How ApoE4 Affects Diabetes Through Body and Brain Processes

Alzheimer’s disease (AD) is the most common form of dementia, an age-related neurodegenerative disorder affecting more than 50 million people worldwide [1,2]. It is clinically characterized by progressive deterioration of cognition, loss of insight, judgment, language, and changes in behavior and, in late stages, physical functioning [3]. Neuro-biochemical assessments of AD patients’ brains revealed selective impairments in the hippocampus and neocortex areas, with damage to mainly cholinergic neuronal populations [4,5]. Along with neuronal loss, AD pathology is characterized by the occurrence of senile plaques, which contain Aβ depositions, neurofibrillary tangles composed of hyper-phosphorylated tau, and extensive synaptic loss [6]. The sex-based prevalence of AD is well documented, with over 60% of AD patients being female [7]. An even greater sex difference has been found in the impact of AD pathology, with each additional unit of AD pathology found to be associated with a nearly 3-fold increase in the odds of clinical AD in men compared with a more than a 22-fold increase in the odds of clinical AD in women [8]. AD is manifested either in a familial or sporadic form; the familial form is characterized by mutations in the amyloid precursor protein (APP) and presenilin 1 and 2, resulting in elevated levels of Aβ deposits and senile plaques [9,10]. Studies of sporadic AD revealed allelic segregation of the apolipoprotein E (APOE) gene to families and individuals with a higher risk of late-onset sporadic AD [11,12,13]. There are three main alleles of the apolipoprotein E, termed apoE2, apoE3, and apoE4, of which the ε4 gene allele (apoE4) has been found to be the most prevalent genetic risk factor for AD [11,12,14]. The frequency of this allele in AD patients is over 50%, whereas in the general population its frequency is about 25% [15]. Each ε4 allele reduces the age of AD onset by 7 to 9 years per allele copy [12,16]. Evidence indicates that the apoE4 risk for AD is greater in women than in men [7]; a single copy of the ε4 allele in women is sufficient to increase the AD disease risk associated with two copies of ε4 in men [17]. Furthermore, it has been shown that carrying the apoE ε4 allele has a larger deleterious effect on neurodegeneration, synaptic plasticity, and adult neurogenesis, and on cognitive performance in females more than males [18].

Neuropathologically, apoE4 is associated with a wide range of brain-related phenotypes, including impaired neurite outgrowth, synaptogenesis, and plastic neuronal remodeling, as well as increased neurodegeneration [19,20,21]. As opposed to the effect of apoE4 on the CNS, there is a lack of understanding regarding its effect on the peripheral nervous system (PNS) [22,23]. This is despite the fact that apoE is expressed throughout the PNS, including at the neuromuscular junction (NMJ) [24,25], where expression levels dramatically increase in response to nerve crush injury and exposure to harmful environmental stimuli [26]. An association between the APOE genotype and disease outcome has been demonstrated for human patients with diabetic neuropathies [27,28], but very few have investigated the influence of the apoE isoform on the PNS in genetically modified animals [26]. In general, although many studies have been carried out regarding AD, the etiology and pathogenesis of the disease are not yet fully understood [29]; therefore, new links that may unravel the answers to these questions are necessary.

