Carbohydrate Metabolism and the Skeleton: Picking a Bone with the β-Cell

Under-carboxylated osteocalcin regulates insulin synthesis, secretion and action in mice; the data in humans are less compelling but require further careful assessment.

Abstract

Novel data show that the bone protein, undercarboxylated osteocalcin, increases islet cell growth and insulin and adiponectin synthesis and enhances glucose metabolism in mice. Human studies, in which undercarboxylated osteocalcin concentrations have been perturbed, show no effect (or a very minor effect) on fasting glucose or insulin concentrations. More extensive and detailed experiments in humans are needed to definitively establish the role of undercarboxylated osteocalcin in human carbohydrate metabolism.

Diabetes mellitus is prevalent (7.8% of the U.S. population) and is responsible for significant morbidity, mortality, and health care costs related to macro- and microvascular complications that occur frequently among patients with the disease (1). Although conventional treatment with diet, oral hypoglycemic agents, and insulin is effective in most patients, new therapeutic interventions that reduce glucose concentrations in the fasting or postcibal state would be of value. Recent findings demonstrating the role of the osteoblast-specific protein, osteocalcin (OC), in regulating glucose homeostasis in mice have attracted considerable interest because they have uncovered a new homeostatic mechanism linking bone to the regulation of pancreatic β-cell function and because they suggest potential new ways in which to regulate glucose concentrations.

In Mice, Bone-Derived Proteins Influence Insulin and Glucose Physiology, and Insulin Receptor Signaling Alters the Release of Proteins Mediating Islet Cell Function

Deletion of the bone- and testis-expressed mouse ptprv (also known as esp) gene results in hypoglycemia, hyperinsulinemia, excessive insulin sensitivity, and reduced adiposity in mutant mice (2). This phenotype is also observed when the ptprv gene is exclusively deleted in osteoblasts, thereby showing that an osteoblast-expressed protein can influence insulin and glucose homeostasis. Interestingly, mice in which the OC (ocn) gene is deleted have the opposite phenotype of that observed in ptprv^−/− mice, and deletion of one allele of the ocn gene corrects the abnormal phenotype seen in ptprv^−/− mice. Coculture of normal osteoblasts with either islets or adipocytes increases insulin (ins1 and ins2) and adiponectin (adipoq) gene expression, respectively, thus demonstrating that osteoblasts alter gene expression in islets and fat cells. Osteoblasts from ptprv^−/− mice increase the expression of ins and adipoq genes more than normal osteoblasts in the coculture experiments, whereas osteoblasts from ocn^−/− mice do not elicit an increase in gene expression. It is possible that ptprv could function by altering OC action by modifying posttranslational γ-carboxylation of OC.

OC is an osteoblast-specific secreted protein that forms 1–2% of bone matrix protein (3). It has three γ-carboxyglutamic residues (one of which is partially modified) that bind calcium. OC γ-carboxylation is vitamin K-dependent and is inhibited by warfarin. Indeed, osteoblasts pretreated with warfarin [which increases the amount of undercarboxylated OC (ucOC)] enhanced adiponectin release from islets. Recombinant uncarboxylated OC enhances ins1 and ins2 expression in β-cells, increases islet cell proliferation, increases adiponectin release from fat cells, and improves glucose tolerance in vivo (2, 4). The precise mechanism by which uncarboxylated OC mediates these effects is uncertain because a receptor for uncarboxylated OC has not been identified in islets or in fat cells. These data are consistent with a role for ucOC in increasing insulin and adiponectin release and improving glucose homeostasis in mice.

Moreover, insulin signaling through its receptor regulates the release of ucOC from bone (5, 6); mice in which the insulin receptor has been ablated in bone have reduced bone resorption, reduced postnatal bone formation, decreased osteoprotegerin (an antagonist of receptor activator of nuclear factor-κB ligand), and low circulating ucOC (5, 6). Additionally, these mice have increased adiposity and decreased islet cell function and area (5, 6). Thus, not only does a bone protein (ucOC) influence islet cell function and insulin release, but the insulin receptor itself appears to influence the release of ucOC in rodent models.

The Role of Such a Homeostatic Bone-Pancreas-Glucose System in Humans Has Not Been Carefully Examined

In humans, a consistent relationship between ucOC and glucose homeostasis has not been shown (7–11). In one report, vitamin K (which increases γ-carboxylation of proteins and reduces the amount of ucOC in serum) administered for 3 yr altered homeostasis model assessment-insulin resistance (HOMA-IR) in men but not women after adjustment for baseline HOMA-IR, body mass index, and body weight change (10). OC and ucOC, however, were not measured. In a recent double-blind, placebo-controlled study, we examined the effect of vitamin K1 (phylloquinone, 1 mg/d vs. placebo) treatment on HOMA-IR in postmenopausal women and observed a 200% decrease in ucOC at 6 and 12 months of treatment with phylloquinone but no change in fasting serum insulin and glucose concentrations, and consequently in HOMA-IR (11). There was no relationship between HOMA-IR and serum ucOC concentrations either at baseline or at 12 months. We did not, however, test the influence of reduced ucOC concentrations on postprandial carbohydrate metabolism, and we did not examine the influence of increasing ucOC (as occurs after the administration of warfarin) on glucose and insulin homeostasis before and after meals. Warfarin, which blocks vitamin K-dependent γ-carboxylation, and thereby leads to a significant rise in ucOC concentrations, has not been reported to alter glucose homeostasis. Given the widespread use of warfarin and the absence of reported hypoglycemia, the relationship between changes in OC γ-carboxylation and glucose homeostasis appears to be absent or, at best, very modest. A case report implicating warfarin therapy in the regulation of blood glucose concentrations (5) merely shows that increased warfarin absorption and a consequent change in the prothrombin time occurred in a patient taking both warfarin and acarbose (a hypoglycemic agent) (12).

Unanswered Questions and Future Directions

Given the novelty of these findings, their importance to the area of diabetes research, and the difference between the observations in mice and humans, we feel that the area needs further investigation to establish its relevance to human physiology and disease. In the realm of human carbohydrate physiology, it would be important to measure OC and ucOC serum concentrations before and after a carbohydrate-containing meal and after an oral glucose load to determine whether concentrations or ratios of ucOC and OC change in response to postcibal changes in blood glucose or other dietary cues. A comparison between the changes in ucOC serum concentrations after an oral vs. an iv glucose load would help discern the role of gut-derived and portal factors in mediating changes in ucOC concentrations. “Insulin clamp” experiments (in which glucose concentrations are varied in the presence of fixed insulin concentrations) would establish the role of glucose as the signal in the release of ucOC independent of insulin. “Glucose clamp” experiments (in which insulin concentrations are varied in the presence of fixed glucose concentrations) would establish a role for insulin in ucOC release. Knowledge of the manner in which ucOC signals in islets and adipocytes (a receptor-dependent or -independent mechanism) might allow the development of novel small molecules that might increase insulin and adiponectin synthesis and release.

It would be helpful to know the composition of the ucOC fraction in blood. OC has several glutamic residues, three of which are subject to γ-carboxylation. Are all ucOCs alike in terms of their bioactivity in islets and adipocytes? In humans it would be helpful to know whether a decrease or increase in serum ucOC as a result of therapy with phylloquinone or warfarin alters model-based, quantitative measures of insulin secretion and action in both the fasting and postprandial state. Such measures are necessary to account for the compartmental kinetics of insulin secretion and hepatic extraction of insulin. If ucOC is important in carbohydrate metabolism, we would expect changes in circulating ucOC concentrations induced by phylloquinone or warfarin administration to be associated with altered insulin and adiponectin concentrations and changes in glucose disposal and endogenous glucose production after a meal. In normal humans, it would be important to assess whether the infusion of increasing amounts of insulin (or glucose independent of insulin) alters OC and ucOC concentrations in normal humans.

Finally, it would be important to ascertain whether a change (decrease or increase) of bone resorption alters ucOC serum concentrations and whether such a change is associated with alterations in carbohydrate metabolism in the fasting state and after the administration of a meal. ucOC concentrations were lower in women receiving antiresorptive therapy for osteoporosis in one study (13), but no information was provided by the authors regarding glucose metabolism. An article by Maugeri et al. (14) showed that insulin consumption decreased after therapy with alendronate in women with a diagnosis of osteoporosis and diabetes mellitus—an effect that is opposite to that predicted by mouse experiments where elevated ucOC had a salutary effect on glucose metabolism.

A careful analysis of the role of OC and ucOC in carbohydrate metabolism is urgently needed in humans because, if present, it will demonstrate a new pathway in the control of carbohydrate metabolism in humans that could be exploited to develop new methods for the treatment of diabetes mellitus.