Noninvasive evaluation of the vascular response to transplantation of alginate encapsulated islets using the dorsal skin-fold model

Alginate encapsulation reduces the risk of transplant rejection by evading immune-mediated cell injury and rejection; however, poor vascular perfusion results in graft failure. Since existing imaging models are incapable of quantifying the vascular response to biomaterial implants after transplantation, in this study, we demonstrate the use of in vivo laser speckle imaging (LSI) and wide-field functional imaging (WiFI) to monitor the microvascular environment surrounding biomaterial implants. The vascular response to two islet-containing biomaterial encapsulation devices, alginate microcapsules and a high-guluronate alginate sheet, was studied and compared after implantation into the mouse dorsal window chamber ( N = 4 per implant group). Images obtained over a 14-day period using LSI and WiFI were analyzed using algorithms to quantify blood flow, hemoglobin oxygen saturation and vascular density. Using our method, we were able to monitor the changes in the peri-implant microvasculature non-invasively without the use of fluorescent dyes. Significant changes in blood flow, hemoglobin oxygen saturation and vascular density were noted as early as the first week post-transplant. The dorsal window chamber model enables comparison of host responses to transplanted biomaterials. Future experiments will study the effect of changes in alginate composition on the vascular and immune responses.


Introduction
Insulin is secreted by β-cells, the predominant cell type within the islets of Langerhans and is the primary hormone responsible for regulating carbohydrate and fat metabolism. The autoimmune destruction of insulin-secreting β cells results in a condition called type I diabetes (T1D) where insulin secretion is deficient, resulting in elevated blood sugar levels. T1D affects over three million children and adults in the U.S. and the incidence is on the rise [1]. Exogenous insulin replacement involving multiple daily injections or the delivery of exogenous insulin via a subcutaneously-implanted pump along with frequent blood glucose monitoring [2] remains the standard of care in the management of T1D. Patients on insulin therapy can experience fluctuating blood glucose levels and severe hypoglycemia, which in some cases, can lead to unconsciousness, seizures, coma or death [3].
Human islet allotransplantation is a low-risk alternative to conventional insulin therapy and can improve glycemic control in diabetic patients. However, a comprehensive review by the CITR (Collaborative Islet Transplant Registry) has reported unsatisfactory long-term success rates [4]. The need for life-long immunosuppression [5], a crippling scarcity of suitable healthy organs from cadaveric donors and inconsistencies in islet yields [6] currently restrict the application of this treatment modality to severe cases of T1D with multiple co-morbidities [7,8]. Newly available encapsulation technologies can help eliminate the need for immunosuppression [9], or help reduce the dose by providing localized immunosuppression at the graft site [10]. Small and large animal studies conducted using alginate macroencapsulation [11] and microencapsulation devices [12,13] have demonstrated prolonged islet allograft and xenograft survival without the need for immunosuppression. These implants provide for nutrient, oxygen, insulin, and metabolite transport by passive diffusion and simultaneously function as an immunoisolation barrier that safeguards islets from cytokine, complement and cell-mediated immune attack [14]. Encapsulated islets may be implanted in a variety of surgically accessible sites in a minimally invasive manner, thereby avoiding the morbidities associated with intra-portal transplantation [15] (see Supplementary Table 2B for a comparative analysis of various sites commonly used in encapsulated islet transplantation).
Porcine islet xenotransplantation is being studied as an alternative to human islet allotransplantation since this option would mitigate donor scarcity issues [16,17] by providing a virtually limitless source of islets [18]. Also, porcine insulin being structurally similar to human insulin [19], has been used to treat T1D patients for decades [20]. While pigs of various strains and ages have been used as islet sources [21], we have expanded on the work done by Korbutt et al. on neonatal pigs [22,23] and developed a simple and effective method of isolating islets from young pigs [24]. Islets isolated using this method were used in our noninvasive imaging studies.
Several techniques are being studied for use in in vivo islet imaging and analysis [25] (See Supplementary Table 2A). Most recently, the mouse and primate anterior eye chamber models (AECM) have generated intense scientific interest in the field of islet transplantation and rejection imaging [26]. Unfortunately, this model does not allow for the study of macroencapsulation devices owing to their macroscopic dimensions, planar configuration or both (See Supplementary Table 2B). Other imaging modalities used in real time implant evaluation include radiological imaging, such as magnetic resonance imaging [27], positron emission tomography [28], computerized tomography [29], or ultrasound imaging [29]. Although noninvasive, such tomographic imaging modalities are time-consuming and require repeated exposure to harmful penetrating radiation (See Supplementary Table 2A). Radiological and bioluminescent imaging techniques [30] are unable to provide highresolution images of the transplanted islets. These techniques also require the injection of several potentially harmful contrast agents if multiple vascular or biocompatibility parameters need to be monitored simultaneously. Thus, there is an urgent need to develop an imaging modality that can evaluate these properties in vivo to expedite the translation of in vitro and small animal studies to large animal research and human clinical trials.
The dorsal windowchamber, first described by Algire in 1943, is an in vivo model that has been used extensively in the study of subdermal microvasculature [31]. This versatile model has been used extensively in the evaluation of angiogenesis [32], tumor physiology [33], targeted biomolecular and laser-based therapies for vascular lesions [34], leukocyteendothelium interaction after muscle injury [35], and islet transplantation [36], and provides a multi-modal platform where bright field microscopy, intravital fluorescence microscopy [37], multispectral imaging, and laser speckle imaging (LSI) techniques [38] can be efficiently used in the noninvasive evaluation of subdermal vascular hemodynamics and implant biocompatibility.
In this study, we compare the host vascular response to xenogeneic islets encapsulated within two alginate-derived biomaterial implants -a planar macroencapsulation device (Islet Sheet) and a spherical microencapsulation construct (UP LVM alginate micro-capsules)and study their efficacy in islet transplantation using the dorsal window model.

Islet isolation and evaluation
Pancreata harvested from young male Yorkshire pigs (14-22 days, S&S Farms) were used for islet isolation. The islets were cultured for 8-10 days using protocols developed in our laboratory as previously described [24] and counted under 25 ×magnification after staining with dithizone [39]. Their viability was assessed using fluorescence microscopy with a mixture of Newport Green DCF diacetate (Life Technologies, NY) and propidium iodide (Life Technologies, NY) [40]. Islet function studies were performed by monitoring insulin release in vitro, after a glucose challenge [41,42]. All animal procedures were performed under approved Institutional Animal Care and Use Committee protocols at the University of California, Irvine.

Dorsal window chamber
The dorsal window chamber model is surgically fashioned out of a segment of dorsal skin between two titanium frames as described in detail by Moy et al. [45]. The skin on one side of the window is excised and the window is then covered using a circular glass slide, allowing the subdermal microvasculature of the opposing skin to be visualized (Fig. 1A, B).

Dorsal window surgery and islet implantation
After encapsulation (as detailed above), the microcapsule and sheet implants (with or without porcine islets) were placed (N = 4 per group) into window chambers that were mounted on male C57BL/6 albino mice. Appropriate controls (blank UP LVM alginate microcapsules and alginate sheets), were included (N = 4 per group).

In vivo imaging
On the day of surgery (Day 0) and on days 2, 4, 7, 10 and 14 after transplantation, wide-field functional imaging (brightfield imaging, and multi spectral imaging) and laser speckle imaging were performed as described previously by Moy et al. [45]. The images obtained were analyzed using MATLAB V.7.4; and the changes in subdermal vascular morphology, vascular caliber, relative blood flow, hemoglobin oxygen saturation [46,47], and functional vascular density [47] were computed using custom-written codes.

Laser speckle imaging
Laser speckle imaging [48] is a technique that allows wide-field image acquisition without the need for complex instrumentation. The technique utilizes the speckle pattern that is observed when moving particles, like red blood cells, are illuminated by laser light. As the cells flow through blood vessels, they scatter the laser and produce a speckle pattern that rapidly changes over time, especially when this pattern is recorded using a camera with relatively long (~1000 ms) exposure times. By acquiring images of the blurred speckle pattern with a camera at specified exposure times and processing the images using customwritten MATLAB codes, relative blood flow information can be acquired after the formulation of speckle flow maps as previously described [49]. Automated computation of speckle flow maps was performed based on images obtained from medium-exposure (100 ms) LSI. These maps were then used to calculate the relative blood flow in a vessel in our region of interest. Automated computation of Functional Vascular Density (FVD) was performed based on images obtained from long-exposure (1000 ms) LSI. FVD is defined as the total length of perfused vessels within a region of interest in a two-dimensional image, divided by the area of that region [50]. It is a metric frequently used to assess changes in blood vessel density, and its measurements are independent of changes in vessel caliber often described as vasodilation and vasoconstriction [51]. FVD was employed to quantify the neovascularization and vessel density. This technique described by White et al. [47], relies on the presence of uninterrupted blood flow to create detailed maps of the subdermal vasculature. Using this technique, FVD can be computed several times faster than calculating the same index manually, yet with comparable accuracy. The FVD code [47] is available online at http://choi.bli.uci.edu/software.html.

Multispectral imaging
Multi spectral imaging is a technique that involves the acquisition of absorption contrast images at multiple wavelengths as described in detail by Moy et al. [45]. The images acquired differ based on the optical properties of the blood within the vessels. The instrumentation in multi spectral imaging is also simple, consisting of a broadband light source, camera, and filter sets for the desired wavelengths. This method is utilized to calculate arteriolar hemoglobin oxygenation saturation [52] in a given vessel and changes in the oxygen saturation in the same vessel are monitored over a two week period.

Histology
At the end of each experiment, the implanted biomaterial was extracted and the underlying subcutaneous tissue was harvested for histological analysis. General tissue characteristics and connective tissue deposition were analyzed using hematoxylin and eosin (H&E) and Masson's trichrome, respectively.

Statistics
Average values are expressed as mean ± standard error of the mean. Changes in the VD, FVD, relative blood flow, and HbOS values were analyzed statistically using two-way analysis of variance (ANOVA) with repeated measures; p-values < 0.05 were considered significant.

Connective tissue deposition
Connective tissue deposition was observed to increase steadily from the day of surgery (day 0) to day 14 in all study groups ( Fig. 2A-D). This phenomenon was better observed in the alginate microcapsule group (Fig. 2D). The blank islet sheet group was noted to have the least connective tissue deposition and any noticeable tissue deposition was restricted to the interstitium surrounding, but not posterior to the sheet implants. Connective tissue deposition was greatest in the group where porcine islet-containing alginate microcapsules were transplanted into the window (Fig. 2D); greater than that noted in the islet sheet implant group (Fig. 2B). Histological analysis confirmed our findings; maximum collagen deposition was noted in the porcine islet microcapsule group (Fig. 4M) compared to the other groups (Fig. 4I, J, K, L). Window chambers implanted with alginate sheets could be imaged for a longer duration without artifacts (13.5 ± 0.5 days) compared to the alginate microcapsule group (10.25 ± 1.03 days), islet microcapsule group (10.11 ± 0.3 days), islet sheet group (12.00 ± 0.82 days) or the control groups (11.00 ± 0.75 days); p = NS. We did not experience any difficulties in imaging the negative controls for up to two weeks post implantation.

Vascular changes
By Day 7, the blood vessels surrounding the implants were dilated ( Fig. 2A-D) and there was a significant increase in vessel diameter (VD) in all groups (Fig. 3A) with a significantly greater percentage rise in arteriolar diameter compared to venous. Neovascularization was also noted ( Fig. 2A-D) and was greater in the microcapsule group (Fig.  2D). All the changes noted above steadily increased in magnitude up until Day 14 ( Fig. 2A-D).
These results were corroborated by functional vascular density (FVD) measurements on Day 7 (Fig. 3B). Although the increase in FVD was greater with the blank alginate microcapsule group (4.97 ± 0.16 mm −1 ) than in the blank islet sheet group (4.53 ± 0.04 mm −1 ), the difference was not statistically significant (p = 0.84).

Relative blood flow
By analyzing speckle flow maps computed from data gathered using LSI [53], relative blood flow was calculated as the ratio of the blood flow in an implant adjacent vessel (V I ) to that in a vessel not in the immediate vicinity of the implant (V C ). It was possible to monitor the changes in the subdermal vascular hemodynamics after biomaterial device implantation by studying changes in the relative blood flow over time. By Day 14, a significantly greater increase in the relative blood flow (Fig. 3C) was evident in the islet containing groups (1.62 ± 0.2, sheet; 1.49 ± 0.9, microcapsule; p = 0.04) compared to the blank (0.96 ± 0.04, sheet; 0.97 ± 0.07, microcapsule; p = NS) and the control groups (0.97 ± 0.1).

Islet function and viability
Preliminary data (Fig. 4C, D, G, H) suggest that islets encapsulated in alginate sheets and microcapsules remain viable (65.44 ± 2.77%, sheet; 68.93 ± 5.6%, microcapsule) and function in vivo (SI = 1.91 ± 0.08, sheet; 1.85 ± 0.06, microcapsule) within the subcutaneous region up to Day 14 post transplantation. No significant drop in viability or function was noted in either the islet sheet or the UP LVM microcapsule groups (Table 1).

Discussion
As early as 1992, Menger et al. [54] demonstrated the utility of the dorsal window chamber model in monitoring subdermal microvascular changes after pancreatic islet transplantation using intravital fluorescence microscopy. Subsequently, several studies by the same group of researchers have used this model to study subcutaneous islet transplantation [36]. However, the model has not been used extensively in the study of biomaterial biocompatibility [55] despite several studies emphasizing the role of in vivo monitoring in the assessment of biomaterial efficacy and transplant outcomes [56,57]. Unlike the intraperitoneal route, transplantation under the skin is minimally invasive, is followed by a shorter recovery period, and has few to no post-surgical complications. Several studies involving the transplantation of encapsulated islets into a subcutaneous site (See Supplementary Table 2B) have demonstrated good results ranging from adequate revascularization and optimal islet function to achievement of long-term nor-moglycemia [11,58]. The surgical and imaging procedures described are a noninvasive, safe way of achieving access to the transplant site without disturbing the implanted device or microvasculature [52].
From our data ( Fig. 2A-L), it is evident that the host response varies depending on the biomaterial device being studied. The significant collagen deposition observed in the alginate microcapsule group (Fig. 4L, M) is expected; previous studies have demonstrated antigen shedding by encapsulated islets into the transplant site [59,60] resulting in a robust host immune response. Nevertheless, when these implants were extracted and analyzed for function (insulin secretion in response to a glucose challenge) ( Table 1) or viability (Fig.  4A-H), they were found to be both functional and viable. We had previously demonstrated this with human islets [61]. Although this is contrary to what would be expected, a previous study has demonstrated that alginate encapsulated islets survive and function after subcutaneous transplantation into immune competent primates despite no immunosuppression [11]. Studies performed in our lab on diabetic athymic nude mice have demonstrated a reversal of STZ (Streptozotocin) induced diabetes for a period of up to 90 days after intraperitoneal transplantation of UP LVM encapsulated porcine islets (data not shown). Compared to planar encapsulation devices like the islet sheet, alginate microcapsules ( Fig. 1E) are mechanically more stable and have a better surface area to volume ratio and immunologic profile [62,63]. Their wall thickness and pore size can be manipulated according to research requirements. Consequently, they remain the most commonly employed biomaterial in islet encapsulation [63]. They can be manufactured using commercially developed air-jet driven droplet technology [64], have been safely used in numerous small and large animal trials [65,66] and can be used safely in clinical trials on human recipients [67]. The islet sheet has been used in islet research since 2001, and we have demonstrated that human islets encapsulated within these devices remain viable and secrete insulin in response to a glucose challenge in vitro [61]. The same study demonstrated function after subcutaneous transplantation. Another study in a pancreatectomized dog demonstrated fasting normoglycemia for 84 days post transplantation after sheets containing encapsulated allogeneic islets were implanted in the omentum [44]. Using our imaging system, we were able to longitudinally and noninvasively monitor the vascular microenvironment surrounding the implant and measure numerous key hemodynamic parameters. The dorsal window model provides excellent real-time visualization of the morphological changes in the subdermal vasculature. Changes in the vascular density, blood flow, and arteriolar oxygen saturation can also be analyzed noninvasively without having to inject fluorescent probes, currently the only method available to image islets in vivo [68]. Changes in vessel diameter were greater in the islet containing groups than in the controls (Fig. 3A). We hypothesize that this vasodilatation was not a response to the environment within the window; rather, a result of the increased metabolic demand from the islets implanted at the site. Unlike Beger and Menger [69], an increase in FVD was noted with islet containing sheet and microcapsule implants (Fig. 3B). We believe that this suggests that alginate sheets and microcapsules trigger neovascularization surrounding the implant. Our data on peri-implant blood flow and arteriolar HbOS clearly show greater flow with islet implants than the blank groups, which would suggest that the greater nutrient and oxygen demand in the islet-containing groups is being met. The host vasculature is able to respond favorably to the metabolic demands of the implanted tissue. This can also be monitored as an indirect measure of islet health -a hypothesis that was confirmed after the implants were extracted and analyzed in vitro on Day 14 (Table 1).
Although several studies have now demonstrated that alginate microcapsules are an effective immunoisolation device in islet xenotransplantation, the technology has produced inconsistent results owing to variabilities in the size of the microcapsules [70], alginate composition [71] and purity [72]. The use of Ca 2+ or Ba 2+ cross-linking also affects the results obtained using this technology. If the chemical composition of the alginate is incorrect or if impurities are not excluded during manufacture, macrophages have been noted to swarm around the capsules [73]. Along with antigen-shedding, these factors can trigger fibroblast activation leading to collagen deposition, implant fibrosis and graft death. The impact of encapsulation on islet oxygenation is also a controversial issue in the field of islet transplantation. Alginate-encapsulated porcine islets have demonstrated oxygen consumption rates (OCR) comparable to that measured in free islets [74] and islets encapsulated in alginate microcapsules demonstrate pO 2 levels comparable to that found in pancreatic tissue (5% O 2 ) post-implantation in non-diabetic recipients [57]. Some studies claim hypoxia secondary to encapsulation adversely affects islet viability and function [75,76]. However, in others, encapsulated islets have demonstrated the ability to cure STZdiabetic recipients and maintain euglycemia for several months [11,77] without the need for immunosuppression, despite significant graft hypoxia being evident on tissue examination. Not all results have been so encouraging. One in vivo study reported that islets encapsulated and transplanted into syngeneic recipients showed large areas of central necrosis as early as two weeks after transplantation [78]. The authors of the study concluded that encapsulation was probably preventing rapid vascularization, resulting in chronic hypoxic stress and affecting insulin release and function. This is plausible, as whenever encapsulated islets are used, a greater islet mass is required to achieve cure than when non-encapsulated islets are implanted [79]. Until further studies address this issue, hypoxia remains a concern with alginate encapsulation.
Using LSI and WiFI we are able to monitor subdermal blood vessel flow and oxygen saturation over larger spatial scales than is possible using optical microscopy. Other advantages include ease of implementation, inexpensive equipment, ease in image acquisition and a simple setup requiring uncomplicated technical equipment [80]. Thus, the dorsal window chamber is a versatile tool to aid in the evaluation of bioartificial devices, drugs, cell types, and even stem cells in in vivo conditions. Several imaging procedures can be performed in quick succession in a short period of time.
Using multispectral imaging, spatial mapping of hemoglobin saturation at the microvascular level ( Fig. 2I-L), is possible. Currently, polarographic microelectrodes are the gold standard in the measurement of tissue pO 2 [33]; however, they are cumber-some, and do not provide enough data to generate a spatial map of oxygen delivery. While phosphorescence lifetime imaging (PLI) can help noninvasively measure tissue oxygenation, some researchers have reported that extravascular accumulation of dye over days impedes the ability to resolve the signal from only the microvessels [81]. Vessel visualization with LSI relies on particle motion; vessels with and without blood flow can be distinguished (See Supplementary Table  2A). This advantage is not afforded by conventional imaging modalities. Since longexposure LSI can identify functional vessels of all sizes, this technique is more widely applicable than comparable video techniques capable of imaging only capillaries [82]. LSI also affords superior temporal and spatial resolution when compared to MRI and micro-CT and a much lower risk of exposure to potentially harmful ionizing radiation (unlike micro-CT).
Using LSI and multi spectral imaging, only the superficial sub-epidermal and dermal microvasculature can be analyzed accurately (See Supplementary Table 2A). This restricts the imaging techniques to a few millimeters from the skin surface [83]. The ability to quantify results using LSI is limited, as the dynamic interaction between light scattering and the host tissue has been over-simplified, resulting in inaccuracies.
Another limitation of our study is the absence of in vivo islet function analysis (assaying Cpeptide levels in diabetic mice transplanted with islets inside the window). This would help demonstrate that our islets and devices are able to function in vivo after transplantation into our model. Since chronic hyperglycemia can lead to microvascular dysregulation [54] and an alteration in the immune response, diabetic animals may not behave in the same manner as the non-diabetic animals used in our study. Another limitation of our study is the difficulty in continuing studies past 14 days. Studies by other researchers using the window model [38,49] demonstrate that under optimal conditions, imaging may be performed for up to 28 days. Studies monitoring implanted bio-engineered devices for biocompatibility analysis have demonstrated that accurate biomaterial evaluation is possible if the device is studied in vivo between 5 and 28 days [38]. Hence, we believe that our model can help analyze biomaterial implants to better understand their effect on host tissues [84].

Conclusion
It is evident from this study that islet encapsulated alginate sheets and microcapsules trigger a qualitatively and quantitatively robust vascular response in the host within the first two weeks after transplantation. We were able to prove this assertion by demonstrating a greater increase in VD, FVD, relative blood flow, and arteriolar HbOS when compared to the control groups (Fig. 3). As encapsulated islets have been reported to be vulnerable to hypoxic injury, we hypothesize that the robust response observed in the subcutaneous milieu could improve islet survival by attenuating islet hypoxia during the first few weeks after transplantation. Our data on islet viability and function post-explant (Table 1) validates this conclusion. In the near future, we expect to conduct studies to characterize the host immune response to subcutaneously transplanted alginate implants and compare alginate devices of varying composition, purity, surface properties and shape to determine the optimal characteristics required for greater success in encapsulated islet transplantation.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. (respectively) that were used in various experiments. We acknowledge our students Nhat Truong, Maria Morkos and Risha Shukla for helping with manuscript editing before submission. Finally, the authors gratefully acknowledge financial and institutional support from Department of Surgery, University of California Irvine; the Hanuman Medical Foundation, San Francisco; the Arnold and Mabel Beckman Foundation; and the National Institutes of Health (P41 EB015890 and R01 HD065536).    Graphs comparing changes in various hemodynamic parameters noted with alginate microcapsule and sheet implants over a 14 day period. Vessel diameter (A), functional vascular density (B), relative blood flow (C) and arteriolar hemoglobin oxygen saturation (D). Four different implant groups were tested -porcine islets encapsulated within an alginate sheet, a blank alginate sheet, porcine islets encapsulated within alginate microcapsules, blank alginate microcapsules, and no implant (negative control). Islet viability and histological analysis. Islet viability analysis using Newport Green/ Propidium Iodide staining is performed on islets before alginate encapsulation (A, E) and shortly after encapsulation within alginate microcapsules (B) and alginate sheets (E) on the day of surgery (Day 0). Viability analysis is also carried out on days 7 (C -microcapsule, G -sheet) and 14 (D -microcapsule, H -sheet). The images represent a projection of 60 optical sections acquired at 4 μm intervals by confocal laser-scanning microscopy. The green-fluorescent Newport Green ™ DCF indicator identifies viable insulin-producing β-cells via binding to intracellular zinc. Histological analysis of the subcutaneous tissue surrounding the implant, performed after day 14 while staining with Masson's trichrome No implant (I), Blank alginate sheet (J), alginate sheet containing porcine islets (K), blank alginate capsules (L) and alginate capsules containing porcine islets (M). 4× Magnification. The white arrows denote collagen deposits that are stained blue. The black arrow denotes the location of the alginate capsule implants. The quantity of insulin (μg/L) secreted by encapsulated porcine islets shortly before implantation (Day 0, n = 3) and post-explant (day 7, n = 3) in response to an in vitro glucose stimulated insulin secretion assay (GSIS) is tabulated above. Stimulation indices for encapsulated islets before implantation and post-explant can be compared with those for unencapsulated islets cultured in vitro. In vitro viability assays of the encapsulated porcine islets were performed on Day 0 (before implantation, n = 3) and Day 7 (post-explant, n = 3) using laser-scanning confocal microscopy after staining with Newport Green (a fluorophore that selectively stains viable beta cells) and Propidium Iodide (an intercalating agent that only stains dead cells; see Fig. 2C-H).