. Author manuscript; available in PMC: 2026 Apr 10.

Published in final edited form as: Nutr Neurosci. 2025 Nov 25;29(5):653–665. doi: 10.1080/1028415X.2025.2590047

Reversal of early-life high fat diet prevents spatial discrimination deficits corresponding with altered brain inflammatory and metabolic profiles in male C57BL/6J mice

Caleb S Bailey ¹, Mckenna C Green ¹, Meggie J Coleman ¹, Linda J Van Eldik ^1,², David J Braun ^1,^2,^*

PMCID: PMC13064915 NIHMSID: NIHMS2162440 PMID: 41288665

Abstract

Objectives:

High-fat diet is well-known to contribute to systemic and central nervous system dysfunction and represents a modifiable risk factor for cognitive decline. A recent surge of new weight loss medications has demonstrated that caloric restriction by reduced overeating has been successful at mitigating peripheral markers of inflammation and metabolic dysfunction. However, less is known regarding such effects of dietary reversal within the brain.

Methods:

Male mice received high-fat diet (HFD; 60%kCal fat) from 6 weeks of age (JAX #380050). At 14 weeks, half of the mice were reversed (Rev = 9) to a standard diet (14%kCal fat) while the other half (HFD = 10) remained on HFD for an additional 2 months. Weight, frailty, peripheral cytokines, and behavior were recorded at baseline and again at 1 and 2 months after dietary reversal. Mice were sacrificed at 22 weeks of age and terminal measures of brain neuroinflammation and metabolism were evaluated.

Results:

In contrast to mice on continuous HFD, the mice in the Rev group lost weight, had lower frailty scores, preserved spatial discrimination, altered peripheral cytokine profiles, lower hippocampal GFAP, increased hippocampal MCP1 and IL4, improved hippocampal and midbrain insulin tone, and differentially altered TRIB3 in the hippocampus and midbrain.

Discussion:

Early dietary reversal was effective at preventing the inflammatory, metabolic, and cognitive effects of sustained HFD with divergent effects on metabolic markers (TRIB3) depending on brain region.

INTRODUCTION

While genetic and environmental factors both contribute to obesity, diet remains a critical modifiable factor in the management of obesity and its corollary comorbidities.¹ Modern approaches to curb the ongoing obesity epidemic involve treatment by way of glucagon-like peptide-1 (GLP1) agonists,² bariatric surgery,³ combined with lifestyle modification programs. These types of interventions have been largely successful at reducing food overconsumption and stabilizing obesity and T2D-related metabolic and inflammatory profiles, demonstrating the effectiveness of sustained caloric restriction/dietary change in management of these conditions⁴. While the majority of studies examine peripheral metrics of intervention, fewer studies have evaluated the effect of such dietary-reversals on brain metabolism and inflammation in the diet-induced obesity (DIO) context. While some indicate recovery,^5,6 there appear to be certain brain responses that are not normalized by dietary reversal alone.⁷ Considering that even early-life excess body weight is associated with reduced midlife cognitive function,⁸ a clearer understanding of how the brain recovers from the stress of high-fat diet (HFD) is necessary to refine how lifestyle and other interventional treatments can be best leveraged to prevent later-life cognitive decline associated with this common condition.

Rodent high-fat diet (HFD) models have long been employed to study inflammatory and metabolic effects of DIO. A chronic HFD in rodents can induce rapid and sustained weight gain, peripheral and central metabolic dysfunction and inflammation, anhedonia, and cognitive impairment.⁹ While prior experiments have used different methods of HFD administration, many of these pathological DIO changes are mitigated by reversal to a standard (lower fat) diet,^5,6,10 consistent with the efficacy of dietary reversal in humans. Nonetheless, dietary reversal is not always sufficient to fully reverse pathology. For example, in the Look AHEAD study of people with midlife diabetes,¹¹ intensive lifestyle intervention successfully helped participants reduce diabetes symptoms up to 10 years later¹² with no effect on subsequent cognitive test scores.¹³ Even in those with obesity without T2D, impaired brain responses to nutrient intake are not reversed by diet-induced weight loss.⁷ To identify additional potentially targetable features in the early central nervous system response to HFD insult, we performed a longitudinal study to investigate how baseline neuroinflammatory and cognitive metrics change in response to a dietary reversal from a HFD (60%kCal fat) to a standard rodent chow (6%kCal fat) in wildtype C57BL/6J male mice placed on HFD shortly after weaning at 6 weeks of age. We hypothesized that improvements in metabolic and neuroinflammatory profiles resulting from dietary reversal would preserve spatial discrimination performance in the mice. We collected plasma and behavioral endpoints at baseline prior to dietary reversal (14 weeks of age, 8 total weeks on HFD) and again at 4 and 8 weeks after dietary reversal in half the animals (or 12 and 16 total weeks on HFD for the continuous HFD group). Directly after the final behavioral endpoints, mice were sacrificed for post-mortem analysis of brain cytokines, insulin tone, tribbles pseudokinase 3 (TRIB3) and glial fibrillary acidic protein (GFAP). Sustained HFD induced prolonged weight gain, spatial discrimination deficits, peripheral and central neuroinflammatory dysregulation, elevated hippocampal GFAP, and dysregulated insulin tone and TRIB3 expression compared to animals with dietary reversal.

METHODS

Design.

This experiment was performed at the University of Kentucky in accordance with University of Kentucky IACUC protocol (#2023–4337). Twenty male mice were placed onto a high fat diet (HFD) at 6w of age (Jackson C57BL/6J DIO, #380050) and delivered to our facility at 12w old. At 14w old, all mice underwent baseline behavioral assessment and submental blood collection after which half of the mice were placed onto a standard mouse chow diet (STD; 6% kcal fat; Teklad Global 18% Protein Rodent Diet) (reversal group – Rev) and the other half remained on HFD (60%kCal fat; Research Diets #D12492) (Figure 1A). One mouse in the Rev group died just after the first submental bleed, for final group numbers of n = 9 Rev and n = 10 HFD. Weekly weights and monthly submental blood collections were performed to monitor systemic effects of dietary change. Observable health (frailty index), spontaneous open field activity (SOFA), reference memory, spatial discrimination, and submental blood draws were recorded during baseline when all animals were still on HFD (14 weeks of age, 8 weeks on HFD), approximately one-month post dietary-reversal (18 weeks of age), and two-months post dietary-reversal (22 weeks of age). Spatial discrimination and reference memory impairment was assessed using the novel spatial recognition (NSR) and spontaneous alternation (SA) tasks in a Y-maze apparatus. At 22 weeks of age, mice were sacrificed for terminal tissue collection.

Figure 1. — (A) Experimental schematic. After 2mo on HFD (6w HFD at JAX, 2w in our facility), baseline blood collections and behavioral assessments were performed when the mice were 14w old. Afterward, half of the mice were reversed to a standard diet (Rev = 9) and the other half remained on HFD (HFD = 10). Blood collections and behavioral battery were performed again 1mo and 2mo following dietary reversal. (B) Weight trajectories for the HFD group continued to increase over 2mo while weights declined in the Rev group and were maintained at approximately 15.97% lower than their baseline HFD weight by 3w into the dietary reversal. Two-way repeated measures ANOVA. (C) Mice in the HFD group had increased frailty indices between baseline and 1mo and baseline and 2mo. Additionally, at 1 and 2mo, mice maintained on HFD were frailer than mice in the Rev group. Two-way repeated measures ANOVA with adjusted Tukey’s multiple comparisons tests. **p <.01; ***p<.001; ****p <.0001; ^#one animal in the Rev group died after baseline submental blood collection.

Longitudinal blood collection.

For non-terminal blood draw, the submental blood collection method was used as published¹⁴. Briefly, the mouse was scruffed to expose the submental region of the head, with vein location ascertained by anatomic landmarks. The chin area was wiped with an alcohol prep pad, and the vessel pricked with a 5.0 mm lancet (Avantor #MSPP-GR5MM). Sample volume up to 50 μL was collected with a 100 μL Microvette EDTA capillary tube (Sarstedt #20.1278.100). Mild pressure was applied with gauze to stop the bleeding, and the mouse returned to its home cage. All non-terminal blood draws were performed two days after the final behavioral endpoint in the battery.

Behavioral assessments.

A behavioral battery was performed at each timepoint (14, 18, and 22 weeks) across three consecutive days to assess health status, spontaneous activity, reference memory, and spatial discrimination. Prior to all behavioral assessments, mice were tube handled for 5 days¹⁵ and allowed 1 h to habituate to the room on any given behavior day. ANY-maze video tracking software was used to document mouse movement in relevant assays (ANY-maze version 7.49; Dublin, Ireland). See Supplement 1 for full behavioral data.

Frailty index.

On the first day of the behavioral battery, the frailty index was performed as previously published.¹⁶ In brief, mice were assessed for severity of frailty using 27 observable metrics. On each metric, a score of 0 indicated typical observation, 0.5 indicated modest degree of health peculiarity, and 1 indicated severe degree of health peculiarity. Summed scores across traits indicated degree of frailty such that higher frailty indices indicated worse observable health.

Spontaneous open field.

On the first day of the behavioral battery, the spontaneous open field assessment (SOFA) was performed as previously described¹⁶ using four adjacent gray opaque open field arenas (40cm × 40cm × 40cm; Stoelting Co.; Wood Dale, IL, USA) to track up to four mice at a time. Mice were allowed to freely explore the arena for 15 min. The mean speed and proportion of time spent in the center (20cm × 20cm area) were recorded for each mouse.

Y-maze novel spatial recognition.

On the second day of the behavioral battery, the novel spatial recognition (NSR) assay was performed using a clear polycarbonate 3-armed mouse Y-maze (Maze Engineers; Boston, MA, USA) with distinct visual cues at the distal wall of each Y-maze arm as previously described.¹⁷ The visual cues were as follows: black-and-white diagonal stripes in the ‘start’ arm, white plus sign against a black background in the ‘familiar’ arm, and a black circle against a white background in the ‘novel’ arm. There were three phases in the NSR assay: 10 min habituation, 5 min intertrial interval (ITI), and 5 min probe. During the 10 min habituation, mice were allowed to freely explore the ‘start’ and ‘familiar’ arms, but an opaque door prohibited entrance into the ‘novel’ arm. Mice were then placed into a holding cage for the 5 min ITI before returning to the Y-maze for 5 min probe phase in which the ‘novel’ arm was then accessible. The Y-maze was sanitized with 70% ethanol solution before habituation, during the ITI, and after the probe for each mouse. A single discrimination index (DI) was calculated for each animal, where a positive DI indicates discriminatory preference toward the novel arm and a negative DI indicates a discriminatory preference toward the familiar arm:

Discrimination Index = \frac{Time in Novel Arm - Time in Familiar Arm}{Time in Novel Arm + Time in Familiar Arm}

Y-maze spontaneous alternation.

