Intranasal Insulin Helps Overcome Brain Insulin Deficiency and Improves Survival and Post-Stroke Cognitive Impairment in Male Mice

Abstract

Obesity increases the risk for stroke and is associated with worse post-stroke outcomes; however, the mechanisms are poorly understood. Diet-induced obesity leads to insulin resistance and subsequently, brain insulin deficiency. The purpose of this study was to investigate the potential impact of brain insulin deficiency on post-stroke outcomes. To accomplish this, brain insulin levels were assessed in male C57BL/6J (B6) mice placed on either a standard diet or 54% kcal high-fat diet, a known model of insulin resistance. Mice were subjected to either a sham surgery (control) or 30-minute middle cerebral artery occlusion to induce an ischemic stroke and administered either intranasal saline (0.9%) or intranasal insulin (1.75 U) twice daily for 5 days beginning on day 1 post-stroke. High-fat diet-induced brain insulin deficiency was associated with increased mortality, neurological and cognitive deficits. On the other hand, increasing brain insulin levels via intranasal insulin improved survival, neurological and cognitive function in high-fat diet mice. Our data suggests that brain insulin deficiency correlates with worse post-stroke outcomes in a diet-induced mouse model of insulin resistance and increasing brain insulin levels may be a therapeutic target to improve stroke recovery.

Graphical Abstract

The purpose of this study was to investigate the potential impact of brain insulin deficiency on post-stroke outcomes. While high-fat diet-induced brain insulin deficiency correlated with worse survival and increased neurological and cognitive deficits, increasing brain insulin levels via intranasal insulin improved these impairments in high-fat diet mice.

Introduction

More than 50% of stroke survivors without diabetes have insulin resistance (1). Insulin resistance, diabetes, and acute hyperglycemia are associated with an increased stroke risk and worse outcomes following stroke (2–6). Insulin resistance was associated with increased mortality following stroke (7). Furthermore, diabetic patients with elevated fasting blood glucose were at increased risk for stroke compared to non-diabetic patients (8). Reducing blood glucose levels via the systemic administration of insulin improves stroke outcomes. For example, insulin administered intraperitoneally or subcutaneously had a beneficial effect on stroke damage (3, 9). Unfortunately, a potential complication associated with peripheral insulin administration is hypoglycemia, which can produce symptoms that mimic acute stroke (10–12).

Insulin resistance is associated with hyperglycemia and increased levels of circulating insulin or hyperinsulinemia (4, 13). Insulin from the periphery is transported to the brain via receptor-mediated uptake (14). Previous studies reported that hyperinsulinemia, which occurs during insulin resistance, saturates this transport system contributing to lower levels of insulin in the cerebrospinal fluid (15–17). High-fat diet-induced obesity leads to hyperinsulinemia and subsequently a reduction in brain insulin transport and levels, evident by a decrease in the levels of insulin in the cerebral spinal fluid (16, 17).

Brain insulin plays a vital role in angiogenesis, neurogenesis, oligodendrogenesis, synaptogenesis, myelination, axonal growth, and cognition (18–23). Brain insulin deficiency can result from insulin resistance and lead to reduced cognitive function (24). Intranasal administration of insulin may facilitate the positive neurological effects of insulin while diminishing the risk posed by peripheral insulin administration. Intranasal administration allows substances to bypass the blood brain barrier to rapidly enter the central nervous system (CNS) through the perivascular spaces, as well as perineural spaces of the olfactory and trigeminal nerves (14, 25). Intranasal insulin reduced lesion volume, decreased neuroinflammation, and improved both cognitive and motor function following traumatic brain injury (26). Furthermore, Intranasal administration of insulin-like growth factor I was protective against cerebral atrophy and focal cerebral ischemia in rats (27, 28). Collectively, these studies suggest that targeting brain insulin could also be beneficial in improving stroke outcomes.

While previous human studies demonstrate the correlation between peripheral insulin and stroke (3, 4, 29), to date, no studies have evaluated the potential benefit of increasing brain insulin levels on stroke outcome. The impact of brain insulin levels on stroke outcomes is not known. The purpose of this study is to demonstrate that brain insulin correlates with worse stroke outcome and that increasing brain insulin levels would improve post-stroke survival, neurological and cognitive deficits. Overall, these studies will reveal the importance of brain insulin levels in post-stroke recovery.

Materials and Methods

Animals

Mice were housed in a pathogen-free temperature-controlled environment and placed on a standard 12 hour light/dark cycle; the protocols and procedures were approved by Medical University of South Carolina Institutional Animal Care and Use Committee. The procedures and protocols were in compliance with the university guidelines, state and federal regulations, and the standards of the “Guide for the Care and Use of Laboratory Animals”. Animal Welfare Assurance Numbers on file with the NIH Office of Laboratory Animal Welfare (OLAW) are A3428-01 (Medical University of South Carolina). The university is accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International (AALAC, Intl). The experiments were reported based on ARRIVE guidelines.

