Plasticity of the β cell insulin secretory competence: preparing the pancreatic β cell for the next meal

Different components of plasticity of the β cell insulin secretory competence

The pancreatic β cell is unique in that it is the only animal cell type that can produce, store and secrete insulin in quantities that are required for growth and survival under the most variable environmental conditions (Rorsman, 1997; Goodge & Hutton, 2000; Rutter, 2001). This physiological task is not trivial, as insulin is not only required for survival (as evidenced by ketoacidosis in untreated type 1 diabetes), but can also be lethal within minutes when circulating at inappropriately high concentrations. To match the secretion of insulin to the physiological demands, the differentiated mammalian β cell has evolved in such a way that it displays a sophisticated degree of plasticity that allows adaptation at various ranges of time, preventing both shortage and excesses of insulin in the circulation. On a rapid time scale, typically occurring when a meal is digested and absorbed, nutrients and the incretin hormones, glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic peptide (GIP), synergize in the acute stimulation of insulin secretion (Jia et al. 1995). This potentiation occurs via a rise in intracellular cAMP in the β cell (Moens et al. 1996) and seems to be required to obtain sufficient insulin for the proper nutrient disposal in various tissues of the organism (Holst & Orskov, 2001). A safeguard against insulin toxicity is that GIP and GLP-1 (henceforth referred to as incretins) only potentiate insulin release as long as extracellular glucose is elevated above the basal concentration (Jia et al. 1995). At the other extreme of the time scale, requiring days to weeks, glucose and also incretins have been proposed to determine the size of the pancreatic β cell pool by affecting the rates at which β cells are generated and destroyed, respectively, by neogenesis and apoptosis (Hoorens et al. 1996; Jonas et al. 1999; Bonner-Weir, 2000; Li et al. 2003; Drucker, 2003).

How the acute stimulation of exocytosis (Rorsman, 1997; Henquin, 2000; Rutter, 2001; Schuit et al. 2001; Newgard & McGarry, 1995), as well as the regulation of β cell number (Bonner-Weir, 2000; Drucker, 2003), is regulated by glucose and incretins has been reviewed elsewhere so that these aspects of plasticity are not discussed in detail in the current text. As for the acute stimulation, Fig. 1A gives a brief overview. In order to stimulate exocytosis, the sugar has to be taken up by the β cells and metabolized. Glucose metabolism in β cells is specialized in signal generation and contains an important mitochondrial phase (Maechler & Wollheim, 1990) with very little anaerobic glycolysis (Schuit et al. 1997). The mitochondrial catabolic flux causes a rise in the cellular pyridine nucleotide redox state (Malaisse et al. 1978), and in the ATP/ADP ratio (Detimary et al. 1998); in concert an anabolic flux called cataplerosis (Farfari et al. 2000; Flamez et al. 2002) is activated, resulting in the export of metabolites such as citrate and malate into the cytosol (Schuit et al. 1997). The further details in signalling are best understood for the rise in ATP/ADP (Fig. 1A). This change closes ATP-sensitive potassium channels (K⁺_ATP channels; Ashcroft, 1988), which will cause depolarization of the plasma membrane and a rise in cytosolic calcium (Herchuelz et al. 1980). It was recently shown that opening of L-type Ca²⁺ channels calcium is responsible for about 80% of calcium-induced exocytosis (Gopel et al. 2004).

The process by which incretin hormones potentiate the nutrient-induced process described above is essentially caused by the fact that both GLP-1 receptors and GIP receptors are functionally coupled to adenylate cyclases, raising cAMP in the β cell (Moens et al. 1996). A signalling cascade initiated by cAMP interacts with the rise in cytosolic Ca²⁺ to induce exocytosis (Ammala et al. 1993; Takahashi et al. 1999). This effect has been shown to require both cAMP-dependent protein kinase (PKA) and a PKA-independent mechanism (Renstrom et al. 1997). The latter has been identified as Epac2 (cAMP-GEFII), a guanine nucleotide exchange factor that interacts with Rim2 and Rab3 in islets (Kashima et al. 2001; Holz, 2004). Notably, generation of cyclic AMP and stimulation of downstream kinase modules by incretin receptor activation has been shown to be independent of ambient glucose concentrations (Hinke et al. 2000; Ehses et al. 2002), yet for incretin potentiation of insulin release, glucose must be elevated. The precise molecular mechanism of this phenomenon is not clearly understood, and although probably an over-simplification, it has been proposed that the glucose dependence of incretin action on exocytosis is a combined manifestation of the need for a high ATP/ADP ratio and energy dependence of granule mobilization by PKA (Gromada et al. 1998). Within a time frame of minutes, the acute insulin secretory competence can be intensified by repetition of the glucose stimulation, either in intact perfused pancreata (Moens et al. 2002) or in perifused isolated islets of Langerhans (Zawalich et al. 1988). As this short-term glucose memory occurs so rapidly, it is likely that this adaptation occurs either via post-translational modifications of already existing protein(s) involved in exocytosis or by the cellular accumulation of signals such as calcium or lipid mediators (Zawalich et al. 1988) that are responsible for triggering the release process.

This Topical Review focuses on an intermediate level of plasticity, taking place within minutes following the onset of stimulation and extending up to many days. At this level, gene expression in existing differentiated β cells is modified, changing the functional behaviour of the cells. During a particular meal, when physiological stimuli acutely trigger exocytosis during the whole period of digestion and absorption (60–120 min), glucose and incretins simultaneously affect multiple other steps in β cell function. The outcome of this effect is that release of insulin granules is balanced (Schuit et al. 1991) by a process of granule formation; the responsible mechanisms comprise transcription and translation of the insulin gene(s), accelerated insulin folding and formation of other insulin secretory granule components (Goodge & Hutton, 2000; Ohneda et al. 2000). It is expected that such adaptation involves enhanced expression of a large number of other genes in addition to the insulin gene(s).

Glucose and incretins regulate plasticity of β cell insulin secretory competence in vitro

In serum-free conditions and using primary rat β cells, Ling et al. (1996) showed that the functional status of β cells was much better preserved in culture medium at 10 mm glucose than in 5 mm glucose, although cellular viability and the size of the insulin storage pool were constant. Translational activity was also significantly higher in the cells cultured at 10 mm glucose, indicating that overall translational rates are correlated to glucose competence. This idea was further supported by two studies using the non-specific protein translation blocker cycloheximide (Garcia-Barrado et al. 2001; Schuit et al. 2002), as such treatment leads to rapid loss of insulin secretory competence. In mouse islets, within 2 h of incubation with cycloheximide, the secretory response to glucose was diminished (Garcia-Barrado et al. 2001). In fluorescence-activated cell sorting (FACS)-purified β cells, 6 h of cycloheximide treatment significantly suppressed insulin secretory competence, without measurable effects upon cellular viability or insulin content (Schuit et al. 2002). This indicates that protein(s) with rapid turnover are implicated in regulating glucose competence of the β cells in a manner that is independent of the amount of stored insulin granules per cell. As glucose itself is a major stimulus for translational activity in β cells (Schuit et al. 1988) and glucose can affect gene transcription as well as mRNA stability in β cells (see next section), it can be proposed that a history of glucose-induced gene expression (transcriptional or post-transcriptional) is necessary for the glucose competence of the cells. Similar evidence is accumulating for the incretin hormones. GLP-1 has been reported to induce several genes in β cells lines, promoting the growth of the cells and inhibiting apoptosis (Li et al. 2003; Drucker, 2003). In a recent study (Delmeire et al. 2004), it was shown that the secretory competence of cultured primary rat β cells was enhanced by intermittent addition of 0.5 nm of GLP-1 and GIP together to the tissue culture medium. As longer incubation with incretins resulted in receptor desensitization, the stimulation was applied as three 1-h periods together with 10 mm glucose, attempting to mimic the periodicity of nutrients and hormones observed in vivo. The enhanced secretory capacity of these terminally differentiated cells could not be explained by altered viability of the cells or a change in the size of the insulin granule storage pool. One can hypothesize therefore that the incretin-induced changes in secretory competence are caused by changes in gene expression in the β cell.

Molecular mechanisms involved in plasticity of β cell secretory competence

Over the time frame of minutes to several hours, insulin secretory competence can be altered by changes at the level of signalling or altered gene transcription and protein synthesis. Some pathways by which this can be achieved are shown in Fig. 1B–D. It can be assumed that changes in this plasticity are mediated by critical protein regulator(s) that are up-regulated (stimulators) or suppressed (inhibitors) following glucose and incretin stimulation. A number of microarray experiments have suggested (Webb et al. 2000; Shalev et al. 2002; Flamez et al. 2002; Schuit et al. 2002) that glucose alters the abundance of a large number of transcripts (depending on the chosen conditions perhaps as many as a few hundred) that are involved in synthesis of new insulin granules, metabolic signalling, second messenger production, signal amplification, exocytosis as well as (post)transcriptional regulation. It is becoming increasingly clear that the specialized phenotype of insulin-secreting β cells is the work of number of β cell-specific proteins from which expression is regulated by a β cell-specific transcription factor network (Fig. 1B). Microarrays searching for gene promoters binding to particular transcription factors present in the β cell, such as HNF1α, HNF4α and HNF6, suggest that a large part of the set of transcripts present in β cells are expressed in this way (Odom et al. 2004). The tight link between maturity onset diabetes of the young (MODY; a monogenic form of diabetes) and a series of transcription factors expressed in the β cell (Fajans et al. 2001) is a medical consequence of this phenomenon. While transcriptional regulation seems extremely important, the β cell secretory competence seems also to be regulated at the post-transcriptional level. This section will discuss some of these mechanisms.

Transcriptional modulation in β cells by hormones and nutrients. 

Glucose and incretins can alter mRNA abundance of critical transcripts by changing rates of transcription (Fig. 1B), either by altering the activity of regulators of transcription that are already present in the cell or by net synthesis of novel transcription factors. Via the latter mechanism, up-regulation of mRNA encoding the sterol element binding protein 1c (SREBP1c) and down-regulation of peroxisome proliferation-activated receptor α (PPARα) in glucose-stimulated β cells might explain suppression of transcripts encoding enzymes of fatty acid oxidation and induction of mRNA encoding lipogeneic enzymes (Flamez et al. 2002). It is not clear whether such a coordinated response enhances metabolic signalling in the β cell or whether it serves a greater need for secretory granule membrane synthesis. Probably the most often encountered transcriptional regulation occurs via changing the functional status of existing transcription factors that are already present or have to be translocated into initiation complexes of transcription. This change can be caused either by second messengers that act allosterically upon transcription factors or by kinases/phosphatases that cause changes in the phosphorylation status of transcription factors (Fig. 1B). Allosteric interactions and (de)phosphorylation can modify the subcellular localization, e.g. translocating the transcription factor from cytosol to the nucleus, or (in)activate transcription factors that are already residing in the nucleus, waiting for a signal to respond. Glucose and incretins synergize at the level of a rise in intracellular calcium ([Ca²⁺]_i) via effects on ATP-sensitive potassium channels (K⁺_ATP channels) and L-type voltage-dependent calcium channels (VDCCs) (Suga et al. 1997a,b), and potentially also via the process of calcium-induced calcium release (Holz et al. 1999). Furthermore, glucose and GLP-1 synergize at the level of intracellular cAMP production, as β cells express adenylate cyclase type VIII (Delmeire et al. 2003). So it seems conceivable, and even likely, that many of the effects of glucose/incretins on the activity of transcription factors may be the result of changes in [Ca²⁺]_i and in intracellular production of cAMP.

Instead of a comprehensive review of all such known regulations of this type in the β cell, we chose to discuss one example of the sophisticated interplay between glucose and incretins in the stimulation of insulin gene transcription. nuclear factor of activated T cells (NFAT) is a well characterized calcium-dependent transcription factor in lymphocytes (Rao et al. 1997; Okamura et al. 2000) that was proposed to mediate a synergistic interplay between glucose and incretin-induced insulin gene transcription (Lawrence et al. 2001, 2002). In resting T cells, NFAT resides in the cytoplasm and is fully phosphorylated; following stimulation, it rapidly becomes dephosphorylated by calcineurin (calcium/calmodulin-dependent protein phosphatase 2B) activated by an increase in cytosolic calcium concentration. Dephosphorylation results in the exposure of a nuclear localization sequence promoting the rapid translocation of NFAT to the nucleus (Luo et al. 1996). In β cells, NFAT has been shown to bind in a calcium/calcineurin-dependent manner to three distinct NFAT-response elements within the insulin promoter and to activate insulin gene transcription (Lawrence et al. 2001; Lawrence et al. 2002). As the major NFAT-binding site in the insulin promoter overlaps with an inverted responsive element for C/EBPβ, a repressor of insulin gene transcription during prolonged exposure of β cells to high glucose (Seufert et al. 1998), competition between different transcription factors for binding to common responsive elements activated by different second messengers further illustrates the plasticity of the secretory response of β cells. Inhibition of calcineurin during post-transplantation therapy by the immunosuppressant calcineurin inhibitors FK506 and cyclosporin A has been shown to decrease pre-proinsulin mRNA, depleting the cells of insulin and increasing the risk for post-transplant diabetes mellitus (Drachenberg et al. 1999), perhaps illustrating the clinical relevance of NFAT-regulated insulin gene expression.

Post-transcriptional control of mRNA abundance and β cell plasticity. 

In addition to transcriptional regulation, changes in mRNA abundance can be regulated post-transcriptionally, for example, differential effects upon mRNA splicing, nuclear export or stability. Following glucose stimulation, exocytotic events are responsible for the loss of granules from the storage pool. To replenish this pool and keep a balance between release and storage (Schuit et al. 1991), the β cell should rapidly enhance the synthesis of new granule proteins and produce new membrane at the level of the endoplasmic reticulum (ER). Protein and mRNA analysis suggest that the abundance of a large number of secretory granule-related genes is up-regulated by glucose in a coordinated manner (Guest et al. 1989; Knoch et al. 2004). As for accelerated protein synthesis in primary β cells, this effect is achieved primarily by a powerful acceleration of general translation (Schuit et al. 1988; Schuit et al. 1991). In addition, in primary β cells, insulin mRNA is translated preferentially as compared to non-insulin encoding mRNAs in these cells (Schuit et al. 1988). Increased translation of proinsulin mRNA results from the stimulation of translation initiation and elongation (Permutt, 1974), as well as from a reduced degradation of insulin mRNA (Welsh et al. 1985) (Fig. 1C). This latter mechanism has recently been explored and involves the rapid nucleo-cytoplasmic translocation of polypyrimidine tract-binding protein (PTB) upon β cell stimulation with glucose (Knoch et al. 2004). Activated PTB has been shown to increase the stability of insulin mRNA by binding to its 3′-untranslated region, which contains a pyrimidine-rich consensus sequence (Tillmar et al. 2002). The coordinated glucose-stimulated synthesis of insulin and other components of dense-core secretory granules (Guest et al. 1989) can be explained (at least in part) by the fact that PTB also binds and stabilizes mRNAs encoding different transcripts related to granule biogenesis, in particular prohormone convertases 1/3 or 2 (PC1/3, PC2), pre-chromogranin A, secretogranin II, synaptobrevin 2, synaptophysin and the tyrosine phosphatase like molecule I-A2 (Knoch et al. 2004). The importance of PTB stabilization for the regulation of dense core biogenesis was further supported by silencing of the endogenous PTB mRNA expression in the β cell line INS-1 (Knoch et al. 2004). Four days after addition of siRNA, transfected cells had virtually no insulin granules left, and the expression of a number of granule-proteins was strongly reduced. It is assumed that an unidentified mechanism, perhaps involving protein kinases and proteases, mediates the effect of glucose/metabolites upon PTB, converting part of the 59 K full length protein into a 27 kDa protein fragment, but the exact molecular events promoting the nucleo-cytoplasmic translocation remain to be elucidated. Interestingly, in neuroendocrine PC-12 cells, cAMP-dependent phosphorylation of PTB has been reported, resulting in its nucleo-cytoplasmic transport (Xie et al. 2003); if this process also occurs in primary β cells, PTB would represent another regulator of gene expression by which glucose and incretins exert their control of β cell plasticity.

The role of regulating protein translation in β cell function. 

A final important step in the regulation of gene expression is mediated by the fact that glucose has a profound effect on overall efficiency of translation in the β cell (Schuit et al. 1988), implying that the mRNAs that are present at a given time point are more rapidly translated as extracellular glucose rises. The exact molecular mechanisms are still partially unknown. Early studies on insulin biosynthesis pointed to accelerated translational initiation (Permutt, 1974) and elongation at the level of signal peptide-mediated arrest (Welsh et al. 1986). Glucose control of initiation may involve phosphorylation of IF4E binding protein (Xu et al. 1998) and eIF2α, the latter being a sensor of the balance between the total protein folding load and folding capacity in the ER compartment (Kaufman, 2002; Ron, 2002). Accelerated synthesis of insulin, other secretory proteins and components of secretory granules (see previous paragraph) impose a burden at the level of protein folding in the ER lumen. Physiologically, such a condition activates the so-called unfolded protein response (Kaufman, 2002). When the biosynthetic load exceeds folding capacity, the ER chaperone BiP is reallocated from the eIF2α-kinase PERK to the ER lumen, complexing the unfolded protein. By dissociation of BiP from the luminal domain of PERK, this kinase is activated (Ron, 2002). PERK activation then rapidly attenuates overall translation by phosphorylating the initiation factor eIF2α. Phosphorylated eIF2α, however, specifically favours translation of mRNA encoding the transcription factor ATF4, thereby increasing ATF4 protein in the cell (Fig. 1D). As ATF4 is involved in transcription of a number of genes that are relevant for protein folding in the ER compartment, the cell can adapt with time to a higher biosynthetic load by increasing the ER folding capacity (Ron, 2002). Recent data have indicated that without such compensatory translational and transcriptional response, the increased load on the biosynthetic apparatus in the β cell would result in severe ER stress and loss of insulin secretory competence (Harding et al. 2001; Scheuner et al. 2001). The pivotal roles of PERK and eIF2α-phosphorylation have been underlined in PERK-deficient mice (Harding et al. 2001) and mice in which the serine-51 phosphorylation site was mutated into an alanine (Scheuner et al. 2001). A direct clinical correlation was provided by the observation that a mutation of the gene encoding PERK in man causes diabetes (Delepine et al. 2000). Thus, translational control of the β cell and its connection to the molecular pathology of diabetes is an interesting emerging field of research. It seems relevant to investigate whether incretin hormones can interact with this pathway and whether nutritional variables can modulate the integrated response to ER stress in the β cell.

Do glucose and incretins regulate plasticity of β cell insulin secretory competence in vivo?

A major question which has not been effectively addressed is whether the in vitro observations of glucose and incretin priming effects occur in vivo. The primary difficulty would be to experimentally differentiate these effects from the well documented effects of nutrients and hormones on β cell neogenesis and proliferation (Drucker, 2003). Furthermore, in intact animals, it may be hard to separate direct from indirect effects, also considering that paracrine effects of other islet hormones may alter β cell glucose competence (Huypens et al. 2000), and that innervation of the endocrine pancreas also plays a large part in controlling the secretory response (Gilon & Henquin, 2001). Two approaches have been made to examine how the nutritional state of an animal alters its future secretory responsiveness: glycaemic clamp studies (in vivo analysis) and perifusion of freshly isolated islets (ex vivo study), in both cases comparing the fed and fasted state. Virtually all reports find greater responsiveness of pancreatic β cells which are exposed to periodic elevations of nutrients and incretin hormones (i.e. ad libitum food access) compared to the fasted state. The molecular mechanism underlying this phenomenon has been a matter of debate for several decades (Zawalich & Zawalich, 2000). It can generally be stated that it does not correlate well to changes in insulin content. The molecular underpinnings of this observation probably occur as the result of a number of alterations in the β cell caused by periodic elevations in nutrients and incretin hormones which induce intermediate and long-term alterations in the secretory phenotype that are based upon altered gene and protein expression in the β cell.

In healthy individuals, glucose is strictly maintained between narrow limits as a result of insulin action during and following a meal, and counter-regulatory mechanisms during fasting. Similarly, bioactive incretin hormone concentrations are also tightly controlled, by the rapid inactivation by dipeptidyl peptidase IV and renal elimination of peptide metabolites over a longer time course (Kieffer et al. 1995; Meier et al. 2004). The truncated metabolites of GIP and GLP-1 appear to have no effect on glucose homeostasis (Hinke et al. 2002; Vahl et al. 2003); however, it cannot be discounted that these metabolites have physiological effects, but the appropriate parameter and experimental model have not yet been examined. It is perhaps an important observation that despite their rapid degradation, measurable intact bioactive incretins can be detected even following prolonged fasting (Meier et al. 2004), indicating that in addition to acting during the postprandial phase, they also may be exerting tonic effects on the β cell. Support for this notion is given by the biological effects described for so-called ‘sustained release’ formulations of GLP-1 or metabolically stable GLP-1 analogues, which have been shown to exert long-term beneficial effects on glucose homeostasis after single injection (Chang et al. 2003; Nielsen et al. 2004). Implicit to this hypothesis is that although the acute actions of incretins are dependent upon elevated glucose concentrations, many studies have shown that the receptor activations of signal transduction cascades do not require glucose.

It seems likely that prolonged stimulation in vivo with bioaccumulating incretin agonists would also cause down-regulation of the receptors on islets (although this has not been shown experimentally); however, most of the data regarding desensitization is based upon in vitro observations. The first studies examining desensitization of the GIP and GLP-1 receptors were reported in 1991, examining HIT-T15 cells by perifusion; results demonstrated both homologous and heterologous desensitization for the incretin receptors (Fehmann & Habener, 1991). Subsequent studies have used both insulinoma cell models and over-expression systems to examine the molecular nature of incretin receptor desensitization. The desensitization of the GLP-1R during prolonged stimulation was found to be PKA independent (Gromada et al. 1996), and homologous and heterologous desensitization (and secondary endocytosis) were shown to occur following phosphorylation of serine doublets found in the intracellular C-terminal tail of the receptor, probably mediated by PKC (Widmann et al. 1996, 1997). Homologous desensitization of the GIP receptor was similarly shown to be PKA independent, and although PKC was implicated in modulating GIP receptor responsiveness from kinase inhibitor studies, its involvement in GIPR desensitization was not supported by the experimental evidence (Hinke et al. 2000). Analysis of the carboxyl tail of the GIPR by site directed mutagenesis was examined in two independent studies, and identified several potential regulatory target residues for receptor internalization (over the course of minutes to hours of stimulation) or chronic down-regulation (24 h of continuous stimulation) (Tseng & Zhang, 1998; Wheeler et al. 1999). The acute control of GIP receptor responsiveness is thought to be primarily controlled by phosphorylation mediated by G-protein receptor kinase 2 (GRK-2) (Tseng & Zhang, 2000); however, similar studies have not been published for the GLP-1 receptor. Recent studies have shown that desensitization of both incretin receptors also occurs in primary purified rat β cells (Delmeire et al. 2004). In addition, as receptor desensitization is unlikely ever to be 100%, a chronic low stimulation of the β cell, such as with incretin analogues with improved biological and circulating half-lives, may foster appropriate handling of glycaemic excursions via priming, recruitment and sensitization of the β cell, which may be mediated by direct post-translational modification (phosphorylation/dephosphorylation) or positive long-term changes in gene expression. Alternatively, the process of receptor desensitization itself may improve the capacity of the endocrine pancreas to respond to glucose, as receptor internalization has been demonstrated to be one of the mechanisms used by G-protein-coupled receptors to transactivate the epidermal growth factor receptor (Pierce et al. 2000; Buteau et al. 2003), and thus may promote β cell function and/or expansion via activation of genes controlling β cell differentiation and proliferation or activation of kinases/phosphatases which mediate the intermediate phase of islet competence.

In coverage of this topic, it must be noted that some studies have been unable to reproduce the priming effects of incretins on subsequent insulin release in human in vivo experiments (Rachman et al. 1997; Brandt et al. 2001; Meier et al. 2003). It is unclear whether the discrepancy between these clinical studies and results from rodents (in vitro and in vivo) are due to experimental design, alteration in neural tone, the result of receptor desensitization (or failure to re-sensitize), or due to differences in the proliferative or regenerative capacity of human endocrine tissue relative to rodents. One possibility which warrants further examination is reproduction of periodicity of stimulation, as this is likely to be a critical element in this process. Thus, in studies using sustained release incretin analogues (Chang et al. 2003; Nielsen et al. 2004), it is likely that the tonic stimulation of β cells was greater than usually present during the interdigestive phase, but periodicity of endogenously secreted incretins was normal, and thus mediated future positive effects on β cell responsiveness. Further examination of the secretory competence of the β cell, especially comparing ex vivo data on gene expression and in vivo function must resolve these questions.

Perspective

Five out of six known forms of MODY are caused by mutations in transcription factors expressed in the β cell (Fajans et al. 2001). Because of this clinical information as well as mechanistic studies in β cell lines (see for example Wobser et al. 2002) the multiple links between diabetes, β cell dysfunction and abnormalities in the operation of transcription factor networks are starting to become understood (Ferrer, 2002). It will be important to search in a systematic manner, e.g. by genome-wide mRNA or proteome analysis, for transcription factors from which the abundance is regulated by glucose, as such factors may be high in the hierarchy of glucose control of gene expression, regulating in turn the expression of numerous glucose-dependent genes. The recently published microarray study on chromatin immunoprecipitated from human islets (Odom et al. 2004) is an important example of this type of research. Mechanistic studies on HNF1α (Wobser et al. 2002; Ferrer, 2002), as well as SREBP1c (Flamez et al. 2002; Wang et al. 2003) and PPARα (Roduit et al. 2000; Flamez et al. 2002) should shed more light on the function of the encoded proteins. A difficult task will be to identify the true master switch genes of glucose control in a complicated interacting network. Perhaps the concept of a ‘master switch’ will prove to be incorrect, while the multiple and complex functionality states of the network itself predominate. As compared to glucose control of gene transcription, relatively few studies have been carried out on post-transcriptional regulation of β cell genes, as well as the influence of incretin hormones on gene expression in the β cell. Consequently, it can be anticipated that these topics may become fascinating areas of future β cell research. Analogous to the MODY paradigm, molecular defects in post-transcriptional regulation in β cells may turn out to be predisposing factors for human diabetes as has been already illustrated by the Wolcott-Rallison syndrome (Delepine et al. 2000). Finally, new pharmacological tools could be designed on the basis of the physiological emerging principle that a synergism between glucose and incretin hormones is required to direct β cell gene expression in preparation for the next meal.