PREreview del Insulin resistance alters cortical inhibitory neurons and microglia to exacerbate Alzheimer’s knock-in mouse phenotypes

In this manuscript, Nicholson et al., investigate how systemic metabolic dysfunction alters brain function in Alzheimer’s disease (AD). Building on prior evidence that both hyperglycemia and insulin signaling deficits contribute to AD risk, a key question addressed by this study is whether it is insulin resistance itself or hyperglycemia alone that drives cognitive decline in AD, or vice versa.

To address this question, the authors use two distinct metabolic perturbations to induce hyperglycemia in human mutant APP and wild-type tau knock-in (DKI) AD mouse model. One is administration of streptozotocin (STZ) to reduce insulin production, and the other is high-fat, high-sugar (HFHS) diet-induced insulin resistance. They demonstrate that both STZ and HFHS treatments induce hyperglycemia, but only HFHS-induced insulin resistance leads to learning and memory deficits. Although overall Aβ deposition and tau phosphorylation remain unchanged, HFHS diet can induce microglia morphological remodeling in DKI mice.

To further investigate the glia regulation under HFHS, they perform single-nucleus RNA-seq to identify a unique glial metabolic impairment state (MinD), while L2/3 Meis2+ inhibitory neurons and excitatory neurons display convergent transcriptional changes affecting synaptic organization and trans-synaptic signaling. These results suggest HFHS-induced insulin resistance, rather than hyperglycemia alone, may exacerbate AD-related functional deficits through cell type specific transcriptional remodeling in both glia and neurons, with consequences on synaptic organization and inhibitory signaling, independent of classical Aβ or tau pathology.

Overall, this study provides a valuable and detailed descriptive framework for how metabolic dysfunction modulates AD progression. The identification of glial MinD states and Meis2⁺ inhibitory neuron sensitivity represents an advance, but the evidence linking insulin resistance to cognitive decline remains primarily correlative. Thus, the strength of support for this conclusion is suggestive but not definitive, and additional mechanistic studies would help clarify mechanistic insights and strengthen the central claims.

Major comments:

1.     Behavioral testing was performed at a single timepoint, and snRNA-seq was done after behavioral deficits, which is reasonable to capture transcriptional changes associated with the phenotype, but it limits the ability to determine whether these changes are causal or secondary. Multiple timepoints would better clarify causality and dynamic progression (Fig. 1, Ext Fig. 1 and Ext. Fig. 2). Additionally, the authors used one behavior assay, which may give a limited view of cognitive function. Adding a second assay, such as novel object recognition, would help confirm whether the impairment is specific to one assay and could help strengthen the conclusions.

2.     Immunofluorescence analysis was performed only in the cortex and did not include the hippocampus. Given that the hippocampus is critical for the cognitive functions assessed, including hippocampal immunofluorescence data, or briefly discussing why it was not examined would improve the study (Fig. 2a-e).

3.     The rationale for using different control groups for comparisons in Fig. 3 is unclear. Fig. 3d-g analyses use DKI-HFHS vs DKI-Lean to identify 62 uniquely altered genes in DKI-HFHS microglia, whereas Fig. 3h compares DKI-HFHS vs WT-Lean. The authors should clarify the rationale for using different reference groups and how these choices affect the interpretation of results, which would help the reader interpret which transcriptional changes are attributable to HFHS diet alone, AD pathology alone, or combined metabolic. This clarification would enhance the internal consistency of the analysis and reduce potential ambiguity about how HFHS-specific signatures are defined.

4.     Fig. 4h, inhibitory synapse density is only compared between WT-HFHS and DKI-HFHS. Including DKI-Lean or WT-Lean would clarify the effect of HFHS diet alone. Clarifying why different comparisons are used across Fig. 4a–e versus Fig. 4g–h would help the reader distinguish the contributions of HFHS diet versus AD pathology to synaptic changes, improving the interpretability of the data and supporting conclusions about the specific impact of insulin resistance-induced metabolic stress on inhibitory synapses.

5.     Neuregulin 3 (Nrg3) emerged as a candidate linking glia transcriptional shifts to inhibitory synaptic signaling. Since Nrg3 primarily binds to ErbB4, which is enriched at inhibitory synapses, the authors may consider manipulating either Nrg3 or ErbB4 to test whether Nrg3-ErbB4 signaling mediated the synaptic deficits observed in Fig. 4. or briefly discuss this pathway as a potential direction for future mechanistic investigation.

6.     Fig. 6 shows that HFHS induces Meis2 upregulation in Layer 2/3 inhibitory neurons, but it is unclear whether this change contributes to cognitive deficits. The authors may consider manipulating Meis2 expression specifically in these neurons and assess effects on synapse and behavior. The experiment is not essential for the current descriptive analysis, but it could clarify the causal link between metabolic stress and cognitive impairment. Alternatively, adding a short discussion about whether Meis2 may play a functional role in metabolic vulnerability could further contextualize the findings.

7.     Electrophysiology recordings targeting L2/3 Meis2+ neurons and excitatory neurons would significantly strengthen the causal inference between HFHS and synaptic deficits shown in Fig. 6 and Fig. 7. While not necessary for the current manuscript, the authors could briefly discuss this as a future direction to test functional consequences of transcriptional changes.

Minor comments:

1.     Ext Fig1, Panels a-c appear mislabeled regarding WT-Lean, WT-HFHS, DKI-Lean and DKI-HFHS, AppNLF/hTau WT label also appears incorrect in Ext Fig 1d. To improve readability, the authors should consider verifying these labels.

2.     The figure referenced for the statement on line 17 of page 3"Performance in the visible platform task remained consistent across all groups (Fig. 1d)" appears to be incorrect. Please verify and update the figure citation to accurately reflect the data supporting this claim.

3.     Ext Fig. 5f, comparing WT-Lean to DKI-Lean would clarify whether DKI mice exhibit synaptic deficits independent of HFHS diet. Performing statistical comparisons across all groups (WT-Lean, WT-HFHS, DKI-Lean, DKI-HFHS) would allow the authors to distinguish the contributions of genotype versus diet and better communicate which changes are specifically driven by HFHS-induced metabolic stress versus AD pathology. This would improve the interpretability of the synaptic data and strengthen the link between metabolic perturbation and synaptic alterations.

4.     Nrg3 is upregulated in microglia, astrocytes, and oligodendrocytes based on snRNA-seq data. Immunofluorescence staining validation to confirm the protein expression changes with cell-specific localization is suggested.

5.     The figure citation for the statement on line 40 of page 4 "First, we focused on microglia to identify transcriptional changes that may underlie the observed Trem2 alterations, despite the equal number of microglial nuclei in the DKI-Lean and DKI-HFHS groups" appears to be incorrect. Please verify and update the figure reference (currently listed as Extended Data Fig. C) to accurately reflect the relevant data.

6.     The figure citation for the statement on line 1 of page 5, “We next cross-referenced gene expression within the MG3 cluster against a curated list of microglia state markers, which were stratified by WT and DKI groups based on disease-associated activation (Extended Data Fig. d)” is incorrect. Please doublecheck.

7.     The authors should cite relevant literature regarding whether patients with Type 1 diabetes have cognitive decline or an increased risk of AD (10.1038/s41598-024-53043-x; 10.1007/s10654-023-01080-7; 10.1002/brb3.3533), which is important to clarify whether STZ-induced findings in AD mouse model are specific to mouse or supported by human data. The authors should provide a discussion addressing this point, integrating existing data to contextualize their findings.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.