Epidemiologic studies revealed that the risk of AD is increased by 50%–100% by type 2 diabetes miletus (T2DM) [30,31,32]. A recent population-based study found that the overall incidence densities of AD for diabetic men and women, respectively, were 0.82 and 1.15 per 1000 person-years, with an especially notable higher hazard ratio of AD in older diabetic women [33]. T2DM is a metabolic disorder of impaired glucose regulation, one of the most common chronic metabolic diseases, with more than 300 million patients worldwide. Clinically, T2DM is characterized by a slow decline in the functions of multiple body organs and systems, and impairments in cognitive functions [34]. Pathological characteristics include peripheral and central impaired insulin action, insulin secretory deficiency, and increased endogenous glucose production [35]. Diabetic peripheral neuropathy (DPN) is a common late complication of diabetes, defined as a symmetric length-dependent sensorimotor polyneuropathy attributable to metabolic and microvascular alterations as a result of chronic hyperglycemia exposure (diabetes) [36,37]. DPN is characterized by loss of small-fiber–mediated sensation, resulting in the loss of thermal and pain perception, and large-fiber impairment resulting in the loss of touch and vibration perception [37]. T2DM and AD have been shown to share many pathophysiological features [38], which include insulin resistance, disrupted glucose metabolism, peripheral oxidative and inflammatory stress, amyloid aggregation, neural atrophy, neurodegeneration, and cognitive decline [30]. Furthermore, key pathological changes that occur in an AD brain, such as amyloid aggregation, resemble those that occur in the pancreas and vasculature of diabetic patients [39,40]. These findings and more led to the hypothesis that impaired neuronal insulin action might be a unifying pathological mechanism in the development of both T2DM and AD [41]. Indeed, insulin resistance and impaired insulin signaling in the central nervous system (CNS) have been linked to the pathogenesis of AD, observations that resulted in terming AD “type 3 diabetes” [30,42,43].

The expression of insulin receptor (IR) units in the CNS displays a widespread but selective regional pattern, and they are found abundantly in brain areas involved in glucose and energy homeostasis as well as cognitive processes [41]. In rodents, one of the areas with the highest density of IRs is the hippocampus [44,45,46]. In the canonical insulin signaling pathway, insulin first binds to its receptor on the plasma membrane and induces signal transduction that includes activation/phosphorylation of Akt on residues Thr308 and Ser473 [35]. In turn, activated Akt inhibits glycogen synthase kinase-3β (GSK-3β) by inhibition/phosphorylation of Ser9 residues [47,48]. GSK-3β is an inhibitor of glycogen synthase (GS), and the Akt-driven inhibition of GSK-3β thus results in the promotion of glycogen synthesis and glucose absorption [35,49]. Additionally, GSK-3β has been found to be activated by auto-phosphorylation of Tyr216, which in AD can result in pathological and abnormal phosphorylation of tau [50]. In the brain, the insulin signaling pathway plays a significant role in neuronal health as well as synapse formation and maintenance. Significant impairments in this pathway have been documented in both postmortem analysis and animal models of AD [51]. Furthermore, insulin resistance may also contribute to cognitive deficits in AD, seeing that healthy brain insulin signaling has been found to be crucial for learning and memory [43,52].

Given the overlap of AD and T2DM, several anti-diabetic agents have been tested for treatment of AD, with many showing reduced benefits in apoE4 patients, such as intranasal insulin treatment and insulin sensitizer Rosiglitazone trials [53,54]. Although both apoE4 carriers and non-carriers with AD appear to have a brain insulin defect, the apoE4 carrier status appears to modulate the relationship between the peripheral and brain insulin metabolism [53]. These findings suggest an association between two major risk factors for AD, apoE4, and T2DM. The ε4 allele has been found to be an independent risk factor for T2DM [55] and associated with the development of T2DM [56]. Furthermore, the presence of apoE4 exacerbates AD neuropathology in the presence of T2DM [57]. For example, apoE4 enhances the differences between T2DM and non-T2DM subjects in the number of hippocampal and cortical neuritic plaques, neurofibrillary tangles, and the load of cerebral amyloid angiopathy [58]. It has also been shown that in apoE4 carriers, the degree of glucose dysregulation (evaluated by fasting blood glucose concentration and mean glycemic value, as measured by the HbA1c concentration [59]) correlates with reduced cortical thickness and that apoE4 carriers with T2DM demonstrate a level of cortical thinning comparable to that of preclinical AD [60]. Metabolic deficiencies, such as abnormally low rates of brain glucose metabolism, that are unrelated to regional amyloid burden [61] were observed in healthy apoE4-positive volunteers as young as their twenties [62]. Additionally, altered expression of insulin signaling parameters has been observed in apoE4 mouse and human brains [63,64]. However, despite the strong body of evidence linking T2DM and apoE4, the mechanism underlying the interaction between these two critical risk factors for AD is yet to be determined [54].