On the third day of the behavioral battery, the spontaneous alternation (SA) assay was performed using the same Y-maze apparatus with the same spatial cues present as in the NSR assay. Mice were placed in the ‘start’ arm and allowed to freely explore all arms for 5 uninterrupted minutes, with alternation measured as the number of 3 consecutive non-repeating arm entries.

Euthanasia, tissue collection and processing.

Two days after the final behavioral endpoint collection, mice were anesthetized by 5% isoflurane and euthanized by exsanguination using transcardial perfusion of 50mL 1X PBS (Cellgro Mediatech, #46–013-CM; Manassas, VA, USA), and tissue collected as previously described.¹⁶ Brain hemispheres were bisected at midline, and the right hemisphere was immediately fixed in 4% paraformaldehyde for 24h then cryopreserved at 4°C in 30% sucrose solution for >5 days before sectioning. The right hemisphere was then coronally sectioned (30μm) on a freezing microtome and stored in a cryoprotectant at −20°C. The left hemisphere was used for biochemical endpoints such that the hippocampus was dissected, laid flat, and bisected lengthwise. The midbrain, including thalamus and hypothalamus, was similarly dissected from the left hemisphere. Tissue was flash frozen using liquid nitrogen and stored at −80°C prior to processing. For cytokine assays, one piece of hippocampus was homogenized to enrich for cytosolic proteins (cytokines) using 1X PBS lysis buffer (1:20w/v) containing 0.2X Halt Protease Inhibitor Cocktail with 0.5mM EDTA (Pierce, #78442; Thermo Fisher, Waltham, MA, USA) and 1mM PMSF. For metabolic endpoints, the other piece of hippocampus and the entire midbrain were homogenized using the same Halt and PMSF concentrations except TPER (#78510; Thermo Fisher) buffer was used (1:20w/v). All samples were homogenized using a Bead Ruptor 24 homogenizer without beads (Omni International Inc.; Kennesaw, GA, USA). Homogenates were centrifuged (12000×g, 20 min, 4 °C) and supernatants were collected for biochemical analyses.

MesoScale Discovery ELISA.

All plates were read using a MesoScale Discovery (MSD) Quickplex SQ120 plate reader (MSD; Rockville, MD, USA) and data analyzed by Discovery Workbench 4.0 Software. All assayed proteins from brain were normalized to total protein loaded per sample as determined by BCA assay. Analytes not detected in ≥20% of the samples were excluded from the final analyses (see Supplement 2 for full MSD data).

MesoScale Discovery Sandwich ELISAs.

Hippocampal supernatants and plasma samples were assayed for cytokines in the PBS-soluble fraction using an MSD V-PLEX Mouse Cytokine 19-Plex Kit (MSD, #K15255D) following manufacturer’s instructions, with the following modifications: for brain, 50μL of supernatant with 25μL of assay diluent buffer was loaded in each well and plasma was diluted 1:4 in assay diluent buffer. Plates were shaken overnight at 4°C. For metabolic endpoints (pIR/pIRS1, pAKT/AKT), the insulin signaling panel (#K15151C) and AKT assay kits (#K15100D) were used, respectively, according to manufacturer’s instructions. Analytes with >20% missing values from being below detection limit were excluded from analysis (IL-12p/70, IL-4, IL-15, IL-17A/F, IL27p28, and IL-9 for plasma, and the same but IL-2 replacing IL-4 for hippocampus).

MesoScale Discovery Direct ELISA.

The tribbles pseudokinase 3 (TRIB3) metabolic endpoint was analyzed by customized MSD direct ELISA assay using MSD high bind plates (#L15XB), with 5 μL of TPER-homogenized lysate coated in each well and incubated at 4°C overnight for liquid evaporation. Plates were blocked with 5% BSA (R93BA) in PBS solution for 1 h at room temperature, with shaking at 1000 rpm. A total of 25 μl per well of rabbit anti-mouse TRIB3 antibody (RayBiotech, #102–21807) was added in antibody diluent (1% BSA in PBS solution) at a final concentration of 1 μg/ml of antibody. Plates were again shaken at 1000 rpm for 1 h at RT. Subsequently, 25 μL of 1 μg/ml MSD sulfotag goat anti-rabbit antibody (MSD, #R32AB) was added in antibody diluent, and plates shaken at 1000 rpm for 1 hour at RT. Plates were washed 3x in PBS with 0.05% Tween-20 between all steps after blocking. Afterward, 150 μL MSD Gold Read Buffer B was add per well and plates immediately read on the MSD Quickplex SQ120.

Immunohistochemical GFAP staining, imaging, and quantification.

Glial fibrillary acid protein (GFAP) was stained using 1:5000 rabbit anti-GFAP (DAKO, #Z0334) with 1:200 goat anti-rabbit peroxidase secondary (JAX #111–035-144). Sections were developed with 0.05% (w/v) 3,3’-diaminobenzidine tetrahydrochloride hydrate (MilliporeSigma, #D5637) and counterstained in Mayer’s modified hematoxylin solution (Abcam, #ab220365) for nuclear visualization prior to coverslipping with DPX mounting medium. Slides containing 3–8 tissue sections through the hippocampus were imaged per animal on an Aperio ScanScope XT digital slide scanner at 20x magnification. Image analysis (percent positive area in hippocampus) was performed by blinded investigator in HALO v.3.5 software (Indica Labs).

Statistical analyses and figure generation.

Mixed-effects models, T-tests, repeated measures ANOVAs, and post hoc analyses were conducted using SPSS (Version 26; IBM Corporation; Armonk, New York, USA) or GraphPad Prism 10.5.0 (GraphPad Software; Boston, Massachusetts, USA). All data figures were generated using Prism 10.5.0 (GraphPad Software). Non-data figures were generated using BioRender.

RESULTS

Sustained high-fat diet increases weight gain and worsens frailty indices.

Weight change and frailty were analyzed by two-way repeated measures ANOVA with corrected Tukey’s multiple comparisons. Mice in the dietary reversal group (Rev) lost nearly 20% of their high-fat diet (HFD) weight and maintained this weight loss while mice maintained on the HFD continued to gain weight across 2mo, p <.0001 (Figure 1B). Further, sustained weight gain corresponded with higher frailty indices for the HFD group compared to the Rev group at 1mo and 2mo post diet reversal, p <.01 and p <.001 respectively (Figure 1C). Mice in the HFD group had higher frailty indices at 1 and 2mo compared to baseline, p <.001 and p <.0001, respectively.

Sustained high-fat diet alters plasma cytokine and metabolic hormone profiles.

Peripheral cytokines and metabolic hormone profiles were analyzed by mixed-effects analysis with corrected Sidak’s multiple comparisons (Table 1); violations to model assumptions were resolved by linear transformations for analysis, but raw values are plotted in Figure 2 for a subset of analytes that showed significant differences in either group from baseline. Certain peripheral cytokines increased with continued time on HFD, such as TNFα, IL-10, and IP-10. Others had more complex dynamics, such as MCP1 which spiked in both groups at the 1mo timepoint and returned to baseline by 2mo. Interestingly, the 1mo spike was higher in the Rev group, p <.01. MIP2 was relatively stable in the HFD group but declined relative to baseline by 2mo in the Rev group, p <.01. IL-6 showed an increase only in the Rev group by 2mo relative to HFD, p <.0001. There were further time by diet interactions in terms of circulating metabolic hormones, such that an insulin elevation occurred in the HFD group at 1mo and 2mo (p <.05) but not in the Rev group, and Rev animals had reduced circulating leptin levels from baseline by 1mo (p <.0001) and 2mo (p <.01). In the HFD group, leptin levels continued to increase at both 1mo (p <.01) and 2mo time points (p <.0001). Consequently, the HFD group had significantly higher leptin levels at 1mo and 2mo time points compared to the Rev group, p <.0001.

Table 1.

Plasma analytes over time.

		Baseline			1mo			2mo
Analyte	Group	Mean	SD	N	Mean	SD	N	Mean	SD	N
IFNγ
	Rev	0.319	0.078	9	0.382	0.219	9	0.45	0.261	9
	HFD	0.308	0.106	10	0.276	0.097	10	0.346	0.075	10
IL10
	Rev	9.081	2.257	9	11.466	2.625	9	12.671 ^# ^*	1.71	9
	HFD	8.089	1.571	10	11.368 ^#	2.812	10	15.957 ^# ^{^} ^*	4.555	10
IL1β
	Rev	0.716	0.37	9	0.565	0.32	9	0.524	0.385	8
	HFD	0.601	0.457	10	0.416	0.23	10	0.473	0.287	9
IL2
	Rev	1.089	0.788	9	1.045	0.332	9	1.254	0.368	9
	HFD	0.983	0.404	10	1.389	0.537	10	1.038	0.347	10
IL5
	Rev	1.759	0.928	9	1.051 ^*	0.312	9	1.553	0.483	9
	HFD	1.347	0.49	10	1.623 ^*	0.517	10	1.515	0.383	10
IL6
	Rev	4.737	1.418	9	4.013	2.066	9	24.657 ^# ^{^} ^*	15.814	9
	HFD	7.753	6.846	10	6.154	1.901	10	5.126 ^*	1.864	10
KCGRO
	Rev	45.103	11.209	9	38.479	9.979	9	43.676	20.941	9
	HFD	50.826	18.584	10	47.353	15.932	10	36.664	10.533	10
TNFα
	Rev	4.272	1.187	9	4.198	1.243	9	4.2 ^*	1.179	9
	HFD	3.781	0.639	10	4.611	1.428	10	6.031 ^# ^{^} ^*	1.483	10
IL33
	Rev	0.51	0.401	8	0.719	0.87	8	0.593	0.303	9
	HFD	0.454	0.423	10	0.269	0.158	7	0.485	0.451	9
IP10
	Rev	18.533	3.595	9	20.496	5.073	9	19.869 ^*	4.588	9
	HFD	18.706	4.193	10	22.307 ^#	5.131	10	24.656 ^# ^*	5.412	10
MCP1
	Rev	10.699	3.357	9	33.319 ^# ^*	7.637	9	12.535 ^{^}	7.636	9
	HFD	10.884	4.282	10	22.377 ^# ^*	5.443	10	10.763 ^{^}	4.568	10
MIP1α
	Rev	0.735	0.283	9	2.261	2.99	9	1.051	1.009	9
	HFD	4.311	8.757	10	1.464	1.278	10	1.007	0.528	10
MIP2
	Rev	5.388	2.673	9	6.166	2.523	9	2.268 ^# ^{^} ^*	1.353	9
	HFD	5.988	3.329	10	8.183	4.499	10	5.976 ^*	2.039	10
Insulin
	Rev	59.828	48.472	9	31.128 ^*	13.662	9	44.328 ^*	34.683	9
	HFD	60.974	65.756	10	364.234 ^# ^*	522.994	10	154.027 ^# ^*	112.489	10
Leptin
	Rev	16721.82	4883.921	9	7600.093 ^# ^*	3519.188	9	9697.418 ^# ^*	4701.943	9
	HFD	14646.7	4451.289	10	19237.29 ^# ^*	2354.849	10	23317.79 ^# ^*	4837.153	10

Open in a new tab

Results from mixed effects models with Sidak’s multiple comparisons. All raw values are reported as pg/ml except Insulin which is reported as μIU/ml. Final sample (N) is less than total group size due to non-detects for those specific analytes.

significantly different between groups;

^{^}

significantly different from prior timepoint;

significantly different from baseline.

