Genetically engineered cells have the potential to solve the cell availability problem in developing a pancreatic tissue substitute for the treatment of insulin-dependent diabetes (IDD). These cells can be beta-cells genetically engineered so that they can be grown in culture, such as the betaTC3 and betaTC tet mouse insulinomas developed by Efrat et al; or, they can be non-beta cells genetically engineered to secrete insulin constitutively or under transcriptional regulation. The aim of this work was to thoroughly characterize and improve the secretion dynamics of pancreatic substitutes based on genetically engineered cells. One issue involved with the continuous beta-cell lines is the remodeling of the cells inside an encapsulated cell system, which may affect the insulin secretion dynamics exhibited by the construct. To evaluate the effect of remodeling on the secretion properties of the construct, we used a single-pass perfusion system to characterize the insulin secretion dynamics of different alginate beads in response to step-ups and downs in glucose concentration. Results indicated that the secretion dynamics of beads indeed changed after long-term culture. On the other hand, data with a growth-regulated cell line, betaTC tet cells, showed that the secretion profile of beads can be retained if the cell growth is suppressed. A major concern associated with genetically engineered cells of non-beta origin is that they generally exhibit sub-optimal insulin secretion characteristics relative to normal pancreatic islets. Instead of relying on molecular tools such as manipulating gene elements, our approach was to introduce a glucose-responsive material acting as a control barrier for insulin release from a device containing constitutively secreting cells. Proof-of-concept experiments were performed with a disk-shaped prototype based on recombinant HepG2 hepatomas or C2C12 myoblasts, which constitutively secreted insulin, and concanavalin A (con A)-based glucose-responsive material as the control barrier. Results demonstrated that the a hybrid pancreatic substitute consisting of constitutively secreting cells and glucose-responsive material has the potential to provide a more physiologic regulation of insulin release than the cells by themselves or in an inert material.

An implantable closed-loop insulin delivery device (INsmart device) containing a glucose responsive gel has been developed within the INsmart research group, over a period of 10 years, to mimic pancreas. In this thesis, the reliability and performance capability of the INsmart device was studied for future clinical use. Investigations into the device material compatibility with insulin solution, assessed by monitoring insulin loss and degradant formation over a period of 31 days using RP-HPLC have shown that stainless steel and titanium are the most compatible materials. Polycarbonate contributes to insulin loss after 11 days, resin might not be the best material and polyurethane is the least compatible for future device designs. To study insulin delivery mechanism and kinetics from the device, fluorescently labelled human insulin (FITC-insulin) was synthesised and characterised using RP-HPLC and MS, to produce a product with predominantly di-labelled conjugate (>75%) with no unreacted FITC or native insulin. Clinically used insulin analogues were also fluorescently labelled to produce predominantly di-labelled FITC-insulin conjugate with potential future biological and in vitro applications. The drug release mechanism from the glucose sensitive gel held in the INsmart device, studied using fluorescein sodium was determined as a Fickian diffusion controlled release mechanism. The diffusion coefficient (D) for FITC-insulin in the non-polymerised dex2M-conA gel (NP gel) determined using mathematical models, QSS and TL slope methods was 1.05 ± 0.02 x 10-11 m2/s and in the cross-linked dex500MA-conAMA gel (CL gel) was 0.75 ± 0.06 x 10-11 m2/s. In response to physiologically relevant glucose triggers in the NP gel, the diffusivity of FITC-insulin increases with increasing glucose concentrations, showing a second order polynomial fit, device thus showing glucose sensitivity and graded response, mimicking pancreas. Rheological measurements further confirmed the gel glucose responsiveness demonstrated by a third order polynomial fit between FITC-insulin D and the NP complex viscosity in response to increasing glucose concentration. The knowledge of FITC-insulin diffusion kinetics in the gel has aided in making some theoretical predictions for the capability and performance of the INsmart device. Alternate device geometry and design optimisation is also explored.