Metformin Therapy Aggravates Neurodegenerative Processes in ApoE –/– Mice

INTRODUCTION

Epidemiological studies identified a strong correlation between type 2 diabetes mellitus (T2DM) and Alzheimer’s disease (AD) [1–3]. Furthermore, Morris et al. [4] showed an insulin resistance in the central nervous system (CNS) in AD patients. Likewise, Craft and co-workers described insulin abnormalities in AD [5, 6]. In addition, they reported that a reduced hippocampal insulin-degrading enzyme in late-onset AD is associated with the Apolipoprotein E (ApoE) status, which in turn influences the risk to develop AD [5]. A well-known insulin-sensitizing drug is metformin, the most widely prescribed oral medication for T2DM in the world. Metformin is also described to have neuroprotective properties [7]. In this context, a longitudinal aging study showed that long-term treatment (>6 years) with metformin among T2DM patients was significantly associated with lowest risk of cognitive impairment [8]. It is thought that metformin activates the AMP-activated protein kinase (AMPK) [9]. Activated AMPK leads to decrease of mammalian target of rapamycin (mTOR) signaling activity by inactivation of the mTOR/p70S6K-mediated negative feedback loop to insulin receptor substrate-1 [10], which enhances the insulin signaling. Beside this, repressed mTOR signal inhibits tau hyperphosphorylation and reduces formation of neurofibrillary tangles as a neuropathological hallmark of AD [11].

Further, it is described that metformin increases circulating fibroblastic growth factor 21, a starvation hormone (Fgf21) [12]. Fgf21 activity occurs when Fgf21 binds to fibroblast growth factors receptor (Fgfr) and β-klotho, a single transmembrane protein [13]. Fgfrs consist of seven major isoforms (1b, 1c, 2b, 2c, 3b, 3c, and 4), whereby the isoform Fgfr1c is the primary receptor of Fgf21 in the mediation of its activity in in vivo studies [14]. Upon receptor binding of Fgf21, AMPK becomes activated [15]. Beside beneficial effects on metabolism through increasing insulin sensitivity and energy expenditure [16], there is growing evidence that Fgf21 exerts neuroprotective effects and improves cognition [17]. Accordingly, we recently showed in ApoE-deficient mice (ApoE– /–) that caloric restriction-associated neuroprotection seems to be Fgf21/AMPK/mTOR-dependent [18]. ApoE– /– mice represent a well-established mouse model of tauopathy with memory deficits [19, 20]. ApoE binds to tau protein and prevents hyperphosphorylation of tau protein [10], leading to a slowdown of formation of neurofibrillary tangles. In turn, ApoE deficiency results in tau hyperphosphorylation.

There are controversial data on the cerebral effect of metformin. Barini et al. [21] and Li et al. [22] reported that cognitive function of metformin-treated mice with tauopathy was restricted despite reduced tau phosphorylation. Furthermore, a clinical study in diabetic patients showed that metformin impairs cognition [23], whereby Ng et al. [8] described the opposite. To clarify this discrepancy, the present study examined to what extent metformin has neuroprotective properties in ApoE– /– mice and focused on the Fgf21/AMPK/mTOR pathway (Fig. 1A).

MATERIAL AND METHODS

Animals

For the experiments, female ApoE deficient (ApoE– /–) mice were used at 10 weeks of age. Female mice were chosen because of the comparability to previous studies [18]. The mice received either drinking water without additives (H₂O group, n = 8) or drinking water with metformin ∼300 mg/kg/day in accordance to Barini et al. [21] (metformin group, n = 8) for 18 weeks (Fig. 1B). All mice were housed individually in standard cages in a temperature-controlled room (22°C±2°C) on a 12 h light/dark cycle with free access to food and water under specific pathogen free (SPF) conditions. The drinking quantity was measured weekly as well as the body weight. The experimental protocol was approved by the local Animal Research Committee (Landesamt für Landwirtschaft, Lebensmittelsicherheit und Fischerei (LALLF) of the state Mecklenburg-Western Pomerania (LALLF M-V/TSD/7221.3-1.1-002/14) and all animals received humane care according to the German legislation on protection of animals and the Guide for the Care and Use of Laboratory Animals (NIH publication 86–23 revised 1985).

OPEN IN VIEWER

Fig. 1.

