Introduction

Alzheimer's disease (AD) is an aging-dependent neurodegenerative disease that is most well-known for causing learning and memory loss. Indeed, much of early research on AD focused on the aggregation of amyloid-beta and tau in brain regions associated with cognition and memory, such as the hippocampus and entorhinal cortex1. However, numerous clinical trials have now succeeded in reducing such aggregations in AD subjects and have generally shown limited or no beneficial effects on cognition2,3. With time, a broadened scope of research has revealed that AD patients harbour a plethora of other brain and body changes, many of which precede cognitive impairment, including dysregulated lipid metabolism4,5,6,7,8,9,10. Intriguingly, lipid droplet accumulation was described in the brains of AD patients by Dr. Alois Alzheimer over 100 years ago, a finding largely ignored due to the complexity of studying brain lipids11.

The development of new tools and technologies has made it increasingly possible to understand the role of lipid changes in health and disease. Lipids have wide-ranging functions in the body, and it is now clear that dysregulation of lipid metabolism is at the core of many diseases. A critical and potentially disease-initiating role of lipids in AD is supported by both genetic and environmental/lifestyle factors. For example, GWAS studies show that many AD risk genes such as APOE, TREM2, APOJ, PICALM, CLU, ABCA7 and ECHDC3 are directly involved in lipid metabolism12. Moreover, individuals with lipid-related metabolic syndromes such as obesity or diabetes are at increased risk of developing cognitive impairments and AD specifically13,14.

In AD, both central and peripheral lipid changes have been reported in patients and animal models4,15,16,17. Interestingly, dysregulation in levels of monounsaturated fatty acids (MUFA) have emerged in studies of AD and other neurodegenerative diseases, with remarkably beneficial effects found by lowering activity of Stearoyl-CoA desaturase (SCD), the rate limiting enzyme in MUFA synthesis15,18,19,20,21,22,23,24,25,26. In AD specifically, we previously demonstrated that infusion of an SCD inhibitor into the brain’s lateral ventricles decreased local MUFA build up and resulted in reversal of core biological features of AD, including microglial activation, hippocampal dendritic spine loss, and learning and memory impairments15,25.

Intriguingly, the brain is a crucial regulator of peripheral metabolism. For example, the hypothalamus regulates food intake and satiety, insulin secretion, hepatic glucose production, and glucose/fatty acid metabolism in adipose tissue and skeletal muscle 29,30. Moreover, in the periphery, inhibiting SCD activity can improve systemic metabolic fitness 27,28, a modifying factor for AD pathogenesis. These considerations raised the question of whether cognitive improvements seen in AD mice following central SCD inhibition might be mediated in part by changes in peripheral metabolism, either by modulation of central control of peripheral metabolism or via direct peripheral diffusion of the inhibitor. Clarifying the involvement of peripheral metabolic changes in the benefits seen following central SCD inhibition may have strategic implications for development of SCD inhibitors, as those that are currently available are not CNS-specific and thus have peripheral side-effects.

Results

Increased adiposity in 3xTg-AD mice

Being overweight in midlife is linked to greater risk of developing AD31,32 and brain changes in obese individuals mirror those of AD33. Thus, we first aimed to characterize body weight-related defects in the slowly progressing 3xTg-AD mouse model. 3xTg-AD mice carry the human APPSwe, tauP301L, and PS1M146V mutations, express elevated levels of soluble amyloid species, and develop symptoms of learning and memory impairments as early as 6 months of age, preceding appearance of amyloid plaques and neurofibrillary tangles34. Female 3xTg-AD mice and wildtype (WT) strain controls were euthanized for analysis at regular intervals from 2–3 to 10–11 months old (mo). 2-way ANOVA revealed significant changes in body weight related to both the age-factor (F(4, 180) = 46.53, p < 0.0001) and strain-factor (F(1, 180) = 16.99, p < 0.0001). Post-hoc analysis indicated that the 3xTg-AD mice exhibited statistically significant increases in body weight at the 6–7 mo (p = 0.0123), 8–9 mo (p = 0.0281) and 10–11 mo (p = 0.0192) timepoints (Fig. 1a).

Figure 1
figure 1

Increased adiposity in 3xTg-AD mice. (a) Cross sectional analysis of body weight in grams (g) between WT and 3xTg-AD mice at 2–3 months (n = 22, p = 0.2870), 4–5 months (n = 21, p = 0.3697), 6–7 months (n = 18, p = 0.0123), 8–9 months (n = 22–25, p = 0.0281), and 10–11 months (n = 10–11, p = 0.0192) of age (2-way ANOVA, Fisher post-hoc test). (b–f) Weights of adipose tissue, liver and spleen in 2–3-month-old WT and 3xTg-AD mice. Adipose tissue weight (grams) expressed as a percent (%) of total body weight for (b) Visceral white adipose tissue (n = 5, p = 0.0689) (c) subcutaneous white adipose tissue (n = 5, p = 0.0980), and (d) interscapular brown adipose tissue (n = 5, p = 0.7756). (e) Liver (n = 5, p = 0.2074) and (f) spleen (n = 5, p = 0.0128) weight (grams). Unpaired t-tests. (g–k) Weights of adipose tissue, liver and spleen in 10–11-month-old WT and 3xTg-AD mice. Adipose tissue weight (grams) expressed as a percent (%) of total body weight for (g) Visceral white adipose tissue (n = 10–11, p = 0.0005) (h) subcutaneous white adipose tissue (n = 10–11, p = 0.1264), and (i) interscapular brown adipose tissue (n = 10–11, p = 0.1519). (j) Liver (n = 10–11, p = 0.0543) and (k) spleen (n = 10–11, p = 0.0005) weight (grams). Unpaired t-tests. Error bars represent mean ± standard error of the mean (SEM). Significance level was set at p ≤ 0.05.*p ≤ 0.05.

Full size image

To determine if increased adiposity was responsible for this weight gain, we weighed white and brown fat pads of 2–3 mo mice (prior to memory impairments) and 10–11 mo mice (memory-impaired). 2–3 mo 3xTg-AD mice showed a tendency for increased percentage of both visceral (abdominal) (p = 0.0689, unpaired t-test, Fig. 1b) and subcutaneous (hindlimb) (p = 0.0980, unpaired t-test, Fig. 1c) white adipose tissue (WAT), with no change in the percentage of the thermogenic and metabolically protective brown adipose tissue (p = 0.7756, unpaired t-test, Fig. 1d). This tendency towards increased peripheral fat in 2–3 mo 3xTg-AD mice was not reflected in changes in circulating plasma levels of total free fatty acids (p = 0.9386, unpaired t-test, Fig S1a) or triglycerides (p = 0.6880, unpaired t-test, Fig S1b); however, the concentration of leptin, a key metabolic hormone that correlates with adiposity and is itself secreted by adipose tissue, was significantly elevated in 3xTg-AD plasma (p = 0.0301, unpaired t-test, Fig S1c). By 10–11 mo, the increase in visceral WAT in 3xTg-AD mice was highly significant (p = 0.0005, unpaired t-test, Fig. 1g). This peripheral lipid accumulation shows an interesting parallel with our previous findings of increased lipid accumulation in the brains of young 2-month-old 3xTg-AD mice and in the post-mortem brains of AD subjects15.

Since lipotoxicity can involve metabolic and immune dysfunction, we also dissected the liver and spleen from 2–3 mo and 10–11 mo WT and 3xTg-AD mice to examine any gross changes in organ weight (Fig. 1e,f,j,k). The liver, a central hub for lipid metabolism that regulates uptake, esterification, oxidation, and secretion of fatty acids showed no weight change between WT and 3xTg-AD mice at 2–3 mo (p = 0.2074, unpaired t-test, Fig. 1e) but tended to be increased in 10–11 mo 3xTg mice (p = 0.0543, unpaired t-test, Fig. 1j). Similarly, the spleen, which plays a key role in blood filtering and regulates the secretion of proinflammatory cytokines, showed a significant enlargement in 3xTg-AD mice at 2–3 mo (p = 0.0128, unpaired t-test, Fig. 1f) that remained significant by 10–11 mo (p = 0.0005, unpaired test, Fig. 1k). In line with this, previous studies in 3xTg-AD mice have shown alterations in the spleen including increased organ weight and T cells at older ages35,36,37,38.

Together, these data show that 3xTg-AD mice develop increased peripheral adiposity, including increased body weight, peripheral fat and leptin levels, as well as hepatosplenomegaly.

Diabetes-like dysfunction in insulin and glucose handling occur in young 3-month-old 3xTg-AD mice

Through its communication with other organs, adipose tissue can regulate diverse processes including insulin sensitivity and glucose metabolism39. Under physiological conditions, pancreatic beta cells secrete insulin post prandially to increase the storage of excess glucose in the liver to reduce blood sugar. However, in type-2 diabetes, a reduction in insulin secretion and a reduced sensitivity to insulin cause an excess of blood glucose leading to chronic hyperglycemia.

Here, we used the glucose tolerance test (GTT) as an indicator of proper peripheral insulin response/secretion. Mice were food-deprived for 16 h prior to the administration of a single intraperitoneal (i.p) dextrose (1 g/kg) injection. Plasma glucose levels were measured prior to dextrose injection (time 0) and then every 15 min for 90 min (Fig. 2a). Repeated measures (RM) analysis of variance (ANOVA) identified a significant time effect (F(6, 72) = 77.03, p < 0.0001), strain effect (F(1, 12) = 22.14, p = 0.0005), and interaction (F(6, 72) = 8.546, p ≤ 0.0001), between WT and 3xTg-AD glucose levels. Post hoc multiple comparison tests showed significant strain differences at t15-75 min post injection (Fig. 2a). In addition, the food-deprived 3xTg-AD mice had higher blood glucose compared to WT mice, prior to dextrose injection (t0) (unpaired t-test p = 0.0021, Fig. 2b). The slope of the first measurement following the dextrose injection was used as an indicator of maximum tolerance to glucose and showed a significantly more pronounced glucose spike by the 3xTg-AD mice (unpaired t-test p = 0.0037, Fig. 2c). Similarly, the area of the curve (AOC)40 showed a significant impairment in clearance of blood glucose, suggesting reduced glucose tolerance by the 3xTg-AD mice (unpaired t-test, p = 0.0024, Fig. 2d). Together, the GTT test showed dysregulation of glucose handling in young 3xTg-AD mice.

Figure 2
figure 2

Diabetes-like dysfunction in insulin and glucose handling occur in young 3-month-old 3xTg-AD mice. (a–d) Glucose tolerance test (GTT), 16-h food-deprived mice were given an intraperitoneal injection of dextrose (1 mg/g body weight). Glucose concentration (mmol/L) was measured from blood sampled from the tail vein at the times indicated (n = 6–8 mice/group). (a) Repeated measures (RM) analysis of variance (ANOVA) identified a significant time (F(6, 72) = 77.03, p ≤ 0.0001), strain (F(1, 12) = 22.14, p = 0.0005), and interaction (F(6, 72) = 8.546, p ≤ 0.0001) between WT and 3xTg-AD glucose levels. Post hoc multiple comparison analysis showed significant differences at t15-75 min post injection. (b) blood glucose levels between WT and 3xTg-AD mice prior to injection of glucose (p = 0.0021). (c) Slope of the first 15 min after glucose injection (p = 0.0037). (d) Area of the curve of the GTT (p = 0.0024). (e–h) Insulin tolerance test (ITT), 16-h food-deprived mice were given an intraperitoneal injection of insulin (0.5U/g body weight). Glucose concentration (mmol/L) was measured from blood sampled from the tail vein at the times indicated. (e) RM-ANOVA identified significant time (F(1.984, 23.81) = 18.89, p ≤ 0.0001) and strain effects (F(1, 12) = 7.555, p = 0.0176) between WT and 3xTg-AD glucose levels, but no significant interaction. Post hoc multiple comparison analysis showed significant differences at t15 minutes post injection. (f) Blood glucose levels between WT and 3xTg-AD mice prior to injection of insulin (p = 0.0021). (g) Slope of the first 15 min after insulin injection between WT and 3xTg mice (p = 0.1574). (h) Area of the curve of the ITT between WT and 3xTg-AD mice (p = 0.3482). (i–j) Unfasted plasma (i) insulin concentration (ng/ml) (n = 8–10, p = 0.2458), (j) glucose concentration (mmol/L) (n = 3, p = 0.0314), between WT and 3xTg-AD mice. Unpaired t-test. Error bars represent mean ± standard error of the mean (SEM). Significance level was set at p ≤ 0.05. * ≤ 0.05, ** ≤ 0.05, **** ≤ 0.0001.

Full size image

Insulin suppresses glucose release from the liver and promotes glucose uptake by muscle. The degree to which blood glucose levels fall in response to insulin administration is indicative of insulin sensitivity. To assess if the 3xTg-AD mice exhibit signs of insulin resistance/insensitivity, we employed the insulin tolerance test (ITT). Mice were fasted 1 h prior to the administration of a single i.p insulin (0.5U/kg) injection. Plasma glucose levels were taken prior to insulin injection for baseline (time 0) assessment and every 15 min for 90 min (Fig. 2e). RM-ANOVA identified a significant time effect (F(1.984, 23.81) = 18.89, p < 0.0001), strain effect (F(1, 12) = 7.555, p = 0.0176), and no significant interaction (F(6, 72) = 0.9370, p = 0.4741) between WT and 3xTg-AD glucose levels. Post hoc multiple comparison tests showed significant strain differences at t15 minutes post injection (Fig. 2e). In addition, the 3xTg-AD mice showed higher baseline glucose compared to WT mice (unpaired t-test, p = 0.0021, Fig. 2f). The slope of the first measurement following the insulin injection showed a trend towards a sharper response by the 3xTg-AD mice (unpaired t-test, p = 0.1574, Fig. 2g). The area of the curve (AOC) showed no significant difference (unpaired t-test, p = 0.3482, Fig. 2h). Unfasted insulin was not significantly changed (unpaired t-test, p = 0.2458, Fig. 2i) while unfasted glucose was significantly elevated in 3xTg-AD mice (unpaired t-test, p = 0.0314, Fig. 2j). The weight of the insulin-secreting pancreas was unchanged at these ages (Fig S1d). In sum, the ITT results showed dysregulation of insulin handling in young 3xTg-AD mice.

Together, these data show that 3xTg-AD mice have increased fat and gluco-regulatory abnormalities at presymptomatic ages making them a good model to study the peripheral effects of the lipid modulating drug SCD.

1-month central SCDi does not alter peripheral fat accumulation

SCDs are central lipogenic enzymes throughout the body, catalyzing the desaturation of the saturated fatty acids (SFA), palmitate and stearate, into the MUFAs, palmitoleate and oleate, respectively. Whole body knock-out of Scd1 in mice improves metabolic fitness including reduced adiposity, plasma leptin and FFAs27,28. It has also been reported that gene expression is increased in the brain for Scd1 in 3xTg-AD mice and for SCD/SCD5 in AD patients, along with associated changes in their fatty acid substrates and products10,15,24,25,41. In previous work, we used a commercially available SCD inhibitor (SCDi) (Abcam, C20H22ClN3O3) with high potency (IC50 4.5 nM) and showed that intracerebroventricular infusion of this inhibitor in 3xTg-AD mice stimulated marked changes in hippocampal gene expression, a decrease in microglia activation, rescue of dendritic spines and structure, and reversal of learning and memory deficits. Similar effects on hippocampal gene expression and dendritic spines were found using a structurally distinct SCDi15,25, suggesting that these are on-target effects. Given the importance of the CNS in regulation of peripheral metabolism, we therefore asked whether the beneficial effects of central SCDi administration might involve improvements in peripheral metabolism.

To assess if central SCD inhibition improves peripheral metabolic features of AD, we employed the same protocol as previously25, infusing the Abcam SCDi ICV for 1-month in memory-impaired 9-month-old 3xTg-AD mice and WT strain controls (Fig. 3a). Comparison of body weight at the end of the 1-month study (Fig. 3b) showed a significant strain effect with 3xTg-AD mice weighing more than WT mice (F(1, 48) = 21.26, p < 0.0001) but no drug effect or interaction. Post hoc analysis showed a significant difference between vehicle treated WT-V/3xTg-V (p = 0.0456) and WT-S/3xTg-S (p = 0.0023) and no significant difference between 3xTg-V/3xTg-S (p = 0.4294), 2-way ANOVA Tukey’s post hoc. Body composition analysis by EchoMRI showed a significant strain effect with 3xTg-AD mice having more fat than WT mice (F(1, 26) = 29.19, p < 0.0001) but no drug effect or interaction. Post hoc analysis showed a significant difference between vehicle treated WT-V/3xTg-V (p = 0.0040) and WT-S/3xTg-S (p = 0.0038) and no significant difference between 3xTg-V/3xTg-S (p = 0.9418), 2-way ANOVA Tukey’s post hoc (Fig. 3c). Lean mass was unchanged between strains, drug, and showed no interaction (Fig. 3d). Fat as a percent of total body weight showed a significant strain effect (F(1, 26) = 37.36, p p < 0.0001) but no drug effect or interaction. Post hoc analysis showed a significant difference between vehicle treated WT-V/3xTg-V (p = 0.0012) and WT-S/3xTg-S (p = 0.0010) and no significant difference between 3xTg-V/3xTg-S (p = 0.8903), 2-way ANOVA Tukey’s post hoc (Fig. 3e).

Figure 3
figure 3

1-month central SCDi does not alter peripheral fat accumulation. (a) Schematic of experimental groups, 9-month-old WT and 3xTg-AD implanted with intracerebroventricular osmotic pumps containing vehicle (V) or SCD inhibitor (S) for 1-month. (b) Body weight in grams (g) (n = 12–14 animals/group) at sacrifice showed a significant strain effect with 3xTg-AD mice weighing more than WT mice (F(1, 48) = 21.26, p < 0.0001) but no drug effect or interaction. Tukey’s post hoc analysis showed a significant difference between vehicle treated WT-V/3xTg-V (p = 0.0456) and WT-S/3xTg-S (p = 0.0023) and no significant treatment effect 3xTg-V/3xTg-S (p = 0.4294). (c–e) Body composition by EchoMRI (n = 7–8 animals/group). (c) 2-way ANOVA showed a significant strain effect with 3xTg-AD mice having more fat than WT mice (F(1, 26) = 29.19, p < 0.0001) but no drug effect or interaction. Tukey’s post hoc analysis showed a significant difference between vehicle WT-V/3xTg-V (p = 0.0040) or drug treated WT-S/3xTg-S (p = 0.0038) and no significant difference between 3xTg-V/3xTg-S (p = 0.9418). (d) lean mass was unchanged between strains, drug, and showed no interaction. (e) fat mass as a percentage of body weight showed a significant strain effect (F(1, 26) = 37.36, p < 0.0001) but no drug effect or interaction. Tukey’s post hoc analysis showed a significant difference between vehicle treated WT-V/3xTg-V (p = 0.0012) and WT-S/3xTg-S (p = 0.0010) and no significant difference between 3xTg-V/3xTg-S (p = 0.8903). (f–h) Adipose tissue weight in grams (g) as a percent (%) of total body weight (n = 11–12 animals/group). (f) visceral white (g), subcutaneous white, (h) interscapular brown. A significant strain effect on (f) visceral (F(1, 42) = 31.87, p < 0.0001), post hoc analysis showing a significant difference between vehicle treated WT-V/3xTg-V (p = 0.0002) and WT-S/3xTg-S (p = 0.0095) and no significant difference between 3xTg-V/3xTg-S (p = 0.5733), 2-way ANOVA Tukey’s post hoc. (g) subcutaneous WAT also showed a significant strain effect (F(1, 42) = 12.04, p = 0.0012) but no drug or interaction. Post hoc analysis showed a significant difference between vehicle treated WT-V/3xTg-V (p = 0.0348) and WT-S/3xTg-S (p = 0.1702) and no significant difference between 3xTg-V/3xTg-S (p = 0.7046), 2-way ANOVA Tukey’s post hoc. Moreover, we did not detect any changes in BAT % in any group (h). (i–k) Plasma concentration of (i) free fatty acids (n = 7–10 animals/group), (j) triglycerides (n = 7–10 animals/group), (k) leptin between WT and 3xTg-AD mice (n = 6–11 animals/group). No effect was observed between vehicle treated WT and 3xTg-AD mice (p = 0.9995, 2-way ANOVA Tukey’s post hoc), however, plasma free fatty acids were elevated in 3xTg-AD mice treated with SCDi compared to Vehicle (p = 0.0043, 2-way ANOVA Tukey’s post hoc). (j) no significant strain or drug effect on total triglycerides was detected, 2-way ANOVA. (k) levels of leptin showed a significant strain effect (F(1, 29) = 8.929, p = 0.0057) but no drug effect or interaction. Post hoc analysis showed no significant differences between any groups, 2-way ANOVA Tukey’s post hoc. Error bars represent mean ± standard error of the mean (SEM). Significance level was set at p ≤ 0.05.*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

