Cell encapsulation is a potential approach to the transplantation of allo- or xenoislets without rejection or autoimmune complications (1, 2). Isolated islets are sequestered inside alginate-based capsules whose walls have pores large enough for nutrients and hormone to diffuse, but small enough to prevent entry of the host immune system. Capsules in this experiment are fabricated using alginate, a crude product polysaccharide comprised of mannuronic and guluronic acid residues that form a hydrogel when cross-linked with divalent cations.
Several variables must be controlled during the fabrication of capsules such as mechanical strength, pore size, and biocompatibility. Particularly nonbiocompatible capsules can elicit fibrosis on the capsule wall capable of causing islet necrosis (3–5). Empty capsules have been shown to elicit immune responses (6–8). Improper purification of the alginate can result in the presence of lipopolysaccharides, which have been shown to elicit host immune reactions (3, 8). Common leaching and swelling of implant materials can result in the slow diffusion of islet debris from the capsules. In addition to the standard foreign body reaction to implant materials, capsules can elicit specific humoral responses if foreign islet cells are exposed by either capsule wall breakage, or if islet cells are caught in the capsule wall during fabrication.
Current procedures for acquiring biocompatibility measurements of encapsulated islets include fabrication and implantation of capsules into the peritoneal cavity of an animal model, incubation for specified periods of time, and retrieval via peritoneal lavage. Labs create quantitative scales dependant on the number of cells or cell layers adhered to the capsule walls to score biocompatibility (3–10). These approaches, which allow only one data point to be acquired per animal, make it difficult to determine a time line of immune response unless many animals are used. Using many mice allows for more error associated with animal variability.
To circumvent these problems, we propose an alternate method of monitoring the inflammatory response. In vivo bioluminescence imaging allows measurements of immune response to be acquired real time. For our studies we used transgenic mice expressing the proximal 5′ human immunodeficiency virus long terminal repeat (referred to as HIV-LTR/Luciferase, HLL mice), a known NF-kB dependent promoter sequence. NF-kB is a transcription factor that plays a critical upstream role in coordinating the inflammatory and wound healing cascades. It initiates the transcription of many cytokines, chemokines, adhesion molecules, and proinflammatory genes (11–14). In this transgenic mouse model, when NF-kB binds to its promoter sequence to initiate transcription of immune response proteins, luciferase enzyme is also produced. In conjunction with ATP, oxygen, and its substrate luciferin, luciferase catalyzes a bioluminescent reaction emitting photons at approximately 560 nm which are detectable by a sensitive charge coupled device camera (11, 15–18). The amount of light (photons/sec/cm2/sr) detected by the camera is therefore indicative of NF-kB activity, which is highly involved in regulating inflammation.
These techniques allow the real time assessment of biocompatibility. Acquiring information from the same mouse repeatedly to construct a time line of immune response allows each new data point an accurate reference to the last, instead of relying on data points acquired from different animals. This is in part made possible because each mouse can be used as its own control. The requirement for fewer animals reduces the reliance on statistical analysis and increases the importance of real time data. Collected data can reveal important information about the immune response time course.
MATERIALS AND METHODS
Islets were isolated from canines using a collagenase digestion method. Islets were suspended in an alginate and cellulose sulfate solution (1% each in concentration). After release from an air-stripping drop generator, the polyanion drops were washed in a loop reactor containing a continuous flow of calcium chloride solution (1% in concentration). The length of the loops dictates reaction time. The capsules were then washed in a poly(methylene-co-guanidine) (PMCG) solution (1% in concentration). The length of this polyelectrolyte complexation step dictates pore size. The capsules were coated in alginate solution (1% in concentration) for 2 min and gelled in calcium chloride for 1 min.
Three types of capsules are assessed in this experiment. One transplant group is the standard capsules, 0.8 mm in diameter. The second group is non-coated capsules, also 0.8 mm in diameter, and are lacking the final alginate coating and gelling steps. The third group is beads. Beads are larger capsules, 1.4 mm in diameter, which contain an average of 2.5 capsules within. They are fabricated by replacing the already fabricated capsules in the polyanion solution and into the drop generator. Each bead is air-stripped in the same manner and gelling occurs in the calcium chloride solution. Table 1 summarizes all transplant groups, fabrication, and rational for transplantation. All finalized implant materials were normalized to capsule surface area prior to transplantation, allowing 18 capsules and 6 beads per mouse. Each capsule and bead was inspected to ensure they contained approximately the same quantity of viable islets (by site), and that no cellular debris was present inside the capsule walls.
Capsules or beads were transplanted into the dorsal-cervical fat pad of HLL mice approximately 7 weeks of age, weighing an average 20–25 grams. The fat pad was chosen for several reasons. Customary transplantation into the peritoneal cavity does not allow for even distribution. Implantation into the fat pad reduces the selection bias associated with capsule retrieval. When capsules are transplanted into epididymal fat pads, capsules stay localized. Additionally, experiments have shown that pericapsular reactions are uniform throughout the pad (5, 20).
Mice were anesthetized via intraperitoneal injection of a 5 mg/ml solution of nembutal in sterile saline solution. Dosage was dependent on body weight, .01 ml/g. Once the mice were anesthetized, they were shaved over the area of implantation, and the operative site covered with Saran wrap to help prevent infection. An incision, approximately 5 mm in length, was made in the wrap, then the skin, and then the membrane underneath that separates the skin from the adipose tissue. Separations in the fat were probed with surgical tweezer tips, and a pocket, approximately 10 mm3 depending on mouse size, was made into which the capsules were transplanted.
Aliquots of capsules or beads were siphoned into a pipette with sterile saline and allowed to gather at the bottom of the pipette tip just prior to insertion. Once the capsules were transferred, a single suture was tied in the adipose to enclose the implant materials in the pocket. Another suture was made joining the incised membrane, and several sutures (as required) were placed to join the incised epidermis. Each mouse was wrapped in gauze for warmth while the effects of the nembutal dissipated.
Anesthetized mice, between 7 and 45 days after transplantation, were immobilized for the 3 min of integration time of photon counting. A 15 mg/ml D-luciferin potassium salt (Biosynth International Inc., Naperville, IL) solution suspended in sterile, distilled water was administered by intraperitoneal injection. Mice were anesthetized as previously described in Surgical Procedures; the amount of luciferin given was derived using the same dosage regimen as the nembutal. Mice were imaged dorsally using the XENOGEN IVIS 100 Imaging System and charge coupled device camera (Spectral Instruments, Tucson Arizona). Photon counts were acquired using Living Image 2.2 software. The settings were on high resolution (binning of 4), field of view 25, flat fielded, cosmic, and background subtracted.
Maximum bioluminescence occurred at different times for each mouse during the weeks of image acquisition. In addition, maximum bioluminescence occurred at different times for the same mouse on different days. Taking this into account, images were acquired using consecutive 3-minute exposures for approximately 30 min. The exposure that shows maximum bioluminescence over the dorsal-cervical fat pad was used for that imaging sessions data point.
Luciferase activity in the dorsal-cervical fat pad was measured by standard luciferase assay. Buffers were purchased from Promega (Madison, WI). Luciferase activity was measured in Relative Light Units using a Monolight 3010 luminometer (BD Pharmingen, San Diego, CA). Results were normalized per gram tissue.
H & E Stain
To visualize cell morphology surrounding transplanted capsules, Hematoxylin and Eosin stains were performed. Each excised tissue sample containing transplanted capsules was suspended in a formulin fixative, provided by the Pathology Core, Vanderbilt University. Once suspended in the formulin fixative, the samples were given to the Pathology Core, Vanderbilt University, for embedding in paraffin blocks. Slides containing cross-sections of the polymer-based capsules were H and E stained to visualize cell morphology.
Results are expressed as a mean ± SEM. Statistical analysis was made with the Mann-Whitney U test. A P value of <0.05 was considered statistically significant. Comparisons were made using the Sigma Stat. software.
Each mouse used in this experiment was first shaved over the dorsal cervical area and imaged five times to establish basal NF-kB activity. The mice were then transplanted with one of the three capsule groups, or subjected to a sham surgery, or not subjected to surgery at all to monitor the effects of repeated nembutal and luciferin injections over time. Postsurgery, mice were allowed to recover for one week before sutures were removed. Each mouse was then imaged every 2–3 days until 45 days postsurgery.
Images were acquired using consecutive three-minute exposures for approximately 30 min. A sample bioluminescent photograph is shown below (Fig. 1).
Maximum bioluminescence occurred at different times for each mouse over the 30 min imaging session, and at different times for the same mouse on different days. The maximum integrated intensity is used because it is the most indicative of the total amount of produced luciferase enzyme present at the site of implantation. The maximum point is theoretically consistent over multiple image acquisition sessions because care was taken to ensure that the limiting reagent in the bioluminescent reaction was the produced luciferase enzyme and not the injected luciferin.
To determine statistically significant differences between transplant groups per imaging session, and to compare total bioluminescent activity between transplant groups over a period of six weeks, normalization procedures were required. Between three and five baseline measurements of the dorsal-cervical fat pad were obtained before surgical procedures for each mouse. These measurements were averaged to represent basal NF-kB activity for each fat pad. This average was used as a base for which every subsequent postsurgical measurement could be inferred as a percent increase over basal NF-kB activity.
Four sessions of transplantations were conducted for this experiment, resulting in a total of 36 mice with capsule transplants that were monitored between 7 and 45 days posttransplantation. To compare total bioluminescent activity between transplant groups of all four sessions, all the data was normalized to the same time duration. This requires an assumption that all transplant groups behave similarly in time. Compiled data suggests this is a reasonable assumption as other transplant groups behave similarly to the capsule group (Fig. 2).
The four sessions of transplantation presented additional complications. The animals of different sessions were from different litters. The surgical procedure proficiency improved in time. These factors resulted in different responses in each session. For instance, two of the four transplantations show maximum NF-kB activity approximately 250% above basal levels for all transplant groups, and the remaining two show maximum NF-kB activity approximately 450% above basal levels. Therefore, transplant group data was normalized within each transplantation session since mice were transplanted with standard capsules in all four transplantations. Second, as described in the Materials and Methods section, capsules are fabricated using techniques that allow for very high reproducibility (19, 21). It is for these reasons that every transplant session was normalized to the standard capsule results in their transplantation. Normalized results from each transplantation were then averaged (Fig. 3).
It can be inferred from Figure 3 that there is little difference in NF-kB activity over a six-week period between the capsule, no coat, and sham groups. Activity for the no coat and sham group was suppressed approximately 5% below capsule NF-kB activity. However, NF-kB activity resulting from bead transplantation was elevated approximately 31% over capsule activity, statistically significant relative to the capsule transplant group. Control group activity was approximately 83% below capsule activity.
The normalized time line of NF-kB activity for the capsule group is presented in Figure 2 (other transplant group data not shown). Statistical analyses reveal no significant differences between transplant groups per imaging session. The only point at which there is a statistically significant difference between the medians of transplant groups is between the bead and sham groups, 8 days postsurgery (data not shown).
To ensure that data obtained from imaging is indicative of varying quantities of luciferase, a luciferase assay was performed on dorsal-cervical adipose tissue from three mice, one transplanted with beads, another with capsules, and the last with non-coated capsules. The following graphs display un-normalized integrated intensities (photon count/3 min) for each mouse and corresponding luciferase assay results (relative light units; Figs. 4–5). The no coat mouse exhibited approximately a quarter to a sixth the integrated intensity of the capsule and bead mice, respectively. Accordingly, the luciferase assay data obtained shows the no coat mouse tissue sample contained approximately half to three fifths the luciferase quantity per gram tissue of the capsule and bead mice, respectively. All three transplant groups showed significantly greater amounts of luciferase than the wild type mouse.
In addition to the luciferase assay, retrieved capsules still embedded in the dorsal-cervical fat pad from the same three mice were extricated and embedded for H and E staining. Photographs of each cross section show a markedly higher congregation of leukocytes, lymphocytes, and some fibroblasts surrounding the bead and capsule walls than the noncoated walls (data not shown). It is important to note that these three mice are not necessarily representative of their perspective transplant group’s data, but rather they verify the correlation between bioluminescent intensities acquired from the camera and mouse histology.
Imaging showed statistically significant differences between pre- and postoperative data points per mouse, as well as differences in NF-kB activity between transplant groups over a six-week time period. This indicates that the imaging modality can be a useful methodology for monitoring immune response real time and in situ using the NF-kB transcription factor as a bioluminescent marker.
NF-kB activity in the sham group shows an initial peak shortly after surgery, and then a slow decrease in NF-kB activity over several weeks back to basal levels. The bead, capsule, and no coat groups show similar responses, indicating that they, too, experienced initial increases in NF-kB activity due to surgical procedures. Both the no coat and sham groups show only a 5% decrease in NF-kB activity compared to the capsule group over six weeks. Because the normalized area under the curve for the capsule group is close to the sham surgery group, it suggests that the trauma of surgery is more significant than host immune responses to the capsules.
The same conclusion can be drawn about the no coat group. This is particularly interesting because even those capsules fabricated for use as a positive control demonstrate a degree of immune response still masked by the traumatic effects of surgical procedures. Within the variables modified during capsule production, capsule wall composition has negligible effects on host NF-kB activity compared to our surgical procedures.
The bead group shows a 31% increase in NF-kB activity over the capsule group and the sham group, indicating that the size parameter effects NF-kB activity in excess of the effects of surgical procedures over a 6-week period. These results support findings by Robitaille et al. who found that larger capsules elicit a greater immune response when normalized to surface area, although the capsules studied were smaller in scale than those used in this experiment (5). Several experiments have been conducted to determine the effects of capsule size on water activity, mechanical strength, capsule fabrication limitations, transport behavior, porosity, in vitro insulin kinetics, and TNF-α permeability, but not in situ biocompatibility studies directly (22–24).
Although the exact mechanisms for the apparent effect of capsule size on biocompatibility are not clear, there are several possibilities. The transplanted beads occupy a larger total volume than the capsules. Larger volumes may provoke greater stress on neighboring cells, and therefore stronger immune responses. It is also possible that larger implant volumes have no more an effect on inflammation than do the standard capsules, but rather a more negative effect on wound healing. The sham surgery group data indicates that wound healing does affect NF-kB activity.
In conclusion, this imaging modality could prove to be a practical methodology for monitoring immune response using the NF-kB transcription factor as a bioluminescent marker. The results obtained from this imaging modality allow the following conclusions to be made about its efficacy. It is sensitive enough to significantly differentiate between preand postoperative data points per mouse for all transplant groups (Fig. 2 and data not shown). It is capable of determining differences between transplant groups over extended periods of time. We found that the surgery itself initiates enough NF-kB activity to mask the effects of varying capsule wall compositions on biocompatibility, if any. Only when comparing capsule size could elevated NF-kB activity greater than surgical trauma be noted. Neither the capsule nor the no coat groups were able to provoke a NF-kB activity greater than that elicited by surgical procedures. Only beads with a greater diameter than the standard capsule and no coat groups were able to elicit host NF-kB activity in excess of those triggered by surgical procedures.
We would like to thank Dr. Timothy Blackwell and Dr. Fiona Yull for providing resources and valuable scientific expertise. We would also like to thank Irina Trenary for all chemical preparation and capsule fabrication contributions. We would like to thank the Vanderbilt Pathology Core for their services.
1. Wang TLR. Bioartificial Pancreas. In: Principles of Tissue Engineering, L.R. Lanza R, Vacant J, Editor Academic Press: San Diego: 495–507.
2. Lim FSA. Microencapsulated Islets as Bioartificial Endocrine Pancreas. Science
1980; 210: 908–910.
3. Strand BRL, Veld P, Kulseng B, et al. Poly-L-Lysine Induces Fibrosis on Alginate Microcapsules via the Induction of Cytokines. Cell Transplantation
2001; 10: 263–275.
4. Wijsman JAP, Mazaheri R, Garcia B, et al. Histological and Immunopathological Analysis of Recovered Encapsulated Allogenic Islets From Transplanted Diabetic BB/W Rats. Transplantation
1992; 54(4): 588–592.
5. Robitaille RPJ-F, Leblond F, Lamoureux M, et al. Studies on Small (<350um) Alginate-Poly-L-Lysine Microcapsules. III. Biocompatibility of Smaller vs. Standard Microcapsules. J Biomed Mat Res
1999; 44(1): 116–120.
6. de Vos PdHB, van Schilfgaarde R. Effect of the Alginate Composition on the Biocompatibility of Alginate-Polylysine Microcapsules. Biomaterials
1997; 18: 273–278.
7. de Vos PvHC, van Zanten J, Netter S, et al. Long-term Biocompatibility, Chemistry, and Function of Microencapsulted Pancreatic Islets. Biomaterials
2003; 24: 305–312.
8. de Vos PdHB, Wolters G, Strubbe J, van Schilfgaarde R. Improved Biocompatibility but Limited Graft Survival After Purification of Alginate for Microencapsulation of Pancreatic Islets. Diabetologia
1997; 40: 262–270.
9. Clayton H, London N, Colloby P, et al. The Effect of Capusle Composition on the Viability and Biocompatibility of Sodium Alginate/Poly-L-Lysine Encapsulated Islets. Transplant Proceed
1992; 24(3): 956.
10. Schneider SFP, Slotty V, Kampfner D,. Multilayer Capsules: A Promising Microencapsulation System for Transplantation of Pancreatic Islets. Biomaterials
2001; 22: 1961–1970.
11. Sadikot RJE, Debelak J, Yull F, et al. High-Dose Dexamethasone Accentuates Nuclear Factor-kB Activation in Endotoxin-Treated Mice. Am J Respir Crit Care Med
2001; 164: 873–878.
12. Blackwell TCJ. The Role of Nuclear Factor-kB in Cytokine Gene Regulation. Am J Respir Cell Mol Biol
1997; 17: 3–9.
13. Boone D, Lee E, Shon L, et al. Recent Advances in Understanding NF-kB Regulation. Inflam Bowel Dis
2002; 8(3): 201–212.
14. Christman JSR, Blackwell T. The Role of Nuclear Factor-kB in Pulmonary Diseases. Chest
2000; 117(5): 1482–1487.
15. Blackwell TSYF, Chen C, Venkatakrishman A, et al. Multiorgan Nuclear Factor Kappa B Activation in a Transgenic Mouse Model of Systemic Inflammation. Am J Respir Crit Care Med
2000; 162: 1095–1101.
16. Sadikot RWL, Jansen E, et al. Hepatic Cryoablation-Induced Multisystem Injury: Bioluminescent Detection of NF-kB Activation in a Transgenic Mouse Model. J Gastrointest Surg
2002; 6(2): 264–270.
17. Ugarova NBL. Protein Structure and Bioluminscent Spectra for Firefly Biolumiscence. Luminescence
2002; 17: 321–330.
18. Oba OOM, Inouye S. Firefly Luciferase is a Bifuntional Enzyme: ATP-dependent Monooxygenase and a Long Chain Fatty Acyl-CoA Synthetase. Fed Eur Biochem Soc Lett
2003; 540: 251–254.
19. Anilkumar ALI, Wang T. A Novel Reactor for Making Uniform Capsules. Biotech Bioeng
2001; 75(5): 581–589.
20. Pariseau J-FLF, Harel F, Lepage Y, Halle J-P. The Rat Epididyman Fat Pad as an Implantation Site for the Study of Microcapsule Biocompatibility: Validation of the Method. J Biomed Mat Res
1995; 29: 1331–1335.
21. Lacik IBM, Anilkumar A, Powers A, Wang T. New Capsule with Tailored Properties for the Encapsulation of Living Cells. J Biomed Mat Res
1998; 39: 52–60.
22. De Vos PDHB, van Schilfgaarde R, Wolters G. Factors Influencing the Adequacy of Microencapsulation of Rat Pancreatic Islets. Transplantation
1996; 62(7): 888–893.
23. Rosinski SGG, Lewinska D, et al. Characterization of Microcapsules: Recommended Methods Based on Round-Robin Testing. J Microencapsulation
2002; 19(5): 641–659.
24. Strand BGO, Kulseng B, Espevik T, Skjak-Braek G. Alginate-polylysine-alginate Microcapsules: Effect of Size Reduction on Capsule Properties. J Microencapsulation
2002; 19(5): 615–630.