Experiment 1: Establishing High-fat diet mouse model of brain insulin deficiency

Diet-induced model of obesity and insulin resistance

Male and female C57BL/6J (B6) mice were purchased from Jackson Laboratory (#000664, RRID:IMSR_JAX:000664, Bar Harbor, ME). At 4 weeks of age, mice were placed on either the standard (STD) diet consisting of 10% kcal from fat from Research Diets Inc. (#D12450K, New Brunswick, NJ) or high-fat diet (HFD) with 54% kcal from lard from Research Diets Inc. (#D17012508, New Brunswick, NJ) ad libitum. Mice were randomly assigned to these feeding groups based on cage number.

Assessment of Cerebrospinal Fluid Insulin Ratio

Cerebrospinal fluid (CSF) was isolated in a subset of mice to assess brain insulin levels in STD (n = 8) and HFD (n = 11) mice at 24 weeks of diet according to previously published protocol with some modifications (L. Liu & Duff (30), 2008). Briefly, mice were anesthetized with isoflurane and secured in the prone position with the head fixed on stereotaxic instrument at 30° angle. A sagittal incision was made, the subcutaneous tissue and neck muscles were separated to expose the dura mater of cisterna magna. The cisterna magna was punctured using a needle tip and a pipette was used to collect CSF. Mice were euthanized immediately following CSF collection. CSF was flash frozen in liquid nitrogen until use for measurements of insulin levels as described below.

Tissue Preparation

Following euthanasia, the brains were dissected per our previously published protocol (31). Mice were anesthetized with isoflurane and transcardially perfused with 1x PBS after 4 hours of fasting. The hippocampi were dissected, flash frozen in liquid nitrogen, and stored at −80 °C until needed according to our previously published protocol for lysate preparation (31). Briefly tissue was homogenized in tissue protein extraction reagent (TPER; #78510, Thermo Fisher; Rockford, IL) supplemented with a protease inhibitor cocktail tablet (EDTA; #11836170001; Sigma-Aldrich; St. Louis, MO) according to the manufacturer’s instructions using a 1mL syringe. Following homogenization, the tissue homogenate was centrifuged for 5 minutes at 14,000 RPM. The supernatant was used for protein assay.

Pierce Protein Assay and Lysate Preparation

For the protein assay, Bovine Serum Albumin (BSA; #A2153; Sigma-Aldrich; St. Louis, MO) standards (0ug, 100ug, 250ug, 500ug, 750ug, 1000ug, 2000ug) were used to generate a standard curve. The standards and samples were mixed with Pierce Protein Assay Reagent (#22660; Thermo Fisher; Rockford, IL) according to the manufacturer’s instructions to assess absorbance using the Epoch plate reader (#1807092; BioTek; Winooski, VT). The protein concentration of the samples was interpolated from the BSA standard curve and lysates with 7 ug of protein were prepared for assessment of hippocampal insulin levels using enzyme-linked immunosorbent assay.

Insulin enzyme linked immunosorbent assay

Insulin levels were assessed in male mice in CSF after 24 weeks of diet or hippocampal lysates prepared from STD (n = 8) and HFD (n = 10) mice after 6 and 24 weeks of diet using the Mouse Insulin Elisa (EMD Millipore Corporation; #EZRMI-13K; Billerica, MA) according to the manufacturer’s instructions. Samples were run in duplicates. Insulin ELISA data was normalized to protein concentration of each sample.

Experiment 2: Examining the impact of brain insulin levels on stroke recovery

Middle Cerebral Artery Occlusion

Middle Cerebral Artery Occlusion (MCAO) was induced in mice at 16 weeks of age in both male (after 12 weeks on either STD; n = 19 or HFD; n = 35) and female (after 12 weeks on either STD; n = 12 or HFD; n = 12) similar to a previously published protocol with (Atkinson (32) et al., 2006) with some modifications. Mice were given saline and buprenex (0.1mg/ml) subcutaneously. A subset of male mice were anesthetized with intraperitoneal injection of ketamine (80 mg/kg) and xylazine (10 mg/kg; STD; n = 9, HFD; n = 22) and another subset of mice were anesthetized with continuously isoflurane vaporizer (Smiths Medical, San Clemente, CA; male: STD; n = 10, HFD; n = 13 and female: STD; n = 12 and HFD; n = 12) to improve survival. Mice were placed under a dissecting microscope from AmScope (#SM-3BZ-80S, Irvine, CA) and body temperature was continuously maintained and monitored during MCAO procedure with a heating pad. A midline incision was made on the neck to expose and clamp the left common carotid artery (CCA). The internal carotid artery (ICA) was ligated and a 6-0 silk suture was placed at the external carotid artery (ECA). The 20-mm nylon monofilament was coated with poly-L- lysine prior to insertion into the ECA and advanced to occlude the middle cerebral artery (MCA). Mice were placed in a heated recovery box and allowed to wake until removal of the filament for reperfusion. Following a 30-minute occlusion, mice were anesthetized with isoflurane and the filament was removed to allow for reperfusion. To confirm proper occlusion and reperfusion, cerebral blood flow (CBF) was measured using Laser Doppler flowmetry (MOOR VMS-LDF) from Moor Instruments, (United Kingdom) before MCAO (baseline), during occlusion, and after reperfusion as previously described (33, 34). Briefly, a small incision was made on the head to expose the skull 1-mm posterior and 5-mm lateral to bregma. Monochromatic light emitted from the probe was used to measure red blood cell movement, while the photodetector was used to measure Doppler shifted scattered light (35). An 80% decrease in CBF for occlusion and a return within 20% of baseline in CBF for reperfusion was the criteria met to confirm successful occlusion/reperfusion (Fig. 1). Once the filament was removed the incision exposing the CCA was sutured with 6-0 silk suture and the incision on top or the head was sealed using wound clips. A SHAM surgery, which involved the same surgical procedure except the MCA was not occluded, was performed on a subset of male mice (STD-SHAM-S; n = 2, HFD-SHAM-S; n = 2) as controls. SHAM-S groups were combined for analysis where no difference with diet existed for behavior and excluded from analysis and data presentation for all other measures. All raw data will be provided in repository. Standard post-operative care was performed on the mice per our approved protocols including monitoring weight daily for 5 days post-stroke, temperature, dehydration, and pain management.

Figure 1. Middle cerebral artery occlusion significantly reduced blood flow during occlusion in male and female mice.

(a) Percentage (%) of blood flow measured using a Laser Doppler during occlusion (white bar) and reperfusion (black bar) normalized to baseline blood flow in (a) male mice (n=24 for occlusion and n=18 for reperfusion); ****P<0.0001 based on unpaired t-test; t=7.854, df=40 and (b) female mice (n=28 for occlusion and n=28 for reperfusion); ****P<0.0001 based on unpaired t-test; t=6.851, df=54. Data presented as mean ± SD.

Intranasal Insulin Administration

Mice were randomly assigned to either saline (S; 0.9%, #786-560, G-Biosciences, St. Louis, MO) or human insulin (I; CAS# 11061-68-0; #19278, Sigma Aldrich, St. Louis, MO) treatment groups based on cage number. For intranasal administration, the mice were anesthetized with isoflurane and given no more than 10μl of intranasal saline or insulin via micropipette. The concentrations were chosen based on previous intranasal insulin studies (26, 36). Hence, 1.75 U of insulin was used in this study; mice were administered either intranasal saline (0.9%) or insulin (1.75 U) twice per day (morning and afternoon) for 5 consecutive days beginning one day post-stroke. The experimental design for Experiment 2: Examining the impact of brain insulin levels on stroke recovery Middle Cerebral Artery Occlusion is summarized in Fig. 2. Animals anesthetized with intraperitoneal injection of ketamine (80 mg/kg) and xylazine (10 mg/kg) were a part of the pilot study and had very low survival among those animals 3 HFD-S animals were excluded day 2 post-stroke for receiving the incorrect intranasal treatment, thus, experiment 2 consists of the following five groups: 1) Sham-S; n = 4 (male) and n = 6 (female), 2) STD-S; n = 10 (male) and n=6 (female), 3) STD-I; n = 6 (male) and n = 6 (female), 4) HFD-S; n = 16 (male) and n = 6 (female), and 5) HFD-I; n = 14 (male) and n = 6 (female).

Figure 2. Study design.

Summary of experimental timeline in mice on either standard diet (STD) or a high-fat diet (HFD) for assessing insulin levels in the brain (Experiment 1) and stroke outcomes following the middle cerebral artery occlusion (MCAO), model of stroke (Experiment 2).

Modified neurological severity score assessment

The severity of impairment following MCAO was assessed in both male and female mice using the modified neurological severity score (mNSS) as previously described (Hata et al., 1998). Briefly, the mice were scored on mobility using a scale of 0 (normal) to 5 (severe neurological deficits; see Table 1 for description) for 5 days post-stroke but reported in manuscript on days 1 and 2 post-stroke by a trained observer, blinded to diet and treatment groups (Male: SHAM-S; n = 4, STD-S; n = 10, STD-I; n = 6, HFD-S; n = 16, and HFD-I; n = 14, and Female: STD-S; n = 6, STD-I; n = 6, HFD-S; n = 6, and HFD-I; n = 6).

Table 1.

Descriptors of Modified neurological severity score (mNSS) assessment.

Score	Description of Assessment
0	Normal
1	Walks straight, rears to contralateral side when held by the tail
2	Normal posture at rest but circling to contralateral side when held by the tail
3	Circling to contralateral side at rest
4	No spontaneous movement
5	Unresponsive to stimulation, moribund, or dead

5 point scale of Modified neurological severity score (mNSS) assessment.

Novel Object Location Assessment

The novel object location task was used to assess cognition. The STD (n = 10) and HFD (n = 14) mice performed this task at baseline (8 weeks of age/4 weeks of diet) and 14 days post-stroke in male (control group: SHAM-S and STD-S combined; n = 5, STD-I; n = 6, n = 5, HFD-S; HFD-I; n = 5) and female (controls group: SHAM-S and STD-S combined; n=6, STD-I; n=5, n=9, HFD-S; and HFD-I; n=6) mice based on a protocol from Barker and Warburton, 2011 (37). During the exploration phase the mice were placed in a circular arena with two similar objects on opposing sides of the arena and allowed to explore for 10 minutes. One hour after the exploration phase the objects were taken out of the arena and cleaned with Rescue fragrance free disinfectant and deodorizer (#002240) from Virox Technologies Inc. (ON, Canada). The objects were placed back in the arena with one of the objects placed in the same location and the other object placed in a new (novel) location (37). Mice were allowed to explore the objects for 10 minutes. The SHAM-S and STD-S mice were combined to as “Controls” for post-stroke NOL. This task was completed with 4 arenas using Microsoft LifeCam HD cameras to record the behavioral tasks. The videos were acquired and the behaviors were tracked using Panlab SMART 3.0 software.

Statistical Analysis

Data analyses were performed using Prism v9 GraphPad Software, Inc (RRID:SCR_000306, La Jolla, CA). All data was reported as mean ± standard deviation (SD) with significance P<0.05. For experiment 1: Establishing High-fat diet mouse model of brain insulin deficiency; an unpaired t-test was used to evaluate difference between STD and HFD hippocampal and CSF Insulin levels.

For experiment 2: Examining the impact of brain insulin levels on stroke recovery: An unpaired t-test was used to evaluate the differences in blood flow percentage during occlusion and reperfusion normalized to baseline. For the interactions between diet, treatment and days post-stroke in body weight, a three-way Mixed-effects model was performed with Tukey’s post-hoc analysis. Curve comparison was performed to determine significance among survival curve data using the log-rank (Mantel-Cox) test. Three-way mixed-effects modeling was also used for mNSS with Sidak’s multiple comparison test. The percentage of total memory was calculated by taking the ratio of the sum of time spent in the old location (reference memory; T_ref) and the novel location (T_novel) to the total time spent exploring all locations including the T_ref, T_novel, and familiar (T_fam) locations [((T_ref + T_novel) / (T_ref + T_novel + T_fam)) *100]. Memory was determined to be intact if mice spent significantly more than 33.33% (by chance with 3 zones: T_ref, T_novel, and T_fam) of time exploring the T_ref and T_novel locations. Two different analyses were performed for NOL. First, a one-sample t-test was performed to ensure that the % of time spent exploring the novel object is greater than “by chance” using 33.33% as the theoretical mean. Second, a comparison was made to compare between STD and HFD at baseline (prior to stroke) using an unpaired t-test or among groups (combined controls: STD-SHAM, HFD-SHAM, and STD-S; STD-I; HFD-S; and HFD-I) at 14 days post-stroke using a one-way ANOVA with Tukey’s multiple comparisons test.

Results

HFD reduced Hippocampal and CSF Insulin Levels in male mice

To validate whether HFD leads to brain insulin deficiency, insulin levels in the hippocampus and CSF were assessed. Hippocampal insulin levels were significantly reduced by 50% in mice on HFD compared to mice on STD after 6 weeks of diet (unpaired t-test = 3.856; n = 8/group; df = 14; P = 0.0017; Fig. 3a). Similarly, hippocampal insulin levels were significantly reduced by 22% in mice on HFD compared to mice on STD after 24 weeks of diet (unpaired t-test = 2.138; n = 10/group; df = 18; P = 0.0464; Fig. 3b). CSF insulin levels were significantly reduced by 66% in HFD mice compared to STD mice after 24 weeks of diet (unpaired t-test = 3.635; STD n = 8; HFD n = 11; df = 17; P = 0.0020; Fig. 3c).

Figure 3. HFD reduced brain insulin levels.

Hippocampal insulin levels in standard (STD) and high-fat (HFD) diet male mice after (a) 6 weeks of diet; n=8 (STD) and n=8 (HFD); **P<0.01 based on unpaired t-test (P=0.0017; t=3.856, df=14) and (b) 24 weeks of diet; n=10 (STD) and n=10 (HFD); *P<0.05 based unpaired t-test (P=0.0464; t=2.138, df=18). (c) Cerebral spinal fluid (CSF) insulin levels in male mice after 24 weeks of diet; n=8 (STD) and n=11 (HFD); **P<0.01 based on unpaired t-test (P=0.0020; t=3.635, df=17). Data presented as mean ± SD.

HFD male mice weigh more than STD despite greater weight loss following stroke

The impact of stroke on body weight was assessed daily for 5 days (Fig. 4). Three-way mixed-effects modeling revealed a significant interaction with days post-stroke (F (1.915, 62.05) = 72.54, P<0.0001), diet (F (1, 46) = 55.10, P<0.0001), between days post-stroke and diet (F (5, 162) = 8.927, P<0.0001), and among days post-stroke, diet, and treatment (F (5, 162) = 4.256; P=0.0012). The HFD-S mice weighed approximately 35% more than the STD-S mice. Tukey’s multiple comparisons test revealed that HFD-S mice weight significantly more than STD-S mice at baseline (P < 0.0001), 1 (P = 0.0226), 2 (P = 0.0249), 3 (P = 0.0359), and 4 (P = 0.0399) days post-stroke. HFD-S mice lost on average 9% more weight than STD-S mice post-stroke. STD-S mice experienced a significant decrease in weight at 1 (P = 0.0327), 2 (P = 0.0048), and 3 (P = 0.0079) days post-stroke compared to baseline (0 day). Whereas, STD-I mice only experienced a significant decrease in weight at 2 (P = 0.0004) and 3 (P = 0.0072) days post-stroke compared to baseline. On the other hand, HFD-S mice weigh less at 2 (P = 0.0016), 3 (P = 0.0001), 4 (P = 0.0004), and 5 (P = 0.0003) days post-stroke compared to baseline (0 day). HFD-I mice weigh significantly less at day 1 (P = 0.0244), 2 (P < 0.0001), 3 (P < 0.0001), 4 (P < 0.0001), and 5 (P < 0.0001) days post-stroke compared to baseline. Overall, there was no significant interaction with insulin treatment on weight in STD or HFD male mice (Fig. 4). All of the comparisons for the Tukey’s post-hoc is provided in the Supplemental Table 1.

Figure 4. Body weight reduced post-stroke.

The body weight in grams (g) in male mice on either a standard diet (STD) or high-fat diet (HFD) following middle cerebral artery occlusion (MCAO); n=10 (STD-S), n=20 (HFD-S), or n=14 (HFD treated with intranasal insulin, HFD-I) and n=6 (STD-I). A three-way ANOVA with mixed-effects modeling reveals a significant interaction with days Post-stroke (F(1.915, 62.05) = 72.54; P<0.0001), diet (F(1, 46) = 55.10; P<0.0001), and between days post-stroke and diet (F(5, 162) = 8.927; P<0.0001). Tukey’s post-hoc analysis reveals that STD-S mice experienced significant decrease in weight at 1 (P = 0.0327), 2 (P = 0.0048), and 3 (P = 0.0079) days post-stroke compared to baseline (0 day), STD-I experienced significant decrease in weight at 2 (P = 0.0004) and 3 (P = 0.0072) days post-stroke compared to baseline. The data is presented as mean ± SD.

Intranasal Insulin improved Post-Stroke Survival and Neurological Deficits in male mice

To evaluate the impact of diet and brain insulin on post-stroke outcomes, survival and neurological severity were assessed. The post-stroke survival of STD-S mice was 80% and HFD-S mice was 47%. On the other hand, SHAM-S, STD-I, and HFD-I mice experienced 100% survival. Comparison of survival curve log-rank (Mantel-Cox; Chi square = 16.91; df = 4; P = 0.0020) test reveals that survival curves are significantly different (Fig. 5a). Neurological severity was assessed using mNSS. Three-way mixed effects modeling of mNSS scores revealed a significant interaction with diet (F (1,28 = 6.262), P=0.0184) and treatment (F (1, 54) = 12.68), P=0.0008; Fig. 5b). Approximately 74% of HFD-S mice had worse neurological outcomes (mNSS ≥ 3) compared to 20% of STD-S mice at Day 1 post-stroke. Following insulin treatment approximately 33% of HFD-I mice and none of the STD-I mice displayed a mNSS ≥ 3 at Day 1 post-stroke. At Day 1 post-stroke, approximately 50% of HFD-S, 20% of STD-S, 0% of STD-I, and 14% of HFD-I mice displayed a mNSS ≥ 3. Tukey’s multiple comparisons test revealed that intranasal insulin treatment significantly reduced mNSS in HFD-S mice compared to HFD-I mice at day 2 post-stroke (P=0.0194) based on Sidak’s multiple comparison’s test. All comparisons are provided in Supplemental Table 2.

Figure 5. Intranasal insulin improves post-stroke survival and neurological deficits.

(a) Percent survival with following SHAM surgery or middle cerebral artery occlusion (MCAO), model of stroke in mice on either a standard diet (STD) or high-fat diet (HFD) treated with either intranasal saline (S) or insulin (I); A comparison of survival curves used Log-rank (Mantel-Cox) test (Chi square = 16.91, df=4, P=0.0020) and Log-rank test for trend (Chi square = 7.600, df=1, P = 0.0058). (b) Modified neurological severity score (mNSS) on a scale of 0 (normal) through 5 (moribund) at Day 1 and Day 2 post-stroke in STD-S (n=10), HFD-S (n=19), STD-I (n=6), and HFD-I (n=14) mice. Three-way ANOVA with mixed-effects modeling revealed a significant interaction with diet (F (1, 28)=6.262, P=0.0184) and treatment (F(1, 54)=12.68, P=0.0008). Tukey’s multiple comparisons test revealed a significant improvement in mNSS scores in HFD-I mice compared to HFD-S mice at 2 days post-stroke (P=0.0194). Data presented as mean ± SD.

Intranasal Insulin Improved Post-Stroke Cognitive Impairment in male mice

The impact of diet and brain insulin on post-stroke cognitive impairment was evaluated using the novel object location task. At baseline (8 weeks of age/4 weeks of diet) both STD (n=10; t = 3.355; df = 9; P = 0.0085) and HFD (n=14; t = 3.670; df = 13; P = 0.0028) mice spent significantly more than 33.33% (by chance) of time exploring the novel location and the former object location (total memory), indicative of normal memory, based on a one sample t-test (Fig. 6a). There is no significant difference between STD and HFD mice based on an unpaired t-test at baseline (t = 0.1007, df = 22; P = 0.9207). At 14-days post-stroke, combined controls (n=5; t = 2.847; df = 4; P = 0.0465) and STD-I (n=6; t = 5.920; df = 5; P = 0.0020) mice spent more than 33.33% of time exploring the novel location and the former object location based on a one-sample t-test (Fig. 6b). While HFD-S mice did not discriminate the novel object location and former object location (n=5; t = 1.346, df = 4; P=0.2495), the HFD-I (n=5; t = 4.674; df = 4; P = 0.0095) mice spent significantly more than 33.33% of the time exploring the novel location and the former object location based on a one sample t-test. A significant interaction was observed based on a one-way ANOVA comparing performance among treatment groups at 14-days post-stroke (F (3, 17) = 3.208; P = 0.0495). Tukey’s multiple comparison test revealed a trending increase in total memory at 14-days post-stroke in the HFD-I mice compared to the HFD-S mice (P=0.0991). All comparisons are provided in Supplemental Table 3.

Figure 6. Intranasal Insulin improved post-stroke cognitive impairment.

(a) Baseline percentage (%) exploration of novel location in male mice on either a standard diet (STD) or high-fat diet (HFD). **P<0.01 based on a one sample t-test (STD: P=0.0085; HFD: P=0.0028); STD; n=10, HFD; n=14. (b) % Total Memory of exploring the novel location in male mice 14 days following middle cerebral artery occlusion (MCAO), model of stroke, in mice on either STD or HFD receiving either intranasal saline (0.9%) or insulin (1.75U) beginning Day 1 post-stroke. *P<0.05 in Controls (P=0.465), **P<0.01 in STD-I (P=0.0020) and HFD-I (P=0.0095) based on one-sample t-test; Controls (n=5), STD-I (n=6), HFD-S (n=5), and HFD-I (n=5). There are no significant differences between groups with Tukey’s multiple comparisons test. Data presented as mean ± SD.

The Impact of Stroke in Age-Matched Females is Unremarkable

The impact of stroke on body weights were monitored for 5 days in female mice (Fig. 7). Three-way mixed-effects modeling revealed a significant interaction with days post-stroke (F (2.203, 39.65) = 42.04; P<0.0001), diet (F (1, 20) = 109.9, P <0.0001), and between days post-stroke and diet (F (5, 90) = 24.33, P<0.0001). The HFD-S mice weigh approximately 43% more than the STD-S mice. HFD-S mice weighed significantly more than STD-S mice at baseline (0 day (P = 0.0409), day 1 (P = 0.0473), day 2 (P = 0.0402), day 3 (P = 0.0445), day 4 (P = 0.0479), and day 5 (P = 0.0379) based on Tukey’s multiple comparison test. Furthermore, STD-S mice weighed significantly less at 1 (P = 0.0181) and 2 (P = 0.0114) days post-stroke compared to baseline (0 day). STD-I weighed significantly less only at 1 day (P = 0.0025) post-stroke compared to baseline (0 day). On the other hand, HFD-S mice weigh significantly less at 1 (P = 0.0257), 2 (P = 0.0075), 3 (P = 0.0065), and 4 (P = 0.0214) days post-stroke compared to baseline (0 day). HFD-I mice experienced a significant decrease in weight at 1 (P = 0.0.0011), 2 (P = 0.0095), 3 (P = 0.0184), 4 (P = 0.0302), and 5 (P = 0.0145) days post-stroke compared to baseline (0 day). All of the comparisons for the Tukey’s post-hoc is provided in the Supplemental Table 4.

Figure 7. Body Weight homeostasis significantly impacted in HFD MCAO female mice immediately after MCAO.

The body weight in grams (g) in female mice on either a standard diet (STD) or high-fat diet (HFD) following middle cerebral artery occlusion (MCAO). (STD-S; n=6), high-fat diet (HFD-S; n=6), or HFD treated with intranasal insulin (HFD-I; n=6) and (STD-I; n=6). Based on a three-way ANOVA with Tukey’s multiple comparison test, STD-S mice experienced significant decrease in weight at 1 (P=0.0181) and 2 (P=0.0114) days post-stroke compared to baseline (0 day). STD-I experienced significant decrease in weight at 1 day (P=0.0025) post-stroke compared to baseline. HFD-S mice experienced significant decrease in weight at 1 (P=0.0257), 2 (P=0.0075), 3 (P=0.0065), and 4 (P=0.0214) days post-stroke compared to baseline (0 day). HFD-I mice experienced significant decrease in weight at 1 (P=0.0.0011), 2 (P=0.0095), 3 (P=0.0184), 4 (P=0.0302), and 5 (P=0.0145) days post-stroke compared to baseline (0 day). The data is presented as mean ± SD.

To evaluate the impact of diet and brain insulin on post-stroke survival and neurological severity were assessed. Both STD-S, and STD-I mice experienced 83% survival, whereas 100% of the SHAM-S, HFD-S, and HFD-I mice survived following MCAO (Fig. 8). There are no significant differences among the survival curves in female mice based on log-rank (Mantel-Cox; Chi square = 2.769, df = 5, P = 0.7355) or trend based on the Logrank test (Chi square = 0.1412, df = 1, P = 0.7071). Neurological impairments were assessed using mNSS. There were no differences with day post-stroke, diet, or treatment in females following MCAO based on three-way mixed-effects analysis with Sidak’s multiple comparisons test (Fig. 8). All comparisons are provided in Supplemental Table 5.

Figure 8. Intranasal insulin improves post-stroke survival and neurological deficits in female mice.

(a) Percent survival with following SHAM surgery or middle cerebral artery occlusion (MCAO), model of stroke in mice on either a standard diet (STD) or high-fat diet (HFD) treated with either intranasal saline (S) or insulin (I). A comparison of survival curves used Log-rank (Mantel-Cox) test (Chi square=2.769, df=5, P=0.7355) and Log-rank test for trend (Chi square=0.1412, df=1, P = 0.7071). (b) Modified neurological severity score (mNSS) on a scale of 0 (normal) through 5 (moribund) at Day 1 and Day 2 post-stroke in STD-S (n=6), HFD-S (n=6), STD-I (n=6), and HFD-I (n=6) mice. Three-way ANOVA with mixed-effects modeling with Tukey’s multiple comparisons test revealed no significant interactions or differences between groups. Data presented as mean ± SD.

Working memory was assessed through performance on novel object location task to determine the impact of brain insulin levels on post-stroke cognitive impairment. At baseline (4 weeks of diet/8 weeks of age) both STD (n=14; t = 12.25; df = 13; P < 0.0001) and HFD (n=14; t = 27.30; df = 13; P <0.0001, Fig. 9) mice spent significantly more than 33.33% of time exploring the novel location based on a one sample t-test. An unpaired t-test revealed that HFD mice at baseline spend significantly more time than STD exploring the novel location (t = 6.536, df = 26, P<0.001). At 14-days post-MCAO, Controls (n=9; t = 7.260; df = 8; P<0.0001), HFD-S (n=6; t = 5.723; df = 5; P = 0.0023), and HFD-I (n=6; t = 3.464; df = 5; P = 0.0180) based on a one-sample t-test. On the other hand, STD-I (n=5) mice do not spend more than 33.33% of time exploring the novel location based on a one-sample t-test. There are no significant differences among the groups with a one-way ANOVA (F (3, 22) = 1.180, P = 0.3399). All comparisons are provided in Supplemental Table 6.

Figure 9. Intranasal Insulin had no impact on post-stroke cognitive impairment in high-fat diet female mice.

(a) Baseline percentage (%) exploration of novel location in female mice on either a standard diet (STD) or high-fat diet (HFD). ####P<0.0001 based on one sample t-test (STD: t=12.25, df=13, P<0.0001; HFD: t=27.30, df=13, P<0.0001); ****P<0.0001 based on unpaired t-test (t=6.536, df=26); STD: n=14, HFD: n=14. (b) % Total Memory of exploring the novel location in female mice 14 days following middle cerebral artery occlusion (MCAO), model of stroke, in mice on either STD or HFD receiving either intranasal saline (0.9%) or insulin (1.75U) beginning Day 1 post-stroke. ****P<0.0001 in Controls, ns (not significant) in STD-I, **P=0.0023 in HFD-S, and *P=0.0180 in HFD-I based on one-sample t-test; Controls (n=9), STD-I (n=5), HFD-S (n=6), and HFD-I (n=6). A one-way ANOVA (F(3,22) =1.180, P=0.3399) with Tukey’s multiple comparisons test revealed no difference between groups. Data presented as mean ± SD.

Discussion

Studies have demonstrated that insulin resistance is not only a risk factor of stroke (4, 38) but is also associated with worse functional outcomes following stroke (7, 39–41); however, the mechanisms are not known. The purpose of this study was to evaluate the potential impact of brain insulin deficiency, a consequence of insulin resistance, on stroke outcomes including survival, neurological and cognitive function. To our knowledge, this is the first study that evaluates the potential impact of brain insulin levels on stroke outcomes. Collectively, we have demonstrated that HFD-induced insulin resistance leads to brain insulin deficiency and worse post-stroke survival, neurological and cognitive impairment. Increasing brain insulin levels improves post-stroke survival, neurological and cognitive function.

HFD is a model of insulin resistance and hyperinsulinemia (31, 42, 43). Our data confirms that the HFD mouse model is also a model of brain insulin deficiency, evident by the reduction of insulin levels in hippocampal tissue and CSF (Fig. 3). This aligns with previous studies that demonstrated that rats on a HFD had reduced insulin uptake from the periphery compared to STD rats (44) and reduced CSF insulin levels (16). Collectively, these studies, along with ours, suggests that diet-induced insulin resistance correlates with brain insulin deficiency.

Given the pleotropic effects of brain insulin on processes such as neuroplasticity, angiogenesis, and cognition (18, 20, 22, 23), a deficiency of brain may correlate with worse stroke outcomes. Thus, next we assessed the impact of post-stroke brain insulin levels on survival, neurological, and cognitive deficits. Studies have demonstrated a link body weight at stroke onset with post-stroke outcomes in what is known as the obesity paradox (45). In brief some studies reveal that obesity is associated with more favorable stroke outcomes (46, 47) whereas other studies have not seen that association (48, 49). Post-stroke weight loss is common (50). Similarly, our data demonstrates that mice on the STD and HFD lose weight following stroke (Fig. 4). Previous studies have reported that brain insulin modulates food intake and weight gain (51, 52). Intranasal insulin treatment did not have an impact on body weight in our study.

The presence of comorbidities is associated with higher mortality (up to ~50%) following ischemic stroke in rodents (53, 54). Our studies also revealed that HFD was associated with a higher mortality (Fig. 5a); however, this was ameliorated by increasing brain insulin levels via intranasal insulin administration (Fig. 5a). Our data suggests that insulin resistance correlated with higher mortality. This aligns with a study on patients in China reported that insulin resistance was associated with higher mortality within 1 year of stroke (7). In contrast, insulin resistance did not increase the risk of 3 month mortality in patients with acute stroke. While both of these clinical studies used the homeostasis model assessment of insulin resistance (HOMA-IR) calculated by multiplying fasting glucose mg/dL by fasting insulin mU/L/405 (40), one key difference was the nationality of the study populations making it difficult generalize the results for different study populations due to different lifestyle habits and cultural differences in these countries (41). Thus, lifestyle habits and cultural differences may contribute to the discrepancy in the correlation between insulin resistance and stroke mortality. Insulin resistance was associated with worse neurological and functional outcomes at a subacute stage and also at a 3 month follow-up (55).

The modified Rankin score (mRS), which is used to determine the severity of disability following stroke, was determined to be higher in patients with a higher HOMA-IR quartile score, indicative of worse neurological outcome (41). We assessed gross neurological impairments using the mNSS. Brain insulin deficiency was associated with more severe neurological deficits, evident by increased scores on the mNSS scale (Fig. 5b) at 1 and 2 days post-stroke. On the other hand, increasing brain insulin levels via intranasal insulin administration (Fig. 5b) ameliorated deficits measured by the mNSS assessment. Collectively, these studies support the data presented here revealing that HFD-induced obesity, animal model of insulin resistance, is associated with stroke-related mortality and worse functional outcome.

Intranasal insulin has been used as a therapy in terms of mitigating cognitive impairment associated with mild cognitive impairment, Alzheimer’s disease, and traumatic brain injury (11, 26, 56); however as of yet no one has explored its therapeutic potential in post-stroke impairment. The hippocampus is highly susceptible to the adverse effects of HFD-induced obesity (57), and the increased proinflammatory responses from stroke may result in secondary damage extending into the hippocampus. The hippocampus is one of the main brain regions responsible for memory processing (58, 59). Intranasal administration can penetrate the hippocampus, which is essential for object location memory (37), via the olfactory pathway with projections through the entorhinal cortex to the CA1 region of the hippocampus (60). The novel object location task did not yield a difference in reference memory associated with diet/brain insulin deficiency prior to stroke (Fig. 6a); however, HFD was associated with post-stroke cognitive impairment. Insulin treatment improved post-stroke reference memory in HFD mice (Fig. 6b). This suggests that intranasal insulin may be useful for improving post-stroke memory deficits.

Overall, our study reveals that brain insulin levels may play a role in post-stroke outcomes. We demonstrate that brain insulin deficiency is associated with worse mortality rates, neurological, and cognitive impairment. On the other hand, increasing brain insulin levels improved post-stroke survival, neurological and cognitive function. Intranasal insulin administration maybe a therapeutic option to improve outcomes post-stroke in patients with diabetes and/or insulin resistance. Additional studies are needed to determine the importance of timing of intranasal administration following stroke. Furthermore, additional studies are also needed to understand the long-term impact of intranasal insulin treatment on post-stroke cognitive impairment.

Study Limitations

One limitation of this study is that the study was performed in young male and female mice. There were no differences observed in the female mice. Due to previously described sex differences with HFD in male and female mice, the data in the current manuscript did not use sex as a variable and institute a 4-way ANOVA. Instead, here we opted to analyze the male and females separately. Previous studies have demonstrated young females have improved CBF and decreased infarct volume in a female diabetic and normoglycemic mouse model (61–63). However, aged females demonstrate worse outcomes following stroke (64). In fact, studies reveal that females are more susceptible after menopause (64, 65). Thus, future studies are needed to also explore the potential impact of brain insulin levels in older female mice after stroke. Another limitation in the study is that brain insulin levels were not assessed in mice post-stroke.

Significant Statement.

High-fat diet induced obesity is linked to worse outcomes following stroke. This study reveals that brain insulin deficiency, a consequence of a high-fat diet, contributes to worse stroke outcomes. Increasing brain insulin levels may be a useful therapy for improving survival and functional stroke recovery.

Abbreviations:

MCAO

Middle cerebral artery occlusion

HFD

high-fat diet

STD

standard diet

CSF

Cerebrospinal Fluid

mNSS

modified neurological severity score

MCA

Middle cerebral artery

CCA

Common carotid artery

ICA

Internal carotid artery

ECA

external carotid artery

CBF

Cerebral blood flow

C57BL/6J (B6)

Black 6 mice

Akt

protein kinase B

pAkt

phosphorylated protein kinase B

HOMA-IR

Homeostasis Model Assessment- Insulin Resistance

Data Availability Statement.

Smith et al., raw data for hippocampal insulin levels, relative CSF insulin levels, blood glucose measurements, male and females post-stroke survival, modified neurological severity scores, pre- and post-stroke novel object location, post-stroke weights, and blood flow is provided in a figshare repository https://figshare.com/s/2b41986c24beacc9d882.