Schematic illustration of the working hypothesis (A) and of the animal experiment (B). (A) The present study hypothesized whether metformin caused indirectly via hepatic Fgf21-induced release or directly the Fgfr1c-activated AMPK/mTOR pathway. An activation of AMPK/mTOR pathway may lead to a reduction of tau phosphorylation and in consequence to a neuroprotection. (B) Ten-week-old ApoE– /– mice received either drinking water without additives (H₂O group, n = 8) or drinking water with metformin ∼300 mg/kg/day (metformin group, n = 8) for 18 weeks. The drinking quantity was measured weekly as well as the body weight. At the end of the experiment, all mice at the age of 28 weeks were exsanguinated by puncture of the vena cava inferior for immediate measurement of blood sugar and separation of plasma, followed by harvest of brain tissue.

RESULTS

Metformin caused a slight body weight reduction

The blood sugar measurement over the time of 18 weeks showed neither hypoglycemic nor hyperglycemic events and revealed no differences with 6.3±0.1 mmol/L upon metformin or with 6.1±0.2 mmol/L upon H₂O application at the end of treatment period. However, the plasma insulin concentration was almost 38% less in the metformin-treated mice. The drinking behavior differed slightly as indicated by an overall 4 mL less liquid intake upon metformin exposure (Fig. 2A). This was taken into account in dose determination of metformin. Remarkably, the body weight of metformin-treated mice was significantly reduced over the observation period of 18 weeks when compared to H₂O group (Fig. 2B, p < 0.001).

OPEN IN VIEWER

Fig. 2.

The drinking quantity (A) and the body weights (B). Ten-week-old ApoE– /– mice received either drinking water without additives (H₂O group, n = 8) or drinking water with metformin ∼300 mg/kg/day (metformin group, n = 8) for 18 weeks. The drinking quantity was measured weekly as well as the body weight. Values are given as means±SEM. Two-Way ANOVA for repeated measurements: ^*p < 0.001 versus H₂O.

Metformin increased systemic concentration of Fgf21

Upon metformin exposure the systemic Fgf21 concentration significantly raised and reached approx. 2-fold higher levels versus H₂O-treated mice (Fig. 3A, p = 0.013). Interestingly, the phosphorylated form of cerebral FgfR1c (Fig. 3B and lower panel) and its co-receptor β-Klotho (Fig. 3C) was found 2-fold downregulated upon metformin administration.

OPEN IN VIEWER

Fig. 3.

Quantitative analysis of plasma Fgf21 (A), pFgf1c positive neuronal cells given as number per high power field (HPF, B, n = 4) and the cerebral protein expression of β-Klotho (C) in the ApoE– /– mice. Representative immunohistochemical images (B), lower panel, original magnification ×400) of pFGFr1c positive neuronal cells in the cortex. Mice received either drinking water without additives (H₂O group, n = 8) or drinking water with metformin ×300 mg/kg/day (metformin group, n = 8) for 18 weeks. Signals were normalized to GAPDH. Values are given as means±SEM. Unpaired Student-t-Test followed by Man-Whitney Rank sum test: ^*p < 0.05 versus H₂O.

Metformin increased tau phosphorylation

Metformin did neither influence phosphorylation of AMPK nor of mTOR when compared to the H₂O group (Fig. 4A–C). However, metformin caused an almost 2-fold increase of cerebral tau phosphorylation (Fig. 4D).

OPEN IN VIEWER

Fig. 4.

Representative western blot and densitometric analysis of cerebral pAMPK (A). Quantitative analysis of pmTOR positive neuronal cells, given as number per high power field (HPF; B) and representative immunohistochemical images (original magnification ×400; C) of cortical pmTOR expression in the brain of ApoE– /– mice. Representative Western blot and densitometric analysis of pTau in the brain of ApoE– /– mice. Mice received either drinking water without additives (H₂O group, n = 8) or drinking water with metformin ∼300 mg/kg/day (metformin group, n = 8) for 18 weeks. Signals were normalized to GAPDH and to AMPK or tau. Values are given as means±SEM.

Metformin decreased neuronal integrity

Metformin led to a 20% reduction of NeuN positive cells (Fig. 5A, C) and to a significant reduction of PSD95 (post-synaptic marker) positive neurons in the cortex (p = 0.015, Fig. 5B, D). In line with this, the protein expression of VAMP2 (synaptic integrity marker; Fig. 6A; p = 0.054) and of Synapsin I a/b (pre-synaptic marker; Fig. 6B) was markedly decreased in metformin-treated mice.

OPEN IN VIEWER

Fig. 5.

Representative immunohistochemical images (original magnification ∼400) of NeuN (A) and PSD95 (B) of brain sections of ApoE– /– mice as wells as the quantitative analysis of cortical NeuN- (C) and PSD95- (D) positive cells, given as number per high power field (HPF). Mice received either drinking water without additives (H₂O group, n = 8) or drinking water with metformin ∼300 mg/kg/day (metformin group, n = 8) for 18 weeks. Values are given as means±SEM. Unpaired Student-t-Test followed by Man-Whitney Rank sum test: ^*p < 0.05 versus H₂O.

OPEN IN VIEWER

Fig. 6.

Representative western blot and densitometric analysis of cerebral protein expression of VAMP2 (A) and of Synapsin I a/b (B) of ApoE– /– mice. Mice received either drinking water without additives (H₂O group, n = 8) or drinking water with metformin ∼300 mg/kg/day (metformin group, n = 8) for 18 weeks. Signals were normalized to that of GAPDH. Values are given as means±SEM.

Metformin induced fat accumulation

Metformin slightly increased plasma cholesterol content (Fig. 7A) and led to significant rise of plasma triglycerides (Fig. 7B; p = 0.008) while the plasma concentration of ß-hydroxybutyrate and of MDA as signs of ketogenesis and oxidative stress remained unchanged (Fig. 7C, D). The mRNA expression of genes, indicative for lipogenesis, i.e., lxrα and srebp1c were almost 2-fold higher in metformin-than in H₂O-treated mice (Fig. 8A, B) whereas the mRNA expression of abca1, a reverse cholesterol transporter, was almost unchanged (Fig. 8C).

OPEN IN VIEWER

Fig. 7.

Analysis of plasma concentrations of cholesterol (A), triglycerides (B), β-hydroxybutyrate (C) and malondialdehyde (MDA; D). Mice received either drinking water without additives (H₂O group, n = 8) or drinking water with metformin ∼300 mg/kg/day (metformin group, n = 8) for 18 weeks. Of note, the significant increase of plasma triglyceride upon metformin administration. Values are given as means±SEM. Unpaired Student-t-Test followed by Man-Whitney Rank sum test: ^*p < 0.05 versus H₂O.

OPEN IN VIEWER

Fig. 8.

Quantitative real-time PCR of the nuclear receptors lxrα (A), srebp1c (B), and abca1 (C) and representative immunohistochemical images (D) of cortical GFAP positive cells in the brain of ApoE– /– mice. Signals were corrected to that of rps18. Notable is the increased number of GFAP positive cells indicating a rise of inflammatory processes upon metformin treatment. Mice received either drinking water without additives (H₂O group, n = 8) or drinking water with metformin ∼300 mg/kg/day (metformin group, n = 8) for 18 weeks. Values are given as means±SEM. Unpaired Student-t-Test followed by Man-Whitney Rank sum test: ^*p < 0.05 versus H₂O.

DISCUSSION

Although it is described that metformin treatment may cause hypoglycemic events [25], the current study showed no serious changes of chronic blood glucose concentrations in metformin-treated mice. However, the concentration of insulin was almost 38% decreased which might indicate increased insulin sensitivity. Accordingly, metformin is characterized as an insulin-sensitizing drug. Furthermore, metformin has been suggested to be neuroprotective. Therefore, it is supposed that especially the tau phosphorylation is reduced by metformin [22, 26], although this does not necessarily increase neuronal integrity [21]. Due to these findings and the controversial results relating metformin-associated neuroprotection in diabetic patients [8, 23], we used ApoE– /– mice, a well-known AD mouse model of tauopathy, and treated them with metformin.

It is reported that metformin stimulates the Fgf21 expression in primary hepatocytes [27] and increases circulating Fgf21 in high-fat-diet rats [12]. Further, release of Fgf21 indicates increased lipolysis and ketogenesis [28, 29]. In line with this, we could previously show that caloric restriction in ApoE– /– mice cause increased lipolysis and ketogenesis, Fgf21-release, as well as activation of the Fgfr1c/AMPK/mTOR pathway in the brain [18]. Although the present study detected a significant increase of systemic Fgf21 in metformin-treated mice, β-hydroxybutyrate was found almost unchanged and activation of the FgfR1c/AMPK/mTOR pathway failed. Moreover, the protein expression of the Fgfr1c co-receptor β-Klotho was reduced.

There are two possible explanations for these unexpected findings. One explanation for failed AMPK activation might be a different regulation of energy balance upon metformin treatment in ApoE– /– mice. It is known that AMPK is involved in the fundamental regulation of energy balance at the whole body level by responding to hormonal and nutrient signals in the central nervous system and peripheral tissues that modulate food intake and energy expenditure [30]. Thus, during energy deprivation, for instance upon caloric restriction, the AMPK pathway is activated [18]. However, metformin-treated mice revealed an energy abundance, probably making due to enhanced lipogenesis. Therefore, the activation of AMPK pathway seems to be dispensable. The present study demonstrated a cerebral increase of lipogenic genes such as srebp1c and lxrα upon metformin exposure. In addition, the mRNA expression of hepatic lxrα was increased upon metformin exposure (unpublished data) suggesting enhanced lipogenesis of the whole body. On the other hand, hepatic mRNA expression of pparα was markedly decreased (unpublished data) indicating a reduction of lipolysis, which in turn enhances the lipogenesis and may explain failed AMPK activation. A further argument for increased fat accumulation would be— besides ApoE deficiency— an almost unchanged abca1 mRNA expression implicating failure of the reverse cholesterol transport [31]. Accordingly, plasma concentrations of both triglyceride and cholesterol were found increased upon metformin treatment. Since it is discussed that fat accumulation in the periphery and therewith the increase of inflammatory processes influences the whole body including the brain [32, 33], it is conceivable to state that metformin-enhanced lipogenesis may cause cerebral gliosis, as observed in the present study. Although an increased lipogenesis was observed, the body weight was significantly reduced upon metformin treatment. Since it is known that lipogenesis is an energy-consuming step to store lipids and an energetically inefficient way [34], it is likely to state that increased lipogenesis causes weight loss although the visceral organs become fatty. A typical example of this scenario gives the NPC1 mutant mouse, also a mouse model of neurodegenerative disease [35]. This mutant mouse shows a strong adiposity of the organs due to a disturbed cholesterol transport, but loses enormously body weight [36, 37].

The second explanation for the absence of Fgfr1c activation is as follows: In line with findings in adipose tissue showing reduced β-klotho expression upon TNFα application [38], inflammatory processes in metformin-treated ApoE– /– mice may be responsible for failed FgfR1c/β-klotho activation. Thus, a Fgf21-resistante state might be present in ApoE– /– mice, similarly to the report that obesity-associated inflammation causes Fgf21-resistance [39]. Fgf21-resistance is frequently accompanied with insulin-resistance and in turn with a downregulation of glucosetransporter-1 (glut-1) [40]. Glut-1 is commonly reduced in case of AD [41]. Thus, AD might be associated with a Fgf21-resistance, particularly as AD is known to be characterized by neuro-inflammation [42]. In addition, it is described that a high fat diet as means of exogenous-induced fat accumulation aggravates amyloid and tau pathologies [43, 44]. Thus, metformin-associated fat accumulation may enhance tauopathy and may worsen AD-like symptoms as shown by increased tau phosphorylation as well as reduced numbers of NeuN-and PSD95-positive cells and reduced protein expression of VAMP2 and Synapsin I a/b. Overall, these results show that not only post-synaptic but also pre-synaptic and intra-synaptic integrity is impaired upon metformin treatment. In support of this view, Barini et al. [21] and Li et al. [22] reported that metformin impairs cognition in mice, which was also observed in a clinical study [23]. In conclusion, metformin-associated lipogenesis and inflammation may lead to enhanced neurodegeneration in ApoE– /– mice (Fig. 9). In light of our findings, the use of metformin as a neuroprotective drug shall be with caution. Finally, the application of metformin in case of ApoE risk groups of AD patients has to be proven in future clinical studies.

OPEN IN VIEWER

Fig. 9.

Schematic illustration summarizes the main results of the present study. In contrary to our hypothesis, the metformin-induced Fgf21 release may not be involved in the AMPK/mTOR activation but rather in stimulation of lipogenetic processes. This might lead to an increase of inflammatory processes and in consequence to enhanced neurodegenerative processes which is represented by increased tau phosphorylation.