Full size image

Since unoperated 3xTg mice had an increase in WAT (Fig. 1), we again dissected fat pads from visceral (abdominal) WAT (Fig. 3f), subcutaneous (hind limb) WAT (Fig. 3g) and interscapular BAT (Fig. 3h). We found a significant strain effect on visceral WAT % (F(1, 42) = 31.87, p < 0.0001), with post hoc analysis showing a significant difference between vehicle treated WT-V/3xTg-V (p = 0.0002) and WT-S/3xTg-S (p = 0.0095) and no significant drug effect 3xTg-V/3xTg-S (p = 0.5733), 2-way ANOVA Tukey’s post hoc. Subcutaneous WAT % also showed a significant strain effect (F(1, 42) = 12.04, p = 0.0012) but no drug or interaction effect. Post hoc analysis showed a significant difference between vehicle treated WT-V/3xTg-V (p = 0.0348) and WT-S/3xTg-S (p = 0.1702) and no significant drug effect 3xTg-V/3xTg-S (p = 0.7046), 2-way ANOVA Tukey’s post hoc. Moreover, we did not detect any changes in BAT % in any group at this older symptomatic age (Fig. 3h).

To determine if plasma metabolites were altered by ICV-SCDi, we measured plasma free fatty acids, triglycerides, and leptin levels. Intriguingly, although there was no baseline strain difference in plasma free fatty acid levels between WT and 3xTg-AD mice, 3xTg-AD mice treated with ICV SCDi showed a significant increase (3xTg-V/3xTg-SCDi, p = 0.0043, 2-way ANOVA, Tukey’s post hoc), suggestive of fatty acid clearance into the blood (Fig. 3i). No significant strain or drug effect on total triglycerides was detected (Fig. 3j, 2-way ANOVA), however, levels of the adipose signalling hormone leptin showed a significant strain effect (F(1, 29) = 8.929, p = 0.0057) with no drug effect or interaction. Post hoc analysis showed no significant differences between any groups (2-way ANOVA, Tukey’s post-hoc, Fig. 3k).

In the mouse, Scd1 is ubiquitously expressed and shows highest expression in highly metabolic tissues. To determine if ICV-SCD altered peripheral organ weight we measured the weight of the liver and spleen. No significant treatment or strain changes were observed on liver weight (Fig S2a). As at younger ages, the spleen (Fig S2b) showed a significant strain effect (F(1, 29) = 8.929, p < 0.0001). Post hoc analysis showed a significant difference between vehicle WT-V/3xTg-V (p = 0.0157) and SCDi treated WT-S/3xTg-S (p = 0.0009) but no significant drug effect, 3xTg-V/3xTg-S (p = 0.5725), 2-way ANOVA Tukey’s post hoc.

Together, these data demonstrate that ICV SCDi for one month does not alter the gross peripheral fat accumulation and associated circulating leptin levels seen in the 3xTg-AD mice.

1-month central SCDi does not improve insulin and glucose handling

Since young 3xTg-AD mice exhibited impaired glucose regulation (Fig. 2) and glucose sensitivity is improved in whole-body Scd1-KO mice, we next assessed the impact of central SCDi infusion on glucose and insulin tolerance.

The GTT test (Fig. 4a) showed significant group (F(3, 25) = 3.078, p = 0.0458) and time effects (F(3.865, 96.62) = 36.24, p < 0.0001) with no interaction by 2-way RM-ANOVA. Indeed, resting glucose levels were significantly higher in 3xTg-AD mice compared to WT (F(1, 25) = 13.45, p = 0.0012). Tukey’s post hoc analysis showed a trend towards higher resting glucose levels in 3xTg-V mice (p = 0.0863), but no drug effect or interaction was observed (Fig. 4b). The slope of the first measurement following the dextrose injection was used as an indication of the tolerance to glucose and showed no strain or drug effect, 2-way ANOVA (Fig. 4c). Similarly, the area of the curve (AOC) showed no strain or treatment effect on the response to glucose, 2-way ANOVA (Fig. 4d).

Figure 4
figure 4

1-month central SCDi does not improve insulin and glucose handling. (a–d) Glucose tolerance test (GTT), 16-h food-deprived mice were given an intraperitoneal injection of dextrose (1 mg/g body weight, n = 7–8 animals/group). (a) Glucose concentration (mmol/L) was measured from blood sampled from the tail vein at the times indicated. 2-way RM-ANOVA showed a significant group (F(3, 25) = 3.078, p = 0.0458) and time effects (F(3.865, 96.62) = 36.24, p < 0.0001) with no interaction. (b) Blood glucose levels between WT and 3xTg-AD mice prior to injection of glucose (time 0) were significantly higher in 3xTg-AD mice compared to WT (F(1, 25) = 13.45, p = 0.0012). Tukey’s post hoc analysis showed a trend towards higher resting glucose levels in 3xTg-V mice (p = 0.0863), but no drug effect or interaction. (c) Slope of the first 15 min after glucose injection between WT and 3xTg-AD mice showed no strain, drug effect, or interaction by 2-way ANOVA. (d) Area of the curve of the GTT between WT and 3xTg-AD mice by 2-way ANOVA showed no strain or treatment effect. (e–h) Insulin tolerance test (ITT), 1-h fasted mice were given an intraperitoneal injection of insulin (0.5U/g body weight, n = 7–8 animals/group). (e) glucose concentration (mmol/L) was measured from blood sampled from the tail vein at the times indicated. The ITT test showed significant group (F(3, 25) = 4.675, p = 0.0100) and time effects (F(3.315, 82.89) = 14.16, p < 0.0001). (f) Blood glucose levels between WT and 3xTg-AD mice prior to injection of insulin (time 0) showed higher baseline glucose in 3xTg-AD mice compared to WT (2-way ANOVA, (F(1, 25) = 9.200, p = 0.0056) but no drug effect or interaction. (g) Slope of the first 15 min after insulin injection between WT and 3xTg-AD mice showed no strain or treatment effect, 2-way ANOVA. (h) Area of the curve of the ITT between WT and 3xTg-AD mice showed no strain or treatment effect, 2-way ANOVA. (i–k) Unfasted plasma (i) insulin concentration (n = 7–10 animals/group) showed a strain effects (F(1, 46) = 7.513, p = 0.0087), (j) glucose concentration (n = 11–14 animals/group) showed a strain effect (F(1, 29) = 7.020, p = 0.0129) (k) TNFa (n = 7–9 animals/group) showed a strain effect (F(1, 27) = 6.326, p = 0.0182), by 2-way ANOVA. No significant drug effect or interaction was observed between WT and 3xTg-AD mice. Error bars represent mean ± standard error of the mean (SEM). Significance level was set at p ≤ 0.05.

Full size image

To evaluate signs of insulin resistance/insensitivity we performed the ITT test (Fig. 4e). The ITT test showed a significant group (F(3, 25) = 4.675, p = 0.0100) and time effects (F(3.315, 82.89) = 14.16, p < 0.0001). The 3xTg-AD mice showed higher baseline glucose compared to WT mice (2-way ANOVA, (F(1, 25) = 9.200, p = 0.0056) but no drug effect or interaction was observed (Fig. 4f). The slope of the first measurement following the insulin injection was used to infer the response to insulin and showed no strain or treatment effect, 2-way ANOVA (Fig. 4g). The area of the curve (AOC) also showed no strain or treatment effect, 2-way ANOVA (Fig. 4h). To assess the unfasted resting glucose and insulin levels we measured resting plasma insulin (Fig. 4i) and glucose (Fig. 4j) levels and found significant strain effects (F(1, 29) = 7.020, p = 0.0129), (F(1, 46) = 7.513, p = 0.0087) respectively, 2-way ANOVA, but no significant effect of treatment or interaction. We also detected no weight difference in insulin secreting pancreas at these ages (Fig S2c).

Increased adiposity causes inflammation and increased cytokine production that can impact both the brain and periphery42. We have previously shown in LPS-treated mouse microglia, that SCDi or Scd1-KO dampens Tumor necrosis factor alpha (TNFa)25. TNFa is a cytokine and adipose-tissue-associated adipokine known to contribute to insulin resistance associated to obesity43. Here, plasma TNFa levels showed a strain dependent increase in 3xTg-AD mice (F(1, 27) = 6.326, p = 0.0182), 2-way ANOVA) however, no significant drug effect or interaction was observed (Fig. 4k). Thus, these data show that the significantly higher peripheral fat accumulation and cytokines in 3xTg-AD mice are not significantly modulated following 1-month central administration of SCDi.

Collectively, these findings show that 3xTg-AD mice have increased resting and fasted glucose levels, that their response to insulin and glucose challenge is moderately affected in mid-life, and that these metabolic features are not modulated by 1-month central administration of SCDi.

Discussion

This study assessed the ability of central SCD inhibition to regulate peripheral metabolism in the 3xTg-AD mouse model. Using a genetic model of AD where dementia-causing mutations lead to cognitive impairment in mid-life, we confirmed that 3xTg-AD mice exhibit some metabolic deficits that precede these cognitive symptoms. Moreover, we show that while 3xTg-AD mice have early and mid-life metabolic changes, including increased peripheral fat accumulation and circulating leptin as well as impaired resting glucose and glucose tolerance, 1-month ICV infusion of SCDi in mid-life does not normalize these parameters.

The ICV infusion paradigm for SCDi used here results in remarkable improvements in brain structure and function of 3xTg-AD mice, including a recovery of learning and memory in mid-life25. Knowing that these mice also have peripheral metabolic deficits, we wondered if the benefits may be due in part to improved metabolism in the periphery. Our present results reveal that 1-month ICV SCDi does not significantly affect the aspects of peripheral metabolism measured here, indicating that improvements of these peripheral metabolic features are not needed for the benefits of central SCDi treatment on the brain. This is important as it suggests that targeting CNS MUFA metabolism may be sufficient to have cellular and cognitive improvements in AD. Intriguingly, since SCDi-induced increases in levels of plasma free fatty acids were detected, it is conceivable that longer term CNS treatment with SCDi may eventually impact peripheral metabolism.

AD and peripheral metabolism

AD patients exhibit significant metabolic symptoms including body weight changes, dyshomeostasis of central and peripheral lipids, diminished brain glucose uptake, and Type 2 diabetes. Although these alterations occur prior to the initial mental decline and progressively worsen as the disease advances, it is difficult to establish whether they are AD prodromes, risk factors, or both. Our data suggests that expression of dementia-causing genes (APP, PS1, Tau) is sufficient to cause peripheral metabolic deficits including increased adiposity and diabetes-like changes in glucose handling and that occur in the absence of changes in dietary or environmental factors. Indeed, other groups have also reported changes in peripheral parameters in human familial AD patients and genetic AD animal models5. These observations are in line with the idea that treating the CNS sufficiently early could interrupt the cascade that leads to neurodegeneration and peripheral metabolic syndromes.

Pleiotropic effects of SCD inhibition in peripheral metabolism

Studies in transgenic mouse models have demonstrated an essential role of Scd1 in regulating cellular processes including lipid synthesis and oxidation, thermogenesis, hormonal signaling, and inflammation. Scd1 was initially identified in the ‘asebia’ mouse that has a naturally occurring mutation in the Scd1 gene and exhibits alopecia, sebocyte hypoplasia, and resistance to leptin deficiency induced obesity44. Indeed, Scd1 itself is a known target of leptin and many of leptin's metabolic effects are mediated through its effects on Scd145,46,47. Subsequently, whole-body Scd1 knock-out mice were engineered and found to be resistant to high-fat diet induced obesity, with reduced hepatic as well as circulating triglycerides and cholesterol esters, dry eye, alopecia, dermatitis, and increased skin barrier permeability28,48. These intriguing results spawned multiple tissue-specific Scd1 deletion studies. For instance, liver-specific Scd1 deficiency protects mice from carbohydrate-induced de novo lipogenesis49. Adipose-specific deficiency results in decreased inflammation in adipocytes and increased insulin-independent glucose uptake50. Interestingly, the obesity-resistant lean metabolic phenotype of global Scd1 deficiency was recapitulated not by liver or adipose-specific Scd1 deletion but by deletion in the skin51. Deficiency of Scd1 in the skin resulted in significant increases in energy expenditure and protection from diet-induced obesity, hepatic steatosis, and glucose intolerance. These mice also had an inability to retain heat due to significant alopecia and poor skin integrity.

Together, this suggests that SCD has far reaching effects on central and peripheral metabolism, increasing the risk of unwanted secondary effects from systemically administered inhibitors.

SCDi as a therapeutic target for neurodegenerative diseases

This study in combination with our prior findings15,25 shows that CNS SCDi is sufficient to have major beneficial effects on core AD symptoms including cognition and inflammation. Indeed, Scd inhibition is showing beneficial effects in multiple neurodegenerative disease models including Multiple sclerosis (MS), Parkinson’s disease (PD), and AD15,18,19,20,21,22,23,25,26. However, the studies in MS and PD have used oral SCDi, making it unclear if the benefits come from the CNS or periphery. A clinical trial using oral administration of an SCDi in PD patients was launched by Yumanity therapeutics in 2021, and the results of this trial will provide vital information on the feasibility of systemically administered SCDi to treat neurodegenerative diseases.

When thinking of translating SCDi into a clinical setting, it is essential to consider the most specific target and most effective mode of administration to minimize off-target and unwanted side-effects. Our data suggests that administration of SCDi into the CNS will be sufficient to have major beneficial effects on AD and potentially other neurodegenerative diseases. Future studies using CNS-specific SCDi in other neurodegenerative diseases will be essential to determine if this is the case. In this regard, delivery systems that are better able to selectively target genes and locations of interest, such as antisense oligonucleotides and nasal sprays, are already used in the clinic for other indications; our data supports developing such modes of administration for SCD inhibition in AD as well.

Methods

Experiments were approved by the Institutional Animal Care Committee of the Centre de Recherche du Centre Hospitalier de l’Université de Montréal (CRCHUM) following the Canadian Council of Animal care and ARRIVE guidelines and regulations.

3xTg-AD mice and strain controls

3xTg-AD mice (Jackson Laboratory MMRRC stock #: 034830) and their WT strain controls (B6129SF2/J, Jackson Laboratory stock #: 101045) were purchased from Jax mice. Briefly, 3xTg-AD mice were originally derived by Oddo and colleagues by co-microinjecting two independent transgenes encoding human APPSwe and the human tauP301L (both under control of the neuron-specific mouse Thy1.2 regulatory element) into single-cell embryos harvested from homozygous mutant PS1M146V knock in (PS1-KI) mice34. Wildtype mice are the PS1-knock-in background strain (C57BL/6J × 129S134. All mice were bred in-house, maintained in identical housing conditions (22 °C, 50% humidity, 12 h Light/Dark cycle OFF 10am ON 10 pm) and given free access to a standard rodent diet (#2018, Harlan Teklad) and water. 3xTg-AD mice undergo a progressive increase in soluble amyloid beta peptide levels, with intracellular amyloid immunoreactivity being detected in some brain regions as early as 3–4 months. Synaptic transmission and long-term potentiation are demonstrably impaired in mice 6 months of age. Evidence of gliosis and inflammation is present by at least 7 months52. Between 12 and 15 months, amyloid plaques and aggregates of conformationally altered and hyperphosphorylated tau are detected in the hippocampus34,53. Female mice were used for all experiments and sex and age matched. Mice were euthanized using a lethal dose of ketamine (Bimeda-MTC)/xylazine (Bayer Healthcare).

