Molecularly Tailored Interface for Long-Term Xenogeneic Cell Transplantation

Abstract

Encapsulation of therapeutic cells in a semipermeable device can mitigate the need for systemic immune suppression following cell transplantation by providing local immunoprotection while being permeable to nutrients, oxygen, and different cell-secreted biomolecules. However, fibrotic tissue deposition around the device has been shown to compromise the long-term function of the transplanted cells. Herein, a macroencapsulation device design that improves long-term survival and function of the transplanted cells is reported. The device is comprised of a semipermeable chitosan pouch with a tunable reservoir and molecularly engineered interface. The chitosan pouch interface decorated with 1,12-dodecanedioic acid (DDA), limits the cell adhesion and vigorous foreign body response while maintaining the barrier properties amenable to cell encapsulation. The device provides long-term protection to the encapsulated human primary hepatocytes in the subcutaneous space of immunocompetent mice. The device supports the encapsulated cells for up to 6 months as evident from cell viability and presence of human specific albumin in circulation. Solutions that integrate biomaterials and interfacial engineering such as the one described here may advance development of easy-to manufacture and retrievable devices for the transplantation of therapeutic cells in the absence of immunosuppression.

Graphical Abstract

A macroencapsulation device for xenogeneic cell transplantation is developed. The device consists of a porous chitosan pouch with an adhesive inner surface that promotes cell viability and a cell repellent outer surface that minimizes fibrosis. The device supported long term (6 months) transplantation of human primary hepatocytes into an immunocompetent mice.

1. Introduction

Cell transplantation therapy holds immense potential for treating a long and growing list of diseases resulting from impaired cell functions.^[1] In addition to contributing to tissue repair through differentiation and paracrine factors, cell transplantation can also be used as a “drug store” to deliver key biomolecules, such as insulin to treat type 1 diabetes or factor VII or factor IX (FIX) to address bleeding disorders, to circumvent compromised cell functions.^[2] Strategies such as vascularization have been used to improve the viability and function of the transplanted cells.^[3] However, the availability of human primary cells cannot begin to meet the clinical need. Advancing technologies that circumvent some of the barriers of allogeneic and xenogeneic transplantation could significantly advance cell transplantation and its clinical outcomes. One of the major hurdles of allogeneic/xenogeneic cell transplantation is the host immunological barriers that are incited by the transplanted cells resulting in their rejection.^[4] Toward this, biomaterial supported cell encapsulation approaches, micro- and macro-encapsulations, have been extensively investigated but with varying outcomes.^[5] These biomaterial-based approaches rely on the premise that the semipermeable barriers can offer local immunoisolation and protect the encased cells from immune rejection, without the use of systemic immunosuppression, while facilitating exchange of nutrients, gases, and other low molecular weight molecules between the cell-laden devices and the host.^[6] Though the encapsulation of cells within a semipermeable device could provide immunoisolation, these approaches often suffer from foreign body response (FBR) elicited by the biomaterials resulting in formation of fibrous connective tissue around the implant.^[7] This leads to implant isolation, nutrient limitation, hypoxic condition—all of which induce cell death within the device leading to transplant failure.^[8] Hence, an ideal biomaterial based approach requires both immunoisolation and minimal fibrous capsule formation to improve the therapeutic outcome of cell transplantation.

Biomaterial devices that support cell encapsulation with adequate volume capacity and optimal barrier properties can be a way forward in cell transplantation. To this end, various approaches such as modification of the biomaterial devices with chemical moieties to mitigate fibrosis,^[9] surface modification with antifouling molecules such as PEGylation or zwitterions,^[10] or immobilization of programmed cell death-1 ligand (PDL-1) to regulate the local immune response^[11] have been exploited in the recent years. Other approaches to support macro-encapsulation include perfusion of the devices with oxygen^[12] or promoting vascularization adjoining to the implant.^[13] While these technological and conceptual advances have offered varying degrees of success, there is still a need for the development of macro-encapsulation approaches enabling allogeneic and xenogeneic cell transplantations.

Here, we describe the development of a macroscopic biomaterial device that supports xenogeneic cell transplantation while housing a large number of cells. Specifically, we have devised a semipermeable chitosan pouch with an adhesive cytocompatible inner surface and a repulsive outer surface limiting fibrosis (Figure 1a). Protein and cell adhesion to the outer surface of the chitosan pouch was limited by the tethering of a layer of 1,12-dodecanedioic acid (DDA) molecules, which has a long alkyl chain –(CH₂)₁₀– terminating with a carboxylic acid (COOH) group. Previously, we have shown that “reach and accessibility” of the alkyl chain terminating with carboxylic acid groups can be leveraged to achieve unique function and regulate interfacial properties including interactions with cells and tissues.^[14] The potential of the modified chitosan pouch to promote xenogeneic cell transplantation was examined by transplanting human primary hepatocytes into immunocompetent mice. We used hepatocyte transplantation as a proof of concept due to its clinical importance. Hepatocyte transplantation that can augment liver functions has emerged as an alternative to liver transplantation or to bridge the gap until a suitable organ is available.^[15] While studies have demonstrated the preliminary efficacy of hepatocyte transplantation in clinics,^[16] maintaining the viability and functions of the transplanted cells is a major obstacle in extending the promising initial success. For example, studies have demonstrated that up to 70% of the transplanted hepatocytes are eliminated by early phagocytic immune response, without significant differences between allogeneic and syngeneic transplantation.^[17] So approaches that support cell transplantation with long-lasting effect will advance the ultimate goal of increasing hepatocyte transplantation efficacy and its clinical applications.

Figure 1.

