13.11 Gene therapy in diabetes mellitus

The distinctions between what has previously been termed cell therapy and gene therapy have become blurred. Cell therapy traditionally implied the in vitro expansion of cells that could subsequently be engrafted into patients to elicit a therapeutic effect, while gene therapy was a term applied to the genetic manipulation of tissues or cells in vivo or ex vivo. With the amazing advances that have been achieved using transcription factors to reprogramme cells, this distinction, at least for regenerative medicine applications, no longer exists. In this chapter, following the statement of the unmet clinical need, we review potential sources of new β cells and approaches to β cell replacement therapy; discuss how recent advances in safety and efficacy of gene transfer technology can augment cellular therapeutic approaches, and summarize pure gene therapy approaches dependent on expression of genes encoding insulin and other glucose-lowering hormones in the recipient’s own cells. Since both type 1 and type 2 diabetes are associated with a decline in β cell mass, cell and gene therapy targeted at the β cell and insulin replacement have potential applications for both forms of the disease.

In type 1 diabetes, uninterrupted compliance with insulin injection therapy is necessary to prevent potentially fatal ketoacidosis. The landmark Diabetes Control and Complications Trial and Epidemiology of Diabetes Interventions and Complications follow-up study have confirmed that chronic hyperglycaemic microvascular and macrovascular complications can be prevented by tight glycaemic control, but this was at the expense of a threefold increase in severe hypoglycaemia—one of the greatest fears of those living with daily insulin injections. Overall, the health implications and economic costs of type 1 diabetes are massive, and increasing annually. There is, therefore, an unquestionable clinical need for new therapeutic options.

While transplantation of whole pancreas together with its blood supply can entirely normalize blood glucose levels, the major surgery required is associated with 5% mortality in the first year, even in the most experienced centres. Isolation and transplantation of purified insulin-secreting islets of Langerhans from a donor pancreas requires only minimally invasive cannulation of the portal vein transhepatically under X-ray guidance. This offers the promise of more widespread implementation restoring excellent control, preventing both long-term complications and severe hypoglycaemia. Capacity will, however, be severely limited by the scarcity of deceased donor organs: currently sufficient for fewer than 1% of those who might benefit from this form of treatment. This has provided impetus to efforts to produce a replenishable supply of glucose-responsive insulin-secreting cells that could be used in transplantation. One potential source might involve the in vitro differentiation of stem cells derived from embryonic and adult tissue.

Type 2 diabetes is marked by both a resistance of target tissue to the effects of insulin and impaired function of the β cell. The major β-cell defects relate to an impaired secretory response to glucose, altered kinetics of secretion including pulsatility, accumulation of islet amyloid polypeptide, an increase in glucagon-secreting α cells, and a decline in β-cell mass. Current therapy for type 2 diabetes involves a combination of drugs directed at improvements in both insulin sensitivity and β-cell function, together with management of associated cardiovascular risk factors. Conventional treatment modalities have not been able to prevent the inexorable progressive loss of β-cell function necessitating insulin replacement in the majority over time, but this is often insufficient to sustainably achieve target glucose levels outwith intensive clinical trials. It is envisaged that novel cell therapy approaches will enable restoration of β-cell mass.

An unlimited supply of insulin-secreting β cells would clearly provide the potential for widespread transplantation in much larger numbers of those with both type 1 and type 2 diabetes. The immune system in an individual with autoimmune type 1 diabetes is, however, uniquely efficient at targeting and destroying β cells leading to recurrent insulin deficiency. At present type 1 islet allograft transplant recipients require antibody induction and ongoing systemic immunosuppression to prevent both allo-rejection and endogenous recurrent autoimmune-mediated destruction of transplanted β cells. Ideally an alternative to lifelong immunosuppression would be induction of tolerance mediated by short-term suppression of effector T (Teff) lymphocytes and induction of tolerogenic regulatory (Treg) lymphocytes. This requires development of novel therapies for induction of such antigen-specific tolerance in parallel with approaches to replace β-cell mass. In type 2 diabetes, β-cell replacement in combination with insulin sensitizers may be sufficient to break the vicious cycle of glucose and lipid toxicity combined with inflammatory effects on the islets that makes the disease presently so difficult to treat. There are several sources of cells from which new β cells could be derived (1).

Embryonic stem cells, derived from the inner cell mass of preimplantation embryos, are pluripotent, meaning that they have the potential to differentiate into any cell type in the body. The most effective protocols used to differentiate embryonic stem cells along a pancreatic lineage are based on our understanding of the events that occur in the developing mouse embryo. There is not space in this chapter for a detailed description of the developmental biology of the pancreas, which can be found elsewhere (2). It is sufficient to explain that the pancreas first appears as dorsal and ventral outgrowths from developing foregut. It then undergoes a period of expansion during which a branching epithelial network forms. Genetic lineage tracing studies provide evidence for a pool of multipotent progenitor cell (MPCs) within the expanding pancreatic bud. These MPCs give rise to all pancreatic cells types including acinar, endocrine, and duct cells. The establishment of the various pancreatic lineages proceeds in a step-wise manner in response to signalling molecules generated from various surrounding cell types, including those present in the mesenchyme and vasculature.

Successful in vitro differentiation of human embryonic stem cells is therefore dependent on the sequential addition of cocktails of growth factors and small molecules that activate or inhibit particular signalling pathways (3). Thus activin A, acting through the nodal signalling pathway, is used to convert a monolayer of embryonic stem cells into definitive endoderm (Fig. 13.11.1). Cyclopamine is another key factor inhibiting sonic hedgehog, and, thus, promoting pancreas-specific development. Retinoic acid further promotes differentiation of the pancreatic lineage, while inhibitors of Notch signalling direct formation of endocrine cells. After 20 or so days, the resultant cells contain a high proportion of β-like cells, most of which express insulin at quantities approaching those measured in adult human β-cells. However, the cells are still not fully differentiated as they do not secrete insulin in response to glucose and many of the insulin-positive cells express more that one hormone. Nevertheless, if early pancreatic progenitors (pancreatic foregut in Fig. 13.11.1) are transplanted into mice, they mature (over a fairly lengthy period of time, i.e. 72 days) into the various endocrine cell types that become organized within islets, and acquire the capacity to secrete insulin (C-peptide) in response to glucose (6). These results are important in that they provide proof of principle that embryonic stem cells can be induced to differentiate into functional islets that can normalize blood glucose levels. However, they provide little insight into the final key signals and mechanisms involved in in vivo end-differentiation that could inform further approaches to the goal of deriving mature β cells in the clinical laboratory for transplantation. It has been considered unlikely that embryonic stem cell-derived islet progenitor cells would have therapeutic applications because of concerns related to the unpredictable outcome in terms of the size and cellular composition of the resultant islet-like structures in additional to potential tumorigenicity of transplanting incompletely differentiated cells. Nevertheless, feasibility and acceptability of phase 1 human safety trials of embryonic stem-derived cells transplanted subcutaneously within a non-degradable solid encapsulation device preventing cellular escape and enabling surgical removal are currently being explored.

Exciting recent advances in cell reprogramming have raised the possibility of generating patient-specific induced pluripotent stem (iPS) cells from skin cells from individual patients (7, 8). iPS cells were first derived from mouse fibroblasts following retroviral-mediated chromosomal integration of four key stem cell transcription factors (Oct3/4, Sox2, c-Myc, and Klf-4) (7). They have since been generated from a wide variety of mouse and human adult cell types. iPS cells are pluripotent, and share many of the properties of embryonic stem cells including similar gene expression profile, the ability to form teratomas and to contribute to all cell types of chimeric mice including the germ line. iPS cells generated from human skin fibroblasts can form islet-like clusters by following a sequential in vitro differentiation approach similar to that used for embryonic stem cells (9).

There are, however, a number of safety issues that would have to be addressed before iPS cells could be considered for cell therapy applications (10). First, there are worries related to the risk of cancer when transplanting genetically modified cells in humans, although these problems are currently being addressed and to date iPS cells have been generated without the use of the oncogene c-Myc or viral integration into the genome. Second, and more difficult to address, is the possibility that iPS cells may accumulate pro-oncogenic genetic mutations that provide a selective advantage in the culture conditions used for reprogramming and differentiation. Finally, there is the problem of teratoma formation through transplantation of cells that are not fully differentiated. This is a potential problem with both embryonic stem- and iPS-derived cell types and will require imaginative approaches to eliminate entirely contaminating undifferentiated cells. Given the careful balance to be struck between benefit and risk in comparison to conventional therapy, any possibility of teratoma formation would largely preclude clinical implementation of pancreatic progenitors for the treatment of diabetes.

Attempts to expand β-cell mass within intact isolated islets have been largely unsuccessful in these differentiated and highly structured organelles. Isolated rodent β cells have a limited ability to proliferate in culture as measured by incorporation of BrdU into newly synthesized DNA, while under similar conditions isolated human β cells appear unable to proliferate (11). Human islets (as opposed to isolated β cells) on the other hand can be expanded through several population doublings in two-dimensional culture when plated on culture dishes coated with extracellular matrix in the presence of growth factors such as epidermal growth factor (EGF) or hepatocyte growth factor (HGF), but they very rapidly lose the ability to express insulin and other differentiated phenotypic markers. Interestingly, when adult human islets are plated on dishes in the absence of matrix and growth factors a population of fibroblast-like cells can be expanded up to 10¹² fold. When serum is removed from the media (certainly at relatively early passage), the cells aggregate to form islet-like clusters that express insulin and other β cell markers (12). It was originally thought that this represented a reversible process involving an epithelial–mesenchymal transition (EMT), whereby the β cells dedifferentiated into mesenchymal cells with an enhanced capacity for expansion in culture but maintained potential to be redifferentiated into functional β cells. Subsequent lineage tracing studies in mice have clearly shown that the mesenchymal-like cells do not originate from β cells at least in murine cells (13). The origin of these cells in human islet cultures remains unknown. It is possible that many of these fibroblast-like cells that grow out of plated adult human islets are derived from mesenchymal stem cells (MSCs) similar to those that can be isolated from bone marrow. Preliminary in vitro lineage tracing studies in human cells provide some evidence for β-cell EMT yielding proliferative cells with a mesenchymal phenotype (14). It is hoped that further research will lead to the development of protocols for expanding cadaveric islet cultures prior to transplantation. There are, however, numerous caveats, including problems in reproducing the methods used to expand and redifferentiate cultured islets, variations in human islet preparations in their capacity to undergo this process, and the low expression levels of insulin in the redifferentiated cultures.

Bone marrow is an important source of easily accessible adult stem cells with a longstanding track record of safety and efficacy following clinical transplantation. It has been proposed that adult bone marrow contains multipotent adult progenitor cells (15) that can be induced to differentiate into a wide variety of cell types including pancreatic β cells. It has, however, been very difficult to replicate these findings, and at this stage (despite numerous published claims) there is no convincing evidence that bone marrow stem cells are pluripotent. Insulin expressing cells have been derived from bone marrow mesenchymal stromal cells (MSCs) isolated from adult human bone marrow, but only after modification with genes that are known to control development of the pancreas. The issue as to whether transplanting bone marrow cells may have a role in replacing β cells is also controversial. One study showed that mouse bone marrow cells could differentiate into β cells with glucose and incretin dependent secretory responses when transplanted into lethally irradiated mice (16). However, at least three further studies have been unable to replicate these findings. Other sources of adult MSCs include adipose tissue and human umbilical cord blood. Although preliminary studies indicate that insulin expressing cells can be derived from both of these sources, it is too early to say whether the efficiency will be in the range required for clinical purposes. There is certainly evidence that donor-derived cells are found in the pancreas following bone marrow transplantation but not within the β-cell compartment (17). A unifying hypothesis may be that MSCs play a role in preventing tissue fibrosis and in supporting endogenous tissue regeneration without themselves transdifferentiating into new β cells. Recent clinical studies of autologous mobilized bone marrow stem cell intravenous infusion have demonstrated the potential for prolongation of ‘honeymoon’ β-cell function in those with nonketotic newly presenting type 1 diabetes (18).

Transdifferentiation is a form of reprogramming that differs from that used in the generation of iPS cells in that the process involves direct transformation of one differentiated cell type into another. An example of this is the spontaneous dedifferentiation and redifferentiation that pancreatic acinar cells undergo when placed in a culture dish. During the redifferentiation process the cells acquire characteristics of ductal cells through a process that mimics early stages of pancreatogenesis. The progenitor cells derived from adult acinar cell cultures can be directed towards an hepatocyte lineage by treatment with the glucocorticoid dexamethasone, while treatment with EGF and leukaemia inhibitory factor (LIF) can induce formation of β cells, albeit at low efficiency. Further work is required to determine whether human acinar cells exhibit the same plasticity as rodent cells, and whether they can be reprogrammed to β cells using similar approaches (19). These findings may also have implications for in vivo β-cell regeneration (see below).

Before describing how residual islet cells in the pancreas of both type 1 and 2 diabetic patients might be targeted it is first important to provide some background on the mechanisms that regulate β-cell mass.

During embryogenesis the expansion of the pancreas occurs from a pool of progenitor stem cells. In adult rodents the final mass of β cells is in part determined by the number of embryonic progenitor cells (20). In humans the ultimate β-cell mass in adults is determined by the rapid expansion in β-cell mass that occurs in the early years of life up to age 5 (21). This emphasizes some of the marked differences between rodents and humans and further complicated by the reliance on ‘circumstantial evidence’ from autopsy material, with all its attendant limitations, to quantify β-cell mass in humans. In both rodents and humans the turnover of β-cells is very slow. There are certain times in life, however, such as during pregnancy and during periods of excessive weight gain when β-cell mass expands to compensate for the increased metabolic demands, underlining the as yet untapped potential ex vivo.

Although the mechanisms are not well understood it is clear that β-cell mass is maintained through β-cell replication and from differentiation of islet progenitor cells—a process termed ‘neogenesis’. That β-cell replication is the principal mechanism involved in the maintenance of β-cell mass, at least in young growing mice, was demonstrated by lineage tracing of genetically marked β cells (22). This was subsequently confirmed using a DNA-analogue-based lineage tracing technique (23). Indeed, autopsy studies in humans provide strong supportive evidence that β-cell replication is the primary mechanism underlying β-cell expansion in childhood (24) and potentially also in obesity/pregnancy. Replicating β cells are, however, only very rarely seen in adult human autopsy studies and there is a significant body of data supporting a role for differentiated adult ductal cells as a source of pancreatic progenitors (25). Increased neogenesis, or budding of endocrine cells from the ducts, has been observed in rodents undergoing partially pancreatectomy, and in response to treatment with exendin-4, B-cellulin or overexpression of interferon (IFN) γ or transforming growth factor α. Lineage tracing of genetically marked ductal cells show that in mice they can give rise to both new islets and acinar tissue after birth and injury. Further support for the presence of progenitor cells in the duct comes from an elegant study in adult mice, whereby new β cells were formed from non-β cells located in the lining of the duct during regeneration of the pancreas in response to duct ligation. Shortly after duct ligation there was an increased number of cells expressing Ngn3, which is not normally expressed in the adult pancreas (26). These Ngn3-positive cells were sorted by flow cytometry and implanted into pancreatic buds from Ngn3^–/– mice. Under these conditions the Ngn3-positive cells from the regenerating adult pancreas differentiated into β and other endocrine cell types. It has been postulated that neogenesis may be particularly important in adult humans for compensatory expansion.

In humans the rate of β-cell division is very slow with less than 1% of the total β-cell number replicating with a 24-h period. Studies in the mouse suggest that progression through the cell cycle is regulated by D cyclins (particularly D2 and D4) acting predominantly at the G1/S transition, although A and B cyclins acting at the G2/M checkpoint may also be important. Very little is known about the mechanisms involved. The expansion of β cells that occurs during pregnancy may involve menin, a protein that acts as a tumour suppressor in neuroendocrine cells (27). Menin suppresses β-cell growth by acting as a component of a histone methyltransferase complex that controls expression of the cell cycle regulators p27^kip1 and p18^INK4c, and during pregnancy the maternal hormones prolactin and placental lactogen inhibit the actions of menin in β cells. Other proteins implicated in the control of the cell cycle in β cells include the transcription factor FoxM1 and survivin, a member of the inhibitor of apoptosis family of proteins. It is hoped that by understanding these mechanisms better we may able to develop novel drugs and expand the use of hormones such as GLP-1, Exendin-4, and cholecystokinin, which are currently known to increase β-cell proliferation.

There is also some evidence that the β-cell expansion that occurs in the duct-ligated rat may be derived from transdifferentiation of acinar cells (see above). Thus, adult acinar cells genetically labelled with an amylase promoter-driven Cre recombinase gave rise to new insulin-secreting cells following treatment in vitro with EGF and nicotinamide (28). The mechanisms may involve disruption and remodelling of cadherin-mediated intercellular contacts by phosphatidyl inositol-3-kinase mediated pathways. The in vivo transdifferentiation of acinar cells to β cells is, however, the subject of some controversy since conflicting lineage tracing studies provide strong evidence that pre-existing acinar cells can give rise to acinar cells but not β cells using several models of pancreatic injury in mice (29). Transdifferentiation can also be achieved by expression of exogenous transcription factors. Thus expression of a ‘superactive’ form of the key β cell transcription factor Pdx1 in the liver of Xenopus laevis converts the liver to pancreas, while adenoviral administration of Pdx1 (alone or in combination with other factors) to mice results in uptake by the liver and transdifferentiation of a small population of cells to insulin-secreting cells (30). The recent demonstration that adenoviral-mediated administration of three transcription factors known to be involved in the development of β cells, i.e. Pdx1, Ngn3, and MafA can convert acinar cells to β cells (31), confirms the plasticity of differentiated exocrine cells, but, as with any approach that relies on random integration of exogenous transcription factors and viral sequences, clinical applications are limited by the unacceptable risk of cancer.

Following realization of the potential for recombinant DNA technology to be harnessed clinically in the early 1970s, gene therapy has followed the classical trajectory of new technologies promising revolutionary changes to conventional medical practice. Initially heralded as a panacea for all disease areas, early set-backs largely related to toxicity of the viral-derived vectors employed to transfer the therapeutic gene into the patients’ cells. Adenoviral vectors can induce a profound acute inflammatory response leading to short-lived expression and at least one death in a phase 1 clinical trial with an early generation vector (32). Retroviral vectors enable long-term uptake and expression in dividing cells through chromosomal integration but, despite evidence of therapeutic benefit, targeting of insertion to the promoter of an oncogene has been associated with leukaemic transformation in children with severe combined immunodeficiency (33). These disappointments led to loss of confidence and replacement of initial hope and perhaps hype with pessimism. Over the past few years, continued measured incremental progress has begun to overcome remaining challenges increasingly leading to meaningful clinical benefit.

HIV-derived lentiviral vectors have been developed which can cross an intact nuclear membrane and thus infect nondividing cells. Potential for oncogene activation at the site of chromosomal insertion and generation of pathogenic HIV viral particles remains a concern for clinical implementation (34). Adeno-associated viruses are nonpathogenic and can mediate long-term transgene expression particularly after in vivo gene transfer to stable end-differentiated tissues such as muscle. This has enabled clotting factor replacement in individuals with haemophilia although circulating levels have remained subtherapeutic to date. Exciting preliminary success has been achieved in individuals with visual loss due to inherited retinal degeneration following subretinal injection of adeno-associated viral vectors encoding normal copies of the mutant gene causing the disease.

Avoidance of viral vectors through employment of bacteria-derived plasmids appears to be an extremely safe approach for clinical translation but has been limited by inefficient gene transfer and short-term expression given maintenance as an episome without chromosomal integration (35). Recently, co-expression of the enzyme transposase with a plasmid encoding the therapeutic transgene flanked by specific transposon base sequences has enabled chromosomal integration without viral vectors. This approach is currently being refined to enable targeting to a specific site in the human genome with the aim of preventing insertional oncogenesis (36).

Gene therapy has potential for clinical translation in virtually all aspects of diabetes management (Box 13.11.1). Many of these approaches are complementary to the cell-based approaches outlined earlier in this chapter. Indeed, the current excitement being generated around induced pluripotent stem cells has led to increased acceptance of inclusion of transcription factor gene transduction within a translational clinical approach. As alluded to above, transfection with key β-cell transcription factors such as Pdx1, NeuroD1, and Ngn3 may play a role in differentiation/neogenesis of embryonic stem cells and adult stem/precursor cells in pancreas, liver or bone marrow. Development of proliferative β-cell lines in the laboratory has proved considerably more challenging from primary human tissue than from small mammal sources. Transfer of Pdx1 in addition to the two dysfunctional components of the K_ATP channel (SUR-1 and Kir6.2) in β cells derived from a patient with persistent hyperinsulinaemic hypoglycaemia of infancy (37) and reversible immortalization of normal human islet β-cells with SV40 large T antigen and human telomerase (38) has enabled physiological glucose-responsive insulin secretion. Concerns around residual tumorigenicity have precluded clinical translation, however.

Transduction of gut endocrine K cells (which secrete glucose-dependent insulinotropic polypeptide in response to nutrient ingestion) with an insulin construct under the control of the GIP promoter has enabled prevention of hyperglycaemia following chemical induction of diabetes in mice with streptozotocin (39). In its most recent as yet unpublished studies the Kieffer group has successfully delivered the vector to gut cells in vivo using transposon plasmid technology and a novel endoscopic approach.

Skeletal muscle is a particularly attractive target for in vivo gene delivery offering the promise of simple injectable DNA therapies as a platform for sustained systemic protein secretion within routine clinical practice. Insulin secretion leading to glucose lowering without dangerous hypoglycaemia has been attained following plasmid-mediated (40) and adeno-associated viral vector in situ insulin gene delivery to muscle (41). Currently the possibility of attaining long-term GLP-1 therapy following a single intramuscular plasmid injection is being explored. In addition to using the recipient’s own cells and thus avoiding alloimmune rejection, it is hoped that insulin expression in nonpancreatic cells will circumvent recurrent autoimmunity due to sufficient differences in antigen expression to endogenous β cells.

Although it remains in its infancy with huge scope for further refinement, reproducible therapeutic success following deceased donor islet transplantation has confirmed that β-cell replacement therapy can restore normoglycaemia while entirely avoiding dangerous hypoglycaemia––a goal unattainable with any exogenous insulin replacement approach including current generation closed loop glucose sensors/insulin pumps. This has provided a pathway for clinical translation of novel cell-based diabetes therapies. Progress towards embryonic stem cell-derived β cells has surpassed all expectations over the past few years and the potential of recipient-specific stem cells by reprogramming skin cells is currently generating considerable excitement. Bone marrow-derived stem cells appear to have real promise for attenuating endogenous β-cell destruction in early type 1 diabetes with more definitive controlled trials eagerly awaited. Progress towards safer gene therapy vectors has facilitated augmentation of cellular therapy by gene manipulation. It is envisaged that these approaches will be implemented clinically in the foreseeable future to enhance function and prevent apoptosis/rejection of transplanted islets. Therapeutic applications of in situ gene delivery without the need for ex vivo cell manipulation or transplantation are increasingly entering clinical trials with considerable future potential for diabetes-specific applications.

KD was supported by the Juvenile Diabetes Research Foundation.