Challenges and emerging technologies in the immunoisolation of cells and tissues

Abstract

Protection of transplanted cells from the host immune system using immunoisolation technology will be important in realizing the full potential of cell-based therapeutics. Microencapsulation of cells and cell aggregates has been the most widely explored immunoisolation strategy, but widespread clinical application of this technology has been limited, in part, by inadequate transport of nutrients, deleterious innate inflammatory responses, and immune recognition of encapsulated cells via indirect antigen presentation pathways. To reduce mass transport limitations and decrease void volume, recent efforts have focused on developing conformal coatings of micron and submicron scale on individual cells or cell aggregates. Additionally, anti-inflammatory and immunomodulatory capabilities are being integrated into immunoisolation devices to generate bioactive barriers that locally modulate host responses to encapsulated cells. Continued exploration of emerging paradigms governed by the inherent challenges associated with immunoisolation will be critical to actualizing the clinical potential of cell-based therapeutics.

1. Introduction

Transplantation of primary or genetically engineered cells of allo- or xenogenic origin has emerged as a promising approach for the localized and regulated ‘de novo’ delivery of a nearly unlimited number of therapeutic agents with potential capacity to cure or treat virtually any disease or disorder, including, but not limited to, those related to the endocrine system (diabetes, hypoparathyroidism, adrenal insufficiency), central nervous system (Parkinson’s, Alzheimer’s, ALS, Huntington’s), as well as cancer, kidney failure, and heart disease [1–8]. However, widespread clinical application of cell transplantation remains limited, in part, by the deleterious side effects of current immunosuppressive regimens necessary to prevent host rejection of transplanted cells. It has long been postulated [9] that systemic immunosuppression could be eliminated if transplanted cells and tissues were physically isolated from the host immune system using a semipermeable membrane, or immunoisolation device. Since the pioneering work by Chick et al. in the development of a bioartificial pancreas [10] and Lim and Sun’s introduction of alginate-encapsulated islets [11], decades of extensive research has focused on the design and application of immunoisolation devices capable of protecting transplanted allo- and xenogenic cells from the host, while facilitating adequate transport of oxygen, nutrients, and secreted therapeutic molecules.

To date, a variety of polymeric and inorganic matrices and membranes have been utilized to produce immunoisolation devices of diverse physiochemical properties and geometries [12], which include vascular perfusion devices, avascular diffusion chambers, macrocapsules, and microcapsules (Fig. 1) [13,14]. The need for vascular surgery, and associated risks of surface-induced thrombosis, has largely limited the development of vascular perfusion devices [14]. Though notable exceptions exist [15,16], clinical application of macrocapsules and avascular diffusion chambers has been hampered by insufficient oxygen and nutrient transport to cells in the center of such devices [14]. Consequently, most immunoisolation strategies have employed microcapsules consisting of cells or cell clusters entrapped within a spherical semipermeable membrane, an inherently favorable geometry for diffusive nutrient transport that can be implanted with minor surgery [13,14]. Though this review will mostly address limitations in current microencapsulation strategies, the challenges and considerations discussed herein are generally applicable to a variety of immunoisolation devices.

Fig. 1.

Devices for cell transplantation have traditionally included vascular perfusion devices, avascular macrocapsules, and microcapsules. In an effort to reduce the size and void volume of immunoisolation devices, conformal coatings, cell-surface PEGylation, and nanoencapsulation have recently been explored.

Transplantation of encapsulated pancreatic islets for the treatment of diabetes mellitus has been the most common application of cell transplant and immunoisolation technology, and, consequently, will be an area of focus within this review. However, widespread clinical application of cell-based therapy requires an inexhaustible source of cells or tissues capable of delivering the therapeutic agent in an appropriate form, often in response to physiological cues. Genetic engineering has emerged as a promising tool for addressing this challenge [17–19], but genetically engineered cells often arise from an allogenic or xenogenic source and will also require protection from the host immune system. Thus, the lessons afforded by efforts directed towards islet encapsulation will prove to be highly valuable as genetically engineered cells gain in clinical relevance.

While far from a comprehensive history of islet encapsulation, Table 1 summarizes the ability of some notable microcapsules to restore normoglycemia in several animal models of diabetes. The large number of reports and variable success rates attest to the promise and inherent challenges of immunoisolation. As non-encapsulated islet allografts are generally destroyed within two weeks and xenografts within one week, immunoisolation significantly improves the survival of transplanted tissue. However, insulin independence, particularly in larger animals or more rigorous small animal models is of limited duration. Consequently, despite remarkable advances in immunoisolation technology, only a few clinical trials have been performed with immunoisolated islets [20–22] or other cell types [1,15,23–25]. While many of these studies continue to fall short of long-term clinical objectives, early evidence of graft function and improved patient health have demonstrated the potential significance of cell-based therapeutics, providing motivation for further elucidation of mechanisms that limit graft survival with a goal of further optimization of immunoisolation strategies.

Table 1.

Maintenance of normoglycemia after transplantation of microencapsulated islets

Donor	Recipient	Capsule Type	Tx site	Islet dose	Normoglycemia (avg. days)	Normoglycemia (max. days)	Normoglycemia (range)	Ref.
Allogenic
B6AF1 mice	NOD mice	Alginate/Ba2+	Peritoneum	900–1000	>350a	>385	350–>385	[56]
DA rat	Lewis rat	Alginate/Ba2+	Peritoneum	5000 IE	54±6	91	15–91	[253]
Lewis rat	AO rat	APA	Peritoneum	3500–4200	102.6±9.7	>200	42–>200	[163]
Canine	Canine	Agarose	Omentum	4300–11,000 IE/kg	20.5±20.9	49	28–49	[276]
Xenogenic
Porcine	Canine	Agarose-based	Peritoneum	4000–45,000/kg	49.7±21.7	119	7–119	[277]
Hamster	Wistar rat	Agarose-based	Peritoneum	3000	124.5±44.6	175	77–175	[278]
Porcine	NOD mice	APA	Peritoneum	11,341+/−3293 IE	13.2±1.2	NR	NR	[248]
Porcine	NOD mice	Alginate/Ba2+	Peritoneum	5000 IE	15±1	184	NR	[239]
Porcine	Primate	APA	Peritoneum	3–7×104b	NR	804	120–804	[51]
Porcine	B6AF1 mice	Alginate/Ba2+	Peritoneum	10,000	NR	140c	NR	[279]
SD rat	Balb/c mice	Alginate/Ba2+	Peritoneum	1800	NR	270d	90–270	[162]
Human	Balb/c mice	Alginate/Ba2+	Peritoneum	1800	NR	268d	115–268	[162]
SD rat	C57BL/6 mice	Alginate/PMCG	Peritoneum	1000	NR	>300	20–>300	[43]
SD rat	NOD mice	Alginate/PMCG	Peritoneum	1000	NR	180	40–180	[43]
Canine	NOD mice	APA	Peritoneum	4×103–1.2×10	11.5±3	17	5–17	[245]
Canine	C57BL/6 mice	APA	Peritoneum	4×103–1.2×10	51 ±8	68	36–68	[245]
Lewis rat	NOD mice	APA	Peritoneum	1800–2000	10±2	17	6–17	[245]
Bovine	Balb/c mice	alginate/Ca2+	Peritoneum	2–4×103	NR	>70	30–>70	[160]
Wistar rat	Balb/c mice	APA	Peritoneum	900–1000	219.8±46.2	308	172–308	[280]
SD rat	Balb/c mice	Agarose-based	Peritoneum	500	79±32	134	47–134	[281]

NR: not reported. APA: alginate/poly(L-lysine)/alginate. PMCG: poly-methyl-co-guanidine. IE: islet equivalents.

^aNatural lifespan of mice precluded further investigation.

^bMultiple transplants used.

^cStudy terminated at 140 days.

^dOngoing graft survival at time of report.

The objective of this review is not to provide a comprehensive survey of over thirty years of immunoisolation research, but rather to present a summary of current principal obstacles to the clinical application of immunoisolation technology. These challenges include mass transport limitations and cell hypoxia, deleterious innate inflammatory responses to device materials and/or encapsulated cells, and immune-mediated inflammatory responses elicited via the indirect antigen presentation pathway. Moreover, this review highlights current areas of investigation that hold particular promise for circumventing these limitations, with an emphasis on biomaterial- and tissue engineered-based approaches.

2. Mass transport limitations in microencapsulation

Although the high surface-to-volume ratio provided by microencapsulation considerably improves mass transport relative to macrocapsules or extravascular diffusion chambers [13,26], the relatively large size of conventional microcapsules, typically 400–800 µm in diameter, continues to impose transport limitations that, if not accounted for, may adversely effect cell survival and function. Experimental evidence and mathematical models demonstrate that oxygen concentration decreases radially within cylindrical or spherical devices due to the consumption of oxygen by the encapsulated cells [27–30]. Therefore, if oxygen levels are insufficient at the site of transplantation, cell density must be reduced as device diameter increases to minimize hypoxia of centrally located cells. Even sublethal levels of hypoxia can have deleterious effects on ATP-dependent cell functions, such as insulin secretion [31], and may also induce expression of inflammatory mediators [32]. Consequently, the number of cells that may be transplanted within a given microcapsule is limited both by device size and the related metabolic profile of the donor cells, often leading to an increase in cell transplant volume and an associated increased incidence of device defects [33] and surgical risk [34].

Effective cell-based therapy often relies on the ability of transplanted cells to respond to physiological stimuli in a concentration- and time-dependent manner [8]. The characteristic time for diffusion through a sphere of radius, R, scales as R2/D, where D is the diffusivity of the solute through the encapsulation matrix and/or immunoisolation membrane [35]. Therefore, cells in the center of the device will experience a given solute concentration at a later time than those on the periphery, leading to a lag in response time [30]. Moreover, depending on the pore size and other physiochemical properties of the immunoisolation membrane and cell immobilization matrix, the diffusivity of important solutes such as glucose, insulin, and oxygen may be substantially less than their diffusivity in water [30,36–39], further delaying responses as compared to those observed for non-encapsulated tissue. This is perhaps most clearly illustrated by microencapsulated islets, where the distance between the capsule surface and outer cell layer of the islet may be on the order of 100– 400 µm, creating a void space which glucose and insulin must cross prior to transport in or out of the device. Indeed, delayed in vitro insulin secretion in response to step changes in glucose have been observed for a variety of different capsule formulations [11,40–43]. Decreasing capsule size has been shown to minimize this delay [44–46].

Arguably more detrimental than diffusion limitations inherent to conventional microcapsules are constraints imposed by the transplant sites necessary to accommodate the volume of microencapsulated cells. For example, current clinical islet transplantation protocols require ∼600–700 thousand islets, a volume of roughly 5–10 mL [47]. In contrast, a current clinical trial using islets entrapped in 500 µm microcapsules requires a transplant volume of 50mL [21], representing approximately a 5– 10 fold increase in transplant volume. Consequently, most microencapsulated cells have been transplanted into sites that have a relatively limited vascular supply, such as the omentum [48,49] or peritoneal cavity [20,21,50,51]. The anatomy of the peritoneal cavity does not facilitate instantaneous transport of insulin or glucose to and from the systemic circulation, as insulin must be absorbed by the peritoneum and extracted by the liver [52,53]. As such, insulin production within the peritoneal cavity results in a delayed systemic response relative to intraportal insulin production [54,55], thereby impairing metabolic control. In response to a meal challenge, Tatarkiewicz et al. observed blunted C-peptide concentrations in animals transplanted with non-encapsulated, syngeneic islets in the peritoneal cavity, indicating that transplantation site is critical to proper maintenance of metabolic processes [41]. Though successful reversal of diabetes has been achieved despite impaired insulin and C-peptide responses [41,42,56], it is unclear whether metabolic control will be sufficiently robust to minimize the chronic complications of diabetes [57–59].

Viability and function of microencapsulated cells transplanted into relatively avascular sites may be further exacerbated by partial pressures of oxygen which are 40% of that found in arterial circulation [27]. Microencapsulated islet autografts retrieved from the peritoneum upon graft failure often have necrotic cores [60], a hallmark of hypoxia [32]. Interestingly, core necrosis may be observed even in the absence of encapsulation [60], corroborating previous findings that the peritoneal cavity provides a suboptimal environment for islet transplantation [61–63].

3. Transplantation of microencapsulated cells into a microvascular bed: intraportal islet transplantation

Though several clinical trials and large animal studies have demonstrated the potential efficacy of intraperitoneally transplanted encapsulated islets [20,21,50,51,64], the International Islet Registry reports that, compared to other sites, transplantation of islets into the portal vein is associated with the highest success rate one year after transplantation [65]. Thus, the portal bed remains the clinically preferred site for islet transplantation [47,66,67]. By means of a minimally invasive procedure, isolated and purified islets are infused into the portal vein of the liver where they lodge at distal portal venules [66]. While direct islet-blood contact has been shown to mediate thrombosis [68– 70], the portal vein offers an oxygen and nutrient rich environment and provides physiologically normal drainage of insulin, minimizing delayed insulin secretion in response to glucose. The encouraging early results of the Edmonton Protocol, which combined intraportal islet transplantation with less diabetogenic, steroid free immunosuppression [47], demonstrated the potential of islet transplantation [71]. However, most conventional microcapsules are not suitable for transplantation into microvascular beds due to their large diameter [72– 74]. Intraportal infusion of 420 µm microparticles has been shown to result in dangerous elevations of intraportal pressure and, in some instances, increased mortality in animal models [73]. Bottino et al. have observed impaired hyaluronic acid clearance after intraportal infusion of both islets and an equivalent volume of microparticles, indicating that portal vein endothelial cells are injured in response to particle infusion in a non-specific manner [75]. Schneider et al. have recently demonstrated impaired engraftment of islets encapsulated in 350 µm alginate/Ba²⁺ microcapsules transplanted into the portal vein compared to non-encapsulated controls, apparently due to occlusion of small and medium sized portal venules and subsequent islet hypoxia (Fig. 2A, B) [74]. Clearly, encapsulation strategies for transplantation of cells into microvascular beds must minimize transplant volume.

Fig. 2.

Transplantation of 350–400 µm microcapsules into the liver microvasculature. Histological evaluation of the liver of a SD rat before (A) and after (B) intraportal transplantation of islets encapsulated within 350 µm alginate/Ba²⁺ microcapsules. Intraportal transplantation is associated with increased inflammatory cell recruitment and occlusion of portal venules (reprinted from [74] with permission of Wiley-Blackwell Publishing Ltd). Portography before (C) and 15 min after (D) intraportal injection of empty 400 µm microcapsules. Injection of capsules decreases the number of small portal veins; however, this reduction is comparable to that observed during intraportal transplantation of non-encapsulated islets (reprinted from [96] with kind permission from Springer Science and Business Media).

4. Improving the vascularization of immunoisolation devices

Though immunoisolation precludes integration of blood vessels with transplanted cells and tissues, effective diffusion of oxygen and nutrients from capillaries to neighboring cells can occur over a maximum distance of about 200 µm [76]. Therefore, to circumvent mass transport limitations associated with transplantation into relatively avascular sites, recent efforts have sought to induce neovascularization of extravascular diffusion devices through control of surface microarchitecture [77–80]. Vascularization of membranes may be further enhanced through infusion or controlled release of angiogenic factors, such as vascular endothelial growth factor (VEGF) [81–84] or basic fibroblast growth factor (bFGF) [85]. Interestingly, despite the chemoattractive properties of VEGF, Sigrist et al. observed significantly reduced macrophage adhesion to an AN69 membrane containing allogenic islets [82,83]. To achieve sustained angiogenesis and support existing vasculature, continuous delivery of factors may be necessary [82,83,85]. Towards this objective, Miki et al. have recently demonstrated enhanced vascularization of an ethylene vinyl alcohol membrane through co-delivery of bFGF-impregnated gelatin microspheres and encapsulated vascular endothelial TMNK-1 cells, which provide continuous biochemical support for the neovascularized membrane [85].

As an alternative to inducing vascularization of the immunoisolation device per se, others have sought to generate highly vascularized subcutaneous, intramuscular, or intraperitoneal sites for cell transplantation using stainless steel [86], polyethylene terephthalate [87], or polytetrafluoroethylene (PTFE) [88] mesh, respectively. Generation of such sites may allow microencapsulated cells to be removed and replaced, and may potentially circumvent complications associated with transplantation of encapsulated cells into native tissue [73,74]. Delivery of angiogenic factors, such as bFGF [87], acidic fibroblast growth factor [88], and VEGF [89] has been used as an adjunct to improve vascularization of these artificial cell-recipient beds. Islet transplantation into prevascularized sites dramatically improves graft survival and function relative to transplantation into non-modified tissue [86–88]. Moreover, syngeneic islets transplanted into prevascularized sites perform comparably to portal vein transplants, though some impairment in glucose clearance was still observed [86,88]. de Vos et al. have sought to minimize this effect by generating vascularized beds in the immediate vicinity of the liver to promote integration with the liver vasculature and facilitate portal drainage of insulin [89]. It remains unclear whether such sites can be rendered large enough to accommodate the large graft volumes associated with conventional microen-capsulation devices.

5. Strategies to reduce microcapsule size

5.1. Microfabrication of capsules

An obvious but non-trivial approach to improving transport properties of microcapsules is to produce smaller capsules by optimizing the process parameters used in traditional microcapsule fabrications. Traditional microcapsules have been mostly commonly produced using a coaxial air flow system [14]. Recently, Sugiura et al. have developed an analogous microfluidics device that utilizes an array of microfabricated silicon nozzles and airflow channels. Using 60 µm nozzles and appropriate airflow rates, genetically modified CHO cells could be encapsulated in nearly monodisperse 150 µm alginate– calcium microcapsules, and subsequently coated with a poly(L-lysine) (PLL) /alginate permselective membrane [90]. While microfabricated nozzles smaller than 50–100 µm are clearly not amenable to islet encapsulation, this approach holds considerable promise for reducing the transplant volume of microencapsulated genetically engineered cells. Ross and Chang have reported increased beta-glucuronidase production by genetically engineered fibroblasts using 100–200 µm alginate microcapsules, owing to tighter packing of smaller capsules and increased viable cell density [91]. In particular, smaller capsules may have significant advantages in the treatment of CNS diseases, where injection of microcapsules into the subarachnoid space or the brain parenchyma [3] excludes large transplant volumes. Additionally, smaller microcapsules might be used to facilitate intraportal transplantation of insulin producing cell lines, such as MIN6 [92] and βTC3 cells [93].

Several groups have addressed the feasibility of intraportal transplantation of microcapsules slightly smaller than conventional microcapsules. Hallé et al. have generated 300–350 µm microcapsules using a high voltage electrostatic pulse system [73,94,95]. Injection of 10,000 315 µm diameter microcapsules into the portal vein of rats resulted in only modest and transient increases in portal pressure [73]. Similarly, Toso et al. intraportally injected 10,000 400 µm alginate/poly(methylene-co-guanidine) hydrochloride microcapsules per kilogram body weight (Fig. 2C, D). While portal pressures were elevated immediately post-implant, the increase was comparable to that observed during clinical islet transplantation and returned to normal after three months [96]. However, these [96] and other intraportally infused microcapsules [97] elicited a significant foreign body reaction. The dose of microcapsules used in both of these studies is comparable to the islet dose used in clinical islet transplantation [47], demonstrating the potential to infuse 300–400 µm capsules in vascularized sites. However, decreasing the size of alginate microcapsules from 800 to 500 µm is associated with a ∼4 fold increase in the percentage of incompletely encapsulated islets [98]. Host response to even 2–10% of encapsulated islets has been shown to result in destruction of 40% of the graft [60,99]. Therefore, complete encapsulation of all transplanted cells remains an important prerequisite for successful immunoisolation.

5.2. Conformal coating via fluidic entrainment and interfacial precipitation

To circumvent limitations associated with random entrapment of cell aggregates within microparticles, several investigators have deposited coatings of defined thickness that conform to the surface of the cell or tissue. Transplant volume is, therefore, defined only by the size of the object being coated and the thickness of the coating, significantly reducing void volume while retaining the presence of a polymer barrier to provide immunoisolation.

Fluidic entrainment of polymer solution around cell aggregates followed by interfacial precipitation of the polymer has been most commonly utilized to generate conformal coatings. In light of the promise and widespread use of alginate-based microcapsules, Zekorn et al. [100], and subsequently Park et al. [101], have fabricated conformal alginate hydrogel coatings on the surface of individual pancreatic islets. To accomplish this, they utilized a discontinuous density centrifugation gradient composed of a top layer of islets suspended in sodium alginate, followed by denser spacer layers composed of dextran or Ficoll, one of which contained a divalent cation (CaCl₂ or BaCl₂). During centrifugation, islets approach and deform the alginate–dextran (Ficoll) interface, entraining a film of alginate around the islet provided drainage between the islet and interface is sufficiently slow. When the islet and entrained alginate cross into the layer containing the divalent cation the alginate is crosslinked, resulting in a 5– 10 µm film of alginate that largely conforms to the shape of the islet. Islets coated in this manner demonstrated normal biphasic insulin secretion in a dynamic perfusion assay, indicating that islet function and viability are preserved and that the thin coating prevents the lag in insulin secretion often observed with larger alginate microcapsules [100]. In a similar manner, Sefton et al. have used density centrifugation to coat islets [102] and HegG2 cell aggregates (Fig. 3C) [103] with water insoluble poly(hydroxyethyl methacrylate-co-methyl methacrylate) (HEMA-MMA) copolymers by interfacial precipitation, generating polymer films as thin as 1.5 µm. The non-aqueous nature of HEMA-MMA is anticipated to improve stability relative to hydrogels, which are susceptible to hydrolytic degradation. However, conformal coating of cells with HEMA-MMA results in a significant degree of cell death, owing to the need to expose cells to an organic solvent (polyethylene glycol, MW 200 Da) [104]. Recently, improved cell viability has been achieved by decreasing coating times and pre-encapsulating cells in agarose beads prior to HEMA-MMA conformal coating to minimize exposure to non-aqueous solvents [104].

Fig. 3.

Conformal coating of islets and cell aggregates. (A) Alginate/Ba²⁺ coating formed on a HEK 293 cell spheroid using an aqueous two phase emulsion (reprinted from [106] with permission of Wiley-Liss, Inc., a subsidiary of John Wiley & Sons, Inc). (B) PEG hydrogel coating fabricated on porcine islets via interfacial polymerization (reprinted from [109] with permission of Wiley-Liss, Inc., a subsidiary of John Wiley & Sons, Inc). (C) Electron micrograph of HepG2 cell aggregates coated with HEMA-MMA via density centrifugation and interfacial precipitation (Reprinted from [103] with permission from Elsevier). (D) Islets coated with PEG hydrogel via selective withdrawal (reprinted from [108] with permission of Wiley-VCH).

