A History of Repeated Alcohol Intoxication Promotes Cognitive Impairment and Gene Expression Signatures of Disease Progression in the 3xTg Mouse Model of Alzheimer’s Disease

Repeated alcohol intoxication by exposure to alcohol vapor hastens the emergence of cognitive impairment in 3xTg-AD mice

3xTg-AD mice develop progressive cognitive impairments in spatial memory (Oddo et al., 2003; Billings et al., 2005). To test the hypothesis that a history of repeated alcohol intoxication hastens cognitive impairment progression in the setting of genetic AD susceptibility, we exposed presymptomatic 3xTg-AD and WT mice (age, 2 months) to repeated cycles of alcohol intoxication by exposure to alcohol vapor with the established CIE paradigm (Becker and Hale, 1993; Becker and Lopez, 2004). We then examined spatial memory in the Morris water maze memory test in 3xTg-AD and WT mice during protracted withdrawal after exposure to repeated cycles of alcohol intoxication (5–6 weeks of total exposure with 2 weeks on, 1 week off, repeated three times) and unexposed control mice of both genotypes when they were 4.25 (males) and 4.5 (females) months of age. The results showed that repeated cycles of alcohol intoxication with the CIE paradigm hastened the onset of cognitive impairment in 3xTg-AD mice (Fig. 1B,C). 3xTg-AD mice not exposed to alcohol vapor and WT mice both exposed and not exposed to alcohol vapor were not impaired and performed similarly in the Morris water maze (Fig. 1B,C). These results were confirmed in independent cohorts of male 3xTg-AD mice and 5xFAD mutant mice, which harbor three APP and two PS1 FAD mutations (Oakley et al., 2006; Extended Data Fig. 1-1). Body weights increased across the experiment in all groups by 1.5 ± 0.2 g, with no differences among groups. These results lend support to the hypothesis that a history of excessive alcohol intake promotes AD progression in the setting of genetic susceptibility and show that repeated alcohol intoxication by exposure to alcohol vapor is a robust method to model the interaction of excessive alcohol intake with AD susceptibility.

Ultra-high-definition molecular survey of the interaction of alcohol with Alzheimer’s disease: snRNA-Seq of 3xTg-AD and WT mice with and without a history of repeated alcohol intoxication

To investigate the gene signature and disease-related transcriptional changes induced by repeated cycles of alcohol intoxication sufficient to promote cognitive impairment in 3xTg-AD mice, we performed snRNA-Seq of the PFC of 3xTg-AD and WT mice exposed to alcohol vapor and unexposed mice of both genotypes at 4.75–5 months of age (4–7 d following behavioral testing; 11–14 d following the final alcohol vapor exposure) on the 10x Genomics platform. We sampled 516,560 cells. After quality control filtering (see Materials and Methods), we obtained 113,242 cells with a mean of 1986.7 detected genes per cell. The Azimuth reference-based cell identity assignment was used for cell type identification, confidently assigning 91,898 cells to 22 cytological types according to established mouse neocortex molecular taxonomy (Yao et al., 2021; Fig. 2A, Extended Data Fig. 2-1). Gene markers for the majority of cell types allowed us to clearly and robustly discern the underlying histology, in particular for non-neuronal cell types, including endothelial cells, oligodendrocytes, and vascular leptomeningeal cells, among others (Fig. 2A,B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Cytological summary of snRNA-Seq of 3xTg-AD and WT mice with and without a history of repeated alcohol intoxication recapitulates cell type-specific marker genes. A, Dot plot of the expression and presence of selected known gene markers for specific cell types (columns) in cells defined by Azimuth reference-based cell identity assignments (rows). B, t-SNE projection of the entire dataset, showing the identified cell identities. C, Stacked bar plot of the number of cells in each cell type, distributed according to the four experimental conditions (Extended Data Fig. 2-1, list of abbreviations of cell types, Fig. 2-2, distribution of cell types in the four experimental conditions).

The visualization of single-nucleus transcriptomes in t-SNE showed that nuclei of the 22 detected cell types were loosely distributed in transcriptionally separated clusters (Fig. 2B). Overall, the main source of variance in our dataset was the cell type, reflected by the cell marker landscape (Fig. 2A). Cells assigned to astrocyte, oligodendrocyte, oligodendrocyte progenitor cell (OPC), and microglia lineages were clearly distinct in well defined clusters, whereas GABAergic neurons (Pvalb, Sst, Sst Chodl, Sncg, Vip, and Lamp5) constitute a separate cell population. The most abundant cell populations were glutamatergic neurons, intratelencephalon projecting, from layers 2, 3, 5, and 6 (Fig. 2B,C).

The distribution of cell types in the four experimental conditions was generally well balanced (Fig. 2C, Extended Data Fig. 2-2). Overall, we found that approximately two-thirds of differentially expressed transcripts were represented by protein-coding genes, and approximately one-third by noncoding genes (Extended Data Fig. 3-1).

Cell type-specific genes associated with a history of alcohol intoxication in the PFC of 3xTg-AD mice highlight candidate mechanisms of AD disease progression

Differentially expressed genes are shown in Figure 3. Cmss1 (Cms1 Ribosomal Small Subunit Homolog) was selectively downregulated in 3xTg-AD mice exposed to alcohol versus WT mice without alcohol exposure but not in 3xTg-AD mice and WT mice exposed to alcohol versus WT controls in multiple neuronal and non-neuronal cell types, including L5/6 nonpeptidergic (NP), L5 external tufted (ET), L2/3 IT, L5 IT, L6b, L6 corticothalamic (CT), VIP, parvalbumin, Meis2, Sncg and SSt interneurons, Lamp5, astrocytes, microglia-perivascular macrophages (PVMs), oligodendrocytes, and OPCs (Fig. 4A, Extended Data Fig. 4-1). Several ribosomal proteins were selectively decreased in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice and WT mice exposed to alcohol versus WT controls in L6 IT neurons, including Rpl23a, Rpl13a, Rpl35a, Rpl36a, Rpl37, and Mrpl33 [Mrpl33 also decreased in Sst and L2/3 intratelencephalic (IT) neurons]. Conversely, ribosomal proteins were generally unchanged or increased in other cell types (Fig. 4B, Extended Data Fig. 4-1). The mitochondrial leucyl-tRNA synthetase (Lars2), implicated in mitochondrial protein synthesis, lactate utilization (Capriglia et al., 2021), episodic memory in aging (Ramanan et al., 2015), bipolar disorder, and schizophrenia (Munakata et al., 2005), was selectively downregulated in multiple neuronal and non-neuronal cell types, including L5 IT, Meis2, VIP, astrocytes, oligodendrocytes, and OPC, in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice and WT mice exposed to alcohol versus WT controls (Fig. 4A, Extended Data Fig. 4-1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Cell type-specific gene expression changes in 3xTg-AD and WT mice with and without a history of repeated alcohol intoxication. A–D, Heatmaps of the most significantly differentially expressed coding genes in the following contrasts: 3xTg-AD mice with a history of repeated alcohol intoxication by exposure to alcohol vapor (ADvapor) versus 3xTg-AD control mice not exposed to alcohol (AD; A); 3xTg-AD mice with a history of repeated alcohol intoxication (ADvapor) versus WT mice with a history of alcohol intoxication (vapor; B); 3xTg-AD mice with a history of repeated alcohol intoxication (ADvapor) versus WT mice not exposed to alcohol (ctrl; C); 3xTg-AD mice without alcohol exposure (AD) versus WT mice not exposed to alcohol (ctrl; D; Extended Data Fig. 3-1, distribution of differentially expressed genes in each cell type in the four contrasts).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Selected and global transcriptomics landscape of cell type-specific differential gene expression in 3xTg-AD mice with a history of repeated alcohol intoxication. A, Heatmap of the top five most differentially expressed coding genes per cell type in 3xTg-AD mice with a history of repeated alcohol intoxication by alcohol vapor exposure versus WT control mice not exposed to alcohol but not significantly differentially expressed in 3xTg-AD mice without histories of alcohol intoxication or WT mice with a history of alcohol intoxication versus WT control mice not exposed to alcohol. B, Heatmap of all significantly differentially expressed genes per cell type in 3xTg-AD mice with a history of repeated alcohol intoxication by alcohol vapor exposure versus WT control mice not exposed to alcohol but not significantly differentially expressed in 3xTg-AD mice without histories of alcohol intoxication or WT mice with a history of alcohol intoxication versus WT control mice not exposed to alcohol. The list is available in Extended Data Figure 4-1. Selected genes are labeled. C, Heatmap of differential expression of genes associated with AD from GWASs (see Materials and Methods) in the contrast of 3xTg-AD mice with a history of repeated alcohol intoxication by alcohol vapor exposure versus WT control mice not exposed to alcohol (Extended Data Fig. 4-1, tabular results reporting significantly differentially expressed genes related to panels A and B).

A broad increase in the transcription of cadherins was seen in excitatory neurons of both deep and superficial layers of 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice and WT mice exposed to alcohol versus WT controls (Fig. 4, Extended Data Fig. 4-1). Particularly, L6 IT neurons showed increases in the transcription of cadherins, including Pcdh15, Pcdh17, Pcdh9, Cdh8, Cdh9, Cdh10, Cdh11, and Cdh13, whereas Pcdh17 and Pcdh19 were upregulated in L6 CT neurons. L5 IT neurons showed increased Cdh10. L2/3 IT neurons showed the increased transcription of Cdh9 and Pcdh17, and L6b neurons showed an increase in Pcdh15. Cdh2 (N-cadherin) was significantly decreased in oligodendrocytes of 3xTg-AD mice exposed to alcohol versus WT controls (adjusted p = 6.87e-09; log₂ fold change = −1.70; Fig. 3C) as well as of 3xTg-AD mice not exposed to alcohol versus WT controls (adjusted p = 3.35e-07; log₂ fold change = −1.60; Fig. 3D) and of 3xTg-AD mice exposed to alcohol versus WT mice exposed to alcohol (adjusted p = 4.54e-05; log2 fold change = −1.40; Fig. 3B). Cdh2 showed upregulation in other cell types (i.e., L6 IT; Fig. 3).

Parvalbumin and VIP interneurons and L6 CT neurons showed the increased transcription of Glra2 (glycine receptor α2; Fig. 4A, Extended Data Fig. 4-1), and VIP interneurons showed the increased expression of Grid2 (glutamate receptor GluRδ2; Extended Data Fig. 4-1) in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice versus WT controls or in WT mice exposed to alcohol versus WT controls.

Contactin-associated protein 5c (Cntnap5c), a member of a cell adhesion protein family implicated in interneuron development and function (Bridi et al., 2017), was selectively upregulated in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice or WT mice exposed to alcohol versus WT controls in parvalbumin, Lamp5, and Sst neurons, and decreased in L6 CT neurons (Fig. 4A). Its paralog Cntnap5b increased in VIP GABAergic neurons in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice or WT mice exposed to alcohol versus WT controls (Extended Data Fig. 4-1). Another member of the protein family, Cntnap2, was decreased in OPC (Fig. 4A, Extended Data Fig. 4-1).

α-Synuclein (Snca) was slightly but significantly increased in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice or WT mice exposed to alcohol versus WT controls of pyramidal neurons in L5 and L2/3 (L2/3 IT, L5 IT; Extended Data Fig. 4-1).

Slc1a3, encoding for the excitatory amino acid transporter 1 (EAAT1), was among the genes significantly increased by repeated alcohol intoxication in astrocytes of 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice or WT mice exposed to alcohol versus WT controls (Extended Data Fig. 4-1). In microglia, Slc1a3 was significantly downregulated in both 3xTg-AD mice exposed to alcohol and 3xTg-AD mice not exposed to alcohol versus WT control mice not exposed to alcohol (Fig. 3C,D, respectively).

In 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice or WT mice exposed to alcohol versus WT controls, complement C4b and metallothionein-1 (Mt1) were selectively increased in astrocytes, whereas among IFN-regulated genes, Ifitm3 was upregulated in astrocytes, endothelial cells, and L2/3 IT, and Ifi27 was upregulated in astrocytes and L5 ET (Fig. 4A, Extended Data Fig. 4-1).

Genes with decreased expression in astrocytes in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice or WT mice exposed to alcohol versus WT controls included the following: the WD repeat-containing protein Wdr17, abundantly expressed in astrocytes and associated with macular degeneration (Stöhr et al., 2002); Smim4, which may be involved in mitochondrial toxicity (Dennerlein et al., 2021); and autophagy-related 16 like 2 (Atg16l2), a proposed peripheral marker of AD (Qin et al., 2022) that has been shown to be a negative regulator of the NOD (nucleotide-binding oligomerization domain)-like receptor family, pyrin domain containing 3 (NLRP3) inflammasome (Wang et al., 2022) and thus may contribute to neuroinflammation (Fig. 4B, Extended Data Fig. 4-1).

Among the most increased genes in microglia-PVM in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice versus WT controls and WT mice exposed to alcohol versus WT controls are contributors to neuroinflammation in the setting of AD and alcohol abuse, including the Toll-like receptor 7 (Tlr7), which was found to be increased in the brains of humans with alcohol use disorder (Coleman et al., 2017); Epsti1, a regulator of inflammatory gene expression in macrophages (Kim et al., 2018); cathepsin S, a lysosomal protease that is also active outside the lysosome implicated in inflammation (Kim et al., 2017) and AD (Lowry and Klegeris, 2018); H2-K1, H2-D1, and the Rho guanosine triphosphatase-activating protein 15 (Arhgap15), a regulator of the β2-integrin Mac-1 (Margraf et al., 2020); the IFN-induced protein Apobec3; and Gpm6b, which was also upregulated in the parietal cortex in AD brains (Patel et al., 2019; Fig. 4A,B, Extended Data Fig. 4-1).

The cyclin-dependent kinase Cdk8 was selectively downregulated in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice or WT mice exposed to alcohol versus WT controls in multiple non-neuronal cell types, including both astrocytes and microglia-PVM, VLMC (ventral laryngeal motor cortex), and oligodendrocytes, and similar trends were present in some neuronal populations. Cdk8 is part of a protein complex that regulates RNA polymerase II, is involved in regulating multiple cellular states, including development, metabolism, and innate immunity (Dannappel et al., 2018), and was also decreased in the liver by chronic alcohol exposure (Song et al., 2017).

In endothelial cells, several upregulated genes in 3xTg-AD mice exposed to alcohol versus WT controls but not in 3xTg-AD mice or WT mice exposed to alcohol versus WT controls were previously found to be upregulated in aged capillaries, including innate immunity genes, such as Ly6a, Ly6c1, B2m, H2-K1, H2-D1, and Ifitm3 (Chen et al., 2020).

Overall, neurons in both superficial and deep layers of the PFC of 3xTg-AD mice were highly transcriptionally responsive to a history of alcohol intoxication (Fig. 4B). Genes previously associated with AD in humans by GWASs were found to be differentially regulated in specific neuronal populations in both deep and superficial layers (Fig. 4C), particularly in L2/3 IT, L5 IT, and L6 IT.

Differentially expressed genes associated with a history of alcohol intoxication in the prefrontal cortex of 3xTg-AD mice indicate that a history of alcohol intoxication promotes AD disease progression

To test the hypothesis that a history of repeated alcohol intoxication promotes disease progression in 3xTg-AD mice, we compared transcriptional responses of our dataset with those generated with age-dependent AD progression data by Yin et al. (2020). We searched for similarities in genes upregulated in a dataset of the insular cortex of 3xTg-AD mice at 14 months of age and 2 months of age in the study by Yin et al. (2020) by overlaying them onto the pseudo-bulk signature in our experiment, defined by the comparison of the PFC of 3xTg-AD mice exposed to alcohol to WT mice without alcohol exposure using the GSEA algorithm (Subramanian et al., 2005; Fig. 5A). The results showed positive and significant similarity between the signature of 3xTg-AD mice at 14 months by Yin et al. (2020) and 3xTg-AD mice exposed to alcohol versus WT control mice in the present study but lower similarity for the signature of 3xTg-AD mice not exposed to alcohol versus WT control mice (Fig. 5A). We then conducted the same analysis with the signature of genes upregulated in 5xFAD mice at 18 and 4 months of age in a cerebral cortex dataset from the study by Forner et al. (2021; Fig. 5B). The GSEA calculated for the signature of 3xTg-AD mice exposed to alcohol versus WT controls in the present study resulted in positive and significant similarity to signatures of 5xFAD mice at 18 and 4 months of age from the study by Forner et al. (2021), whereas 3xTg-AD mice not exposed to alcohol versus WT controls in this work were significantly enriched for genes characterizing the signature of 5xFAD mice at 4 months but not at 18 months. However, the reference set of Forner et al. (2021) at 4 months corresponds to 46 genes compared with 400 elements at 18 months.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Transcriptional evidence that a history of alcohol intoxication promotes AD progression. A, GSEA analysis was used to compare genes that were significantly upregulated in 3xTg-AD mice (Yin et al., 2020) at 14 months (top) and 2 months (bottom) with the transcriptome-wide signature defined by our contrast of 3xTg-AD mice with a history of repeated alcohol intoxication by alcohol vapor (ADvapor) versus WT control (left) or 3xTg-AD mice not exposed to alcohol versus WT control (right). All cell types were used to define the signature, expressed as -log₁₀(p)*sign(log₂FC). The results indicated greater similarity between 3xTg-AD mice at 14 months by Yin et al. (2020) for the ADvapor versus WT control contrast than for the AD versus WT control contrast. B, GSEA analysis was used to compare genes that were significantly upregulated in 5xFAD mice at 18 months (top) and 4 months (bottom) from the study by Forner et al. (2021) with the transcriptome-wide signature defined by our contrast of 3xTg-AD mice with a history of repeated alcohol intoxication by alcohol vapor (ADvapor) versus WT control (left) and 3xTg-AD mice not exposed to alcohol versus WT control (right), like in panel A. The analysis showed positive and significant similarity between the AD model at 18 months for the ADvapor versus WT control contrast but not for the AD versus WT control contrast.

These comparisons indicate that the gene expression signature induced by repeated cycles of alcohol intoxication in presymptomatic 3xTg-AD mice in the present study are significantly more similar to 3xTg-AD mice (Yin et al., 2020) and 5xFAD mice (Forner et al., 2021) that are older and have more advanced disease and cognitive impairment than the gene expression signature of 3xTg-AD mice not exposed to alcohol. Therefore, and with the caveats presented by cross-dataset comparison analyses, this analysis suggests that repeated alcohol intoxication promotes transcriptional changes consistent with AD progression.

Pathway analysis by GSEA in the PFC of 3xTg-AD mice with a history of alcohol intoxication indicates increased inflammation, cell type-specific vulnerability, and activation of compensatory and adaptive mechanisms

GSEA revealed biological pathways that were significantly enriched in the four contrasts (Fig. 6, selected nonredundant pathways, Extended Data Figs. 6-1, 6-2, 6-3, 6-4, report of all significantly enriched pathways, Extended Data Fig. 6-5, report of pathways enriched in 3xTg-AD mice exposed to alcohol vs WT mice but not in 3xTg-AD mice or WT mice exposed to alcohol vs WT controls). The expression of gene sets related to type I and II IFN was increased in astrocytes and microglia-PVM of the PFC of 3xTg-AD mice exposed to alcohol versus 3xTg-AD mice and WT mice with and without alcohol exposure (Figs. 6, 7). In astrocytes, gene sets related to calcium signaling were downregulated. In microglia-PVM, pathways related to immune activation and proliferation were upregulated (Fig. 6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Cell type-specific pathway analysis by GSEA in 3xTg-AD and WT mice with and without a history of repeated alcohol intoxication. A–D, Pathway analysis by GSEA in 3xTg-AD mice with a history of repeated alcohol intoxication by exposure to alcohol vapor (ADvapor) versus 3xTg-AD control mice not exposed to alcohol (AD; A); 3xTg-AD mice with a history of repeated alcohol intoxication (ADvapor) versus WT mice with a history of alcohol intoxication (vapor; B); 3xTg-AD mice with a history of repeated alcohol intoxication (ADvapor) versus WT untreated mice (ctrl; C); and 3xTg-AD mice without alcohol exposure (AD) versus WT mice without alcohol exposure (ctrl; D). For tabular results reporting the significant enriched pathways, see the following: Extended Data Figure 6-1 for panel C; Extended Data Figure 6-2 for panel B; Extended Data Figure 6-3 for panel D; and Extended Data Figure 6-4 for panel A. For the contrast of 3xTg-AD mice with a history of repeated alcohol intoxication versus WT mice without a history of alcohol intoxication but not significant in 3xTg-AD mice without alcohol exposure and WT mice with a history of alcohol intoxication versus ctrl contrasts, see Extended Data Figure 6-5.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Induction of type I and II IFN in 3xTg-AD mice by a history of repeated alcohol intoxication. A, B, GSEA results for the IFN-α response pathway [the most significantly upregulated pathway in 3xTg-AD mice exposed to alcohol versus 3xTg-AD mice (ADvapor vs AD contrast)] in astrocytes (A) and microglia (B). C, D, GSEA results for the interferon-γ response pathway (the most significantly upregulated pathway in the ADvapor vs AD contrast) in astrocytes (C) and microglia (D). Top panels indicate the GSEA analysis in the specific contrast and cell type. Bottom panels use box plots to indicate the differential expression of genes belonging to the relative indicated pathways in the four following contrasts: 3xTg-AD mice exposed to alcohol versus 3xTg-AD mice (ADvapor vs AD; red); 3xTg-AD mice exposed to alcohol versus WT mice exposed to alcohol (ADvapor vs vapor; green); WT mice exposed to alcohol versus WT mice not exposed to alcohol (vapor vs WT ctrl; blue); and 3xTg-AD mice versus WT (AD vs WT ctrl; cyan). E, Heatmap showing differential expression values for genes in IFN-α and IFN-γ response pathways. Genes were selected because of being in the GSEA leading edge in at least one of the following four contrasts: ADvapor versus AD_Astrocytes; ADvapor versus AD_Micro-PVM; ADvapor versus vapor Astrocytes; and ADvapor versus vapor_Micro-PVM. Total: 118 genes. The square defines a subset of IFN-α and IFN-γ pathway genes that show an additive effect in ADvapor.

Pathways related to synaptic membrane components were selectively upregulated in a cell type-specific manner in neurons of 3xTg-AD mice exposed to alcohol versus 3xTg-AD mice and WT mice with and without alcohol exposure, including in L6 CT, L6b, Lamp5, and parvalbumin, and Meis2 interneurons. Increases in synaptic elements and synaptic and metabolic activity are believed to be compensatory and adaptive mechanisms of the natural history of early AD (Bobkova and Vorobyov, 2015; Merlo et al., 2019).

Gene sets related to ribosomal RNA decreased in a cell type-specific manner in L6 IT neurons of 3xTg-AD mice exposed to alcohol versus 3xTg-AD mice and WT control mice, whereas they increased in most other cell types (Fig. 6A,C, respectively). A decrease in gene sets related to oxidative phosphorylation, mitochondrial function, and Parkinson-associated genes was seen in Meis2 interneurons, possibly underlying cell type-specific vulnerability (Extended Data Fig. 6-5).