Development of an immune protective cell encapsulation pouch. a) Schematic illustration of the chitosan pouch. The semi-permeable pouch membrane allows nutrients to diffuse in and cell-secreted products to diffuse out. The outer wall of the pouch is modified with DDA molecules to minimize fibrosis. b) Schematic depicting fabrication of the chitosan pouch along with digital images of the pouch (Scale bar, 0.5 cm) and live/dead staining of encapsulated hepatocytes within the pouch, Green: live cells and Red: dead cells (Scale bar, 100 μm). c) Computational analyses of oxygen distribution within the pouch for the optimized wall thickness and pouch diameter showed no hypoxic region. d) Diffusion of encapsulated 70 kDa and 150 kDa FITC-dextran from the chitosan pouch for 24 h.

2. Results

2.1. Fabrication and Characterization of the Cell Transplantation Device

The cell transplantation device is a chitosan pouch with controllable wall thickness having an adhesive inner and a repulsive outer surface and a tunable reservoir to house cells (Figure 1a). We have harnessed the tunable solubility (which allows seamless casting of membranes), and intrinsic adhesive properties (that enable bonding between the membranes) of chitosan to create the device. Specifically, two layers of chitosan membranes were cast around a sacrificial template (e.g., paraffin or polytetrafluoroethylene (PTFE tube)). The removal of the sacrificial template resulted in a pouch with a hollow interior and an inlet that can be used to load the cells (Figure 1b; Figure S1a, Supporting Information). The fabrication process also allows formation of pouches with different dimensions and shapes (Figure S1b, Supporting Information). For a specific dimension, the thickness of the pouch wall can be tuned by varying the volume of the chitosan solution (Figure S1c, Supporting Information).

Nutrient and oxygen transport are major limiting factors in device-assisted cell transplantation where diffusion is the sole mode of transport.^[12b] The pouch wall is expected to facilitate mass transfer due to the short diffusion distance across the wall. We modeled the diffusion of small molecules across the pouch by using oxygen as a model system. We chose oxygen molecule because of the extensive experimental data available for oxygen transport and its consumption by cells. Also, hypoxia within the cell device is a serious problem affecting viability of the transplanted cells.^[12b,18] The pouch was modeled as a cylinder of 11 mm length with an outer wall having a thickness of t as shown in Figure S2a, Supporting Information, where r₁ is the radius of the interior of the pouch, which is filled with cell-laden fibrin. We compared the oxygen concentration at the center of the pouch for r₁ varying from 0.75 to 1.5 mm and t varying from 100 to 300 μm while keeping the length constant. The solution of the mathematical model showed a steady state oxygen concentration gradient between the center and the periphery of the pouch with the oxygen concentration lowest at the center (Figure S2b, Supporting Information). As shown in Figure S2c, Supporting Information, the model predicted that the pouch with dimensions of r₁ = 1 mm and t = 150 μm ensures the oxygen concentration at the center to be 0.031 mol m⁻³, which is greater than 0.01 mol m⁻³ (oxygen concentration considered to be detrimental to cells) (Figure 1c; Figure S2d, Supporting Information).^[19] Based on these results, we used pouches with a dimension of 1 mm radius and 150 μm wall thickness for further experiments. The diffusion profile of the pouch was experimentally determined from the diffusion of fluorescein isothiocyanate (FITC) conjugated dextran molecules with molecular weights of 70 and 150 kDa. As seen in Figure 1d, majority of 70 kDa dextran molecules were diffused out of the semipermeable pouch within 24 h, while minimal diffusion of 150 kDa dextran across the pouch was noticed (Figure 1d). The dextran diffusion suggests that the chitosan pouch is permeable to nutrients and cell secreted molecules such as insulin (5.7 kDa), albumin (63 kDa), FIX (55 kDa), and alpha 1 antitrypsin (54 kDa). Previous studies involving islet cells have shown that alginate hydrogels that would prevent diffusion of molecules above 250 kDa provide adequate barrier properties and support viability of the cells within the device.^[20]

Next, we modified the outer surface of the pouch with DDA molecules and created a layer of bristle brush structures (Figure 1a). The DDA tethering to the chitosan surface, as described in methods, was achieved via amide-coupling by using N-ethyl-N′-(3-dimethylaminopropyl) carbodiimide hydrochloride/N-hydroxysuccinimide (EDC/NHS) chemistry where the primary amine groups of the chitosan reacted with the terminal carboxylic acid group of the DDA (Figure 2a). Successful modification of chitosan surface with the DDA was confirmed by Fourier-transform infrared (FTIR) spectroscopy and thionin acetate (THA) calorimetric assay. FTIR analysis showed appearance of peaks at 1656 cm⁻¹ and 1561 cm⁻¹, which correspond to the amide I and amide II stretching vibrations of –NHCOCH₃ groups of chitosan (Figure 2b). The spectra further showed a broad peak at 1631 cm⁻¹ which corresponds to the terminal carboxylate group of DDA indicating successful conjugation of DDA to the chitosan surface (Figure 2b). Chitosan surfaces with varying extent of DDA grafting were generated by controlling the DDA concentration (2–16 mg mL⁻¹) in the reaction mixture, which resulted in surfaces with a grafting density of 8.5 × 10⁻¹⁰ to 4.0 × 10⁻⁹ mol cm⁻². As indicated by the THA dye adsorption assay, increasing the amount of DDA molecules beyond 8 mg mL⁻¹ in the reaction mixture had no significant effect on DDA conjugation (Figure 2c; Figure S3a, Supporting Information). We have also determined the effect of DDA modification on chitosan degradation as the DDA modification could improve the stability of the chitosan, which will in turn offer long-lasting barrier properties in vivo.^[21] To this end, both the unmodified and DDA-modified chitosan membranes were incubated in 0.2 mg mL⁻¹ lysozyme solution and monitored for weight change as a function of time. The DDA-modified chitosan membranes exhibited minimal degradation by four weeks compared to the corresponding unmodified chitosan membranes (Figure 2d). Similar to unmodified chitosan, the DDA-modified chitosan pouches allowed the diffusion of 70 kDa FITC-dextran while minimal diffusion of 150 kDA FITC dextran was observed (Figure S3b, Supporting Information). The calculated diffusion coefficients (Table S1, Supporting Information) showed no difference between the unmodified and DDA-modified chitosan pouches for the 70 kDa FITC dextran, while the diffusivity of the 150 kDa FITC dextran was found to be lower for the DDA-modified pouches. The observed lower diffusion across the DDA-modified membranes could be due to the presence of the DDA molecules; the layer of DDA molecules could serve as an additional diffusion barrier.^[22]

Figure 2.

Modification of the chitosan surface. a) Reaction scheme of chitosan modification with DDA molecules b) FTIR analysis of the chitosan membranes before and after DDA modification. c) Quantification of DDA grafting density using thionin acetate (THA) calorimetric assay. d) In vitro degradation of chitosan membranes. Remaining weight percentage of unmodified and DDA-modified chitosan membranes incubated in 0.2 mg mL⁻¹ of lysozyme solution as a function of time.

2.2. DDA Modification Reduces In Vitro Cell Adhesion on Chitosan Surface

Foreign body response (FBR) to the biomaterial is another serious constraint limiting the success of device-assisted cell transplantation strategies. Protein adsorption to the biomaterial and subsequent cell adhesion are considered key to FBR.^[23] The DDA modification-dependent changes in protein and cell adhesion were determined in vitro as a function of time. Among major blood proteins, fibrinogen adsorption onto implants has been highly implicated in the initiation of immune cell recruitment and their activation while collagen is the major component of fibrotic tissues.^[24] As shown in Figure 3a, the amount of adsorbed fibrinogen and collagen was decreased monotonically with DDA conjugation. The chitosan surfaces having a DDA surface density of 3.5 × 10⁻⁹ mol cm⁻² showed minimal fibrinogen and collagen adsorption. We next examined the effect of DDA modification on cell adhesion by using mouse primary mononuclear cells (MNCs) and NIH 3T3 fibroblasts. Concomitant with the protein adsorption, the DDA modification prevented cell adhesion albeit with subtle differences (Figure 3b,c). Initial attachment of monocytes, from the infiltrated mononuclear cells, to foreign materials and their differentiation into macrophages are key initial events in the inflammatory response to biomaterial implants.^[23,25] To this end, freshly isolated MNCs from mouse bone marrow were seeded onto chitosan membranes with different amounts of DDA conjugation and cultured under macrophage differentiation media for 6 days and their adhesion onto the chitosan membranes was studied. Figure 3b shows the adhesion of MNCs onto the chitosan surfaces with varying extent of DDA modification. While unmodified chitosan surfaces supported the adhesion of differentiating macrophages, the DDA-modified surfaces prevented their adhesion in a grafting density dependent manner; chitosan surfaces with 1.8 × 10⁻⁹ and 3.5 × 10⁻⁹ mol cm⁻² DDA supporting the least cell adhesion (Figure 3b). Moreover, adhered cells on unmodified chitosan surfaces showed a well spread morphology while those on DDA-modified surfaces largely displayed a rounded morphology (Figure S4a, Supporting Information). Adhesion and growth of fibroblasts showed a similar behavior that the cell adhesion and proliferation reduced significantly with the DDA modification (Figure 3c; Figure S4b,c, Supporting Information). While unmodified chitosan membranes supported adhesion of fibroblasts (45.25 ± 8.42 cells cm⁻²), significantly less cells were found on DDA-modified chitosan membranes albeit with differences. The modified chitosan membranes with a surface density of 1.8 × 10⁻⁹ mol cm⁻² DDA molecules had a cell number of 32.75 ± 6.84 cells cm⁻², which was reduced to 13.5 ± 2.08 cells cm⁻² on chitosan surfaces with 3.5 × 10⁻⁹ mol cm⁻² of DDA. Furthermore, the cells on the unmodified chitosan membranes showed a ≈5-fold increase in cell number by day 7 post-plating. Although the chitosan membranes with a DDA grafting density of 1.8 × 10⁻⁹ mol cm⁻² supported lower initial cell adhesion, the adhered cells were able to proliferate. In contrast, the fibroblasts on chitosan surfaces with higher DDA grafting (3.5 × 10⁻⁹ mol cm⁻² of DDA), did not proliferate and maintained a rounded shape. These results demonstrate DDA dependent changes in the surface properties of chitosan membranes with the DDA-modified chitosan surfaces having a grafting density of 3.5 × 10⁻⁹ mol cm⁻² supporting the least cell adhesion and proliferation. Hence, DDA-modified chitosan surfaces with a grafting density of 3.5 × 10⁻⁹ mol cm⁻² were used for all the subsequent in vivo experiments.

Figure 3.

Effect of DDA modification on protein adsorption, cell adhesion, and Foreign Body Response. a) Adsorption of fibrinogen and collagen on chitosan surfaces with varying DDA grafting density. (*, †, ‡ p < 0.05 compared to 0, 8.5, 1.8 mol cm⁻²). b) Bright field images of mononuclear cells on different chitosan surfaces under macrophage differentiation media and quantification of adhered cells. (*, † p < 0.05 compared to 0, 8.5 mol cm⁻²). c) Bright field images showing NIH/3T3 cell adhesion at day 1 on chitosan membranes with varying DDA grafting densities and their quantification over a period of 7 days. (*, † p < 0.05 compared to day 1 and day 4). d) Immunofluorescent staining of macrophage marker (F4/80, red) and nucleus (blue) on retrieved chitosan membranes at days 3 and 14 following implantation. The white dotted lines delineate the membrane-tissue interface. e) Cellular deposition on implanted chitosan membranes was quantified by counting the total number of nuclei on each membrane using ImageJ software. f) Immunofluorescent images of fibrotic marker, α-SMA, on retrieved chitosan membranes 14 days after implantation. The white dotted lines delineate the membrane-tissue interface g) Masson’s trichrome staining of the membranes for collagen deposition. The black dotted lines denote membrane-tissue interface. h) Collagen thickness of fibrous tissues around the chitosan membrane was quantified by using ImageJ software. Scale bar, 200 μm.