In the present study, we employed a dietary regimen of a high-fat diet (HFD), a well-established and robust animal model for studying impaired glucose tolerance and early T2DM in mice [65,66], compared to a regular diet, on an APOE-targeted replacement (TR) mouse model expressing either apoE3 or apoE4. Utilizing this model, we investigated the effects of the APOE genotype and HFD on glucose metabolism and insulin resistance, and the extent to which these are associated with apoE4-driven peripheral or central pathologies.

Following weaning at 7 weeks of age, mice were subjected to either a HFD or regular chow diet for 18 weeks, as described in the materials and methods, utilizing n = 17–18 mice per group. Throughout these 18 weeks, body weight and fasting blood glucose levels were monitored once a week. The results thus obtained are presented in Figure 1A,B. As can be seen in Figure 1A, there was an age-dependent increase in the weight of apoE3 and apoE4 control diet mice, which was similar in both genotypes (23.23 ± 0.4 and 21.71 ± 0.4 at 18 weeks on diet; respectively). The HFD induced a marked and similar increase in body weight of apoE3 and apoE4 mice (34.89 ± 1.4 and 32.72 ± 0.9 at 18 weeks on diet; respectively). It is important to note that starting at week 15 on the HFD, apoE4 HFD-fed mice showed a trend of decreased weight gain in comparison to apoE3 HFD-fed mice. As can be seen in Figure 1B, the HFD also induced an increase of fasting blood glucose, which was similar between apoE3 and apoE4 mice, in accordance with the corresponding body weight measurements (160 ± 4.5 and 147.3 ± 3.6 at 18 weeks; respectively), whereas control diet apoE3 and apoE4 mice did not differ in fasting blood glucose (122.7 ± 3.2 and 126.7 ± 2.5 at 18 weeks; respectively).

The effects of the HFD and apoE genotype on glucose and insulin tolerance were next evaluated. Glucose tolerance was assessed via an intraperitoneally (IP) glucose tolerance test (GTT) as described in the materials and methods (n = 8 mice/group). As can be seen in Figure 1C, apoE4 control diet mice displayed delayed recovery compared to the corresponding apoE3 control diet mice, suggesting impaired glucose tolerance in apoE4 mice. HFD treatment resulted in increased glucose levels in both the apoE3 and apoE4 mice, indicating a pathological glucose tolerance. Areas under the curve (AUC) of GTT measurements were calculated using the trapezoidal rule [67], and the results are depicted in Figure 1D. Two-way ANOVA analysis revealed a significant effect for genotype (*** p < 0.0001) and interaction diet x genotype (p = 0.01), Bonferroni post-hoc comparisons showed significantly increased glucose levels in apoE4 mice relative to apoE3 mice (33 ± 5 and 20 ± 6, respectively; * p < 0.05), while apoE3 and apoE4 HFD mice’s AUC was elevated in a similar manner (45 ± 3 and 40 ± 3, respectively). ApoE3 control diet mice displayed significantly lower (p < 0.001) glucose levels relative to apoE3 HFD mice.

Insulin tolerance was assessed via the IP-insulin tolerance test (ITT) (n = 10 mice/group). In view of the findings that the differences in glucose tolerance between apoE3 and apoE4 mice were more pronounced in control diet mice, these experiments were performed on mice maintained on a control diet. As can be seen, in Figure 1E, the injection of insulin in both groups resulted in decreased levels of blood glucose. This decrease was more gradual and less pronounced in apoE4 mice. Quantitation of these results in terms of total glucose levels by measurements of AUC revealed increased levels of glucose in apoE4 compared to apoE3 mice. Accordingly, the Student’s t-test for AUC of the ITT results revealed a significant effect (*** p < 0.001) for the comparison of apoE3 and apoE4 control diet mice (46 ± 4 and 64 ± 2, respectively) (Figure 1F). These results suggest that apoE4 is associated with increased insulin resistance relative to apoE3 mice.

Insulin levels in the plasma were assessed at baseline, 15 and 30 min following glucose injection in the GTT as described in the materials and methods, and the results are depicted in Figure 1G. As can be seen, the levels of insulin in the plasma are elevated in apoE3 mice relative to apoE4 mice at baseline (0.514 ± 0.03 and 0.41 ± 0.03, respectively; * p = 0.04) and 15 min following the glucose injection (0.62 ± 0.12 and 0.35 ± 0.01, respectively; p = 0.05), suggesting that there is a deficit in the ability of the apoE4 mice to secrete insulin in response to elevated blood glucose levels. Together with the GTT and ITT results, the results obtained suggest that the pathological effects of apoE4 are mediated by two complementary mechanisms: reduced ability to secrete insulin and increased insulin resistance.

The effects of the HFD and APOE genotype on cognitive and sensorimotor parameters were next examined.

Short-term memory was assessed via the novel object recognition (NOR) paradigm, in which the mouse’s tendency to approach a novel object was measured (n = 10 mice/group). Mice were exposed to a novel object 2.5 h following initial exposure to two similar objects (in which no preferences existed in either of the groups). As can be seen in Figure 2A, control diet apoE4 mice displayed no preference and visited the new and old objects similarly, whereas control diet apoE3 mice spent more time near the novel object, suggesting an intact short-term memory of the old object. This difference between the control diet apoE3 and apoE4 mice was abolished by the HFD, seeing that under these conditions, both apoE3 and apoE4 mice displayed no preference for the novel object. Statistical analysis via two-way ANOVA revealed a significant effect for genotype (p = 0.034). Further Bonferroni post-hoc comparisons revealed that the decreased preference of control apoE4 mice to the new object relative to that of the control apoE3 mice was significant (0.42 ± 0.04 and 0.65 ± 0.09, respectively; * p < 0.05), as was the effect of the HFD on apoE3 mice relative to apoE3 control diet mice as seen by a lack of preference of apoE3 HFD mice (0.43 ± 0.05; * p < 0.05). The HFD did not affect the performance of the apoE4 mice.

The effect of the APOE genotype and HFD on nociception was assessed by testing the latency of reactivity to thermal pain in the hot plate test (52 °C) (n = 10 mice/group). As can be seen, apoE4 control diet mice displayed hypersensitivity compared to apoE3 control diet mice while the HFD manipulation elevated sensitivity in apoE3 mice, rendering them similar to apoE4 mice. Analysis of the results via two-way ANOVA revealed a significant effect for the interaction between genotype x diet (p < 0.05). As can be seen in Figure 2B, apoE4 control diet mice showed statistically increased heat sensitivity relative to apoE3 control diet mice (11.11 ± 1.2 and 14.94 ± 1.4, respectively; * p < 0.05) while, similar to the effect seen in the NOR paradigm, the HFD abolished these differences and increased the thermal sensitivity of the apoE3 mice but did not affect the apoE4 mice (10.62 ± 0.5 and 11.54 ± 1.28, respectively).

Motor coordination and learning were tested using the rotarod apparatus, in which the amount of time mice spent on an accelerating rod was documented. The results thus obtained during the first trial are presented in Figure 2C (n = 10 mice/group). As can be seen, a similar pattern to the results of the NOR and hot plate tests was observed, where apoE4 mice displayed impaired motor skills relative to apoE3 mice in the control diet, while the HFD impaired motor skills in apoE3 mice without affecting apoE4 mice. Two-way ANOVA revealed a significant effect (p < 0.0001) for the interaction between genotype x diet. Bonferroni post-hoc comparisons revealed that apoE4 control diet mice displayed significantly decreased latency to fall relative to the apoE3 control diet mice (6.44 ± 1 and 19.13 ± 2, respectively; *** p < 0.0001). The HFD decreased the performance of apoE3 mice while no effect was seen in apoE4 HFD mice (5.4 ± 0.8 and 6.68 ± 1.3, respectively). This effect, however, was diminished over the four subsequent trials (Figure 1D), suggesting an effect on basal motor coordination, which was abolished in time by learning. The average performance of both genotypes throughout the five trials was similar in the control diet and tended to be decreased by the HFD. Indeed, statistical analysis revealed a significant effect (p < 0.05) for diet, whereas post-hoc comparisons showed no significant differences between groups.

Taken together with the results of the NOR, the hot plate and first trial of the rotarod test revealed a similar pattern in which apoE4 mice were impaired relative to apoE3 control diet mice, and the HFD damaged the performance of the apoE3 mice, rendering them similar to apoE4 mice, which were not affected by the HFD.

The extent to which the observed effects of the APOE genotype and HFD on cognitive and sensorimotor parameters correlate with peripheral neuropathy was next assessed via skin IENFD and sciatic nerve electrophysiological measurements. Accordingly, we assessed footpad intra-epidermal nerve fiber density (Figure 3A) following 18 weeks on either a control or HFD (n = 8–10 mice/group). There were no significant effects of either the APOE genotype or diet on these parameters. Compound muscle action potential (CMAP), nerve conduction velocities, and distal latencies of the sciatic nerve were next examined (Figure 3B) (n = 10 mice/group). Similar to the skin IENFD results, these measurements revealed no effect of either the APOE genotype or HFD on the said parameters. Taken together, these findings suggest that the demonstrated behavioral effects of apoE4 and HFD are not associated with peripheral small fiber neuropathy or sciatic nerve electrophysiological pathologies.

We next focused on the effects of the HFD and APOE genotype on insulin signaling parameters in the hippocampus. This was performed by utilizing immunoblots of the insulin receptor (IR), and the subsequent signaling molecules Akt and GSK-3β.

The results of the total and activated IR levels are presented in Figure 4A. As can be seen, apoE3 and apoE4 mice displayed similar levels of total IR under basal control diet conditions while the HFD decreased levels or IR in apoE4. Two-way ANOVA of the total IR levels revealed a significant effect for the interaction between genotype x diet (p = 0.009). Post-hoc comparisons showed that the total IR levels were similar in the control diet groups and that the HFD significantly reduced levels in apoE4 mice relative to apoE4 control diet mice (0.74 ± 0.1 and 0.96 ± 0.1, respectively; * p < 0.05). Furthermore, apoE3 HFD mice revealed significantly higher IR levels relative to apoE4 HFD (1.1 ± 0.1 and 0.74 ± 0.1, respectively; *** p < 0.001).

Measurements of the activated IR, as presented by the ratio of phosphorylated IR to its total level, revealed elevated levels in apoE4 mice relative to apoE3 mice in the control diet groups while the HFD abolished these differences, rendering apoE3 mice similar to apoE4 mice. Two-way ANOVA revealed a significant effect for genotype (p = 0.0075), and Bonferroni post-hoc comparisons showed that apoE4 control diet mice displayed significantly higher IR activity relative to apoE3 control diet mice (1.45 ± 0.1 and 1 ± 0.08, respectively; ** p < 0.01). The results of all IR immunoblots represent two cohorts, total n = 15–22 mice/group.

Akt is activated by insulin-mediated phosphorylation of Thr308 and Ser473, with Ser473 phosphorylation also driven by mechanistic target of rapamycin complex 2 (mTORC2) [68]. The effects of the APOE genotype and HFD on total levels and said phosphorylation states of Akt are depicted in Figure 4B. As can be seen in the left graph, the total levels of Akt were not changed by either the APOE genotype or diet, which is consistent with previous observations [69] (Figure 4B). In contrast, the extent of Akt activation, as measured by the ratio of p-Akt Ser473 to total Akt, revealed a significant effect for interaction between genotype x diet (p = 0.024). In accordance with previous findings [69], apoE4 control diet mice displayed a trend of decreased Ser473-mediated activation of Akt relative to apoE3 control diet mice (0.664 ± 0.1 and 1 ± 0.1, respectively). Interestingly, this trend was turned around in the HFD mice, where apoE4 mice showed a trend of increased activation relative to apoE3 mice (1.21 ± 0.3 and 0.76 ± 0.14, respectively). The corresponding analysis of the extent of Akt activation, as measured by the ratio of p-Akt Thr308 to total Akt, revealed a significant effect for diet (p < 0.0001). Thr308-mediated Akt activation revealed a small increase in apoE4 control diet mice relative to apoE3 control diet mice, and HFD resulted in a marked similar increase in both genotypes. Statistical analysis revealed significantly higher levels of Thr308 activation in the apoE4 HFD mice relative to apoE4 control diet mice (1.92 ± 0.1 and 1.18 ± 0.1, respectively; *** p < 0.001), as in apoE3 HFD compared to apoE3 control diet mice (1.87 ± 0.1 and 1 ± 0.05; *** p < 0.001, respectively). n = 6–11 mice/group.

GSK-3β is generally active in the absence of exogenous signals and is thus acutely inactivated by Akt via inhibitory phosphorylation on Ser9. Activating autophosphorylation of Tyr216 has been shown to further regulate the kinase’s activity. The results of the GSK-3β immunoblot measurements are depicted in Figure 4C. As previously shown [69], the levels of total GSK-3β were similar in apoE3 and apoE4 control diet mice, and no significant differences between genotypes were found following the HFD (Figure 4B). In terms of activation-phosphorylation of GSK-3β on Tyr216, apoE4 control diet mice showed significantly higher activation than apoE3 control diet mice (1.42 ± 0.2 and 1 ± 0.1, respectively; * p < 0.05) as previously shown [69]. Interestingly, this effect was abolished by the HFD. Two-way ANOVA revealed a significant effect for the interaction between genotype x diet (p = 0.0425). The activity of phosphorylated Ser9, which is a marker of inhibited GSK-3β, is presented in the right panel in Figure 4C. As can be seen, the inhibited activity of GSK-3β via Ser9 phosphorylation was higher in apoE4 control mice relative to apoE3 control mice, and HFD administration resulted in a marked increase in apoE3 mice, rendering them equal to apoE4 mice, which were not affected by the HFD. Two-way ANOVA statistical analysis revealed significantly higher levels of activation in apoE4 control diet mice relative to apoE3 control diet mice (2.23 ± 0.3 and 1 ± 0.1, respectively; * p < 0.05) while the HFD abolished these differences and significantly increased the levels of Ser9-mediated GSK-3β inhibition in apoE3 mice to the levels of the apoE4 control diet mice but did not affect levels of activity in apoE4 mice. Post-hoc comparisons revealed apoE3 control diet mice displayed significantly lower levels relative to apoE3 HFD (1 ± 0.1 and 2.3 ± 0.5, respectively; * p < 0.05) mice. The results of all GSK-3β-related parameters represent two cohorts, total n = 11–17 mice/group.

Taken together, the insulin signaling measurements demonstrate that apoE4 induces insulin and glucose metabolism-related deficits and that these effects are associated with marked behavioral effects.

The effects of the APOE genotype and HFD on apoE levels in the brain and plasma were assessed via immunoblots of hippocampal homogenates and plasma extracts from apoE3 and apoE4 control and HFD-fed mice. In accordance with previous findings [70], the total hippocampal apoE levels were significantly lower in apoE4 control diet mice relative to apoE3 control diet mice (0.55 ± 0.03 and 1 ± 0.03, respectively; *** p < 0.001). Interestingly, the HFD significantly decreased (*** p< 0.001) the levels of hippocampal apoE in apoE3 HFD mice relative to apoE3 control diet mice, rendering them closer to the levels of apoE4 control mice without affecting apoE4 mice (Figure 5A). Still, hippocampal apoE levels in apoE4 HFD mice remained significantly lower (** p < 0.01) relative to apoE3 HFD mice (0.55 ± 0.04 and 0.763 ± 0.06, respectively). Two-way ANOVA revealed a significant effect (p < 0.0001) for genotype and a significant effect (p < 0.01) for diet and the interaction between genotype x diet. The results of the hippocampal apoE immunoblots represent two cohorts, total n = 15–17 mice/group.

Plasma apoE levels were measured next. This revealed that similar to the brain phenotype and as previously described [71], the levels of plasma apoE in control apoE4 mice were significantly lower than those of the corresponding apoE3 mice plasma apoE relative to the apoE3 control diet (0.84 ± 0.05 and 1 ± 0.04, respectively; * p < 0.05), as revealed by post-hoc comparisons of two-way ANOVA. In contrast, there was no difference in the plasma levels of HFD-fed apoE4 and apoE3 mice, which was due to a small increase in the levels of plasma apoE in apoE4 mice. The results of the plasma apoE immunoblots represent two cohorts, total n = 8–10 mice/group.

The effects of the APOE genotype and HFD on apoE lipidation in the brain were assessed via blue native gel, and the results can be seen in Figure 5C, n = 3 lanes per group, where each lane represents a pull of three mice from the same group. In accordance with previous findings [70,72], under control diet conditions, apoE4 mice displayed hypolipidation of apoE relative to apoE3 mice. Importantly, the extent of lipidation of apoE in apoE3 and apoE4 mice was increased in mice exposed to the HFD, rendering the lipidation of apoE4 HFD mice similar to that of the apoE3 control diet mice.

Taken together, these observed results show that under normal diet conditions, the levels of apoE in apoE3 mice are significantly higher than those of apoE4 in the brain and plasma, an effect that correlates with hypolipidation of apoE in apoE4 mice. The HFD increases apoE lipidation in both apoE3 and apoE4 mice, and reduces brain apoE levels in apoE3 mice without further affecting apoE4 mice, but has no significant effect on the levels of apoE in the plasma of neither apoE3 nor apoE4 mice.

Endogenous mouse APOE was replaced by either human APOE3 or APOE4, in order to create APOE-TR mice by gene targeting as previously described [101]. These mice were purchased from Taconic (Germantown, NY, USA), and were backcrossed at Taconic for eight generations after their preparation. To minimize possible genetic drifting between the apoE4 and apoE3 mice, which were offspring of the homozygous apoE4 and apoE3 mice generated by Taconic in 2001, they were further crossed by us with Harlan C57Bl/6JOlaHsd mice, which unlike the standard Jackson laboratory C57Bl/6J ApoEtm1.1(APOE∗4)Adpmc mice (Jackson Laboratories, Bar Harbor, ME, USA) were α-syn−/−. The resulting mice were then crossbred to yield apoE4 and apoE3 homozygous mice on an α-syn−/− background. Accordingly, the present experiments were performed with apoE3 and apoE4 homozygous mice on an α-synuclein −/− background, in which we have previously shown that the apoE4 phenotype is more pronounced than in α-synuclein +/+ expressing mice [70]. These mice are referred to herein as apoE3 and apoE4 mice, respectively. The APOE genotype of the mice was confirmed by PCR analysis (as can be seen in supplementary materials), as previously described [102]. All experiments were approved by the Tel-Aviv University Animal Care Committee, approval number 04-17-058 (date of approval expiration: 07/11/2021). Every effort was made to reduce animal stress and to minimize animal usage.

In the current study, we chose to focus on female mice, seeing that females are more susceptible to AD than males, and the interaction between gender and apoE4 has been demonstrated in numerous animal and human studies, resulting in the finding that apoE4-related pathology is more pronounced in females [18,69]. Furthermore, recent findings in our lab showed that apoE4-TR female mice have a more robust pathological phenotype than the corresponding male mice [70]. The results presented correspond to 2 female cohorts, consisting of 4 groups: 2 genotypes (apoE3 or apoE4) X 2 diets (control or HFD). The first cohort contained 10 mice/group and the second cohort contained 7–8 mice/group.

After weaning, at 7 weeks of age, the mice were randomized by body weight and assigned to either a standard rodent chow diet (Teklad 2018, Envigo, Huntingdon, Cambridgeshire, UK) or an HFD (60% kcal, D12492, Research Diets) consumed ad libitum. Weight and blood glucose were measured weekly following a 6-h fast, as described in the following sections. According to previous studies [103], diabetic symptoms appear in C57BL/6 mice after a minimum of 11 weeks on the HFD. Therefore, starting at 15 weeks on the diet, the mice underwent cognitive tests and either metabolic or motor and sensory tests. At the age of 6 months, the mice were anesthetized with ketamine and xylazine, after which blood samples were collected from the vena cava and mice were perfused transcardially with phosphate-buffered saline (PBS). Their brains were then removed, and the hippocampi were further processed for biochemical analysis.

The mice were subjected to the GTT 15 weeks following diet initiation, and in the following week to the ITT.

The tests were initiated 15 weeks post the beginning of the diet administration. The mice were first subjected to the NOR test. After a 1-week interval, they underwent the rotarod test and 3 days later the hot plate test.

Electrophysiological tests were performed on control and HFD-naive mice when mice reached 6 months of age following 18 months on either a control or HFD immediately prior to being sacrificed, as previously described [73]. Accordingly, mice were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg). Body temperature was maintained as warm by placing the mice on a heating mat. Temperature differences were minimized by conducting the study as soon as the anesthesia had taken effect. Electrophysiological studies were conducted on the sciatic nerve in the prone position. CMAP responses were recorded from the gastrocnemius muscle with an active needle electrode and a reference electrode placed at the center of the foot. Stimulation of the nerve was performed at the ischial notch and knee with a pair of blunt needle electrodes, with the distal cathode 10 mm proximal to the recording electrode. The ground electrode was placed between the stimulating and recording electrodes. Supra-maximal stimulation, at a range of 3–5 mA was employed, and the low and high frequency filters were set at 10 Hz and 10 kHz, respectively. To calculate the motor nerve conduction velocity (MNCV), the distance between the stimulation sites was divided by the latency difference. CMAP amplitudes were measured from the baseline to the negative peak.

Hind footpa skin biopsies were collected, immersed for 6–8 h at 4 °C in Zamboni’s fixative (2% paraformaldehyde, 0.2% picric acid in 0.1 M phosphate buffer), rinsed in 30% sucrose in PBS solution overnight, cryoembedded in mounting media (OCT), and sectioned at 14 μm thick before being processed for immunohistochemistry. Sections were incubated at 4 °C for 16–24 h with rabbit anti-primary PGP9.5 antibody (1:200; Sigma-Aldrich, Singapore). Sections were then rinsed 3 times in PBS and incubated with secondary anti-rabbit AlexaFluor 488 antibody, before being rinsed and mounted. To confirm that there was no nonspecific immunoreaction, additional sections were incubated with primary or secondary antibodies alone. Fluorescent images were collected on an Olympus microscope with an ObserverZ1 imaging system (Zeiss, Oberkochen, Germany). Six sections were measured for each footpad, and the average linear density of IENF was calculated according to current guidelines [107].

The experimental design consisted of 2 genotypes (apoE3 and apoE4) and 2 diets (control and HFD). All data are presented as mean ± standard error of measurement. The results of each parameter were normalized relative to the apoE3 control diet mice and analyzed via GraphPad Prism 5.3 software. Multiple groups and/or time points were analyzed utilizing two-way ANOVA, and when appropriate two-way ANOVA repeated measurements (time x groups). This was followed by a planned Bonferroni post-hoc test to determine the differential effects of diets in apoE3 and apoE4 mice. ITT measurements were analyzed by Student’s t-test in order to determine differences between APOE genotypes. Values of p < 0.05 were considered statistically significant. All graphs were prepared by GraphPad prism 5.3 software.