Emulsification has also been used to generate a conformal coating on islets and cell aggregates [105,106]. Calafiore et al. placed islets into an alginate/polyethylene glycol (PEG)/Ficoll emulsion, whereby alginate-containing Ficoll droplets suspended in a continuous PEG phase coalesced on the islet surface, engulfing individual islets in a layer of alginate, which was subsequently crosslinked with calcium chloride and coated with a poly(L-ornithine)/alginate bilayer [105]. This conformal barrier prevented direct cell-islet and antibody-islet contact, and did not impair insulin secretion in response to glucose stimulation [72]. Leung et al. have recently optimized this emulsion process to minimize the incidence of incomplete and non-uniform coating, and have coated cell clusters with 20–25 µm thin films of alginate (Fig. 3A) [106]. Importantly, conformally coated canine islets transplanted intraportally into a porcine model were viable and free of inflammatory reaction 15 days post transplant; however, a dense inflammatory infiltrate was observed 15 days later [107]. Nonetheless, these studies demonstrate the feasibility of using conformal barriers in intravascular cell transplantation.

Selective withdrawal of one fluid through a second immiscible fluid has recently been used to encapsulate pancreatic islets (Fig. 3D, Fig. 4A) [108]. Islets suspended in PEG-diacrylate were layered onto a more dense, immiscible oil phase, and fluid was withdrawn through a tube placed immediately below the interface. At an appropriate flow rate, the PEG-diacrylate solution, containing islets, was entrained in a thin spout within the oil phase. When the diameter of an islet is greater than that of the spout, surface tension causes the spout to break at both ends, leaving a thin layer of polymer solution around the islet. PEG-diacrylate was subsequently photocros-slinked, resulting in ∼10 µm thick coatings independent of the size of the encapsulated islet. The investigators found it necessary to repeat this process in order to prevent coating defects, ultimately generating ∼20 µm thick coatings. In principle, sequential coating may allow for the generation of composite coatings with each layer having independent properties designed to elicit a preferred biological response. For example, the authors cite examples in which the outer layer may contain anti-inflammatory or pro-angiogenic molecules, while the inner layer may contain molecules to improve islet function. The authors found that two-layer conformal coatings inhibited the transport of a 140 kDa macromolecule and enabled normal dynamic insulin secretion in response to glucose stimulation (Fig. 4B). Despite these promising results, ∼75% of islets were lost during the coating process, a problem that must be remedied in light of donor shortage for human islet transplantation. Moreover, scalability of the technique to allow for encapsulation of a clinically relevant number of islets in a timely manner must be addressed.

Fig. 4.

Selective withdrawal of one fluid through a second immiscible fluid can be used to fabricate conformal coatings on pancreatic islets. (A) Schematic representation of encapsulation apparatus. Islets in the upper fluid are enveloped with a thin film of precursor solution (PEG-diacrylate and photoinitiator) as they are drawn across the interface into a withdrawal tube. The thin film is crosslinked upon exposure to an argon-ion laser. (B) Islets coated with a conformal PEG coating demonstrate normal dynamic insulin secretion behavior in response to glucose and membrane depolarization (reprinted from [108] with permission of Wiley-VCH).

5.3. Conformal coating via interfacial polymerization

Conformal coating strategies have also utilized the islet surface as a template upon which coatings may be grown or chemically deposited. Hubbell et al. have generated ∼35–50 µm thick PEG-diacrylate coatings (Fig. 3B) on both porcine [109–111] and human [112] islets through a process of interfacial polymerization. In this polymerization scheme, eosin Y, a photoinitiator, is non-specifically adsorbed to the islet surface, and islets are placed in a solution containing PEG-diacrylate and triethanolamine. Upon illumination with light, eosin Y is excited and donates an electron to triethanolamine, which initiates the free-radical polymerization of PEG-diacrylate at the islet-macromer interface [110]. Through parametric optimization of key process variables, greater than 90% islet viability and encapsulation efficiency was reported [110], and conformally coated islets were shown to behave comparably to non-coated islets in a dynamic glucose perfusion experiment and intraperitoneal glucose tolerance test [109]. Preclinical trials in diabetic cynomolgus monkeys and baboons demonstrated function of subcutaneously transplanted conformally coated islets for up to 20 months, despite discontinuation of low dose immunosupression (cyclosporine) one month post-transplant [112]. This technology is currently the basis for phase I/II clinical trials by Novocell for encapsulated human islet allografts, which began in 2005 [113]. Although patients are currently receiving transplants in a subcutaneous site, the low void volume of the graft and the high blood compatibility of PEG may also facilitate transplantation into the portal bed.

5.4. Molecular camouflage and nanoencapsulation

While conformal coatings on cell aggregates offer a significant decrease in void volume relative to conventional microcapsules, immunoisolation may, in principle, be accomplished using coatings or membranes of submicron, or nanoscale, thickness. A common approach to generating such barriers has been through immobilization of PEG chains to the cell or tissue surface, creating a molecular barrier of PEG to prevent molecular recognition between cell surface receptors and soluble ligands [114–120]. This has generally been accomplished through covalent coupling of PEG to amines of cell surface proteins or carbohydrates, or by direct insertion of PEG-lipid conjugates into the cell membrane [119]. PEG is a hydrated, flexible polymer chain due to repeating, highly mobile ether units, which allows the polymer chain to act as a steric barrier on the cell surface [121]. Through proper control of polymer chain length and surface grafting density, cell surface PEGylation has been shown to camouflage antigenic sites, alter surface charge, and attenuate cell-cell and receptor– ligand interactions [122]. In an effort to create a universal red blood cell (RBC) donor, PEGylation of RBCs has been shown to mask major and minor blood group antigens from host antibodies [116,123,124]. Evidence has recently emerged that conjugation of PEG to human peripheral blood mononuclear cells [115] and isolated murine splenocytes [114] can interrupt a number of receptor–ligand interactions important in allorecognition, including weakening CD28-B7 costimulation, resulting in T cell apoptosis. Furthermore, transplantation of PEGylated C57BL/6 splenocytes into lethally irradiated Balb/c mice significantly abrogated donor T cell proliferation and improved survival rates in this model of graft versus host disease relative to non-PEGylated controls [114].

Based on such promising findings, several groups have demonstrated that PEG can be grafted to islets without compromising viability or function [117,118,125] and have began to explore whether or not PEG grafting provides a mechanism of preventing, or at least attenuating, host response to both allo-[126–131] and xenografts [120,125]. Lee et al. have recently reported some efficacy of this strategy in a rat allograft model of islet transplantation into the kidney capsule [128–131]. In their most recent experience, islets were serially PEGylated three times in an effort to increase PEG surface density and improve uniformity [128]. In contrast to non-PEGylated controls, which all mice rejected within one week, 3 of 7 recipients transplanted with PEGylated islets provided maintenance of normoglycemia for more than 100 days without any immunosuppression. Histological evaluation demonstrated that PEGylation prevented graft infiltration by host immune cells, but did not prevent host cell recruitment to the graft site. Similarly, Contreras et al. have used islet surface PEGylation in a xenogenic model of intraportal islet transplantation [120]. While the authors did not monitor engraftment beyond two weeks, animals that received islets treated with PEG presented significantly better control of blood glucose levels than animals receiving non-modified islets. PEG with a molecular weight of 5 kDa performed slightly better than 2 kDa PEG, and capping of surface grafted PEG with albumin proved most efficacious, which was attributed to the enhanced capacity of thicker films and/or larger steric barriers to shield islets from complement and xenoreactive antibodies [120,125].

Despite these encouraging results, it is unclear whether or not surface grafted PEG will remain stable enough to provide protection for the anticipated lifetime of the graft. Covalently modified cell surfaces are likely to be turned over and remodeled with time due to endocytosis and/or shedding of PEG-conjugated cell surface macromolecules. Moreover, PEG on the surface of cells with a finite lifecycle is likely to be lost and replaced with fresh tissue, thereby restoring immunogenicity [121]. Lee et al. have demonstrated via avidin staining of biotinylated PEG that PEG is present for at least one month [131], but further time points were not explored.

The efficacy of PEGylation may also be limited, in part, by the lack of a defined pore structure and dependence on a steric exclusion effect to provide an immunoprotective barrier. Therefore, several groups have recently began to explore the possibility of constructing nano-thin films of controlled permeability and surface chemistry directly on the surface of cells and tissues via layer-by-layer polymer self-assembly [132], effectively creating conformal, permselective membranes of nanoscale thickness. Elbert et al. assembled an alginate/PLL polyelectrolyte multilayer (PEM) film on gelatin to limit cell adhesion to the proteinaceous surface [133]. Likewise, Thierry et al. have coated damaged endothelium with PEM films consisting of chitosan and hyaluronic acid to inhibit platelet deposition [134]. PEM films have also been assembled on the surface of gluteraldehyde-fixed red blood cells [135], yeast [136,137], and bacteria [138], but their assembly on viable mammalian cells and tissues has proven more challenging due to the well documented toxicity of most polycations [139–141]. Germain et al. have explored the possibility of encapsulating MELN and HeLa cell lines with various polycations and anionic poly(styrene sulfonate) (PSS) [142]. Most polycations explored were extremely cytotoxic, though nine bilayers of poly (diallyldimethylammonium chloride)/PSS could be assembled with only a ∼25% decrease in cell viability. Coated cells demonstrated impaired receptor mediated endocytosis of transferrin (80 kDa) and a lectin (64 kDa), suggesting that multilayer films may provide a barrier for transport of these molecules. Similarly, Veerabadran et al. have recently reported the assembly of PLL/hyaluronic acid multilayer films on mouse mesenchymal stem cells [143]. Krol et al. have attempted to extend this concept to coat human pancreatic islets, demonstrating a minimal loss of islet function and viability when coated with a poly(allylamine hydrochloride) (PAH)/PSS/PAH multilayer film [283]. This finding is surprising given the reported toxicity of PAH to pancreatic islets [128]. A closer analysis of their data indicate that the PAH is localized both intra- and extracellularly, consistent with polycation-induced pore formation, diffusion of polymers into the cytoplasm, and subsequent cell death [140]. This apparent discrepancy deserves further consideration and exploration. Miura et al. have attempted to assemble an alginate/PLL/alginate multilayer on islets by first inserting a lipid–PEG–NH₂ conjugate into islet cell membranes, thereby generating a positively charged islet surface to facilitate electrostatic binding of negatively charged alginate [144]. Islets coated with this membrane appear to function normally in a static glucose stimulation assay, but no attempts to characterize the barrier capacity of this coating were made. Nonetheless, deposition of thin, multilayer films directly on the surface of cells and tissues provides a promising approach to minimizing void volume while generating a true permselective membrane.

Though PEGylation and nanoencapsulation remain immature technologies, the nanoscale thickness of such coatings offer a number of potential advantages relative to traditional immunoisolation or conformal coating approaches. Stuhlmeir and Yin demonstrated that PEGylation of endothelial cells inhibited binding of immunoglobulins and TNF-α, and reasoned that perfusion of hearts with reactive PEG might attenuate hyperacute xenograft rejection [145]. Though in vivo results were discouraging, this study exemplifies the potential utility of nanoassembled coatings to immunoprotect whole organs.

6. Early inflammatory events in transplantation of immunoisolated cells and tissue

Non-specific inflammatory responses to immunoisolated cells and tissue may also result in limited graft survival and function (Fig. 5). The inflammatory response to immunoisolated cells may be mediated by a foreign body response to the device itself, the surgical procedure necessary for implantation of the device, or the release of inflammatory mediators from the encapsulated cells. Cellular overgrowth of devices generates additional physical and metabolic barriers to the effective transport of metabolites, further exacerbating the aforementioned mass transport limitations. Moreover, localized acute or chronic inflammation can generate a sea of cytotoxic molecules capable of diffusion across immunoisolation devices, potentially damaging the encapsulated graft.

Fig. 5.

Cytotoxic inflammatory responses may be initiated and/or propagated via surgery associated with device implantation, the foreign body response to the implanted material, and/or the release of inflammatory mediators by immunoisolated cells.

6.1. Inflammation induced by devices and surgery

The inflammatory, or foreign body, response to implanted biomaterials, reviewed thoroughly elsewhere [146–149], is generally described as a dynamic biochemical process, initiated by non-specific adsorption of proteins to the material surface, followed by recruitment of neutrophils and macrophages to the implant site, and the subsequent attachment and overgrowth of the device by macrophages, foreign body giant cells, and fibroblasts [150]. While the severity of foreign body responses to immunoisolation devices is dependent on transplantation site and material properties, such as surface charge, porosity, roughness, surface chemistry and free energy, and implant size [146], this generalized response has been observed on a variety of immunoisolation devices [147], including the commonly employed alginate/PLL/alginate (APA) microcapsule [151,152] and the TheraCyte device produced by Baxter Healthcare [153]. For many years, the inherent biocompatibility of immunoisolation devices was implicated as the principle cause for capsular overgrowth and subsequent graft failure [154]; accordingly, many groups focused their efforts on improving the biocompatibility of immunoisolation membranes and encapsulation devices. Surface PEGylation has been used to improve the biocompatibility of synthetic immunoisolation membranes [155,156], as well as APA microcapsules [157]. Use of highly purified alginate of appropriate composition has improved the biocompatibility of alginate-based microencapsulation devices [76]. The use of polycations, in particular PLL, in membrane forming processes has been shown to mediate adhesion of fibroblasts and macrophages [158] and induce cytokine production [159], and many groups have recently abandoned use of polycations in microcapsule formulations [56,160–162]. However, several polycation-containing microcapsules have been optimized such that fibrotic overgrowth of empty capsules is minimized [43,163–165]. Most notably, poly (L-ornithine) (PLO) and PLL containing alginate capsules are currently being utilized in clinical trials of encapsulated islets [21,22].

The surgical procedure associated with device implantation, though often minimally invasive, generates an inflammatory response necessary to facilitate normal wound healing [146]. Indeed, Robitalle et al. have recently observed significant neutrophil infiltration and inflammatory cytokine expression (IL-1β, IL-6) in the 17 h immediately after intraperitoneal injection of saline [152]. Similarly, Roth et al. have reported early and transient NF-κB expression during sham saline injections into fat pads [166]. While it is unclear if this ‘intrinsic’ inflammatory response is capable of adversely effecting graft function, it likely initiates or propagates a foreign body response to the device [146,152], as well as inflammatory responses to the immunoisolated cells [154], as discussed below.

6.2. Inflammation mediated by immunoisolated cells

Encapsulated cells may produce low molecular weight inflammatory mediators capable of diffusion across immunoisolation membranes, potentially triggering inflammatory cell recruitment and activation [40,56,167–170]. Isolated pancreatic islets, for example, have been shown to produce a number of proinflammatory molecules (Table 2) [32,171–184]. Consequently, transplantation of non-encapsulated islets is associated with elevated inflammatory responses in the immediate post-transplant period, characterized by intense macrophage infiltration [185,186], increased expression of endothelial cell and leukocyte adhesion molecules (E-selectin, ICAM-1) [186], and elevation of proinflammatory mediators (TNF-α, IL-1β, and nitric oxide) [75,186,187]. Significantly, recent data suggests that levels of inflammatory molecules produced by islet preparations pre-transplant may correlate with the outcome of clinical islet transplantation [188].

Table 2.

Soluble inflammatory mediators expressed by pancreatic islets

Inflammatory mediator	Molecular weight (Da)	Ref.
TNF-α	51,000a	[176–179]
IL-6	21,500–28,000	[176,180]
IL-1β	17,000	[175–177,179]
Macrophage migration inhibitory Factor (MIF)	12,000	[175]
CXCL9 (MIG)	11,700	[181]
CXCL10	10,000	[181]
IL-8	8000	[175,180]
CCL5 (RANTES)	8000	[181]
MIP-1α	7800	[178]
MCP-1	6000–7000	[175,179,180,182,183]
CXCL2 (MIP-2α)	6000	[181]
Nitric oxide	30	[177,184]

^aSoluble trimeric form.

Not surprisingly, similar mechanisms appear to be operative when islets are encapsulated. Co-culture of isolated peritoneal macrophages and syngeneic islets encapsulated in APA microcapsules mediates islet-specific macrophage activation accompanied by TNF-α and IL-1β release, suggesting that factors other than alloantigens are capable of diffusing across the capsular membrane to activate macrophages [189]. Similarly, co-culture of conformally coated islets and macrophages results in the production of several inflammatory mediators, including MIP-2, IL-1, TNF-α, and IL-6 [190].

6.3. Effects of the inflammatory environment on immunoisolated cells

Generation of such an inflammatory milieu may be highly detrimental to graft performance and longevity [172,187,191]. This is exemplified in an experimental model of syngeneic islet transplantation into non-autoimmune diabetic mice, whereby ∼60% of transplanted islet tissue was lost 3 days post transplantation [192], demonstrating that early islet destruction is not allo- or autoantigen-specific, but rather mediated, in large part, by generation of cytotoxic inflammatory molecules [172,187]. Elegant studies by de Vos et al. suggest that comparable mechanisms may be active in the failure of encapsulated islet grafts [60,99,189]. Examination of encapsulated islet allografts retrieved from the peritoneum revealed that only ∼10% of capsules were overgrown with fibrotic tissue [60,99,163,193]; however, this was accompanied by a 40% loss of viable cells within the first 4 weeks of transplantation [60]. Notably, encapsulated islet autografts performed comparably, suggesting that graft failure was mediated by non-specific mechanisms [163,193]. Histological evaluation indicated that activated macrophages were the predominant cell type attached to overgrown capsules, and co-culture of macrophages and encapsulated islets resulted in macrophage activation, cytokine production, and impaired islet function [60,189]. Similarly, de Groot et al. co-cultured encapsulated islets overgrown with host cells retrieved from the peritoneum with freshly encapsulated islets at a 1:9 ratio (i.e. 10% overgrowth) [194]. Impaired insulin secretion in response to glucose, decreased beta cell replication, and increased cell necrosis occurred after 48 h of co-culture. IL-1β, TNF-α, and nitrite, a marker for nitric oxide (NO), were elevated in the culture media, and analysis of mRNA expression profiles of encapsulated islets suggested that NO mediated islet damage [194]. While some immunoisolation membranes have been reported to protect cells from IL-1β and/or TNF-α [40,169], blockade of free radical diffusion is not likely. Indeed, Wiegand et al. and Chae et al. have demonstrated that, despite its short half-life, NO can destroy encapsulated islets [195,196]. This has recently been supported by a mathematical model of free radical diffusion through a spherical matrix containing pancreatic islets [197]. Not surprisingly, depletion of macrophages improves engraftment of both encapsulated and non-encapsulated cells [75,198]. Hence, significant evidence has emerged that limited survival and function of immunoisolated cells is at least partially mediated by non-immune, inflammatory responses.

7. Strategies to abrogate inflammatory responses

7.1. Systemic delivery of anti-inflammatory agents

The inflammatory response to encapsulated islets has been reported to be complete within 1–4 weeks [99,151,158], and, therefore, temporary anti-inflammatory therapy may be sufficient to dramatically reduce early inflammatory events associated with transplantation of immunoisolated cells. Single-dose or short term administration of drugs that inhibit cytokine actions or macrophage function have been found to dramatically improve islet engraftment in experimental models of islet transplantation (Table 3) [75,178,186,199–204]. Despite parallels between primary islet non-function and limited survival of immunoisolated islet grafts, surprisingly little consideration has been given to adjunctive anti-inflammatory therapy for the transplantation of immunoisolated cells and tissues. Omer et al. delivered clodronate liposomes to deplete peritoneal macrophages, reducing capsular overgrowth and improving islet function [198]. While macrophage depletion is not a clinically viable approach, delivery of agents that inhibit or modulate macrophage activation, such as 15-deoxyspergualin [205], in the immediate post-transplant period could prove valuable in abrogating early inflammatory responses and subsequent destruction of immunoisolated tissue.

Table 3.

Anti-inflammatory therapies for improving islet engraftment

Therapeutic molecule	Treatment regimen	Ref.
15-deoxyspergualin	IP injection (day-1 to day 9)	[203]
ICAM-1 antisense oligodeoxynucleotide	IP injection (7 days)	[202]
anti-ICAM-1 mAb	IP injection (7 days)	[202]
anti-IFN-γ mAb	IP injection (day 0,2,4)	[204]
anti-IL-1β mAb	IP injection (day 0,2,4)	[204]
Acetylsalicyclic acid	Orally (daily for entire experiment)	[199]
IL-1ra	IP injection (day-1 and 0, and 4 h)	[199]
Pravastatin	Orally (daily for first 14 days)	[200]
Anti-MCP-1 mAb	IP injection (3×/week for 2 weeks)	[201]
Activated protein C	IV injection (1 h pre-Tx)	[186]
Dichloromethylene diphosphonate	IV injection (2 days pre-Tx)	[75]
α1-antitrypsin	IP injection (day-1, every 3 days post Tx)	[178]

IP: intraperitoneal, IV: intravenous.

7.2. Bioactive immunoisolation barriers that locally attenuate inflammatory responses

To avoid potential adverse consequences associated with systemic delivery of anti-inflammatory molecules, recent efforts have focused on integrating anti-inflammatory capabilities into immunoisolation devices. Loading of microcapsules with anti-inflammatory molecules, for example dexamethasone [206], offers a simple approach, but may be limited by undesirable release kinetics and/or cytotoxic intracapsular concentrations. Co-encapsulation of cells and drug delivery vehicles offers a rational alternative for controlled delivery of anti-inflammatory agents. Luca et al. co-encapsulated cellulose acetate microspheres (30–70 µm) containing the antioxidant vitamin D3 with rat islets in alginate/PLO microcapsules [207]. Similarly, Ricci et al. found that microcapsules charged with 5 µm polyester microspheres releasing the non-steroidal anti-inflammatory drug ketoprofen reduced the foreign body response to polycation coated microcapsules [208].

Encapsulated cells may be further protected through immobilization of cells or molecules that scavenge, inhibit, or metabolize cytotoxic molecules that diffuse across the immunoisolation barrier. Wiegand et al. have demonstrated that coencapsulation of islets with autologous erthyrocytes within alginate capsules provided nearly complete protection from macrophage-mediated cell lysis due to the capacity of erthyrocytes to scavenge NO and/or convert it to nitrate [196]. Recently, Chae et al. have extended this concept, replacing erthyrocytes with hemoglobin crosslinked with PEG. APA capsules containing crosslinked hemoglobin protected rat islets and RINm5F insulinoma cells from NO-mediated cellular damage [195]. Significantly, after transplantation of a suboptimal mass of encapsulated islet xenografts these microcapsules were found to prolong normoglyemia and improve glucose clearance relative to capsules formulated without crosslinked hemoglobin [209]. While this effect may have also been mediated by improved oxygen tension in the capsule [209,210], this study exemplifies the potential efficacy of actively antiinflammatory immunoisolation devices.

Our lab has postulated that the native endothelium provides a structural and biochemical model for the design of immunoisolation membranes through its ability to act as a template for the assembly of complex macromolecular structures that regulate inflammatory processes. In this regard, we developed a substrate-supported, polymerizable, cell membrane-mimetic thin film, which can be readily assembled on the surface of conventional APA microcapsules (Fig. 6) [165,211,212]. We have recently shown that microcapsules coated with a membrane-mimetic film are stable and largely free of pericapsular overgrowth after four weeks in the peritoneal cavity [165]. Moreover, membrane-mimetic films have been assembled on microcapsules containing CHO cells [212] and pancreatic islets [211] with minimal loss of cell viability or function. Importantly, we have previously demonstrated the ability of membrane-mimetic films to serve as a template for the assembly of membrane-bound macromolecules that may abrogate inflammatory responses to encapsulated cells, specifically heparin and the transmembrane enzyme thrombomodulin [213–217]. Though perhaps better known for its anticoagulant properties, immobilized heparin can inhibit the formation of NO through its capacity to bind superoxide dismutase [218], and has been shown to limit complement activity by inhibiting the formation of C3 convertase and the assembly of C5b-9 [219–221].

Fig. 6.

Construction of a polymerized, self-assembled, membrane-mimetic thin film on the surface of alginate/poly(L-lysine) microcapsules. (A) Alternating layers of poly(L-lysine) and alginate are first assembled on an alginate/Ca²⁺ hydrogel microsphere, followed by adsorption of an ampiphilic terpolymer with anionic anchoring groups. Following monolayer fusion of mono-acrylated phospholipids, photoinitiated polymerization is performed to crosslink the lipid layer. The film may be rendered anti-inflammatory by incorporating surface-bound heparin and thrombomodulin. (B, C) By doping mono-acrylate PC films with 0.1 mol% Texas Red acrylate PE, membrane-mimetic films can be readily observed on the surface of microcapsules (reprinted from [165] with permission from Elsevier). (D) Islets encapsulated in capsules coated with a membrane-mimetic film are viable as evidenced by ethidium homodimer and calcein AM staining. (E) Empty capsules retrieved from the peritoneal cavity of mice after 30 days remain free of cellular adhesion and capsular overgrowth (reprinted from [165] with permission from Elsevier).

Thrombin, invariably generated during device implantation [146], is an important conductor of cellular responses during inflammation [222]. Thrombin acts as a chemoattractant, directing neutrophils and monocytes to the injured site [222], can trigger expression of cell adhesion molecules necessary for inflammatory cell migration [223–225], and can mediate expression of proinflammatory cytokines, IL-6 and IL-8, and platelet activating factor, a potent neutrophil activator [223]. Thrombomodulin (TM) binds thrombin in a 1:1 ratio acting as a thrombin sink [222]. Therefore, TM immobilized on the pericapsular surface may sequester thrombin, preventing its participation in inflammatory processes. Perhaps more importantly, TM is able to redirect the catalytic activity of thrombin towards generation of activated protein C (APC), a potent antiinflammatory molecule [226,227]. Though the complete antiinflammatory mechanism of APC has yet to be fully elucidated, APC has been shown to block F-κB nuclear translocation, an important event in many inflammatory cascades [228]. Indeed, APC has been shown to inhibit macrophage activation, production of proinflammatory cytokines (TNF-α, IL-1β, IL-6, IL-8) [229–233], and endothelial cell expression of E-selectin and ICAM-1 [234,235]. Notably, systemic administration of APC delayed xenograft rejection in a guinea pig to rat cardiac transplant [233], and significantly improved islet engraftment in a mouse autograft model of intraportal islet transplantation [186]. Given the capacity to incorporate such proteins and carbohydrates, membrane-mimetic films offer a route to generate robust, biocompatible, and actively anti-inflammatory immunoisolation membranes.

8. Graft failure due to activation of indirect antigen presentation pathways

The design features of an immunoisolation barrier are driven by the intent to limit the effects of rejection pathways initiated after implantation of donor cells. Recent advances in immunobiology have elucidated alternative mechanisms of immune system activation with important implications for the design of effective immunoisolation barriers. In particular, the role of indirect antigen presentation pathways in the rejection of xenografts, and to a lesser, but notable extent, allografts, has challenged the premise that an effective barrier can be created based solely on the selection of physical dimensions necessary to exclude cytotoxic molecules [236–238].

Allograft rejection is primarily mediated by cytotoxic CD8+ T cells activated by donor MHC-peptide complexes expressed on the surface of graft-derived antigen presenting cells;a process that has been referred to as direct antigen presentation [236,237]. Immunoisolation barriers that prevent cell–cell contact between donor cells and host immune cells are, therefore, expected to block the direct presentation pathway. By contrast, rejection of non-vascularized xenograft tissue is characterized by a response in which CD4+ helper T cells play a major role. In this pathway, termed indirect antigen presentation, host antigen presenting cells display peptides scavenged from free donor proteins to engage CD4+ T cells, which develop into Th2 cells. In turn, Th2 cells produce cytokines that stimulate maturation of B cells into plasma cells, which secrete xenoantigen-specific antibodies [236,237]. While many immunoisolation barriers prevent passage of antibodies, this may not be a prerequisite to protect cells from a humoral response [239], provided the required immunocellular and complement components, many of which are larger than antibodies, are excluded [146]. However, even if not in direct contact with the donor cell or tissue, immune complexes generated by the binding of newly formed antibodies to shed antigens may lead to activation of macrophages and recruitment of neutrophils through activation of the complement cascade or by direct binding of antigen–antibody complexes to leukocyte cell-surface Fc receptors [240,241]. Upon initiation of such a response, T cells and activated macrophages in the region of the graft secrete low molecular weight cytokines and free radicals that, as discussed previously, may be highly destructive to encapsulated cells (Fig. 7) [172,187,191]. Indeed, compelling evidence indicates that indirect antigen presentation is an operative mechanism in the destruction of immunoisolated xenogenic cells and tissues, as evidenced by production of xenoantibodes, CD4+ T cells, and macrophages at the graft site [242–249]. Recently, Orlowski et al. have shown that encapsulation of xenoislets in APA microcapsules does not prevent, but merely delays, provocation of the second set phenomenon, a clear indicator of host sensitization and a potential obstacle if subsequent cell transplantation is necessary [250].

Fig. 7.

Host recognition of immunoisolated cells initiated via the indirect antigen presentation pathway. AgX, shed foreign antigen; MHC, major histocompatibility complex; MHC+pep, MHC presenting foreign peptide; TCR, T-cell receptor; FcR, Fc receptor; Ig, immunoglobulin; IL, interleukin; TNF, tumor necrosis factor; NO, nitric oxide (reprinted from [282] with permission from Birkhäuser Verlag AG).

The degree to which indirect antigen presentation plays a role in graft rejection is related to disparity between shed and host epitopes [237], which likely explains the common observation that immunoisolated xenogenic tissue is rejected more rapidly than allogenic tissue (Table 1). In principle, however, this mechanism is also operative in immunoisolated allograft tissue [237], which may explain several studies in which encapsulated allografts were rejected more quickly than autografted controls [251–253]. The significance of indirect antigen presentation in immunoisolated allograft rejection remains controversial and inadequately investigated. While it is unclear what, if any, shed alloantigens are responsible for initiating host responses to immunoisolated cells, some evidence indicates that peptides derived from MHC molecules themselves are the most likely candidates [237]. Clearly, the design of a barrier that prevents indirect antigen presentation by preventing the release of graft protein or peptide antigens, while simultaneously permitting passage of nutrients and therapeutic agents, remains a difficult hurdle to overcome.

9. Strategies to attenuate immune responses to immunoisolated cells

9.1. Adjunctive immunosuppressive therapy

While complete elimination of immunosuppression remains the paramount objective of immunoisolation, the immunoprotective capacities of many existing technologies may facilitate use of modified immunosuppressive therapies with important clinical benefits. Encapsulation has allowed several investigators to significantly reduce the dose of harmful immunosupressive agents necessary to protect allografted tissue [129,131,254,255]. For example, Lee et al. have reported combining covalent conjugation of PEG to allogenic islets and low dose cyclosporine (Fig. 8) [131]. Initial administration of cyclosporine for 2 weeks at 3 mg/kg/day followed by 1 mg/kg/day thereafter protected PEGylated rat islet allografts for more than 1 year. Non–modified islets receiving the same immunosuppressive treatment, as well as PEGylated islets transplanted in the absence of immunosupression were rejected at approximately 12 days.

Fig. 8.

Islet PEGylation and low dose immunosuppression act synergistically to protect pancreatic islets for one year in a rat allograft model. (A) Proposed mechanisms of immunological protection. (B) Maintenance of normoglycemia for one year post-transplant (reprinted from [131] with permission from Elsevier).

In a recent series of studies [256–258], Aebischer et al. have shown that combining cell encapsulation and transient immunosupression can induce host acceptance of xenogenic cells. Short-term administration (1–4 weeks) of tacrolimus significantly prolonged the survival of encapsulated genetically engineered myoblasts relative to non-encapsulated controls [256]. Importantly, this initial immunosuppressive treatment also prolonged survival of a subsequent encapsulated graft, indicating that cell immunoisolation and transient immunosupression may generate a state of host unresponsiveness to xenografts [257]. This finding has important implications for cell-based therapy, as encapsulated xenogenic cells could be replaced without the need for further immunosuppression. Interestingly, in a clinical trial of APA encapsulated islets by Amcyte, patients will receive a low dose, short term course of the immunosupressive agent Rapamune® [22].

Improved understanding of immune responses to encapsulated cells has helped elucidate targets for immune intervention, leading to the development of more specific and less harmful immunosuppressants. In particular, blocking antibodies and soluble antagonists targeted towards pathways necessary for T cell activation via the indirect antigen presentation pathway have emerged as promising adjuncts to cell encapsulation [153,248,249,259]. Safley et al. have observed that treatment with CTLA4-Ig, a CD28 antagonist, and anti-CD154, an inhibitor of CD40/CD40-ligand interactions, resulted in prolonged survival of encapsulated porcine xenografts in spontaneous diabetic mice, accompanied by reduced capsular overgrowth, inflammatory cell infiltration, and inflammatory cytokine production [248].

9.2. Bioactive immunoisolation barriers that locally attenuate immune responses

In addition to preventing the cell–cell interactions that underlie the initiation of the direct antigen presentation pathway, current immunoisolation strategies might be enhanced by incorporation of cells or molecules that actively limit immune responses to encapsulated cells. The rationale for a number of such barriers has emerged from the observation that co-transplantation of testicular Sertoli cells with allo- [260– 262] or xenogenic [263,264] tissue significantly abrogates graft rejection. In a controversial but noteworthy study, Valdes et al. co-transplanted isolated neonatal porcine islets and Sertoli cells in a vascularized subcutaneous chamber into diabetic human patients. At four year follow up, porcine insulin was present in many patients. Half of the patients showed significant reduction in exogenous insulin requirements, and two were insulin independent for several months [265].

While the mechanisms through which Sertoli cells confer protection to foreign tissue is not well understood, recent data suggests that they may locally modulate immune responses through secretion of bioactive factors such as TGF-β and clusterin [266]. Hence, cell–cell contact between Sertoli and immune cells may not be necessary to confer immunoprotection, though close proximity to grafted tissue appears to be required [260]. Korbutt et al. co-encapsulated Sertoli cells with allogenic pancreatic islets in alginate microcapsules, and reported improved encapsulated graft survival in the peritoneum, extending normoglycemia from 10.4±0.6 to 85±4.9 days [267]. Yang et al. have extended this concept to xenogenic immunoisolation, co-encapsulating fish islets and syngeneic or xenogenic Sertoli cells [268]. In accord with Korbutt et al., microcapsules containing Sertoli cells prolonged survival of intraperitoneal islets grafts (46±6.3 days) relative to capsules containing islets alone (21±6.7 days). It has recently been hypothesized that graft function might be prolonged through replenishment of Sertoli cells at the site of transplantation. In this regard, Luca et al. [269] have produced microcapsules for optimized Sertoli cell viability that may be transplanted along side immunoisolated cells. It is unclear whether or not this approach will confer the same benefit as co-encapsulation of both cell types.

Cells of immunoprivileged tissues express a type-II membrane receptor known as Fas ligand (FasL), which induces apoptosis of autoreactive lymphocytes or infiltrating inflammatory cells through its interaction with Fas (CD95) [270]. The efficacy of the Fas/FasL pathway in conferring immunoprotection at ectopic sites is unclear and reports are conflicting [271– 274]. Lau et al. found that co-transplantation of islets with myoblasts genetically engineered to express FasL dramatically improved allograft survival [271]. Yolcu et al. observed similar results using splenocytes chemically modified to present a novel chimeric streptavidin-FasL molecule [272]. By contrast, islets engineered to express FasL on their beta cells did not protect grafts from rejection and were more susceptible to inflammation-mediated destruction [274]. Nonetheless, presentation of immunoregulatory molecules on the surface of immunoisolation devices may provide a mechanism to specifically and/or locally prevent immune responses to immunoisolated cells. Cheung et al. have recently functionalized PEG-diacrylate hydrogels with anti-Fas antibodies (FasAb) using cell compatible chemistry [275]. FasAbs can induce T cell apoptosis in a manner similar to FasL via crosslinking of Fas receptors on T cells. FasAb functionalized hydrogels were shown to induce apoptosis of Fas sensitive Jurkat T cells, whereas non-modified barriers did not. Though the efficacy of this approach remains to be seen, this work establishes an important step towards the design of stable, biologically active immunoisolation barriers that address operative mechanisms of immunoisolated graft rejection.

10. Conclusions

Protection of transplanted cells from deleterious host immune responses will be necessary to exploit the full clinical potential of cell-based therapeutics, and immunoisolation stands to play a pivotal role towards this end. Over thirty years of intensive research directed at developing and optimizing the performance of immunoisolation devices has yielded both promise and inherent challenges, several of which have been discussed in detail within this review. Maintenance of adequate oxygen and nutrient transport is of paramount importance for success of any transplant, but particular attention must be given to immunoisolated cells due to often limited vascularization of the device or at the intended site of transplantation. Early inflammatory responses are intrinsic to any implanted cell-material composite, and such responses, if not inherently detrimental, can initiate chronic inflammation or otherwise intensify immune responses to transplanted cells and tissues. Finally, recent advances in immunobiology have identified fissures in the prevailing dogma of immunoisolation, particularly as applied to xenogenic tissue. Though advances in membrane technology may ultimately make it possible to prevent such host responses through proper control of membrane transport properties, the possibility that indirect antigen presentation may be operative should be recognized and new paradigms in the development of immunoisolation barriers must be explored (Fig. 9). With these limitations in mind, barrier design must ultimately be governed by clinical objectives and the inherent pathophysiologic mechanisms associated with the underlying disease process. Continued collaboration between engineers, biological and physical scientists, and clinician-scientists will be essential for defining realistic objectives and overcoming existing obstacles.

Fig. 9.

Emerging paradigms in the immunoisolation of cells and tissues include implantation of encapsulated cells into prevascularized sites or native tissue microvasculature, minimization of transplant volume through use of conformal coatings or nanoencapsulation, and local attenuation of host inflammatory and immune responses to encapsulated cells using bioactive membranes