Brain glucose transporters, O-GlcNAcylation and phosphorylation of tau in diabetes and Alzheimer’s disease

Epidemiological studies have suggested that type 2 diabetes mellitus (T2DM) is a risk factor for Alzheimer’s disease (AD), the most devastating progressive neurodegenerative disease that causes dementia and eventually death of the affected individuals. T2DM is characterized by insulin resistance and the resultant metabolic abnormalities in the periphery. It was previously thought that insulin did not regulate glucose metabolism in the central nervous system. However, recent studies indicate that insulin signaling not only regulates glucose metabolism in the brain, but also plays important regulatory roles in neural development and neuronal activities and affects learning and memory (Gerozissis 2008).

Abnormal hyperphosphorylation of microtubule-associated protein tau causes neurodegeneration and the formation of neurofibrillary tangles, a hallmark brain lesion of AD (Wang and Liu 2008; Iqbal et al. 2009). We recently found that altered O-GlcNAcylation, a unique type of O-linked glycosylation of nucleocytoplasmic proteins by a monosaccharide β-N-acetylglucosamine (O-GlcNAc), of tau links the impairment of brain glucose metabolism to hyperphosphorylation of tau (Liu et al. 2004; Li et al. 2006). On the basis of these findings, we hypothesize that the impairment of brain glucose metabolism contributes to neurodegeneration via down-regulation of O-GlcNAcylation, which is regulated by glucose metabolism, and the resultant abnormal hyperphosphorylation of tau (Gong et al. 2006). This hypothesis is further supported by our recent studies showing that the decreases in GLUT1 and GLUT3, the two major brain glucose transporters (GLUT) responsible for the transport of glucose from blood into the neuron, correlate to the decrease in protein O-GlcNAcylation and hyperphosphorylation of tau in AD brain (Liu et al. 2008). It is possible that T2DM contributes to the increased risk for AD by the same mechanism, i.e., by decreased brain glucose metabolism and O-GlcNAcylation and the resultant abnormal hyperphosphorylation of tau. However, whether such type of changes occurs in T2DM brain has not been investigated.

In this study, we determined the levels of major brain GLUTs, protein O-GlcNAcylation and tau phosphorylation in the postmortem brains of individuals with T2DM and compared them with those in AD brain. We found that the neuronal GLUT3 was decreased to a bigger extent in T2DM brain than in AD brain and that O-GlcNAcylation was also decreased in T2DM brain as seen in AD brain. We further observed that tau was also abnormally hyperphosphorylated at several sites in T2DM brain. These findings provide new insight into the likely mechanism by which T2DM increases the risk for AD.

Materials and methods

Results

Neuronal GLUT3 is decreased in diabetic brains

We first determined the levels of three major brain GLUTs (GLUT1–3) in 11 T2DM and seven age- and postmortem delay-matched control brains by quantitative western blots. We found that the level of GLUT3 was significantly decreased in T2DM brains as compared with controls (Fig. 1), whereas there was no significant difference in GLUT1 and GLUT2 levels between the two groups. To investigate whether the decrease in neuronal GLUT3 results from down-regulation of HIF-1α, the major transcription factor regulating expression of GLUTs, in T2DM brain, we determined its level by quantitative western blots. However, no significant difference in the level of HIF-1α was found between T2DM and control brains (Fig. 1a and c).

Comparison of brain GLUTs between diabetes and AD

We previously reported down-regulation of GLUTs in AD brain (Liu et al. 2008). We thus compared the levels of GLUTs in T2DM brain with those in AD brain as well as in brains of individuals who suffered from both T2DM and AD (T2DM-AD). We found that GLUT3, which is the major neuronal GLUT, is decreased markedly in both T2DM and AD, whereas GLUT1 and GLUT2 levels were different in T2DM brain from those in AD brain (Fig. 2). GLUT1, which is localized in the endothelial cells of the blood-brain barrier, was decreased only in AD, but not in T2DM brain. GLUT2 level was unchanged in T2DM brain, but was increased dramatically in AD brain. In the T2DM-AD group, the GLUT1 and GLUT3 levels were similar to those of T2DM group, whereas the GLUT2 level was indistinguishable from that of the AD group.

O-GlcNAcylation of global proteins and of tau is down-regulated in diabetes and AD brains

Deficient glucose transport to the brain in T2DM and AD could have led to a decrease in protein O-GlcNAcylation. Therefore, we determined the level of protein O-GlcNAcylation by quantitative immuno-dot-blots and western blots developed with the monoclonal antibody RL2 against O-GlcNAc. We found that the protein O-GlcNAcylation level in the brain extracts of T2DM, AD and T2DM-AD cases was reduced to approximately half of the control group (Fig. 3a). To learn if the level of O-GlcNAcylation of tau protein was also decreased similarly, we purified tau from human brains by an immuno-affinity purification method and then determined the O-GlcNAcylation level of the purified tau preparations. We found that tau O-GlcNAcylation level was significantly decreased in T2DM as compared with controls (Fig. 3b). The decrease was in a larger degree in AD and T2DM-AD groups, which reached to ∼50% of the controls.

Tau phosphorylation is increased in diabetic and in AD brains

To investigate if the decrease in GLUT3 and O-GlcNAcylation could contribute to tau pathology, we compared tau phosphorylation levels at several known major phosphorylation sites in the brain between T2DM and controls by using quantitative western blots developed with phosphorylation-dependent and site-specific tau antibodies. We found that, among 10 phosphorylation sites studied, the mean tau phosphorylation levels at eight of these sites (except Thr205 and Thr231) were higher in T2DM than in control cases (Fig. 4). The increase in tau phosphorylation at four (Ser202, Thr217, Ser262 and Ser396) of these eight sites reached statistical significance. These results are consistent with the higher tangle scores and Braak stages in T2DM brain than in control brains (Table 1). Although the extents of tau phosphorylation in T2DM brains were much smaller than in AD brains, these results indicate a general increase in tau phosphorylation in T2DM.

We also studied the levels of site-specific tau phosphorylation in brains of individuals with AD alone and with T2DM plus AD (T2DM-AD). As previously demonstrated (Yu et al. 2009), tau was found to be hyperphosphorylated at all the 10 sites examined in AD brain. Although tau was markedly hyperphosphorylated in both AD and T2DM-AD groups, the levels of the site-specific phosphorylation of tau were different between the two groups. As compared to AD brain, tau phosphorylation levels at sites located at the N-terminal half of tau (Thr181, Ser199, Ser202, Thr205, Thr217 and Thr231) were found to be much lower, whereas those at the C-terminal half (Ser262, Ser396, Ser404 and Ser422) were much higher in T2DM-AD than in AD brains (Fig. 4).

Discussion

We have recently demonstrated that O-GlcNAcylation regulates phosphorylation of tau negatively and that impaired brain glucose metabolism leads to hyperphosphorylation of tau via down-regulation of O-GlcNAcylation (Liu et al. 2004; Li et al. 2006). These findings led to a new hypothesis of a possible mechanism by which the impaired brain glucose metabolism contributes to neurodegeneration in AD (Gong et al. 2006). Because T2DM, which is characterized by impairment in the utilization of glucose, increases the risk for AD, it is interesting to explore whether T2DM could contribute to the increased risk for AD by the same or similar mechanism, i.e., via down-regulation of O-GlcNAcylation and thus abnormal hyperphosphorylation of tau. In this study, we found, for the first time, that GLUT3 and tau O-GlcNAcylation were indeed decreased and tau phosphorylation was increased in T2DM brain. These findings provide direct evidence that supports our hypothesis. It is in consistent to our hypothesis that some diabetic changes resulted from insulin resistance, such as the Maillard-reaction-related modifications and protein glycation, had been observed in AD brain (Smith et al. 1994, 1996; Yan et al. 1995). Because insulin signaling is impaired in AD brain (de la Monte and Wands 2005; Steen et al. 2005; Zhu et al. 2005), there are similar abnormalities in the brains between T2DM and AD. Actually, AD was recently proposed to be the type 3 diabetes by de la Monte and her colleagues (Steen et al. 2005).

Brain neurons are unable to synthesize or store glucose, and thus they are fully dependent on glucose transport across the blood-brain barrier, which is facilitated by GLUTs. In mammalian brain, GLUT1 and GLUT3 are the predominant GLUTs responsible for glucose transport (McEwen and Reagan 2004). GLUT1 is highly expressed in the endothelial cells of the blood-brain barrier and is responsible for transporting glucose from blood into the extracellular space of the brain (Qutub and Hunt 2005; Schubert 2005). GLUT3 is the major neuronal GLUT and helps transport glucose from the extracellular space into the neurons (Dwyer et al. 2002). In addition, GLUT2 is expressed in astrocytes of the brain and is responsible to glucose transport into astrocytes. Decreased GLUT1 and GLUT3 have been observed in AD brain (Kalaria and Harik 1989; Simpson et al. 1994; Harr et al. 1995; Mooradian et al. 1997), and this decrease correlates to hyperphosphorylation of tau (Liu et al. 2008). In this study, we found that the GLUT3 level was also decreased in T2DM brain and this decrease was even more marked than that in AD brain, suggesting that the brain glucose uptake/metabolism might be severely compromised in T2DM. It is noted that the GLUT alterations in T2DM brain were different from that in AD brain, in which both GLUT1 and GLUT3 are decreased (Liu et al. 2008). These observations suggest that impaired brain glucose uptake/metabolism in T2DM may result from a mechanism different from AD. Unlike AD brain, in which the decrease of GLUT1 and GLUT3 appears to result from decreased HIF-1α (Liu et al. 2008), the major GLUT-regulating transcription factor (Maxwell et al. 1997; Zelzer et al. 1998; Chen et al. 2001; Hayashi et al. 2004), the level of HIF-1α was found to be unchanged in T2DM brains.

We noticed large variations in GLUT levels in both control and the diseased groups. We examined the variations against the case history and tissue handling, including age, gender, postmortem delay, and the severity of disease. We did not find any significant factors within our cohort which could explain the individual variations. Large individual variations of proteins, including GLUTs (Kalaria and Harik 1989; Mooradian et al. 1997), are actually common in human brain samples, probably because of variable individual genetic, metabolic, and environmental backgrounds. Nevertheless, with our sample size of 7–11 cases per group, we observed a decreased level of GLUT3 in T2DM brain and of both GLUT1 and GLUT3 in AD brain with statistical significance.

In contrast to GLUT1 and GLUT3, the level of GLUT2 in AD brain was found to be markedly increased, which is consistent to our previous report (Liu et al. 2008). The increase in GLUT2 in AD brain does not appear to indicate an increased expression of GLUT2 in astrocytes, because this increase disappears after the level of GLUT2 is normalized by the level of the astrocyte marker, glial fibrillary acidic protein (Liu et al. 2008). Thus, the increase of GLUT2 level in AD brain homogenates is most likely due simply to astrocyte over-activation, which is a well known phenomenon in AD brain (Skaper 2007). Instead, we speculate an impaired glucose uptake into astrocytes in AD brain, because this process requires both GLUT1, which transports glucose from the blood stream into the extracellular space and is decreased in AD brain, and GLUT2, which transports glucose from the extracellular space into astrocytes. In the present study, we did not find any significant change in GLUT2 level in T2DM brain, as compared to control brains. These results are consistent with our observation of no astrocyte over-activation in T2DM brain (data not shown).

We previously reported a decrease in O-GlcNAcylation of global proteins and of tau in AD brain (Liu et al. 2004, 2009). The present study reveals a similar decrease in T2DM and T2DM-AD brains. Thus, it appears that impaired brain glucose uptake/metabolism, although it might result from different mechanisms, leads to decreased O-GlcNAcylation in both T2DM and AD. Furthermore, increased phosphorylation of tau at several phosphorylation sites, as seen in AD brain, was also seen in T2DM brain. Although the extents of the decrease in global protein O-GlcNAcylation levels were similar between T2DM and AD, the specific tau O-GlcNAcylation level was found to be much lower in AD than that in T2DM (Fig. 3b). Because tau phosphorylation is negatively regulated by tau O-GlcNAcylation (Liu et al. 2004, 2009), the lower tau O-GlcNAcylation level in AD brain is consistent with the larger extents of tau hyperphosphorylation in AD brain than in T2DM brain. Taken together, our observations suggest that T2DM may increase the risk for AD via a decreased brain glucose uptake/metabolism because of down-regulation of GLUT3, leading to down-regulation of O-GlcNAcylation and consequent abnormal hyperphosphorylation of tau, which is crucial to the pathogenesis of AD (Fig. 5).

When we compared site-specific tau phosphorylation levels between AD and T2DM-AD, we found that, although tau is abnormally hyperphosphorylated in both groups, there were clear site-specific differences between pure AD and T2DM-AD groups. Notably, complication of AD with T2DM decreased hyperphosphorylation at the N-terminal half and exacerbated hyperphosphorylation at the C-terminal half of tau molecule, as compared to AD alone. The detail molecular mechanism leading to these discriminating impacts on tau hyperphosphorylation is currently unknown, but it might be related to possible deregulation of insulin signaling in T2DM brain (Moroz et al. 2008). It is worthy to note that these C-terminal phosphorylation sites of tau are among those that are most affected in a mouse model of decreased brain glucose metabolism induced by fasting (Liu et al. 2004; Li et al. 2006). Phosphorylation of tau at these sites also makes tau lose its biological activity to bind to microtubules (e.g., Ser262) (Biernat et al. 1993; Liu et al. 2007; Wang et al. 2007) and self-aggregate into neurofibrillary tangles, (e.g., Ser396/Ser404), one of the hallmark brain lesions of AD (Abraha et al. 2000; Haase et al. 2004; Liu et al. 2007). Therefore, T2DM might not only increase the risk for AD by promoting tau phosphorylation, but also accelerate AD by exacerbating tau hyperphosphorylation at the critical abnormal phosphorylation sites.

Most AD cases are sporadic and are believed to result from multiple etiological factors including genetic susceptibility and environmental and metabolic factors. It is important to note that our proposed mechanism – that impaired brain glucose uptake/metabolism contributes to neurodegeneration via down-regulation of O-GlcNAcylation and the resultant abnormal hyperphosphorylation of tau (Fig. 5) – could be just one of several mechanisms by which T2DM increases the risk for AD. Because glucose metabolism provides the primary fuel for the brain, impairment of brain glucose metabolism in T2DM could lead to deficient neurotransmitter synthesis, release and recycling, as well as other neuronal activities, which are all highly dependent on ATP produced via glucose metabolism. T2DM-induced cerebral vasculopathy could also contribute to AD and dementia (Cechetto et al. 2008).

In summary, in the present study we found that GLUT3 and tau O-GlcNAcylation were decreased and phosphorylation of tau was increased in T2DM brain. We further observed complication with T2DM affected abnormal hyperphosphorylation of tau in a site-specific manner. These findings suggest a possible mechanism by which T2DM may increase the risk for AD by facilitating abnormal hyperphosphorylation of tau.