2.3. DDA-Modified Chitosan Reduces Foreign Body Response

To examine whether DDA modification could mitigate FBR in vivo, both unmodified and DDA-modified (3.5 × 10⁻⁹ mol cm⁻²) chitosan membranes were implanted into the subcutaneous space of immunocompetent C57BL/6J mice. C57BL/6J mouse strain is shown to exhibit strong FBR to implants and similar to those observed in humans.^[26] The chitosan membranes were retrieved at day 3 and day 14 post-implantation and characterized for FBR. Previous studies have shown well developed fibrous tissue formation around the implant within two weeks.^[23] Initial inflammatory response to the chitosan membranes was examined on day 3 post-implantation by immunostaining for macrophage marker (F4/80) and cell nuclei (DAPI). DAPI staining of the unmodified chitosan membranes showed a thick layer of cells throughout the implant-host tissue interface (Figure 3d). In contrast, significantly less cell deposition was observed on DDA-modified chitosan membranes. Similarly, the F4/80 staining revealed very few macrophages on DDA-modified chitosan surfaces compared to the unmodified one. Characterization of the membranes at day 14 showed significantly higher cell deposition compared to day 3 (Figure 3d,e). Though some level of cell deposition was observed on DDA-modified chitosan membranes at day 14, the amount was significantly less compared to the unmodified chitosan membranes. The retrieved membranes were also stained for fibrotic markers, alpha-smooth muscle actin (α-SMA) and collagen-1, which showed significantly higher levels of α-SMA positive myofibroblasts on unmodified chitosan membranes compared to the DDA-modified one (Figure 3f). Concurrent with the presence of activated fibroblasts, a thicker collagen rich fibrotic tissue was observed around the unmodified chitosan membranes compared to the DDA modified one (Figure 3g,h).

2.4. DDA Modification Supports Long-Term Xenogeneic Transplantation of Hepatocytes

We next sought to examine the ability of DDA-modified chitosan pouches to support xenogeneic transplantation of human primary hepatocytes in immunocompetent mice. Both unmodified and DDA-modified pouches were loaded with the cells and implanted into subcutaneous space of immunocompetent C57BL/6J mice. The viability and function of the transplanted cells were assessed by quantifying the human hepatocyte secretory protein albumin in the peripheral blood as a function of time (Figure 4a). Analyses of peripheral blood following 1-month implantation showed presence of human specific albumin in both the cohorts (received unmodified and modified chitosan pouches). While both the cohorts showed presence of human albumin in circulation, animals that received DDA-modified pouches had significantly higher human specific albumin compared to those which received unmodified chitosan pouches. Furthermore, two of the recipients that received unmodified chitosan pouches showed no detectable levels of albumin in circulation. Analyses at 3-months post-implantation detected no significant human specific albumin in cohorts that received unmodified pouches except in two animals. On the other hand, all animals that received DDA-modified pouches showed significant human albumin in circulation similar to 1-month post-implantation. At 6-months post-implantation, while presence of human albumin in circulation was diminished, animals that received DDA-modified group still showed significant presence of human albumin in all animals while unmodified pouches implanted group showed no human albumin in the circulation (Figure 4a). The pouches were retrieved at 6 months post-implantation, and the cells were replated onto type I collagen-coated plates for 24 h to enable cell adhesion, fixed, and stained for hepatocyte markers. Consistent with the presence of albumin in circulation, very few cells were found within the unmodified pouch; the few cells present were loosely adhered and washed off easily from the culture plates. On the other hand, cells from DDA-modified pouches were adhered onto the dishes after replating. The adhered cells were positive for human hepatocyte markers, albumin and human cytokeratin 18 (CK18) (Figure 4b,c).

Figure 4.

DDA-modified chitosan pouches supported long-term xenogeneic transplantation of hepatocytes. a) Measurement of human albumin in the peripheral blood of mice that received unmodified or DDA-modified pouches as a function of time up to 6 months. Immunostaining of hepatocytes retrieved from the DDA-modified pouches following 6-months of implantation for b) CK18 and c) albumin. Scale bar, 100 μm. *p < 0.05 compared to unmodified group.

The retrieved pouches were also characterized for long-term fibrotic tissue formation which showed limited fibrosis around the DDA-modified pouches at 6 months. The gross appearance of the retrieved implants revealed that the unmodified chitosan pouches were encased by a thick fibrotic tissue whereas the DDA-modified pouches had minimal fibrotic tissue deposition (Figure 5a). Masson’s trichrome staining corroborated the above observations; the unmodified chitosan pouches were surrounded by thick, dense collagenous fibrotic tissues (Figure S5a, Supporting Information). Average fibrotic tissue thickness around the unmodified chitosan pouches post 6-month implantation was 927.96 ± 412 μm while that for the DDA-modified pouch was 211.45 ± 100 μm (Figure S5a, Supporting Information). Immunostaining for α-SMA further confirmed that the DDA-modified pouches had relatively less myofibroblasts compared to the unmodified pouches (Figure S5b, Supporting Information). We also analyzed the presence of blood vessels within the fibrous tissue around the implant by staining for endomucin. Both unmodified and DDA-modified chitosan pouches showed vascularization but with significant differences. Blood vessel density around the DDA-modified pouches was significantly higher compared to the unmodified pouches (Figure 5b). Moreover, limited blood vessels were observed at the pouch-host tissue interface adjacent to the pouch for the unmodified groups while DDA-modified groups showed dense vascularization at the interface, adjacent to the pouch (<150 μm from the pouch to the host tissue) (Figure S5c, Supporting Information). Finally, we have investigated whether the barrier properties of implanted pouches were breached after long-term implantation by analyzing macrophage invasion into the pouches. The presence of macrophages was apparent within the unmodified pouches (Figure 5c). In contrast, no macrophages were found within the DDA-modified pouches. Invasion of host inflammatory cells into the encapsulation device has been previously shown to affect viability of the transplanted cells in a rat model.^[27]

Figure 5.

Barrier properties of the unmodified and DDA modified pouches following 6-months Sub-Q implantation. a) Digital images of the retrieved pouches six month after implantation. Scale bar, 10 mm. b) Immunofluorescence images of endomucin (EMCN) showing presence of blood vessels around the pouches. White dotted line indicates the periphery of pouch interior. Scale bar, 100 μm. Quantified EMCN-positive area relative to the total deposited tissue area from the stained images is also shown. c) Immunofluorescence staining images of human albumin and F4/80. White dotted line indicates the periphery of pouch interior. Scale bar, 250 μm. d) COMSOL modeling results showing radial oxygen concentration profile of modified and unmodified pouches. e) Schematic comparing the barrier properties of the unmodified and DDA-modified pouch implants following implantation. *p < 0.05 compared to unmodified group.

We have also examined the impact of fibrous tissue layer around the chitosan pouch on mass transport of oxygen across the pouch. Toward this, we have used the experimentally determined thickness of fibrotic capsule (see Figure S5a, Supporting Information) around the unmodified and DDA-modified chitosan pouches. The COMSOL analyses suggest that the oxygen concentration at the center of the unmodified pouch with a fibrotic outer layer of ≈1000 μm thick is ≈0 mol m⁻³ (Figure 5d; Figure S6b, Supporting Information). In the case of DDA-modified pouch that had a thin (≈300 μm) fibrotic tissue layer, the oxygen concentration at the center of the pouch was found to be 0.02 mol m⁻³ (Figure S6b, Supporting Information), while it is less than the design criterion used (0.03 mol m⁻³), the value is higher than the pathological hypoxia of 0.01 mol m⁻³.^[19] Figure 5e summarizes the differences in barrier properties of unmodified and DDA-modified pouches following implantation and its potential effect on transplanted cells.

3. Discussion

In this study, we described the development and validation of a cell encapsulation device, chitosan pouch, with differential surface properties to support long-term xenogeneic cell transplantation. The device is fabricated by a simple method involving a sacrificial layer which allows easy tailoring of size and shape with adequate volume capacity to house the cells. That the chitosan membranes are soluble in water below pH<6 while being insoluble at physiological pH^[28] was leveraged to generate insoluble chitosan membranes. The semipermeable chitosan membranes exhibited intrinsic barrier property that prevented the infiltration of host cells including the immune cells while allowing transportation of nutrients and hepatocyte-secreted factors. To limit the host tissue response to the device, the chitosan surface interfacing the host tissue was modified with 1,12-dodecanedioic acid (DDA). Previously, we have shown that surface modification with molecules containing long hydrophobic alkyl chains terminating with carboxylic acid groups can limit cell adhesion.^[14a] The DDA molecule has a 10-carbon long alkyl chain terminating with carboxylic acid (COOH) groups and the long hydrophobic chain would collapse onto the chitosan surfaces in a manner reminiscent to phase separation. This would decrease the accessibility of the carboxyl groups at the interface to limit protein adsorption and cell adhesion. Consistent with these findings, chitosan surfaces modified with DDA molecules showed minimal cell adhesion and protein adsorption in vitro. Protein adsorption onto the biomaterial surface is a key event that dictates the initial inflammatory cell adhesion and triggering of FBR.^[23] Consistent with the in vitro findings, we observed a decrease in initial cell adhesion on DDA-modified chitosan pouches in vivo. This initial low inflammatory response to the biomaterial could explain the relatively thin and vascularized fibrous tissue around the DDA-modified chitosan pouches. Although subcutaneous space is generally considered poorly vascularized,^[29] the host-implant was vascularized in all groups but the DDA modified groups having more blood vessels around the implants and in the proximity of the implant. Confinement of the vascularization to the implant proximity while limiting its direct contact with the implant not only provides local immunoisolation to the cell-laden pouches but also contributes to transport of nutrients. The thin vascularized fibrous tissue around the modified pouches might have significantly contributed to the viability of the transplanted cells. Prior studies have shown that generation of vascular environment adjacent to the implant promotes the survival of transplanted cells by reducing the distance between the transplant and source of oxygen and nutrients.^[30] DDA modification also improved the barrier properties of chitosan structures by reducing its degradation, which is known to degrade under in vivo conditions within a few months.^[31]

4. Conclusion

In summary, our results show that the DDA-modified chitosan pouches supported long-term survival of human hepatocytes in immunocompetent mice following subcutaneous implantation. Subcutaneous implantation of the device allows minimally invasive interventions and enables a straightforward and easy applicability in terms of implantation, and retrievability. While significant progress has been made in cell transplantation, we still have many bottlenecks with the device design in order to overcome translational challenges in long-term survival and immunoprotection of donor cells. In this report, we designed a chitosan-based semipermeable pouch and evaluated its ability to provide local immunoisolation without compromising the mass transport to sustain the function and viability of human hepatocytes in immunocompetent mice. The DDA modification conferred a distinct surface that limited cell adhesion and function, and subsequent fibrosis. The modified chitosan pouches that minimize the fibrosis enabled long-term survival of the encapsulated hepatocytes following subcutaneous implantation by maintaining barrier properties and physiological oxygen level. To the best of our knowledge, this study shows the longest survival of xenogeneic hepatocytes in an immune competent animal model. Biomaterial enabled long-term function of the transplanted hepatocytes can be used to treat a variety of liver-based metabolic disorders and acute liver failures (ALF). Long term viability and function of transplanted hepatocytes is also important to bridge the gap until an organ becomes available for transplantation. Since fibrosis is a common bottleneck of implants and devices, we believe the molecular engineering of the device/implant-host interface could be widely used to improve the function of a wide variety of implants and cell encapsulation devices. Easy implantation of the device via a minimally invasive surgery and convenient retrieval, if needed, would be particularly attractive for translation.