Metabolic profiling

Insulin tolerance test (ITT)/Glucose tolerance test (GTT)

For the glucose tolerance test (GTT), mice were fasted 16 h (5 pm-9am) followed by i.p injection of dextrose (1 g/kg of body weight). For insulin tolerance tests (ITT), mice were fasted for 1 h (12-1 pm) prior to intraperitoneal (i.p.) injection of insulin (0.5U/kg). For both GTT and ITT, glucose concentration (mmol/L) was measured using blood glucose strips and corresponding glucometer (FreeStyle Lite, Abbott) from blood sampled from the tail vein at 0, 15-, 30-, 45-, 60-, and 90-min post injection. Slope of the first 15 min was achieved by linear regression fitting in GraphPad Prism Version 8 (GraphPad Software, Inc). Area of the curve (AOC) was calculated by using the pre-injection glucose concentration (time 0) of each mouse as baseline and including peaks below zero for ITT and only positive peaks for GTT.

Body composition

Fat and lean mass was measured using an EchoMRI-100 body composition analyzer (version 2008.01.18, EchoMRI LLC) at the rodent cardiovascular core facility of the CRCHUM. % fat mass and % lean mass was obtained by dividing by total body weight (g).

Organ weights

Visceral (abdominal) white adipose tissue, subcutaneous (hind limb) white adipose tissue and brown (interscapular) adipose tissue were dissected and weighed. % organ weight was calculated by dividing the organ weight by the total body weight (g). Whole liver, spleen and pancreas were dissected and weighed.

Blood collection

Blood was collected from the posterior vena cava at the time of sacrifice. Samples were transferred to a microtainer containing K2EDTA (BD Biosciences) inverted 20 times, and the plasma extracted after centrifuging at 1000 g for 20 min at 4 °C. All samples were stored at − 80 °C until use.

Insulin, Leptin and TNFa plasma measurements

AlphaLISA assay kits (#AL204 for insulin, AL521 for Leptin and AL505 for TNFa) from PerkinElmer (Waltham, MA) and an Envision 2104 plate reader from the same supplier were used to assess plasma levels of the corresponding analyte using white 96-well microplates. A standard curve was run on each plate containing samples. Omnibead calibrators (PerkinElmer) were utilized to ensure proper operation of the assay and instrument. The analyses were carried out at the Cellular Physiology core facility of the CHUM Research Center.

Free fatty acids and triglycerides measurements

Plasma Triglycerides (TG) were evaluated using a commercial kit (GPO Trinder, Sigma-Aldrich, Saint Louis, USA) in duplicate. Calculations were performed to quantify triacylglycerol levels using triolein equivalent as per manufacturer recommendations. Plasma FFA were assessed using the NEFA C kit in duplicate (Wako Chemicals, Neuss, Germany) according to the manufacturer's protocol. The analyses were carried out at the Cellular Physiology core facility of the CHUM Research Center.

SCD inhibitor

SCD inhibitor used in this study: ab142089 (Abcam, C20H22ClN3O3, 4-(2-Chlorophenoxy)-N-[3 [(methylamino)carbonyl]phenyl]-1-piperidinecarboxamide, M.W. 387.87).

In vivo surgical procedures

Intracerebroventricular (ICV) osmotic pumps

For ICV infusions of SCDi or vehicle, mice were locally injected with buprivacaine (Hospira) and operated under isoflurane anesthesia (Baxter). Brain cannulae attached to Alzet osmotic pumps were stereotaxically implanted at 0.0 mm antero-posterior and 0.9 mm lateral to Bregma and the pumps placed under the back skin according to manufacturer's instructions. The 28-day Alzet osmotic pumps used in these experiments (0.11 μl/h infusion rate, model 1004; Durect) were primed for 48 h and began pumping immediately when implanted. The SCD inhibitor was dissolved in DMSO (Sigma-Aldrich) and infused at a final concentration of 80 µM in sterile aCSF (148 mM NaCl, 3 mM KCl, 1.7 mM MgCl2, 1.4 mM CaCl2, 1.5 mM Na2PO4, 0.1 mM NaH2PO4). Vehicle pumps contained the same volumes as the SCDi pumps (0.8% DMSO/aCSF).

Statistical analyses

All statistical analyses were performed using GraphPad Prism, Version 8 (GraphPad Software, Inc). When comparing one independent variable a t-test was used, otherwise a 2 × 2 experimental design (strain and treatment) or 3 × 2 experimental design (strain, treatment and timepoint) was chosen and analyzed by ANOVA followed by post-hoc multiple comparisons as indicated in the legends. Error bars represent mean ± standard error of the mean (SEM). Significance level was set at p ≤ 0.05 as indicated in the figure legends. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.