Lessons from glucokinase activators: the problem of declining efficacy

Abstract

The concept of activation of glucokinase (encoded by the Gck gene) as a potential therapy for type 2 diabetes has been explored by several pharmaceutical companies. Small-molecule Gck activators (GKAs) were found to be effective at increasing glucose disposal by hepatocytes and lowering blood glucose in animal models of diabetes during acute or chronic exposure and in human type 2 diabetes during short-term exposure. However, several clinical trials of GKAs were discontinued because of declining efficacy during chronic exposure or other issues. In some cases, declining efficacy was associated with an increase in plasma triglycerides. Accordingly, increased hepatic triglyceride production or steatosis was inferred as the likely cause for declining efficacy. However, other mechanisms of tachyphylaxis need to be considered. For example, elevated glucose concentration causes induction of glucose 6-phosphatase (G6pc) and repression of Gck in hepatocytes. This is best explained as an adaptative mechanism to maintain intracellular phosphometabolite homeostasis. Enhancement of G6pc induction and Gck repression by GKAs because of perturbed phosphometabolite homeostasis could explain the decline in GKA efficacy during chronic exposure. Progress in understanding the mechanisms of intracellular phosphometabolite homeostasis is crucial for development of better drug therapies and appropriate dietary intervention for type 2 diabetes.

Keywords::

• glucokinase
• glucokinase activators
• liver metabolism
• type 2 diabetes

The development of better therapies for type 2 diabetes continues to present a major challenge. Glucokinase (GK, encoded by the Gck gene), the hexokinase that is expressed in the liver and in islet β-cells, has been a major focus of drug discovery programs for type 2 diabetes in recent years [1]. This research effort generated several small-molecule GK activators (GKAs) with remarkable short-term efficacy but with declining efficacy during chronic treatment in man [1,2]. This highlights the hurdles that lie ahead in the development of better therapies for diabetes.

GK has unique molecular properties, compared with the other hexokinases which catalyze the first reaction in glucose metabolism [1,3]. GK has a more flexible hinge linking the bilobal structure surrounding the catalytic cleft that allows greater conformational changes from the ‘closed’ high-affinity state to a ‘super-open’ low-affinity state with a range of intermediate states [3]. This accounts for its unique sigmoidal kinetics and low-affinity for glucose, that make GK ideally suited to function as the glucose sensor for insulin secretion in islet β-cells. GcK is not inhibited by physiological concentrations of its product glucose 6-phosphate (G6P) and an increase in glucose within the physiological range generates an increase in the intracellular concentration of G6P [3]. In the liver, GK is inhibited by the GK regulatory protein (GKRP; encoded by the GCKR gene), which is present predominantly in the nucleus, and inhibits GK competitively with respect to glucose conferring a lower affinity for glucose of GK in liver compared with islet β-cells [4]. The dual regulatory system in the liver comprising GK and GKRP enables both rapid activation of liver GK by the elevated glucose in the portal vein after a meal, and also chronic changes in the affinity of the liver cell for glucose [3,4]. The intrinsic properties of GK and GKRP, in conjunction with other system properties such as the activity of the glucose transporters and downstream enzymes, enable GK to function as the major site of control of glucose metabolism flux within the liver [3]. This property makes GK a potential drug target for control of liver glucose metabolism [1]. Further support for the validity of GK as a potential target for blood glucose control has come from GCK gene mutations in man that stabilize the high-affinity conformation causing a lower blood glucose [1].

The small-molecule GKAs identified by the drug discovery programs bind near the hinge stabilizing the high-affinity (closed) conformation, and mimicking the effect of activating mutations [1]. These GKAs activate GK in hepatocytes and lower blood glucose in animal models and in man [1,2]. More than 20 GKAs progressed to human clinical trials. However, of these only 5 remain active, the rest were terminated in Phase I or Phase II [2]. Two potential hazards with the concept of GK activation as a therapy for diabetes that were flagged at the outset were: i) risk of hypoglycemia, by over-stimulating islet β-cell GK; and ii) increased plasma or hepatic triglycerides, by over-stimulating liver GK [1]. However, an unexpected problem that was not predicted either at the outset or from the preclinical animal studies was that GKAs that were very effective acutely in man would decline in efficacy rather rapidly during chronic exposure in man [5,6]. This raised questions regarding the validity of the animal models or their diets and the underlying mechanism(s) for the decline in efficacy in man. To minimize the risks of hypoglycemia, some groups focused on partial activators or liver-selective activators that are substrates for liver transporters [2,7]. Another approach is to indirectly activate GK by targeting GKRP [8]. Despite concern about the risk of increased plasma or liver lipids [1,2,5], it currently remains debatable whether the decline in glycemic efficacy in man is consequent to the raised lipids.

The clinical data shows either a small (< 20%) increase in plasma triglycerides with MK-0941 [5] or no change with AZD1656 in conjunction with metformin [6]. For both these GKAs efficacy on glycemic control in man declined with time. The animal studies for these GKAs showed no effect on plasma or liver triglycerides [9,10], while studies with other GKAs showed a negligible to moderate increase in plasma triglycerides with liver-selective GKAs [7,8] or a rapid increase in hepatic lipid [11]. The interpretation of the latter study is contentious because the GKA with the greatest effect on steatosis had off-target effects [12]. Plasma triglyceride levels in man are elevated by ∼ 80% on a diet containing 30% fat/55% carbohydrate compared with 45% fat/40% carbohydrate [13]. Arguably therefore, the ∼ 20% increase in plasma triglycerides with a GKA [5] is small relative to the effect of a high-carbohydrate low-fat diet, that is still widely regarded as a healthy diet.

While a raised plasma triglyceride per se may not be the cause of the loss in glycemic efficacy, it may be a marker for other hepatic changes linked to loss of efficacy. Two mechanisms can be considered for the raised plasma triglycerides. The first is a rapid mechanism with a similar time course as the stimulation of glucose phosphorylation by the GKA. This is expected to raise the cell content of G6P and the regulatory metabolite, fructose 2,6-bisphosphate, which activates glycolysis by feedforward allosteric activation mechanisms leading to an increase in flux to fatty acids and triglycerides (Figure 1). The second is a slow process resulting from changes in gene expression linked to raised phosphorylated metabolites (Figure 1) [14,15].

Figure 1. Phosphorylated intermediates regulate both metabolic flux by feedforward allosteric activation and intracellular homeostasis by feedback repression of the glucokinase gene. At basal glucose (∼ 5 mM) GK is sequestered in the nucleus bound to GKRP in its open inactive state. An increase in blood glucose and also GKAs release glucokinase from GKRP causing its translocation to the cytoplasm, stimulation of glucose phosphorylation and an increase in cell. Cytoplasmic G6P concentration correlates directly with GK activity and inversely with the activity of G6pc which is located in the ER. (1) Glucokinase interacts with PFKFB2 forming F26P₂, which allosterically activates PFK1and its product allosterically activates Pklr resulting in feedforward activation of glycolysis, which generates substrate for lipogenesis. (2) Raised levels of F26P₂ also cause translocation of Mlx-linked transcription factors (e.g., ChREBP) to the nucleus and recruitment to gene promoters causing induction of G6pc, Pklr and lipogenic enzymes (e.g., Fasn) and repression of the Gck gene [14,15]. The proteins encoded by these genes have long half-lives [14] and changes in protein expression manifest with a much slower time course than allosteric mechanisms. Because GKAs have opposite effects on glucokinase activity (rapid activation and slow Gck gene repression), a net decline in glucokinase activity would manifest when acute activation can no longer compensate for the decline in protein expression.

It is well established that diets high in carbohydrate and low in fat induce enzymes of de novo lipogenesis, for example, fatty acid synthase and also certain enzymes of glycolysis such as pyruvate kinase [16]. The latter is induced by elevated glucose through the transcription factor carbohydrate response element binding protein (ChREBP), which binds to gene promoters as a complex with Max-like protein X (Mlx) [16]. ChREBP is activated when the cellular concentrations of phosphorylated intermediates of glucose metabolism are elevated [17]. This can occur either at raised concentrations of glucose or in the presence of inhibitors of glucose 6-phosphatase (encoded by the G6pc gene) [14,15] which is a major negative regulator of the cell concentration of glucose 6-phosphate [14,15,17]. The G6pc gene, is an important target gene of ChREBP–Mlx and is markedly induced by elevated phosphometabolites of glucose [14,17]. Although ChREBP function was initially conceptualized as a mechanism for efficient conversion of dietary carbohydrate to fat [16], an alternative hypothesis is that the function of ChREBP is to maintain intracellular homeostasis of phosphorylated intermediates of glucose metabolism [17]. Two sets of evidence support this hypothesis: first, that ChREBP knockdown models are associated with elevated concentrations of phosphorylated intermediates [17]; second, that metabolic conditions that cause activation of ChREBP are associated not only with induction of G6pc, which is a negative regulator of the cell G6P concentration, but also with repression of the “Gck gene because GK protein” is a positive regulator of the G6P concentration [14,17].

Expert opinion

In conditions of elevated glucose, ChREBP activation is markedly enhanced by inhibitors of glucose 6-phosphatase [14] and it is also predicted to be enhanced by GKAs if these were to increase the concentrations of phosphorylated intermediates that cause ChREBP translocation to the nucleus [15] as occurs during overexpression of GK [18]. A recent study on rats treated with a single dose of GKA in conjunction with an oral glucose load showed translocation of ChREBP to the nucleus and induction of G6pc mRNA and repression of GK mRNA by the GKA treatment [7]. A key issue is why repression of GK protein was not observed in animal models exposed chronically to GKAs [19,20]. One possibility is that they were fed high-fat diets [19,20] and that cell phosphometabolites exceed the threshold for ChREBP activation by the combination of a high-carbohydrate load and the GKA.

Liver GK activity is low in human type 2 diabetes [3]. Restoration of this low activity to normal by acute activation was the rationale for GKA therapy in diabetes [1,2]. Liver GK gene expression is regulated by insulin and glucagon, which induce and repress gene transcription, respectively [3]. Accordingly, the simplest explanation for the low GK activity in type 2 diabetes is that it is due to a low insulin-to-glucagon ratio if there is failure of insulin to suppress glucagon secretion within the pancreatic islets, resulting in relative glucagon excess. Such a mechanism would be expected to be rectified by a GKA, which would increase insulin secretion in the pancreatic β-cells by causing a left-shift in the response to glucose. However another explanation for the low GK activity in diabetes is repression of the GK gene by raised G6P and down-stream phosphate-ester intermediates of glucose metabolism [14,15], because of either raised blood glucose or excess dietary carbohydrate [17], such a mechanism would be enhanced by a GKA which further raises G6P the product of the GK reaction. Insofar that GKAs are predicted to have two opposing effects on transcription of the liver GK gene, a positive effect through stimulation of insulin secretion and a negative effect through an increase in hepatic G6P and downstream metabolites, then loss of glycemic control by the GKA would be expected to manifest as a late event when the inhibitory component resulting from gene repression by elevated metabolic intermediates becomes dominant over the acute activation of GK flux and increased insulin secretion.

Three factors need to be considered for the loss in efficacy of a GKA, if this occurs through repression of liver GK: i) the GKA itself; ii) the carbohydrate content of the human diet; and iii) underlying predisposing factors linked to either common gene variants or the progression of type 2 diabetes, which may compromise intrahepatic phosphometabolite homeostasis. The extent by which a GKA would elevate metabolites involved in gene regulation such as fructose 2,6-bisphosphate would be in part compound-dependent because diverse GKAs may have different effects on the interaction between GK with PFKFB2, which generates fructose 2,6-bisphosphate [1,3]. Hepatocyte assays testing for GK gene expression in conditions of elevated glucose could help select for GKAs with minimal effects on GK gene repression. Noninvasive robust assays to monitor liver GK activity in man during chronic therapy with GKAs would be very valuable to establish mechanisms for the decline in efficacy in man. If GKA efficacy in man declines on a high-carbohydrate diet but not on a low-carbohydrate diet, then very important lessons may yet emerge from the GKA studies on the appropriate dietary carbohydrate content for type 2 diabetes, a disease due to compromised glucose homeostasis. The loss in efficacy of GKAs during chronic therapy in man was unexpected, possibly because of insufficient knowledge at the time, on the mechanisms that maintain intrahepatic homeostasis of phosphometabolites such as G6P and downstream intermediates [17]. Addressing this gap in knowledge is crucial for the development of therapies for type 2 diabetes with better chronic efficacy.

Declaration of interest

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Notes