5. Experimental Section

Fabrication of Chitosan Pouch:

The chitosan pouch was fabricated using a solvent casting method as shown in Figure 1b. A layer of chitosan (Sigma-Aldrich; Cat # 448877) membrane was initially cast on a petri dish by pouring 22.5 mL of 2% chitosan solution in acetic acid (Sigma, Cat # 695092) into a glass petri dish (10 cm diameter). A sacrificial template was placed which could be removed after curing the chitosan layer to generate a hollow interior for accommodating the cells. In this study, either paraffin or polytetrafluoroethylene (PTFE) tubing were used as a sacrificial template. The desired size and shape of paraffin template was made and attached to a needle using a hot paraffin glue that could be later used as an inlet for cell loading. PTFE mold was made by cutting desired dimensions of PTFE tubing. The templates were placed on the cured first layer and chitosan solution was poured on top of the sacrificial template to cast a second layer. This layer was cured overnight, and the cured chitosan membranes were neutralized by adding 2.5 M sodium hydroxide (NaOH) to the dish for at least 10 min. The excess NaOH was thoroughly washed with phosphate buffered saline (PBS). A spatula was used to carefully lift the chitosan structures from the petri dish. The pouches were then cut out and PTFE tubing was removed. In the case where paraffin was used as a sacrificial layer, the chitosan system with the paraffin filling was immersed in a water bath at 60 °C to melt the paraffin which was easily diffused out. The resulting chitosan pouch was sterilized by multiple ethanol washes followed by multiple sterile PBS washes to remove ethanol. The final sterilization wash was done with pen-strep in PBS (5% v/v). The thickness of the pouch was manipulated by varying the volume of chitosan solution used to cast each layer. For imaging purpose 1% wt/v of 200 nm diameter green fluorescent particles were added to the chitosan solution before casting chitosan membranes. In each well of a 6-well plate either 3 mL, 2 mL, or 1 mL of chitosan solution was poured. The solution was allowed to cure and neutralized with NaOH solution followed by washing with PBS. The membranes (i.e., walls of the chitosan pouch) were then transferred to a mold filled with OCT (Tissue-Tek O.C.T. Compound; Sakura, Cat # 4583) and frozen with 2-methylbutane and liquid nitrogen and stored at −80 °C until sectioning. For cryosectioning, frozen tissue blocks were sectioned to 20 μm thicknesses with a cryotome cryostat (at −20 °C).

Modeling of Oxygen Diffusion to Optimize Pouch Dimensions and Understand the Effect of Fibrous Tissue:

Computational analyses were used to determine the effect of wall thickness and inner diameter of the pouch on mass transport using oxygen as a model system. While the initial analysis used pouches without the fibrous tissue layer, further analyses were carried out to incorporate the experimentally determined fibrous tissue thickness following in vivo implantation.

The pouch was modelled as two concentric cylinders as shown in Figure S2, Supporting Information, where the inner cylinder (radius r₁) was the cell-laden fibrin hydrogel while the outer cylinder (radius r₂) was the chitosan membrane/hydrogel (thickness t = r₂−r₁). The oxygen concentration at r = r₂ was specified as oxygen content in the peripheral blood. COMSOL 5.4 was used to solve reaction-diffusion equation for oxygen. The equations, parameters, and boundary conditions were:

For interior of the pouch:

$$\frac{\partial C}{\partial t}=D_F(\nabla \cdot \nabla C)-R$$ (1)

where $R=\frac{V_{\max }\cdot C}{K_m+C}$

Boundary condition: At $r=0,\frac{\partial C}{\partial r}=0$

For pouch wall:

$$\frac{\partial C}{\partial t}=D_C(\nabla \cdot \nabla C)$$ (2)

Boundary condition: At r = r₂, C = C_O

In the above equations, C represents the oxygen concentration, C_O is the blood oxygen concentration at the pouch periphery, r is the radial distance from the center, D_c and D_F are the oxygen diffusion coefficients in chitosan and fibrin respectively, R is reaction rate (rate of consumption), V_max is the maximum volumetric oxygen consumption rate by hepatocytes and K_m is the Michaelis constant. The diffusion coefficients used were taken from prior publications.^[32] Oxygen consumption rate of hepatocytes was approximated from previous studies with hepatocyte cell line at similar cell density.^[33] A complete list of parameter definitions and values of the constants are summarized in Table S2, Supporting Information. The steady state solution of the above model was determined for r₁ varying from 0.75 to 1.5 mm and t varying from 100 to 300 μm. Spatial distribution of oxygen within the pouch was assessed by plotting oxygen concentration against radial distance. Previous studies have used varying definitions of oxygen levels (0.01–0.03 mol m⁻³) to define pathological hypoxia.^[19,34] A concentration on the higher end (0.03 mol m⁻³) was selected as the design criterion to allow for any experimental deviations arising from the model’s assumptions.

To incorporate the fibrous tissue layer around the pouch, a third concentric cylinder was added to the previous system of two concentric cylinders. For the modified pouch, the thickness of the outermost cylinder (t_F) was considered as 300 μm while the same for the unmodified pouch was selected as 1000 μm based on measured thickness from cross-sections (Figure S6a, Supporting Information). Although vascularization was observed within the fibrous tissue, the model was simplified to include blood oxygen supply at the periphery of the third layer. Here, Equations (1) and (2) were used with modified boundary conditions. Specifically, the boundary equation pertaining to the periphery of the pouch was now moved from the outer wall of the pouch to the fibrous tissue.

Equation for the fibrotic layer:

$$\frac{\partial C}{\partial t}=D_{\text{F}}^{\prime }(\nabla \cdot \nabla C)-R^{\prime }$$ (3)

where $R=\frac{V_{\max }^{\prime }\cdot C}{K_{\text{m}}^{\prime }+C}$

Boundary condition: At r = r₃, C = C_O

Here, D_F’ represents the oxygen diffusion coefficient of the fibrotic layer and R’ represents the rate of oxygen consumption by the cells in the fibrotic layer. The fibrotic tissue was assumed to be composed of collagen I with 10⁶ cells mL⁻¹ density of fibroblasts.^[35] Additionally, oxygen consumption rate was assumed to be saturated (Km’<<C) in the fibrotic layer.^[36] Complete list of parameter definitions and values of the constants are provided in Table S2, Supporting Information.

FITC Dextran Diffusion through Chitosan Pouches:

Chitosan pouches were loaded with 37.5 μL of 15 mg mL⁻¹ FITC dextran solutions (70 kDa or 150 kDa, Sigma, Cat # FD 150S) and sealed with a cyanoacrylate-based glue. FITC dextran loaded pouches were incubated in PBS at 37 °C (n = 3) and supernatants were collected at 0.1, 0.3, 1, 2, 6, and 24 h time points. The amount of FITC dextran diffused into the PBS from the pouch was quantified by measuring fluorescence intensity of supernatants using a spectrophotometer (Tecan Infinite 200 PRO) with excitation and emission wavelengths of 480 and 520 nm respectively.

Surface Modification of Chitosan with DDA and Characterization:

For the surface modification of chitosan membranes, we initially performed graded dehydration of the neutralized chitosan membrane with dimethyl sulfoxide (DMSO) (Acros organic, Cat # A0393327). DDA (Sigma, Cat # D1009) was dissolved in DMSO and subsequently reacted with 1.2 equivalent of EDC. HCl, Sigma, Cat # E7750 for 15 min followed by 1.1 equivalent of (NHS, Sigma, Cat # 130672) and stirred for another 15 min. Finally, the dehydrated chitosan pouches or membranes were immersed in the activated DDA solution and incubated at room temperature for overnight with continuous shaking. After the incubation period, modified surfaces were washed with DMSO, rehydrated in graded DMSO/PBS mixture, and washed with PBS. Surface chemical analysis of the unmodified and DDA-modified chitosan membranes were performed using Fourier Transform Infrared (FTIR) spectrometer (Thermo Electron Nicolet 8700) via attenuated total reflection (ATR) mode using diamond crystal. A thionin acetate dye adsorption assay was performed to quantify the amount of DDA modification (Sigma, 88930). All the chitosan membranes (n = 3) were equilibrated with ethanol prior to dye exposure. Membranes were immersed in thionin acetate dye solution (0.01 mg mL⁻¹ in ethanol) with continuous shaking for 10 h at room temperature. The adsorbed dye was then extracted from the membrane surface by treating with 0.01 m HCl for 2 h. Fluorescence intensity of the extracted dye solution was determined using spectrophotometer with excitation and emission wavelengths of 580 and 620 nm respectively. Serially diluted thionin acetate solution was used as a calibration curve to calculate the extent of surface modification.

In Vitro Degradation of Chitosan:

Both unmodified and DDA-modified chitosan membranes were dried in a vacuum oven at 60 °C, and weight of each membrane was determined (n = 3). Samples were then incubated in 200 μg mL⁻¹ of lysozyme solution in PBS. Lysozyme solution was changed two times a week during the experimental period. Degradation of each chitosan membrane was monitored by determining the change in dry weight at 1, 3, 7, 14, 21, and 28 days.

Protein Adsorption Assay:

Bovine fibrinogen (Alfa Aesar, Cat# AAJ6327603) was dissolved in PBS (100 μg mL⁻¹) prior to the experiments. Chitosan membranes (n = 3) were then incubated with fibrinogen solution for 6 h at 37 °C. After the incubation period, the amount of fibrinogen in the supernatant solution was quantified by using Bradford assay kit. Amount of protein adsorption on the chitosan membrane was calculated as a difference between the original and supernatant solutions. For the collagen adsorption analysis, fluorescein isothiocyanate (FITC)-labeled collagen (Bio vision, Cat# M1304) was dissolved in PBS (pH 7.4) at a concentration of 100 μg mL⁻¹. Surface-modified chitosan membranes with varying degree of DDA were incubated in collagen solution for 6 h at 37 °C (n = 3). Amount of unadsorbed FITC collagen in the supernatant was measured using a spectrophotometer (Tecan Infinite 200 PRO) with excitation and emission wavelengths of 480 and 520 nm, respectively, and adsorbed amount of collagen on each chitosan membranes was quantified by subtracting unadsorbed collagen concentration from the initial solution.

Cell Adhesion on Chitosan Membranes:

For cell adhesion analysis, all membranes were washed with distilled water and incubated in 70% ethanol for 1 h. Membranes were then extensively washed with PBS (6–7 times) followed by incubation in PBS containing 2% pen-strep for at least 48 h with occasional changing (3–4 times).

NIH/3T3 cells were trypsinized and plated in a cell culture medium containing high-glucose DMEM (Gibco, Cat# 11995065), 10% v/v fetal bovine serum (HyClone, Cat# SH3007103HI), and 50 U mL⁻¹ pen-strep (Gibco, Cat# 15140122) at a seeding density of 25 000 cells per well and imaged for cell attachment at day 1, 4, and 7 days. A media change was performed at every 3–4 days. Mononuclear cells (MNCs) from mouse bone marrow were harvested as previously reported.^[37] Harvested MNCs were seeded onto chitosan membranes and cultured in macrophage differentiation media containing 10 ng mL⁻¹ M-CSF (PeproTech, Cat # 315–02), 10⁻⁷ M PGE2 (Santa Cruz Biotechnology, Cat # sc-201225), 10% FBS, and 1% Pen/Strep in alpha MEM and imaged after 6 days of culture. Three images from different areas were captured per sample (n = 3) and the number of cells were quantified and averaged. The average cell area on each group was analyzed using ImageJ ((National Institutes of Health, Bethesda, MD, USA).

Subcutaneous Implantation of Chitosan Membranes:

All animal procedures were approved by the Institutional Animal Care and Use Committee of the Duke University and performed in accordance with the NIH and national and international guidelines for laboratory animal care. Recipient mice (C57BL/6J mice) were anesthetized with isoflurane and the fur on the back was shaved. Mice were then placed on a heating pad, injected with analgesic (Buprenorphine SR-LAB) and a 1 cm-long incision was made in the back and two subcutaneous space was created using a blunt spatula. Each modified and unmodified chitosan membranes were implanted at both sides (n = 4) and the skin was sutured. The implants were retrieved after 3 or 14 days.

Analysis of Implanted Membranes:

After retrieval, implants were fixed for 20 min using 4% paraformaldehyde at room temperature and washed with PBS. Samples were then incubated in OCT (Tissue-Tek O.C.T. Compound; Sakura, Torrance, CA) at 4 °C overnight and subjected to OCT embedding following a standardized protocol. For cryo-sectioning, frozen tissue blocks were sectioned with a cryotome cryostat (at −20 °C) to 10 μm thicknesses. For immunofluorescent staining, sections were blocked with blocking buffer containing 2% w/v bovine serum albumin, 2.5% normal serum, and 0.1% Triton X-100 v/v in PBS. Subsequently, sections were treated with primary α-SMA antibody (1:150, Sigma, Cat # ABT1487), and F4/80 (1:100, eBioscience, Cat# 14480182) and incubated at 4 °C for overnight. Appropriate secondary antibody was then added along with Hoechst 33342 (2 μg mL⁻¹; Thermo Fisher) for 1 h at RT. Samples were imaged with a fluorescent microscope. The cell number was measured by counting cell nuclei at three different locations from each sample (n = 4) using ImageJ software. For collagen staining on the implant, sections were subjected to Masson’s trichrome staining according to manufacturer’s protocol (Sigma, Cat# HT15–1KT). Images were captured at three different locations for each sample (n = 4) and thickness of the collagen deposition was analyzed using ImageJ software.

Cell Loading in Pouches, Implantation, and Analyses:

Primary human hepatocytes (LONZA, Cat# HUM 4100) were thawed in thawing media (LONZA, Cat# MCHT50) and centrifuged for 10 min at 100 g and dispersed into maintenance media. The loading mixture was prepared by mixing 5% collagen-I solution (rat tail, corning, Cat# 354249), 9.2 mg mL⁻¹ of fibrinogen, and desired number of cells. The loading mixture containing hepatocytes was then infused into the pouch through the inlet using a syringe. Immediately, 25 U mL⁻¹ of thrombin was administered through the pouch wall to ensure homogenous gelation. Finally, the pouch was sealed with an acrylate-based glue and cultured in maintenance media (LONZA, Cat# MM250) for 24 h. Subcutaneous implantation of hepatocyte-loaded pouch was performed as mentioned in the previous section. Two pouches were implanted in each mouse (n = 6) with each pouch containing 2.5 million hepatocytes. Mouse blood from tail vein were collected at pre-determined time intervals. Finally, implants were recovered at the termination of experiments and processed for different characterizations and analyses. For human albumin enzyme-linked immunosorbent assay (ELISA) analyses, peripheral blood was collected into an anticoagulant tube and centrifuged at 14900 × g at 4 °C for 15 min to separate the serum and plasma and extract the serum. The serum was assessed for human albumin using the Human Albumin ELISA Quantification kit (Bethyl Labs, Cat# E80–129) according to the manufacturer’s protocol. Simultaneously, retrieved implants were cryosectioned and stained for human albumin (1:250, Bethyl Labs, Cat# A80–129F), F4/80 (1:100), α-sma (1:150), and endomucin (1:100, Abcam, Cat# ab106100) using staining methods described earlier. Quantification of vascularization was performed using ImageJ and presented as the percentage of EMCN positive area to the area of total tissue layer. Blood vessel density at the pouch periphery was calculated by analyzing EMCN positive area within 150 μm from pouch outer periphery. At least four images from each sample were used for analysis (n = 4). Sections were also subjected to Masson’s trichrome staining according to manufacturer’s protocol (Sigma, Cat# HT15–1KT) and the thickness of collagen tissue on the implants was analyzed using ImageJ software (n = 4). Additionally, the encapsulated hepatocytes from the retrieved pouch were replated onto collagen coated petri dish and cultured for 24 h. Cells were then fixed and stained for hepatocytes markers, human albumin (1:250, Bethyl Labs, Cat# A80–129F) and human cytokeratin 18 (1:300, Abcam, Cat# ab32118).

Statistical Analysis:

All statistical analysis was performed using GraphPad Prism 7. All graph bars represent mean with SEM (standard error of the mean). All the data were subjected to Two-tailed Student’s t-test for comparing two groups, one-way analysis of variance (ANOVA) with Tukey’s post hoc test for three or more groups, and two-way analysis of variance for data having multiple variables. Any p-value less than 0.05 was considered statistically significant.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.