Roles for osteocalcin in brain signalling: implications in cognition- and motor-related disorders

One of the recent important discoveries on bone biology is that bone can act as an endocrine organ [1]. It is an organ with a delicate structure that contains various cells of mesenchymal origin, such as the osteoblasts, chondrocytes, bone marrow stromal cells and adipocytes [2]. Among these, osteoblasts are the main type of bone cells that regulate the formation of bones [3], wherein osteocalcin (OCN) acts as a marker for osteoblast activity and bone formation [2]. OCN is a non-collagenous, vitamin K-dependent protein that contains three gamma-carboxyglutamic acid (Gla) motifs, which in the presence of calcium, facilitates mineral deposition and bone remodeling [2, 3]. OCN also acts as a regulator of the activity of osteoclasts and their precursors [3]. However, while studying the role of OCN in bone mineralization, a more striking observation on the role of OCN in fat mass regulation was found, which eventually led to the discovery of OCN in energy metabolism [1]. Lee et al. (2007) [4] for the first time highlighted the roles of OCN in glucose metabolism by using OCN knockout (Ocn^−/−) mouse models, which showed elevated levels of hyperglycemia and glucose intolerance, decreased β-cell function and insulin secretion, decreased insulin sensitivity and adiponectin expression, and increased fat mass and serum triglyceride level. Subsequent studies have further uncovered the role of OCN in regulating male fertility and exercise adaption, further expanding the functions of OCN in endocrine regulation. Lower circulating testerone levels, decreased testis size and weight, reduced number of spermatocytes and increased germ cell apoptosis were found in Ocn^−/− and osteoblast-specific osteocalcin-deficient (Ocn_osb^−/−) male mice [5], while 3 month-old Ocn^−/− and Ocn_osb^−/− female mice showed reduced exercise capacity compared with WT mice [6].

Mounting evidence from animal studies has demonstrated that OCN, especially in its metabolic active form, i.e. the uncarboxylated OCN, can act on the pancreas, adipose tissue, male gonads and muscle to stimulate insulin secretion, improve insulin sensitivity, male fertility and muscle power through its peripheral receptor G protein-coupled receptor 6a (Gprc6a) [6‐9]. Mice lacking Gprc6a (Gprc6a^−/−) replicate many metabolic and fertile phenotypes of Ocn^−/− mice. For example, mice lacking Gprc6a in the β-cell lineage are glucose intolerant due to an impaired ability to produce insulin, the regulation of which occurs both in the prenatal β-cell proliferation and during adulthood [7]. Its depletion in leydig cells or myofibers of mice results in the reduced testes size and weight, epididymis and seminal vesicle weights, sperm counts and circulating testosterone levels [5], and a decreased exercise capacity that is of equal severity as seen in Ocn^−/− mice [6]. All such evidence indicates the importance of OCN in maintaining the metabolic homeostasis and normal fertility and exercise capacity, providing a basis for its translational potiential in the related metabolic, reproductive and movement disorders.

In addition to these above-mentioned peripheral effects, OCN also has been found to have unexpected central roles in brain development and cognitive functions [10]. OCN, mainly in its uncarboxylated form, can pass through the blood-brain barrier (BBB) and accumulate in the brainstem, thalamus and hypothalamus, and can bind specifically with neurons in these areas influencing various neurotransmitters synthesis and signaling. In the brainstem, OCN is detected to bind in the dorsal and median raphe nuclei that contain serotonergic neurons, thereby increasing serotonin synthesis, inducing calcium flux and activating action potential (AP) frequency. Besides, both γ-aminobutyric acid (GABA) synthesis and AP frequency of GABAnergic interneurons are decreased after OCN treatment. In the midbrain, OCN binds to neurons in the ventral tegmental area (VTA) and facilitates the synthesis of dopamine. In the hippocampus, OCN binds to a Gpr158/Ga_q complex in the neurons of CA3 region and functions in part by inositol 1,4,5-trisphosphate (IP3) and brain-derived neurotrophic factor (BDNF) [11]. A recent study found that the histone-binding protein RbAp48, a molecular determinant of age-related memory loss, could control the beneficial actions of OCN by regulating the expression of Gpr158 and BDNF in the mouse hippocampus [12]. Upon binding, OCN can signal in these neurons to influence the expression of genes that regulate neurotransmitter synthesis. Ex vivo and in vitro experiments confirmed that OCN could act on neurons to decrease the expression of enzyme involving in GABA biosynthesis glutamate decarboxylase 1 (Gad1), increase tryptophan hydroxylase 2 (Tph2) and tyrosine hydroxylase (Th) expression in the brainstem and midbrain explants, which are the key enzymes responsible for the synthesis of serotonin, and dopamine as well as norepinephrine, respectively. An animal study again revealed that when OCN was administered through the intracerebroventricular (ICV) route to Ocn^−/− mice, without leakage to peripheral circulation, it normalized Tph2, Th, Gad1 and Gad2 expression. These results provide the physiological bases for OCN to act directly on the brain [10].

It was obvious that the phenotypes displayed by Ocn^−/− mice were similar to that of clinical T2DM, and the beneficial effects on energy metabolism through exogenous administration of OCN in both cellular and rodent models potentiate its therapeutic possibilities in T2DM and other related metabolic diseases. Daily injections of recombinant OCN at either 3 or 30 ng/g/day could improve the glucose tolerance and insulin sensitivity in mice through increased β-cell mass and insulin secretion [13]. Moreover, ex vivo experiments showed that 1 ng/ml and 3 ng/ml undercarboxylated OCN, but not its carboxylated form, significantly stimulated the expression of cyclin D1 and insulin in the islets as well as the expression of adiponectin, one of the major adipokines that facilitates insulin sensitivity, in adipocytes [4]. A later study by the same team found that OCN intervention not only ameliorated glucose metabolism impairements in Ocn^−/− mice, but also regulated β-cell gene expressions and significantly alleviated the deleterious effect on body mass and glucose metabolism of gold thioglucose-induced hyperphagia and high-fat diet in WT mice [14]. More importantly, OCN could not only regulate β cells function in rodents, but also promote the proliferation and differentiation of β cells and thus enchance insulin sensitivity in cultured human islets at a dose ranging from 1.0–15 ng/ml [15]. Besides the direct regulation on β cells, intraperitoneal administration at 7 μg/kg or oral administration at 10 μg/kg of uncarboxylated OCN could also promote the release of glucagon-like peptide-1 (GLP-1) in intestinal epithelial cells via Gprc6a, thus indirectly stimulating insulin secretion [16].

Given the beneficial metabolic effects of OCN found in mice, series of cross-sectional and longitudinal studies have explored its potential associations in humans. A large number of studies demonstrated the negative associations between circulating OCN level and fasting blood glucose level, fasting insulin level, insulin resistance and fat mass, and positive associations with insulin secretion, serum adiponectin level and fat-free mass in normal subjects or in patients with type 1 or type 2 diabetes, gestational diabetes, metabolic syndrome, and polycystic ovarian syndrome [17‐21]. These endocrine roles of OCN have provided translational possibilities for its potential uses as a predictor or therapeutic target of some metabolic diseases. For example, a long-term study showed that serum OCN was an independent risk factor for the development of diabetes in patients who were not diagnosed with diabetes at baseline but developed T2DM during a 10-year follow-up [22]. Mounting evidence also suggests the relationship between serum OCN and cardiovascular risk factors. Higher baseline total OCN concentrations were reported to be associated with lower abdominal arotic calcification progression and lower mortality [23], whereas a low level of serum OCN was a significant predictor of increased cardiovascular disease events [24]. Moreover, roles of OCN have been found in the regulation of atherosclerosis (low level associated with increased event of atherosclerosis) [25], anti-tumor activity [26], and brain functions (discussed in the later parts of the article).

Carrier-mediated glucose uptake with minimal insulin regulation and the fundamental requirement for a glycolytic substrate with minimal intracellular storage make neurons susceptible to glucose toxicity [27]. The brain hippocampus, which is mainly responsible for storage, conversion and orientation of long-term memory, is one of the brain areas with the highest expression of glucose transporter (GLUT) 1 and 4 [28]. Both acute and chronic disturbances in glucose homeostasis can cause transient or permanent neuronal damage and thereby cognitive impairments. As glucose uptake in neurons is insulin-independent, intracellular glucose levels are abnormally elevated during hyperglycemic milieu, a condition in which glucose is oxidized to form free radicals and reactive carbonyls, further activating mitogen-activated protein kinases (MAPK) and thereby altering the cellular phenotypes [27]. These molecular changes finally lead to a series of functional outcomes of diabetes, including nerve conduction deficits [29], pain and allodynia [30], and impaired axonal regeneration [31], which are collectively referred to as glucose neurotoxicity.

Ever since 1854, brain was described as a crucial organ for glucose homeostasis in animal models [32]. Over the past decades, accumulated evidence has demonstrated brain as an insulin-sensitive organ with insulin responses in specific areas including the hypothalamus, fusiform gyrus, prefrontal cortices and the hippocampi [33]. The hippocampi are central to the processing of memory content, especially memories of a declarative nature [34]. Evidence from human studies suggests that intranasal administration of insulin may increase cognitive ability in healthy volunteers [35] and patients with T2DM [36] and Alzheimer’s disease (AD) [37]. Brain insulin resistance is associated with over-weight, peripheral insulin resistance, increased aging processes, maternal metabolic dearrangement and genetic background, which cause not only a reduced capacity to lose weight, weakened feeling of satiety and the exacerbated metabolic diseases such as T2DM [32], but also mediate the possible consequences for cognitive dysfunction and an increased risk of developing AD [37, 38].

T2DM is characterized by insulin resistance and/or relative defects in insulin serection and resulting hyperglycemia. Mounting evidence demonstrated an unequivocal link between T2DM and cognitive decline. T2DM is reported in epidemiological studies to increase a 50–100 % risk of AD and a 100–150 % risk of vascular dementia [39]. In addition, a longinitudial research showed that higher glucose levels in elderly participants were associated with an increased risk of dementia occurring years before the dementia diagnosis [40]. Factors such as endocrine, metabolic and vascular abnormalities contribute to the link between T2DM and cognitive dysfunctions, including ischemic cerebrovascular diseases, glucose neurotoxicity, changes in insulin and amyloid metabolism, increased oxydative stress and inflammatory factors [41]. Chronic hyperglycemia and microvascular diseases cause cognitive dysfunction in both T1DM and T2DM, and both the two disorders are associated with cognitive and motor slowing, decrements of similar magnitude on attention and executive functioning, neuronal growth arrest, increased cortical atrophy and microstructural abnormalities in white matter tracts [42]. Thus, maintaining energy homeostasis is especially important for favoring normal cognitive function and the powerful regulations of OCN on energy metabolism as stated above provide the basis for its indirect regulations on cognition.

Recent studies have demonstrated putative roles of OCN in the regulation of cognition. To test the central effects of OCN on behaviors and neurotransmitters synthesis, 3 female mice models were generated: Ocn^−/−, Ocn_osb^ERT2 (a mice model with half of serum OCN levels after tamoxifen treatment), and Gprc6a^−/− mice. The first two mouse models demonstrated significant less locomotor activity, increased anxiety- and depression-like behaviors as compared with controls. These passivity behaviors were considered to be caused by increased GABA biosynthesis in the brain due to the facts that the GABA content in the brain, the expression levels of Gad1 and Gad2 in Ocn^−/− mice and tamoxifen-treated Ocn_osb^ERT2 mice were elevated. Other neurotransmitters, such as brain serotonin, norepinephrine and dopamine contents were also decreased in the midbrain, cortex and striatum of the OCN-mutant mice, along with the decreased Tph2 and Th expression. However, all these behaviors, the contents of neurotransmitters, and the expressions of the genes responsible for their synthesis in the brain remained normal in Gprc6a^−/− mice. These findings indicated that independent of its known peripheral receptor Gprc6a, OCN impacts directly or through an undetermined second receptor on mice behaviors and neurotransmitters production [10].

As stated above, 10 ng/h OCN delivered via ICV infusion could reverse the anxiety- and depression-like behaviors fully, and partially the spatial learning and memory deficit in Ocn^−/− mice. Such a favorable function of OCN on learning and memory can also be exerted postnatally and is critical for fetal brain development [10]. As compared with Ocn^−/− mice or even with Ocn^+/− mice, tamoxifen-treated Ocn_osb^ERT2 mice displayed less affected spatial learning and memory capacity, and kept an almost intact corpus callosum and hippocampal dentate gyrus structure. It was found that circulating OCN was detectable 2 days (E14.5) before the expression of this gene in the developing skeleton, suggesting that maternal OCN can reach fetal blood circulation. This hypothesis was proved to be true, that all the OCN found in an embryo before E18.5 comes from maternal origin. Further experiments confirmed that only the maternal uncarboxylated OCN could cross placenta efficiently, enter the fetal blood stream, and prevent fetal neuronal apoptosis in the hippocampus. The beneficial role of maternal OCN on behavior, learning and memory were then established in their adult offspring [10].

In another study, it was described that systemic exposure to young blood counter-balanced the age-related decline in cognitive function in old mice [43]. However, it was later shown that when the plasma from 3-month-old Ocn^−/− mice or young mice plasma with depletion of OCN by its antibody was injected into 16-month-old wild-type mice, no improvement of cognitive performance was observed. Instead, the plasma from 3-month-old wild-type mice or plasma from young Ocn^−/− mice with the addition of recombinant uncarboxylated OCN, so called “spiked plasma”, could elicit the improvement in cognitive function and alleviate the anxiety-like behaviors in old mice [11]. Moreover, long-term (2 months) peripheral delivery of OCN in old mice also fully recovered the cognitive deficits, possibly through the stimulation of BDNF expression in the hippocampi of old mice, thus providing a direct evidence that OCN is important for maintaining normal cognitive function [11].

Then it comes an immediate question, which receptor mediates this central functions of OCN? It is apparently not Gprc6a, because the passivity of Ocn^−/− mice was not observed in Gprc6a^−/− mice [10]. With the use of 3 criteria as a filter, 1) same as Gprc6a, this new receptor should belong to a G protein-coupled receptor; 2) the receptor should be present in the CA3 region of the hippocampus where it had been previously shown to bind with OCN [10]; and 3) the receptor should not be expressed in any cells where Gprc6a mediated OCN signaling, a new receptor, Gpr158, for OCN was identified [11]. A series of subsequent experiments in Gpr158^−/− mice, in shRNA-Gpr158 -treated mice, which had more than 60 % decrease in Gpr158 protein level, and in compound heterozygous mice (Gpr158^+/−; OCN^+/−) confirmed that it is Gpr158 which mediates the regulatory effect of OCN on hippocampal-dependent memory and anxiety-like behaviors [11]. It is noteworthy that Gpr158 is not expressed in serotonergic neurons of the dorsal and medial raphe of the brainstem where OCN binds, suggesting the possibility of the existence of a third receptor for OCN.

The link between the endocrine functions of osteoblasts and their brain functions were further strengthened with the evidence that mice with haplo-insufficient for Runx2, a master gene responsible for osteoblast differentiation and main regulator for OCN expression, also had deficiency in spatial learning and hippocampal memory, and increased anxiety-like behaviors [44]. Therefore, these series of work, for the first time suggested the link between bone, or more specifically, the OCN and brain cognitive function.

SIRT1 is one of the seven sirtuins that act as NAD⁺-dependent histone deacetylase. Resveratrol is a naturally occurring compound, mainly found in red wine obtained from grapes, that mediates its effects on human health through the activation of SIRT1. In the recent years, many studies have confirmed that intake of red wine or direct treatment with isolated resveratrol in humans as well as laboratory animals improve cognitive functions. The compound resveratrol is uptaken by the BBB and is well tolerated in humans [60]. In rat models of vascular dementia, resveratrol has been shown to improve cognition through reduction of oxidative stress [61]. With a 52-week treatment of resveratrol in mild-moderate patients of AD, it was implicated that the treatment could be beneficial for prevention of neurodegenerative disorders through modulation of neuroinflammation [62]. However, a recent meta-analysis revealed that resveratrol might be only useful for elevating mood, but not for improving cognition [63]. This could be possible due to an issue related to the bioavailability of resveratrol, and therefore better derivatives of resveratrol such as pterostilbene, or improved drug delivery tools for the compound might be needed [64, 65]. Nevertheless, the current studies still support the notion that there might be a putative role for this compound on improving cognitive functions, which could be confirmed by long-term prospective studies with resveratrol derivatives with better bioavailability and/or efficacy. Now, the question remains whether the action of resveratrol on cognition has anything to do with OCN or not. SIRT1 activation by resveratrol has been reported be associated with OCN-mediated bone function. In rat calvarial osteoblast-like (ROB) cells, it was found that resveratrol could increase mRNA expression of OCN [66]. In multiple myeloma disease, wherein bone formation is dcreased, resveratrol could induce bone formation through inducement of OCN expression in human bone marrow mesenchymal stem cells and by decreasing nuclear translocation of the NF-κB [67], the transcription factor that is often reported to be inducing the pro-inflammatory cytokines in neuroinflammatory diseases [68]. However, opposing results have also been obtained from other studies [69]. Moreover, there is no direct evidence on whether uncarboxylyted OCN is correlated with the effect of SIRT1. Nevertheless, resveratrol has been found in many studies and has been suggested to be beneficial for bone formation through mechanisms related to OCN [70]. If we summarize the role of SIRT1 on the regulation of OCN and cognition, we can find several common pathways that are in common, e.g., improved insulin sensititvity, improved glucose metabolism, reduced oxidative stress and inflammation through the suppression of NF-κB. Therefore, it is highly possible that SIRT1 may play a crucial role in both bone and cognitive health, which could be complementary to each other, through the improvement of these pathways that involves a common regulatory hormone, i.e. OCN (Fig. 2).

Metformin, which is used primarily as an anti-diabetic therapeutic, has recently been studied largely in many other diseases, including those of in the brain and the bones. Given the role of diabetes in the development of age-related cognitive decline, many anti-diabetic medications, especially metformin, recently have been tested for their efficacy on cognitive improvement [71]. In a longitudinal follow-up study in diabetic patients, long-term use of metformin was found to decrease the risk of developing cognitive decline [72]. Similarly, in the Diabetes Prevention Program (DPP), metformin-treated subjects were found to be at lower risk of developing cognitive malfunction compared to the placebo group. Moreover, the treatment was also found to be cognitively safe, which was, however, found otherwise in other studies [73]. In animal models of cognitive impairments and dementia, metformin has been shown to be fruitful through a number of factors. For example, in a recent study on mice, metformin was shown to alleviate spatial memory loss through an enhanced number of post-mitotic NeuN⁺ neurons (i.e. enhanced neurogenesis) [74]. Another study showed that metformin, either alone or in combination with ursolic acid, ameliorated stress-induced changes in behavioral changes, accompanied with insulin sensitivity, inflammatory and oxidative changes in mice [75]. Interestingly, metformin has also been implicated to be beneficial in PD-related mouse model [76]. Insulin sensitivity, which is an important factor in regulation of cognition, can be increased by higher OCN level in clinical diabetic patients [77, 78]. Now the question remains whether there could be a putative link between such beneficial effects of metformin and OCN signaling in cognition. Interestingly, OCN and its related bone diseases have been shown to be affected by metformin. Liu et al. (2018) found that in a model of ketogenic diet-induced osteoporosis, metformin could attenuate bone loss which correlated with increased level of OCN [79]. The drug has also been shown to improve osteogenic functions of adipose-derived stromal cells, which show increased bone regeneration capability through increasing expression of OCN, via a mechanism related to AMPK [80], which is a well known regulator of brain energy metabolism [81] that regulates both cognition and motor coordination [82]. AMPK is known to inhibit NF-κB signaling and inflammation [83], which are related to cognitive and motor impairments, as discussed already before. Moreover, it is of an interesting note that metformin is effective in regulation of behavior through the upregulation of BDNF [76, 84], a brain neurotrophic factor that is related to the beneficial actions of OCN in age-related memory loss [11, 12]. These findings suggest that there might be a putative link between metformin therapy and brain functions wherein OCN may act as a facilitator of the improved cognitive functions by metformin through improvement of neurotrophic signaling and energy metabolism, and through modulation of inflammatory reactions (Fig. 2). However, it should be noted that metformin therapy has also been found to be adverse or ineffective for cognitive functions [85] as well as for the regulation of OCN levels [86]. Therefore, more studies with acute and chronic treatment of the drug in models of cognitive loss and/or its relation with OCN would be needed to understand the role of metformin on OCN/cognition regulatory axis.

A role for OCN in exercise-induced cognitive improvement may be implicated as well. Physical exercise intervention is a well known preventive measure for dementia and aging [87]. Both in young and aged subjects, exercise mediates beneficial effects on cognitive functions regardless of their ages [88]. Using a multi-modal MRI technique, Den ouden et al. (2018) found that aerobic exercise training for 12-week increased left hippocampal volume and facilitated immediate verbal recall performance in humans [89]. Cognition-related disorders, such as AD, are known to be affected by lowered insulin sensitivity and increased inflammatory load in the brain; presence of diseases such as obesity and metabolic syndrome gives rise to systemic inflammation which eventually passes through the BBB. In such scenario, pro-inflammatory cytokines such as TNF-α and IL-1β turn to be dominative while expression of anti-inflammatory cytokines such as IL-4 and IL-10 are decreased [90]. Therefore, shifting the load toward anti-inflammatory dominance from pro-inflammatory dominance may be an useful strategy for improving cognitive function and other brain processes [91]. Now, factors that can improve metabolic status in the brain and muscle tissues may reduce such inflammation. Interestingly, IL-6, which possesses both pro- and anti-inflammatory properties [92], has been found to be related to mediating exercise-induced OCN-mediated energy utilization in muscle [6]. The investigating team on this found that OCN not only could improve the muscle utilization of glucose and fatty acids, which are important factors in controlling cognition and behavior [93], but it could also induce secretion of IL-6 from the muscle fibers [6]. IL-6 is a known regulator of energy and glucose homeostasis that in obese mice increases energy sensitivity through a mechanism related to glycoprotein 130 (gp130) [94], which is a regulator of OCN level in osteoblasts [95]. This could be another plausible explanation of why obese animals show low cognitive performance, and interventional strategies such as exercise can improve cognitive ability in obese animals [96]; it might be possible through an exercise-induced activation of the IL-6/gp130/OCN axis that in turn improves energy homeostasis and thereby improves cognition (Fig. 2).