Postmortem brain tissue samples were obtained from the New South Wales Brain Tissue Research Centre (BTRC) in Sydney, Australia. The NSW BTRC and its associated donor program have ethics approval from the NSW Government Health authority/University of Sydney to bank postmortem brains from deceased subjects with documented histories of alcohol abuse or normal controls (ref# X11-0107&HREC/11/RPAH/147). Under Australian law, people aged 18 or older are considered adults. The NSW BTRC has ethics approval that allows any adult to consent to brain donation. The present study includes cases between 40 years and 70 years of age. Prospective donors were provided with written and verbal information about the nature of the study, the procedures, and the evaluations involved. Prospective participants were screened for eligibility. Once they or their next of kin (in cases where consent was obtained via the coronial office) understood the nature and purpose of the study and what was being requested of them, they were asked to read and sign a written informed consent form agreeing to all aspects of the study. In essence, written informed consent was always obtained prior to entry into the study. All study data are kept confidential, and no information is revealed to any other sources. Incentives and compensations are not used for enrollment or continued participation. The research herein was conducted in accordance with the rules and regulations of the Institutional Review Boards at Brown University Health and Rhode Island Hospital in Providence, RI (USA), and at the University of Sydney, NSW, Australia. Permission for the investigators at Brown University Health to use the postmortem human tissue was obtained from the NSW Brain Bank, which includes the NSW BTRC. The research was performed with banked human brain tissue as part of an ongoing collaborative project with the University of Sydney. The use of deidentified human postmortem tissue for research meets Exemption Criteria 4 under 45CFR Part 46. The samples were de-identified prior to transfer from the BTRC to Brown University Health. A Tissue Transfer Agreement, outlining the conditions of tissue usage, was required to be completed prior to making the tissue samples available.

This human tissue research was approved by the BTRC Scientific Advisory Committee, the University of Sydney Human Research Ethics Committee (2018/HE000477), and the Brown University Health Institutional Review Board (CMTT/PROJ:#013024). All donors were free of other substance use disorders, and none of the participants had been enrolled in clinical trials. Relevant aggregate demographic, clinical, and postmortem data are provided in Table 1. The AUD and control groups each included 6 male subjects with similar mean ages (Years ± S.D.) (AUD: 57.33 ± 7.37; Control 57.83 ± 6.68) and age ranges (AUD: 50–70; Control: 50–69). Despite the similar number of years (±S.D.) of alcohol consumption (AUD: 34.0 ± 7.11; Control: 26.2 ± 4.27), the lifetime quantity (Kg) of alcohol consumed was significantly greater in the AUD (2,784 ± 1,465) than in the control (42.3 ± 35.5) group (p < 0.0001). Smoking was the only noted cofactor. Four of 6 AUD and 3 of 6 controls had smoking histories. The postmortem AUD mean brain weight (1,335 ± 133 g) was significantly lower than control (1,505 ± 115) (p = 0.02). The mean postmortem intervals to autopsy and brain pHs were similar for the two groups. Cores (6-mm diameter) of fresh frozen anterior cerebellar vermis were stored at −80 °C for later micro-dissection and processing for molecular and biochemical studies [6, 67].

Using a TissueLyser II instrument (Qiagen, Germantown, MD, USA) and 5-mm diameter stainless steel beads, two sets of fresh frozen brain tissue samples (100 mg each) from the anterior vermis were homogenized in 5 volumes of weak lysis buffer (50 mM Tris (pH 7.5), 150 mM NaCl, 5 mM EDTA (pH 8.0), 50 mM NaF, and 0.1% Triton X-100) or Radioimmunoprecipitation Assay (RIPA) buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate). The lysis buffers were supplemented with protease (1 mM PMSF, 0.1 mM TPCK, 2 μg/mL aprotinin, 2 μg/mL pepstatin A, 1 μg/mL leupeptin, 1 mM NaF, 1 mM Na₄P₂O₇) and phosphatase (10 mM Na₃VO₄) inhibitors. The supernatants generated by centrifuging the samples at 14,000 rpm for 10 min at 4 °C were aliquoted and stored at −80 °C for later immunoassays. Protein concentrations were determined using the bicinchoninic acid (BCA) assay.

Immunoreactivity to oligodendrocyte-glial proteins was assessed using duplex ELISAs [54]. The glial protein analyses were clustered into groups corresponding to immature oligodendrocyte/myelin proteins: 2′,3′-cyclic nucleotide 3′ phosphodiesterase (CNPase), proteolipid protein 1 (PLP), Platelet-derived growth factor receptor, alpha peptide (PDGFRA), and Group-specific component Vitamin D Binding (GALC); mature oligodendrocyte/myelin proteins: myelin-associated glycoprotein 1 (MAG), myelin oligodendrocyte glycoprotein (MOG), and myelin basic protein (MBP); and astrocyte proteins: nestin, vimentin, and glial fibrillary acidic protein (GFAP). Supplementary Table S1 lists the antibody sources, concentrations used, research resource identifier (RRID) numbers, and commercial validation methods. The duplex ELISA results were normalized to large acidic ribosomal protein (RPLPO) as the loading control because RPLPO immunoreactivity increases linearly with protein content between 10 ng and 80 ng/well [68]. To perform the duplex ELISAs, triplicate 50 ng protein samples, each in 50 µL bicarbonate binding buffer, were robotically distributed (EpMotion 330) into 96-well MaxiSorp plates. After overnight adsorption at 4 °C, non-specific binding sites were masked with Superblock TBS, and then the samples were incubated with primary antibodies (0.2–5.0 μg/mL) overnight at 4 °C. Immunoreactivity was detected with horseradish peroxidase (HRP)-conjugated secondary antibodies and the Amplex UltraRed soluble fluorophore. Fluorescence intensity was measured (Ex 530 nm/Em 590 nm) in a Spectra-Max M5 Multimode Plate Reader (Molecular Devices, Sunnyvale, CA, USA). After rinsing in Tris-buffered saline (TBS), the samples were incubated with biotin-conjugated anti-RPLPO, followed by streptavidin-conjugated alkaline phosphatase, and RPLPO immunoreactivity was detected with 4-Methylumbelliferyl phosphate (4-MUP) (Ex 360 nm/Em 450 nm). Fluorescence was measured in a SpectraMax M5. The calculated ratios of target protein to RPLPO were used for statistical comparisons.

Neuroinflammatory/metabolic effects in the brain were assessed using commercial magnetic bead-based cytokine/chemokine 11-Plex plus 10-Plex panels (BioRad; Supplementary Table S2). Akt pathway signaling was assessed using 7-Plex total and phosphoprotein commercial magnetic bead-based ELISAs (Supplementary Table S3). The assays were performed according to the manufacturers’ protocols. In brief, after incubating the samples with antibody-coated beads to capture the antigens, immunoreactivity detected with biotinylated secondary antibodies and phycoerythrin-conjugated streptavidin was measured in a Luminex MAGPIX instrument (Diasorin, Austin TX USA) with xPONENT software. MAGPIX calibration and verification standards were used throughout, and standard curves were generated for each analyte.

Total RNA was extracted from fresh frozen tissue using QIAzol Lysis Reagent (Qiagen, Germantown, MD USA). The samples were analyzed for mRNA expression using two custom Quantigene 2.0 Multiplex panels. The 10-Plex Human Glial panel (QuantiGene 2.0 Plex Set Cat# 312185) was used to measure 2′,3′-cyclic nucleotide 3′ phosphodiesterase (CNP), Chondroitin Sulfate Proteoglycan 4 (CSPG4), GFAP, Kallikrein-related peptidase 6 (KLK6), KLK8, MBP, MOG, PLP1, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), Hypoxanthine phosphoribosyltransferase 1 (HPRT1), and Ribosomal Protein L13a (RPL13A) mRNA expression (Supplementary Table S4). The 20-plex assay for human Insulin/Notch pathways (Affymetrix Inc., Santa Clara, CA USA, Cat# 312177) measured mRNA transcripts corresponding to insulin (INS), insulin-like growth factor 1 (IGF1), IGF2, insulin receptor (INSR), IGF1R, IGF2R, insulin receptor substrate 1 (IRS1), IRS2, IRS4, ASPH, NOTCH1, Jagged 1 (JAG1), Hairy and enhancer of split-1 (HES1), HES-related family bHLH transcription factor (HEY1), and hypoxia-inducible factor 1-alpha (HIF1α) (Supplementary Table S5). Quantigene 2.0 Multiplex (QGP) Assays permit direct mRNA quantification using xMAP Luminex beads and reporter signal amplification. RPL13a served as the internal control gene for normalizing the results. Cooperative hybridization and quantification were performed following the manufacturer’s protocol.

In brief, capture beads suspended in lysis buffer, blocking reagent, and an RNA probe set were distributed in 96-well plates. Total RNA (1 µg) was incubated with the reaction mixtures overnight with the xMAP fluorescent beads. Sterile nuclease-free water was used as the negative control. The samples were first incubated with a set of oligonucleotide probes (pre-amplifier, amplifier, and biotin-label), followed by streptavidin-conjugated R-Phycoerythrin (SAPE). The resulting fluorescent signals were detected with a Luminex MAGPIX instrument (Diasorin, Austin TX USA). MAGPIX calibration and verification standards were used throughout, ensuring the levels of SAPE fluorescence were proportional to RNA transcript abundance captured by the beads. After subtracting the probe-related background from the target median fluorescence intensity (MFI), the results were normalized to RPL13a.

Biochemical and molecular assays were performed using two separate tissue cores per case, with all assays performed in triplicate and under code. Each set of triplicate data points per sample was averaged to generate a single value per case. The results from 6 AUD and 6 control samples were used to generate graphs and calculate the between-group differences in mean protein or mRNA expression with GraphPad Prism 10.4 software (GraphPad Software Inc., Boston, MA). Two-way/Mixed Model Analysis of Variance (ANOVA) tests with post hoc Šídák’s multiple comparison tests were used for data analysis. Data for the individual assays passed the D’Agostino & Pearson test for normality (alpha = 0.05). Violin plots display the distribution of results, including the median, first and third quartiles, and upper and lower data points. Additionally, between-group comparisons were made using heatmaps to compare the effects of AUD within the glial, inflammatory, trophic factor, Akt pathway signaling protein biomarker panels, and glial, Insulin pathway, and Notch pathway mRNA biomarker panels. Software-generated statistically significant (p ≤ 0.05) between-group differences are displayed in the Tables and Graphs.

The ELISA MaxiSorp 96-well plates, Bicinchoninic acid (BCA) reagents, horseradish peroxidase (HRP)-conjugated secondary antibodies, and Superblock (TBS) were purchased from Thermo-Fisher Scientific (Bedford, MA USA). The soluble fluorophores, Amplex UltraRed and 4-Methylumbelliferyl phosphate (4-MUP) were from Life Technologies (Carlsbad, CA, USA). Vector Laboratories Inc. (Newark, CA, USA) was the source of the Proton Biotin Protein Labeling Kit and Alkaline Phosphatase-conjugated Streptavidin. The 11-Plex+10-Plex MILLIPLEX MAP Human Cytokine Magnetic Bead Panels were purchased from Millipore (Burlington, MA, USA). The human Total 7-plex and Phospho-7-plex Akt Pathway kits and reagents were from ThermoFisher/Invitrogen (Camarillo, CA, USA). Vector Laboratories Inc. (Newark, CA, USA) was the source of the Protein Biotin Protein Labeling Kit and Alkaline Phosphatase-conjugated Streptavidin. All other fine reagents were purchased from CalBiochem/Millipore Sigma (Burlington, MA, USA), Pierce Chemical (Dallas, TX, USA), or Sigma-Aldrich Co. (St. Louis, MO, USA).

Duplex ELISAs measured immature (CNPase, PLP, PDGFRA, GALC) and mature (MAG, MOG, MBP) myelin oligodendrocyte proteins, and astrocyte proteins (Nestin, vimentin, GFAP). Immunoreactivity was normalized to RPLPO. Comparative effects of ethanol exposure on glial protein expression within each cluster were analyzed with two-way ANOVA tests. The analyses revealed significant AUD and glial biomarker effects on immature and mature myelin/oligodendrocyte proteins (Table 2). Additionally, significant glial biomarker effects were observed with respect to astrocytic proteins. Post-hoc multiple comparisons tests revealed significantly higher levels of CNPase (Figure 1A) and MOG (Figure 1B) in AUD relative to control cerebellar tissue. There were no significant effects of AUD on PLP, PDGFRA, GALC, MAG, MBP, Nestin, Vimentin, or GFAP immunoreactivities (Figures 1A–C).

A commercial custom multiplex RNA hybridization assay measured immature (CSPG4, KLK6, KLK8, and CNPase) and mature (MBP, PLP, and MOG) oligodendrocyte/myelin mRNA transcripts, GFAP, and GAPDH, with results normalized to the HPRT housekeeping gene. Two-way ANOVA demonstrated significant AUD, glial biomarker, and AUD x glial biomarker interactive effects on glial mRNA expression (Table 3). The post hoc multiple comparison tests detected significant AUD-related increases in KLK6 (Figure 2B), CNPase (Figure 2D), MBP (Figure 2E), PLP (Figure 2F), and MOG (Figure 2G). In contrast, there were no significant effects of AUD on CSPG4 (Figure 2A), KLK8 (Figure 2C), GFAP (Figure 2H), or GAPDH (Figure 2I), an insulin-responsive gene [69]. The corresponding heatmap concisely depicts the broadly upregulated expression of immature and mature oligodendrocyte/myelin genes (Figure 3A).

Multiplex ELISAs were used to measure proinflammatory cytokines, proinflammatory chemokines, and anti-inflammatory cytokines in cerebellar tissue homogenates. The objective was to assess the role of persistent inflammation as a mediator of ARBD. Two-way ANOVA detected significant inflammatory biomarker and AUD x inflammatory biomarker interactive effects on proinflammatory cytokines. Significant inflammatory biomarker and AUD x inflammatory biomarker interactive effects were observed for pro-inflammatory chemokines, and a significant inflammatory biomarker effect was observed with respect to anti-inflammatory cytokines (Table 2). The aggregate results, displayed as a heatmap (Figure 4), reveal that most inflammatory biomarkers were expressed at low levels in the cerebellar tissue. Among the 21 factors, only one significant between-group difference was detected, namely reduced levels of IL-16 (a pro-inflammatory cytokine) in AUD versus control cerebellar tissue. It is noteworthy that IL-16 was the most abundantly expressed of the inflammatory factors. Next in abundance was IL-18, which was similarly expressed in AUD and control samples. Re-analysis of log-transformed data to adjust for large differences in mRNA expression also failed to detect significant inter-group differences beyond IL-16. In essence, the other inflammatory mediators were either similarly expressed in AUD and control samples, or they had relatively high within-group variances, resulting in considerable overlap between groups.

The trophic factors measured by multiplex ELISA included NGF, HGF, VEGF, SCF, FGF, and SDF (Figure 5). