultured islets were aspirated into a sterile PE50 tube and packed by centrifugation (160 ×g for 2 min). The length of packed islets gives a specific islet number based on our established standard curve, where the length of packed rat islets in a PE50 tube is correlated with islet number. The diffuser portion of an O₂ transporter was incubated in 250 μg/ml of fibronectin solution at 4°C overnight to make the diffuser surface hydrophilic, as previously described [48]. Autoclaved 1% UltraPure Low Melting Point Agarose gel (ThermoFisher, Waltham, MA) was prepared at 37°C and mixed with islets (20 μl for 600 islets, per O₂ transporter, Fig1H). Islets were loaded on the transporter and incubated at room temperature for 10 min for gelation (Fig.1I).

Measurement of OCR of rat isolated islets

Isolated rat islets were cultured for 2–3 days for OCR measurements. Fifty islets/well were applied to a Seahorse XF24 Islet Capture Microplate (Agilent Technologies, Santa Clara, CA, USA). Because islet metabolism is known to change based on the glucose concentration of the microenvironment [49], the OCR was measured in islets cultured in 3 mM glucose Krebs-Ringer buffer (low glucose solution) for 45 min, followed by 20 mM glucose Krebs-Ringer buffer (high glucose solution) for 60 min. The average OCR of 3 consecutive measurements was calculated for both the low glucose solution and high glucose solution. OCR measurements were performed using islets isolated from 4 different rats.

In vitro pO₂ measurement of the microenvironment around islets seeded in the O₂ transporter

The O₂ microenvironment was examined in vitro, by using the same islet preparation placed on the devices as shown in Fig.1H and Fig.1I. To mimic the O₂ microenvironment of the SC site of rats, hypoxic PBS (pO₂ = 45 mmHg, the same pO₂ as in the rat SC site [22]) at 37°C was prepared; this pO₂ and temperature were maintained throughout the experimental period. An O₂ transporter (or Neg-CTL device) loaded with islets was placed on a spatula, and the diffuser portion was submerged in the hypoxic PBS, with the end of the cannula exposed to the ambient air. To stabilize the O₂ gradient in the measured region, the O₂ transporter and Neg-CTL devices loaded with islets were preincubated in hypoxic PBS for 2 h prior to measurements; according to simulations, this generated a stable O₂ gradient in the range of 0–9,295 μm from the surface of the transporter. An O₂ probe was placed on a rotary positioner (Parker Hannifin, Irwin, PA) and the tip of the probe was placed adjacent to the O₂ transporter. pO₂ was measured stepwise between 0 μm (through the graft embedded in gel) and 2,000 μm in distance from the surface of the transporter. Three individual experiments using islets isolated from different rats were performed per group.

Biocompatibility of transporter material with isolated islets

Isolated rat islets loaded on the O₂ transporter were cultured in RPMI medium supplemented with 5 mmol/L glucose (Hospira, Lake Forest, IL) for one week under normal islet cell culture conditions as described above (Isolation of pancreatic islets). For live/dead cell staining, fluorescein diacetate and propidium iodide were used to stain viable cells (green) and dead cells (red), respectively [50–52]. Images of islets were captured using bright field as well as fluorescent microscopy (IX50 fluorescence microscope, Olympus, Tokyo, Japan).

Experimental Animals

Male LEW rats weighing 400–500 g were used as pancreas donors, except for in vivo imaging experiments, in which Luciferase transgenic LEW rats (Firefly rats) were used as donors [53]. Syngeneic male LEW rats weighing 300 g were used as islet recipients. The use of animals and animal procedures in this study was approved by the City of Hope/Beckman Research Institute Institutional Animal Care and Use Committee.

Islet Transplantation with the O₂ transporter or Neg-CTL device

Diabetes was rendered in recipient rats by a single intravenous injection of streptozotocin (STZ, 60 mg/kg, Sigma-Aldrich). Diabetes was confirmed by measuring blood glucose levels >400 mg/dl in two consecutive measurements. A prevascularized SC graft bed was prepared by implanting an agarose-basic fibroblast growth factor (bFGF) disc, as described [22, 54]. Briefly, a lyophilized agarose disc made from 4.5% agarose solution (Sigma-Aldrich) with bFGF solution [500 μg/ml bFGF (Gold Biotechnology, St. Louis, MO) in saline] and heparin sodium solution [250 μg/ml (Sigma-Aldrich)] was implanted in the dorsal SC site of recipient rats. A week after implantation, a prevascularized capsule had formed around the disc, and the disc was removed through a small opening made on the capsule. Freshly isolated islets were used for SC transplantation. Islets loaded either on an O₂ transporter (n=4) or Neg-CTL device (n=4) were transplanted into the space inside of the prevascularized capsule. The capsule opening and skin wound were separately sutured to close, except for the cannula portion of the device penetrating through the opening. These surgical procedures were performed under general anesthesia.

Observation of islet recipients and graft removal

Recipients were monitored for two weeks after transplantation. Non-fasting blood glucose levels and body weight were measured twice a week. On day 7, in vivo viability and functional assessments were performed (described below). On day 8, the graft, including the O₂ transporter or Neg-CTL device and the surrounding tissue, was resected. The functional assessments were repeated on day 14.

In vivo viability assessment of transplanted islets

In vivo viability assessment of the transplanted islets was performed on day 7, as previously described [55]. Briefly, the bioluminescent intensity of the transplanted area (Luciferase (+) islet grafts) was measured using Lago × platform (Spectral Instruments Imaging, Tucson, AZ, USA), after injection of 15 mg luciferin (PerkinElmer Waltham, MA)/kg of body weight via tail vein under general anesthesia. The area under the curve of bioluminescent intensity between 1 and 5 min was calculated as the viability of transplanted islets. Assessments were performed on day 7 in both the O₂ transporter group (n=4) and Neg-CTL device group (n=4).

In vivo functional assessments of transplanted islets

An intraperitoneal glucose tolerance test and serum C-peptide (a byproduct of insulin production) measurement were performed as in vivo functional assessments of transplanted islets. To assess glucose tolerance, recipient rats were fasted for 8 h before an intraperitoneal administration of glucose solution at 2 g glucose/kg body weight [56]. Blood glucose was measured at 0, 15, 30, 60, and 120 min after glucose injection. Rat serum C-peptide was measured at 30 min after glucose injection using a rat C-peptide ELISA kit (Mercodia, Uppsala, Sweden). These assessments were performed on day 7 (with islet graft) and day 14 (without islet graft) in both the O₂ transporter group (n=4) and Neg-CTL device group (n=4).

Histology

Islet grafts were resected, together with devices and surrounding tissue, on day 8 post-transplantation. The graft tissue was examined under a dissecting microscope (IX50, Olympus, Tokyo, Japan) and fixed in modified Davidson’s fixative [57] for immunofluorescent staining. Guinea pig anti-insulin (Agilent Technologies, Santa Clara, CA) was used as primary antibody (1:1,500 dilution). Images were captured using an IX50 fluorescence microscope (Olympus).

Data analysis

Data were analyzed using JMP 9.0.0 (SAS Institute, Cary, NC) and reported as mean ± standard error (SEM). For statistical comparisons between two groups, we performed Student’s t-tests to compare means. p < 0.05 denotes statistical significance.

Results

Simulation reveals maintenance of O₂ within the graft layer on the O₂ transporter

The balance of O₂ consumption and O₂ supply is the major factor that determines pO₂ in the microenvironment of transplanted islets. Because O₂ consumption by the graft varies physiologically, it is critical to design an O₂ transporter that provides sufficient O₂ to meet the variable OCR of the graft. The OCR of pancreatic islets changes depending on the glucose concentration [49]. We measured the OCR of isolated rat islets in low glucose followed by high glucose solution (Suppl.1A). The high glucose environment induced a higher OCR than the low glucose environment; the average OCR in low and high glucose solution was 6.67 ± 1.74 and 8.85 ± 1.49 pmol/min/islet, respectively (Suppl.1B). We used this OCR data and other parameters shown in Table 1 to perform a pO₂ simulation of an islet layer on the O₂ transporter or the Neg-CTL device in low or high glucose environments (Fig.2A, left). Detailed pO₂ simulations at the level of an individual islet (150 μm in diameter) on the O₂ transporter are also shown (Fig.2A, right). pO₂ was greater on the O₂ transporter than on the Neg-CTL device regardless of the glucose environment. Although the islet graft consumes more O₂ in a high glucose environment, pO₂ on the O₂ transporter was similar in both high and low glucose environments, which indicates that the O₂ supply is overwhelmingly sufficient to meet the OCR need mediated by the glucose environment. We graphed pO₂ on the O₂ transporter and Neg-CTL device as a function of distance from the transporter surface, taking OCR fluctuations (average OCR value ± 1 SEM, calculated in Supplemental Figure 1) into account (Fig.2B). We showed that within the graft layer (0–150 μm), the O₂ transporter provided >120 mmHg pO₂ regardless of glucose concentration (average pO₂ in the graft layer was 138 and 133 mmHg in the low and high glucose environments, respectively). In contrast, average pO₂ in the graft layer on the Neg-CTL device was 5.6 and 4.0 mmHg in the low and high glucose environments, respectively. The steepest pO₂ drop occurs throughout the islet region, indicating that the transporter operates with minimal diffusional resistance. To analyze the graft capacity of the device, we employed pO₂ simulation of standard (600 islets) and high graft density (1,200 or 2,400 islets) in a 150 μm-graft layer on the O₂ transporter under high glucose conditions (Fig.2C). Despite having twice the density of the standard 600 islets, the O₂ transporter maintained >100 mmHg pO₂ in the entire graft layer of 1,200 islets in simulation; for 2,400 islets, pO₂ dropped relatively sharply to 50 mmHg at the top layer of the graft. It should be noted that pO₂ inside the device remained relatively high under these conditions, indicating that a diffusion-limitation, rather than a supply-limitation, occurs within the dense graft layer.

Figure 2: Simulation reveals maintenance of O₂ within the graft layer on the O₂ transporter.

(A) pO₂ simulations of an islet graft (600 islets) on the O₂ transporter or Neg-CTL device in low or high glucose environments as indicated. Overviews show a cross section of the O₂ transporter (left) in which the graft is considered to be a flat layer, and detailed views show islet-level scale (right, islets ~150 μm in diameter are depicted with black circles). Colored scale indicates range of pO₂ in mmHg. pO₂ on the O₂ transporter was higher than on the Neg-CTL device regardless of glucose environment. (B) pO₂ simulation graphed as a function of distance from the transporter. The simulation employed the OCR of 600 islets measured in low and high glucose conditions described in Supplemental Figure 1, with OCR values ranging ± 1 SEM (indicated by error bars in the graph). The Neg-CTL device showed a minimal pO₂ range (i.e. no error bars are seen). The O₂ transporter provided >120 mmHg pO₂ in the graft layer (0–150 μm wide, hatched in gray), regardless of glucose environment. In contrast, pO₂ in the graft layer on the Neg-CTL device was <6.5 mmHg, and the high glucose environment induced a slightly lower pO₂. (C) pO₂ simulation of varying graft densities (~600 (standard), 1,200, or 2,400 (high) islets in the same area) revealed the graft capacity of the device. The simulation employed islet OCR measured in high glucose conditions. The O₂ transporter maintained >100 mmHg pO₂ in the entire graft layer of 1,200 islets, which is two times higher than the graft density than the 600 islets used for in vitro and in vivo studies.

O₂ transporter provides a high O₂ environment for an islet graft in hypoxic solution in vitro

We used an O₂ probe to measure the actual pO₂ of 600 islets contained in the O₂ diffuser or Neg-CTL device in a hypoxic solution for 2 h. Fig.3A, 3B, and 3C show a schematic and images of the experimental set-up. We submerged the O₂ diffuser in a hypoxic solution at 45 mmHg pO₂, mimicking the rat SC site, and exposed the end of the cannula to the ambient air. For accurate measurement of the microenvironmental pO₂ on and near the transporter, we meticulously adjusted the placement of the O₂ probe so that the tip was situated a specific distance from the transporter surface (ranging from 0–2,000 μm). We found that the O₂ transporter maintained significantly higher pO₂ than the Neg-CTL device at distances of 0, 500, and 1,000 μm (Fig.3D). On the device surface (0 μm), the O₂ transporter provided 141 mmHg pO₂ compared to 14.0 mmHg for the Neg-CTL device (p<0.0001). Islets seeded on the O₂ transporter were fully covered by an oxygenated layer, with pO₂ >125 mmHg. In contrast, islets that received no O₂ supply on the Neg-CTL device were exposed to hypoxic conditions of <25 mmHg pO₂. These in vitro data validate the computational simulation shown in Fig.2B. Lastly, we examined the in vitro biocompatibility of islets with the O₂ transporter material (PDMS coated with fibronectin) by culturing them on the transporter for 7 days in a conventional 21% O₂ culture condition (Fig.3E), showing well-maintained rat islets.

Figure 3: O₂ transporter provides a high O₂ environment for an islet graft in hypoxic solution in vitro.

(A) Schematic of the in vitro measurement set-up. The O₂ diffuser was submerged in hypoxic PBS (pO₂ = 45 mmHg, the same pO₂ as previously measured in the rat subcutaneous site) at 37°C, and the end of the cannula was exposed to ambient air for 2 h. pO₂ was measured on and near the O₂ transporter or Neg-CTL device by placing the tip of an O₂ probe at a specific distance from the device surface, ranging from 0–2,000 μm. The oxygen probe was set on a rotary positioner for precise measurement. (B) Photograph of the measurement set-up. The O₂ transporter device was placed on a spatula to stabilize it. (C) Photographs of the set-up with the O₂ probe tip at indicated specific distances. pO₂ was measured at stepwise distances between 0 and 2,000 μm from the surface of the device. (D) pO₂ as a function of distance from the transporter surface. The O₂ transporter maintained significantly higher pO₂ than the Neg-CTL device at distances of 0, 500, and 1,000 μm (n=3 per group, * p<0.05). The islet graft (0–150 μm wide, hatched in gray) was fully covered by a >125 mmHg pO₂ layer on the O₂ transporter, but experienced limited pO₂ on the Neg-CTL device. (E) Representative images of in vitro biocompatibility testing of the O₂ transporter material (PDMS coated with fibronectin) with rat islets. Islets were cultured on the O₂ transporter for 7 days in a tissue culture incubator under air plus 5% CO₂ (pO₂ in the culture media was approximately 160 mmHg). Evaluation using live/dead staining (green, live cells; red, dead cells) indicated the material was biocompatible with rat islets. Scale bar: 1 mm.

Islets engrafted with the O₂ transporter show improved survival and function over those engrafted with the Neg-CTL device in vivo

Following demonstration of higher pO₂ on the O₂ transporter than the Neg-CTL device in computational simulation and in vitro experiments, we performed islet transplantation using the O₂ transporter and examined graft viability and function in vivo. Because more than half of the islets are typically lost in the hypoxic in vitro environment within a week [51], we assessed improvements in short-term (1 week) islet graft survival using the O₂ transporter. We prepared 600 islets on the O₂ transporter (or blocked Neg-CTL device) as shown in Fig.1H and Fig.1I. The islets/device were transplanted into the SC site on the back of syngeneic diabetic rats (n=4 per group), and the end of the cannula was left exposed to ambient air (Fig.4A). Following transplantation, blood glucose levels remained high in both the O₂ transporter and Neg-CTL groups (Fig.4B). Body weight on days 3 and 7 was significantly increased in the O₂ transporter group compared to the Neg-CTL (112.3% vs. 104.7% on day 7, p=0.0350); however, after removal of the graft, body weight decreased in both groups and showed no significant difference on day 10 or 14 (Fig.4C). In vivo viability tests on day 7 demonstrated significantly higher graft survival in the O₂ transporter group compared to the Neg-CTL group (Fig.4D and 4E, p=0.0444). We assessed graft function by performing a glucose tolerance test on day 7 (with graft) and on day 14 (after graft removal). Better glucose tolerance with the islet graft was shown on day 7 compared to that without the graft on day 14 in O₂ the transporter group (Fig.4F, p=0.0242 at 15 min and p=0.0197 at 30 min), which indicates the functional engraftment. On the other hand, no improvement was seen in glucose tolerance of the graft in the Neg-CTL group, indicating that the islet graft function was limited. This result was also backed up by serum C-peptide, a byproduct of the insulin produced from islet beta cells. It was significantly increased on day 7 by islet transplantation with the O₂ transporter but not the Neg-CTL device (Fig.4G, p=0.0104). We resected and examined the graft on the O₂ transporter on day 8; it showed islet graft in recipient tissue, with neovascularization toward the graft (Fig.4H). Immunohistochemistry against insulin, which is predominantly expressed in the beta cells in islets, revealed that the graft on the O₂ transporter contained more viable beta cells than that on the Neg-CTL device (Fig.4I).

Figure 4: Islet graft shows improved survival and function on the O₂ transporter compared to the Neg-CTL device in vivo.

To confirm graft viability and function in vivo, pancreatic islets were placed onto the O₂ transporter or Neg-CTL device and transplanted into the dorsal subcutaneous tissue of syngeneic diabetic rats (n=4/group). (A) Appearance of the graft site after transplant surgery. The diffuser portion with the graft was placed in the subcutaneous tissue, and the end of the cannula was left exposed to the ambient air for O₂ uptake (arrowheads). (B) Blood glucose was measured on indicated days, and showed consistently high glucose levels in both the O₂ transporter (solid line) and Neg-CTL (dotted line) groups. (C) Body weight was significantly increased in the O₂ transporter group (solid line) compared to the Neg-CTL group (dotted line) on days 3 and 7 (* p<0.05). (D) Representative in vivo bioluminescent intensity images of cell viability on day 7. (E) Quantitative analysis of the viability shown in (D) demonstrated significantly higher graft survival in the O₂ transporter group than the Neg-CTL group (p<0.05). (E) Quantitative analysis of the viability shown in (D) demonstrated significantly higher graft survival in the O₂ transporter group than the Neg-CTL group (*p<0.05). (F) Graft function examined by glucose tolerance test on day 7 (with graft, solid line) and on day 14 (without graft, after removal, dotted line). Islet grafts on the O₂ transporter showed improved glucose tolerance (* p<0.05 at 15 min and 30 min), whereas those on the Neg-CTL device did not. (G) Graft function examined by serum C-peptide on day 7 demonstrated a significant increase in graft function with the O₂ transporter compared to the Neg-CTL device (* p<0.05). (H) Resected tissue on the O₂ transporter showed islet graft on the recipient tissue, including neovascularization toward the graft. Representative image; scale bar: 5mm (left figure) and 500 μm (right figure). A schematic (lower panel) shows the preparation of the microphotographs. (I) Representative immunohistochemistry micrographs revealed that islet graft (brown staining indicates insulin-producing cells, arrowheads) was better maintained in the graft on the O₂ transporter (lower panel) than the Neg-CTL device (upper panel). Dotted lines indicate the boundary between O₂ transporter device (or Neg-CTL device) and islet graft. Asterisks (*) indicate the gel embedding islet grafts. Scale bar: 500 μm.

Discussion

In this study, we fabricated a simply structured O₂ transporting system to improve cell transplantation in the SC site. We utilized SC transplantation of pancreatic islets as a model to test the O₂ transporter in this study because 1) islets are known as hypoxia-susceptible cells and 2) the SC site has a harsh O₂ environment, which has hindered the establishment of islet transplantation in this site [18–20]. Our O₂ transporter demonstrated effective O₂ transport properties in computational simulations and in vitro experiments. Importantly, islet transplantation with the O₂ transporter maintained better graft viability and function in the SC site in vivo, compared to the Neg-CTL device.

Our O₂ transporter is less than 1 mm in thickness including the rim structure, making it flexible enough to implant into the SC site with minimal irritation. To maintain the patency of the gas compartment within the diffuser, an array of PDMS micropillars span the top and bottom membranes. As the diameter of the micropillars is 300 μm, occupying <15% of the area of the membrane surface, they have negligible effect on O₂ supply to the graft. The dimensions of the device can be expanded to accommodate more cells or higher O₂ demand as needed. Regarding the O₂ range that can be supplied to the transplanted graft, this device can supply up to atmospheric O₂ levels (159 mmHg) depending on graft O₂ consumption. The O₂ level in the native SC site where we transplanted pancreatic islets in this study is 45 mmHg [22]. Therefore, the O₂ transporter drastically improves the surrounding O₂ environment of the graft.

We used pancreatic islets in this study. Isolated pancreatic islets are characterized by high metabolic activity and are susceptibility to hypoxia [58–60]. This susceptibility is caused by the metabolic activity as well as the structure of the islets. An islet is a cell aggregate that is typically 150 μm in diameter [46] and can be isolated from native pancreas parenchyma. Once islets are isolated and until the revascularization that occurs following successful transplantation and engraftment, they lack connections to blood vessels, which creates a diminishing O₂ gradient toward the islet core [1, 44, 60]. These characteristics are ideal for demonstrating proof-of-concept for improved local O₂ supply and transplantation efficiency by the O₂ transporter. However, variety in islet size and graft aggregation are still the issues that may hinder uniform oxygenation [60], therefore, preventive method for the aggregation in SC site cell transplantation would be another critical issue to solve. Recent advances in stem cell research show the promise of generating functional cell aggregates such as pseudo-pancreatic islets [2–4]. Cell transplantations using such cell aggregates are expected to have similar issues of O₂ depletion in the transplanted site, because they also lack vascular connections in the early stage after transplantation. Thus, the implications drawn from use of the O₂ transporter in this islet model are relevant to development of an O₂ supporting system for future cell replacement therapy.

Pancreatic islet metabolism changes according to the glucose environment [49]. In this study, we utilized diabetic rats, whose blood glucose is abnormally high and unstable, as transplant recipients. A similar situation occurs in diabetic patients after therapeutic islet transplantation. The metabolic fluctuations in islet O₂ demand may be a challenge for creating a consistently oxygenated site. We simulated the O₂ environment on the O₂ transporter using a variable range of islet graft OCR before fabrication (Suppl.1A,B and Fig.2). Such simulations are important for estimating variance in the actual transplants and ensuring that the transporter supplies sufficient O₂ under various conditions. Physiological changes in cell metabolism due to multiple factors within a graft also occur in other types of cell transplantation, and may induce variance of the graft O₂ environment and unexpected outcomes. In the simulations of this study, O₂ supplied by the transporter was found sufficient to meet the metabolic demands of cells in both high and low glucose environments. In fact, the high and low glucose environments demonstrated little difference in O₂ environment in the simulations (Fig.2A,B). This may be explained by the overwhelmingly high O₂ supply provided by the transporter compared to the OCR of the islet graft; the device was still operating in a consumption-limited mode, not a supply-limited mode, in the high glucose environment.

We measured in vitro pO₂ within islet grafts on the O₂ transporter, using the same number of the islets on the transporter as for in vivo transplants. The in vitro data fit the simulation data well, with similar curves, which validated the accuracy of the computational simulation; both experiments demonstrated a high O₂ environment (pO₂ >120 mmHg) in the graft layer. A discrepancy observed between the simulation and in vitro measurements is explained as follows. A steep drop in pO₂ in the narrow graft layer was represented only in the simulations but not observed in in vitro measurements due to the resolution. Compared to the simulation data, our in vitro data demonstrated a slightly more pronounced decrease in O₂ by distance from the O₂ transporter surface; pO₂ at 1,000 μm (from the O₂ transporter surface) was approximately 75 mmHg in vitro (Fig.3D), whereas it was around 100 mmHg in the simulation (Fig.2B). This may be because 1) hypoxic PBS in vitro was circulating rather than static and 2) islet metabolism was higher in vitro than in the simulation. Overall, in silico simulation was a powerful tool for estimating pO₂ at high resolution and informing design of the transporter.

In this study, diabetes was induced before islet transplantation into recipient rats by administration of STZ, which specifically destroys insulin-secreting beta cells in the native pancreas. Therefore, the beta cells in the transplanted islets played a primary role in secreting insulin and lowering blood glucose. Although blood glucose did not differ significantly between O₂ transporter and Neg-CTL groups (Fig.4B), we demonstrated enhanced viability and functionality of islet grafts on the O₂ transporter (Fig.4 D,E,F,G). In addition, increased body weight in the O₂ transporter group indicates a recovery from STZ-induced diabetic toxicity [61–63]. These data indicated that 600 islets were not sufficient to reverse diabetes in this model, but clearly demonstrated the beneficial effects of oxygenation by the O₂ transporter. In addition, serum C-peptide secretion in the Neg-CTL group demonstrated no significant difference between day 7 (with graft) and day 14 (without graft), indicating that the graft was mostly destroyed within 7 days. This is supported by histology results that showed fewer islet grafts remained in the Neg-CTL device, compared to islet grafts maintained on the O₂ transporter (Fig.4I).

We employed short-term in vivo experiments in which the O₂ transporter was implanted for only one week. Given that revascularization of islet grafts takes ~10–30 days to complete [64, 65], longer oxygen support may be beneficial. On the other hand, using an in vivo rat SC transplant platform exactly the same as this study, we showed that a three-day O₂ inhalation treatment (which increased SC pO₂ to 140 mmHg, similar to the O₂ level provided by the O₂ transporter) significantly improved islet engraftment [22]. Furthermore, prolonged oxygenation may have adverse effects on revascularization because hyperoxia suppresses angiogenesis [66]. Therefore, the optimal duration of post-transplant oxygenation remains under debate.

We developed our simply structured O₂ transporter as an expandable device with potential for clinical applications. Because the diffuser portion is made of biocompatible PDMS, removal of this component is not required after engraftment; only removal of the cannula is required to close the wound permanently (Suppl.2A). In this case, the inner surface of the diffuser should be kept sterile until closure of the skin, to prevent contamination and infection after removal of the cannula. This issue can be solved by placing a filter system connected to the end of cannula instead of open-ended tubing. We fabricated several options for this purpose, including a filtration system with 0.22 μm polyethersulfone filter membrane (Suppl.2B), and a small scratch-resistant receiver made of hydrophobized Vycor (Corning, Corning, NY) with 5–10 nm pores (Suppl.2C). This pore size is sufficiently small to prevent bacterial contamination into the O₂ transporter, but large enough for O₂ molecules to pass without hindrance. Furthermore, those systems are water-resistant because 1) the filter materials (polyethersulfone and Vycor) are hydrophobized [67] and 2) the inner and outer pressures of the device are balanced, which hinders the infiltration of water into the inside of the device.

For clinical application of the O₂ transporter device, we propose the human abdominal SC site as a candidate transplant site (Suppl.2A). We calculated the estimated size of the O₂ diffuser when islets are seeded at the same density as in this study: a 16.4-cm-diameter diffuser would be required to hold 300,000 human islet equivalents (IEQ) [46, 68], the minimal islet number required for human islet transplantation. However, given the high oxygenation capacity of the device, which we demonstrated maintains >100 mmHg pO₂ even at twice this graft density (Fig.2C), this will theoretically allow an 11.6-cm-diameter diffuser to accommodate 300,000 IEQ. We consider the abdominal SC site as a promising candidate not only because it has a great enough area to accommodate the device, but also because of the relative ease of access for postoperative care. Although additional simulations are required before fabrication of a large-scale device, our transporter demonstrates great potential to supply high-demand O₂.

We demonstrated the beneficial effects of the O₂ transporter using a SC site transplant model and a correspondingly short cannula. This cannula may be extended to transport O₂ from the ambient air to more distant sites, such as the abdominal cavity, by replacing the stainless steel cannula with flexible tubing (e.g. silicone). Parylene-C, a clinically proven chemical vapor deposition polymer, can be selectively deposited on portions of the device such as the flexible tubing and backside of the diffuser to make those surfaces practically O₂ impermeable, while maintaining biocompatibility and flexibility [69]. However, the inner diameter and length of the tubing should be taken into consideration, as they affect O₂ transport efficiency. The O₂ diffuser component is inherently resistant to collapse due to external pressure, by virtue of the internal array of spacing pillars. In summary, the simple structure of this O₂ transporting system will allow various modifications to support a wide variety of clinical applications that require changing a hypoxic microenvironment into an oxygen-rich site.

Data availability statement

The raw data required to reproduce the findings are found in the methodology and the supplemental file.