Figure 2. — (A) Mice in the HFD group had elevated TNFα, IP10, MIP2, and decreased IL6 at the 2mo post-reversal timepoint. IL10 increased with time in both the HFD and Rev groups but stabilized in the latter by 1mo but was not significantly different between the groups at 2mo. At 1mo, both the HFD and Rev group had elevated MCP1 compared to baseline; however, mice in the Rev group had elevated MCP1 compared to the HFD at the same 1mo timepoint. (B) At 1mo and 2mo post-reversal, HFD mice had elevated Insulin above baseline and Rev levels. At 1mo and 2mo post-reversal, the HFD group had elevated Leptin above baseline and Rev levels. Additionally, Leptin levels decreased below baseline across the same 1mo and 2mo span for mice in the Rev group. All peripheral cytokines and metabolic hormones were analyzed by mixed-effects analysis with corrected Tukey’s multiple comparisons. *significantly different between groups; ŝignificantly different from prior timepoint; ^#significantly different from baseline.

Sustained high-fat diet causes spatial discrimination deficits in the NSR task.

NSR performance was quantified by calculation of a discrimination index for each animal, followed by one-sample t-tests to determine whether the mean discrimination index by group was significantly different from 0 (no discrimination). Prior to dietary reversal, when all mice were on high fat diet for a total of 8 weeks, both groups were able to perform the spatial discrimination task, p <.01 (Figure 3A). At 1mo after dietary reversal, the Rev group was still able to discriminate the novel and familiar arms (p <.05), but the sustained HFD group was not (Figure 3B). This was also the case at the 2mo post-dietary reversal timepoint where the Rev group maintained a non-zero discrimination index (p <.05). Interestingly, at this time point the Rev group displayed a change in preference to the ‘familiar’ rather than ‘novel’ arm. No diet-associated differences between groups were detected in open field or spontaneous alternation assays; however, there was a divergence between the groups assigned to remain on HFD or reverse to standard diet in the open field assay at baseline (see Supplement 1 for full behavioral data). That these effects disappeared across the subsequent two time points, and there were no baseline divergences in any other measure, we suspect these initial open field observations were due to chance.

Figure 3. — (A) At baseline, discrimination indices were significantly higher than 0 for both the HFD and Rev group, indicating a discriminatory preference toward the Novel arm. (B) At 1mo after dietary reversal, the discrimination index was significantly higher than 0 only for the Rev group, indicating a discriminatory preference toward the Novel arm. (C) At 2mo after dietary reversal the discrimination index was significantly lower than 0 only for the Rev group, indicating a discriminatory preference toward the Familiar arm. NSR performance was analyzed by one-sample T-test against 0 at each timepoint. *p <.05; **p <.01.

Sustained high-fat diet increases hippocampal GFAP and reduces MCP1 and IL4 expression.

The same cytokine panel run for plasma was also used to measure hippocampal inflammatory changes (see Supplement 2 for full cytokine data), along with immunohistochemistry for astrocyte activation (GFAP), and data were analyzed by unpaired t-tests. As shown in Figure 4, the HFD group had higher GFAP expression compared to the Rev group, p <.01, consistent with increased astrocyte activation. The only cytokines measured that showed significant changes in hippocampus were MCP1 and IL-4, which were both lower in the HFD relative to the Rev group, p <.01 and p <.05, respectively.

Figure 4. — (A) Representative image of GFAP in the hippocampus, as quantified in (B). Mice maintained on HFD had elevated %positive GFAP staining in the hippocampus. (C) Both MCP1 and IL4 were decreased in the hippocampus for mice maintained continuously on HFD. Hippocampal GFAP, MCP1, and IL4 were independently analyzed by unpaired t-tests. **p <.01; ***p<.001.

Sustained high-fat diet disrupts insulin tone and alters TRIB3 expression in the hippocampus and midbrain.

Insulin tone and TRIB3 in the hippocampus and midbrain were analyzed by unpaired t-tests. In both the hippocampus and midbrain, insulin tone (as measured by the ratio of phosphorylated insulin receptor to phosphorylated insulin substrate 1 – pIR/pIRS-1), was reduced in the HFD group, p = .05 and p <.05, respectively (Figure 5A–B). Ratio of pAKT to total AKT levels was non-significantly reduced in both hippocampus (p = .083) and midbrain (p = .051). Finally, TRIB3 expression was reduced in the hippocampus but elevated in the midbrain in the HFD group, p <.05 and p <.01, respectively (Figure 5C–D).

Figure 5. — (**A-B**) Insulin tone was disrupted in the hippocampus and midbrain of mice maintained on HFD, reflected as a lowered ratio of phosphorylated insulin receptor to phosphorylated insulin receptor substrate 1 (pIR/pIRS-1). (C) TRIB3 was reduced in the hippocampus of mice maintained on HFD. (D) TRIB3 was elevated in the midbrain of mice maintained on HFD. Insulin tone and TRIB3 levels were independently analyzed using unpaired t-tests. **p <.01; ***p<.001.

DISCUSSION

We found that early and sustained high-fat diet (HFD) administration induces spatial discrimination deficits by 12 weeks total time on HFD, corresponding with altered brain inflammatory profiles, reduced central insulin tone, and changes in TRIB3 expression in the hippocampus and midbrain after 16 weeks total time on HFD. As expected in this model, the HFD regiment also resulted in exaggerated weight gain, increased frailty, and dysregulated plasma cytokine and metabolic hormone (insulin, leptin) levels over time. Importantly, we found that reversal of HFD after early-life exposure was sufficient to prevent weight gain and cumulative frailty, spatial discrimination deficits, GFAP overexpression, and insulin receptor dysregulation in the hippocampus and midbrain of male mice. Notably, the present study was conducted in the JAX B6 DIO 380050 strain, available for purchase only in males. Future studies should investigate if these effects occur in female mice which have differential sensitivity to dietary manipulation¹⁸. For the present study, the dynamic effects and implications of HFD and its reversal in male mice are explored below.

Changes in peripheral cytokine levels (IL-6, IL-10, IP-10, TNFα) were akin to those previously reported in a similar study.¹⁹ TNFα, IL10, and IP10 followed similar trajectories such that plasma levels increased from 8 to 16 weeks on HFD in the continuous versus Rev group; however, in the Rev group, IL6 remained relatively constant after 4 weeks of STD and increased by 8 weeks on STD. The implication of this is unclear but consistent with what has been previously reported¹⁹. In addition, we found that MIP2 levels declined in mice over time after dietary reversal, suggesting an inflammatory recovery over time relative to the sustained HFD group. We found that circulating MCP1 is increased from baseline to 1mo in both the HFD and Rev groups, and that MCP1 expression was higher in the Rev group compared to the HFD group at this time. Across both diet conditions, MCP1 returned to baseline levels at the 2mo post-reversal timepoint. Although speculative, this pattern suggests that developmental dynamics may primarily drive peripheral MCP1 levels over this age span, and these developmental dynamics may be altered by HFD. Taken together, these data are consistent with prior literature that reversal of HFD changes a broad array of circulating cytokines in a complex fashion that is likely due to the variable cellular sources of these cytokines in the plasma.

In terms of peripheral metabolic changes, we observed effects in line with the JAX ‘Phenotype Information for B6 DIO (380050)’ strain datasheet. This is true of the gross weight increase over time, as well as increases in insulin and leptin on this diet. Circulating leptin levels continued to increase with time for mice on HFD, whereas they decreased from baseline over time in the Rev group. Additionally, a large increase in insulin levels is observed between 2 and 10 weeks on this high fat diet, that was prevented by switching to standard diet in the Rev group. Interestingly, we did not detect a decrease from baseline in the Rev group insulin levels, which might indicate only a partial metabolic recovery. However, we did not have a group of mice maintained on standard diet in this study, and so this interpretation remains speculative.

In our study, mice in the Rev group maintained their baseline ability to discriminate between arms when tested again at both 4 and 8 weeks after reversal to STD, whereas the HFD group’s discrimination was impaired by 12 total weeks on HFD and remained impaired at 16 total weeks on HFD. Interestingly, at the final assessment, mice in the Rev group preferred the familiar arm over the novel arm. The change in arm preference was unexpected, but a ‘preference’ nonetheless is predicated on the ability of the animal to discriminate (for a review on discrimination learning, see²⁰) and by this metric, the Rev group remains unimpaired. One likely interpretation is that habituation effects may be influencing the behavior. By performing the same task repeatedly, the ‘novel’ arm may no longer be novel, therefore altering the motivational basis of the task. Although the presence of the discrimination ability (i.e. a significantly non-zero discrimination index) indicates intact spatial recognition memory in the Rev group, our test cannot distinguish the reasons for altered arm preference. One way to separate cognitive from non-cognitive contributions to behavioral output is to use operant-based tools that enable quantification of value²¹, and it is a limitation of the assay employed that we are unable to mechanistically explain why this preference shift occurred. Nonetheless, it is worth highlighting that the large peripheral insulin spike coincides with the timing of initial impairment in the HFD group. It may be that a certain level of systemic metabolic disruption is required before effects are manifest in the brain and its output, and this level of disruption was avoided in the Rev group in both the systemic and functional behavioral domains. Unfortunately, one further limitation of this study is that we did not include cross-sectional evaluation of brain changes, but we did detect brain insulin tone and metabolic disruption in the HFD group by the end of the study. Further, MCP1 and IP10 were the only peripheral cytokines that were altered by the time of detectable HFD-induced spatial discrimination deficits in the NSR, and of these, only MCP1 was significantly altered in the brain. In contrast to the periphery, it was decreased in the continuous HFD group as compared to the Rev. This reduction was further associated with both an increase in hippocampal GFAP staining, previously reported in the HFD context,^22,23 and a reduction in levels of IL-4. MCP1 can drive IL-4 production,²⁴ and the simultaneous reduction of both suggests these observations are linked. IL-4 administration has further been reported to reduce brain metabolic dysregulation associated with HFD,²⁵ and the relatively higher levels of MCP1 and IL-4 in the Rev group may therefore represent recovery processes in response to early HFD exposure.

In line with this interpretation of a metabolic recovery, insulin tone was improved in both the hippocampus and midbrain of the Rev versus HFD group, although levels of AKT pathway activation were not significantly increased at this timepoint. In an attempt to further define the metabolic changes in these regions, we measured levels of TRIB3, a pseudokinase known to regulate cellular metabolism and responses to stress across contexts, including diabetes and cancer.²⁶ TRIB3 has been shown to be increased with HFD and contribute to associated pathologies in the periphery,²⁷ as well as the hypothalamus.²⁸ Intriguingly, while continuous HFD was associated with higher levels of TRIB3 in the midbrain in accordance with the literature, levels were lower in the hippocampus despite reductions in insulin tone in both regions. This may indicate a heretofore undescribed and novel role of TRIB3 in normal hippocampal function, hippocampal recovery from HFD, or both; possibilities worth exploring in future studies across myriad neurological contexts featuring metabolic dysregulation. TRIB3 has been associated with neurodegeneration in various contexts, including Parkinson’s²⁹ and Alzheimer’s³⁰ diseases. Perturbations of this pathway may therefore contribute to the risk of cognitive decline associated with obesity. Given its known role in regulating crosstalk between inflammatory and metabolic pathways, TRIB3 may represent a modifiable target in the context of HFD-induced neuropathology to help restore central nervous system function. Indeed, one of the best treatments for metabolic dysfunction associated with obesity – exercise – directly reduces hypothalamic TRIB3 levels²⁸. We speculate that the targeting of brain TRIB3, whether with pharmacological (Hesperidin^31,32) or lifestyle interventions (exercise), might be a promising approach for preserving cognitive function in the face of common metabolic diseases. Future work directly investigating these questions is therefore warranted.

Supplementary Material

Supplement 1

NIHMS2162440-supplement-Supplement_1.docx^{(163.2KB, docx)}

Supplement 2

NIHMS2162440-supplement-Supplement_2.xlsx^{(34.8KB, xlsx)}

ACKNOWLEDGMENTS

We thank the University of Kentucky Division of Laboratory Animal Resources for their mouse husbandry, and the Sanders-Brown Center on Aging Mouse Behavior Facility in which the behavioral endpoints were obtained.

FUNDING

This work was supported by a pilot award under NIH/NIGMS parent grant P30 GM127211 (DJB), a T32 training grant NIH/NIA AG078110 (CSB), and an internal CURE Fellowship from the University of Kentucky (MCG).

Biographies

Caleb S. Bailey is a postdoctoral scholar at University of Kentucky, with an interest in mechanisms of aging and cognitive resilience.

McKenna C. Green is an undergraduate student at University of Kentucky, majoring in psychology and public health.

Meggie J. Coleman is a research scientist at University of Kentucky, specializing in veterinary care and rodent research.

Linda J. Van Eldik is Director and Professor at University of Kentucky, specializing in neuroinflammation and therapeutic development.

David J. Braun is Research Assistant Professor at University of Kentucky, with interests in glial-neuronal interactions in central nervous system disorders.

Footnotes

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author(s).

DATA AVAILABILITY STATEMENT

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials. Additional data is available upon reasonable request.

REFERENCES

1.Parmar RM & Can AS in StatPearls (2025).
2.Xie Y, Choi T & Al-Aly Z Mapping the effectiveness and risks of GLP-1 receptor agonists. Nat Med 31, 951–962 (2025). 10.1038/s41591-024-03412-w [DOI] [PubMed] [Google Scholar]
3.Gulinac M et al. Long-term effectiveness, outcomes and complications of bariatric surgery. World J Clin Cases 11, 4504–4512 (2023). 10.12998/wjcc.v11.i19.4504 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Huang YN et al. Long-term safety and efficacy of glucagon-like peptide-1 receptor agonists in individuals with obesity and without type 2 diabetes: A global retrospective cohort study. Diabetes Obes Metab 26, 5222–5232 (2024). 10.1111/dom.15869 [DOI] [PubMed] [Google Scholar]
5.Sal-Sarria S, Lopez-Taboada I, Gonzalez-Pardo H & Conejo NM A shift to a standard diet after exposure to a high-fat, high-sucrose diet from gestation to weaning restores brain metabolism and behavioral flexibility in adult rats. Behav Brain Res 467, 115020 (2024). 10.1016/j.bbr.2024.115020 [DOI] [PubMed] [Google Scholar]
6.Braga SP, Delanogare E, Machado AE, Prediger RD & Moreira ELG Switching from high-fat feeding (HFD) to regular diet improves metabolic and behavioral impairments in middle-aged female mice. Behav Brain Res 398, 112969 (2021). 10.1016/j.bbr.2020.112969 [DOI] [PubMed] [Google Scholar]
7.van Galen KA et al. Brain responses to nutrients are severely impaired and not reversed by weight loss in humans with obesity: a randomized crossover study. Nat Metab 5, 1059–1072 (2023). 10.1038/s42255-023-00816-9 [DOI] [PubMed] [Google Scholar]
8.Chiesa ST, Norris T, Garfield V, Richards M & Hughes AD Early-life cumulative exposure to excess bodyweight and midlife cognitive function: longitudinal analysis in three British birth cohorts. Lancet Healthy Longev 5, e204–e213 (2024). 10.1016/S2666-7568(24)00005-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Freeman LR, Haley-Zitlin V, Rosenberger DS & Granholm AC Damaging effects of a high-fat diet to the brain and cognition: a review of proposed mechanisms. Nutr Neurosci 17, 241–251 (2014). 10.1179/1476830513Y.0000000092 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Ferreira NB et al. Improving dietary patterns in obese mice: Effects on body weight, adiposity, anhedonia-like behavior, pro-BDNF expression and 5-HT system. Brain Res 1838, 148996 (2024). 10.1016/j.brainres.2024.148996 [DOI] [PubMed] [Google Scholar]
11.Ryan DH et al. Look AHEAD (Action for Health in Diabetes): design and methods for a clinical trial of weight loss for the prevention of cardiovascular disease in type 2 diabetes. Control Clin Trials 24, 610–628 (2003). 10.1016/s0197-2456(03)00064-3 [DOI] [PubMed] [Google Scholar]
12.Pi-Sunyer X The Look AHEAD Trial: A Review and Discussion Of Its Outcomes. Curr Nutr Rep 3, 387–391 (2014). 10.1007/s13668-014-0099-x [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Rapp SR et al. Effect of a Long-Term Intensive Lifestyle Intervention on Cognitive Function: Action for Health in Diabetes Study. J Am Geriatr Soc 65, 966–972 (2017). 10.1111/jgs.14692 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.George AJ, Harmsen BJ, Ford JA, Tadepalli SR & Horton ND Evaluation of Submental Blood Collection in Mice (Mus musculus). J Am Assoc Lab Anim Sci 62, 92–98 (2023). 10.30802/AALAS-JAALAS-21-000136 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Braun DJ et al. Early chronic suppression of microglial p38alpha in a model of Alzheimer’s disease does not significantly alter amyloid-associated neuropathology. PLoS One 18, e0286495 (2023). 10.1371/journal.pone.0286495 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Braun DJ et al. Therapeutic treatment with the anti-inflammatory drug candidate MW151 may partially reduce memory impairment and normalizes hippocampal metabolic markers in a mouse model of comorbid amyloid and vascular pathology. PLoS One 17, e0262474 (2022). 10.1371/journal.pone.0262474 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Frazier HN et al. A small molecule p38alpha MAPK inhibitor, MW150, attenuates behavioral deficits and neuronal dysfunction in a mouse model of mixed amyloid and vascular pathologies. Brain Behav Immun Health 40, 100826 (2024). 10.1016/j.bbih.2024.100826 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Casimiro I, Stull ND, Tersey SA & Mirmira RG Phenotypic sexual dimorphism in response to dietary fat manipulation in C57BL/6J mice. J Diabetes Complications 35, 107795 (2021). 10.1016/j.jdiacomp.2020.107795 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Kiran S, Rakib A, Kodidela S, Kumar S & Singh UP High-Fat Diet-Induced Dysregulation of Immune Cells Correlates with Macrophage Phenotypes and Chronic Inflammation in Adipose Tissue. Cells 11 (2022). 10.3390/cells11081327 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Nilsson SR, Alsio J, Somerville EM & Clifton PG The rat’s not for turning: Dissociating the psychological components of cognitive inflexibility. Neurosci Biobehav Rev 56, 1–14 (2015). 10.1016/j.neubiorev.2015.06.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Gilroy SP Interpretation(s) of essential value in operant demand. J Exp Anal Behav 119, 554–564 (2023). 10.1002/jeab.845 [DOI] [PubMed] [Google Scholar]
22.de Paula GC et al. Hippocampal Function Is Impaired by a Short-Term High-Fat Diet in Mice: Increased Blood-Brain Barrier Permeability and Neuroinflammation as Triggering Events. Front Neurosci 15, 734158 (2021). 10.3389/fnins.2021.734158 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Hou J et al. High fat diet-induced brain damaging effects through autophagy-mediated senescence, inflammation and apoptosis mitigated by ginsenoside F1-enhanced mixture. J Ginseng Res 46, 79–90 (2022). 10.1016/j.jgr.2021.04.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Gu L et al. Control of TH2 polarization by the chemokine monocyte chemoattractant protein-1. Nature 404, 407–411 (2000). 10.1038/35006097 [DOI] [PubMed] [Google Scholar]
25.Lin SY et al. Interleukin-4 Improves Metabolic Abnormalities in Leptin-Deficient and High-Fat Diet Mice. Int J Mol Sci 21 (2020). 10.3390/ijms21124451 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Mondal D, Mathur A & Chandra PK Tripping on TRIB3 at the junction of health, metabolic dysfunction and cancer. Biochimie 124, 34–52 (2016). 10.1016/j.biochi.2016.02.005 [DOI] [PubMed] [Google Scholar]
27.Yeo H et al. Diet-induced obesity and aging-induced upregulation of Trib3 interfere with energy homeostasis by downregulating the thermogenic capacity of BAT. Exp Mol Med 56, 2690–2702 (2024). 10.1038/s12276-024-01361-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Rodrigues Bde A et al. Acute Exercise Decreases Tribbles Homolog 3 Protein Levels in the Hypothalamus of Obese Rats. Med Sci Sports Exerc 47, 1613–1623 (2015). 10.1249/MSS.0000000000000585 [DOI] [PubMed] [Google Scholar]
29.Aime P et al. Trib3 Is Elevated in Parkinson’s Disease and Mediates Death in Parkinson’s Disease Models. J Neurosci 35, 10731–10749 (2015). 10.1523/JNEUROSCI.0614-15.2015 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Lorenzi M et al. Susceptibility of brain atrophy to TRIB3 in Alzheimer’s disease, evidence from functional prioritization in imaging genetics. Proc Natl Acad Sci U S A 115, 3162–3167 (2018). 10.1073/pnas.1706100115 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Xiong H et al. Hesperidin: A Therapeutic Agent For Obesity. Drug Des Devel Ther 13, 3855–3866 (2019). 10.2147/DDDT.S227499 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Chen L et al. TRIB3 promotes malignancy of head and neck squamous cell carcinoma via inhibiting ferroptosis. Cell Death Dis 15, 178 (2024). 10.1038/s41419-024-06472-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials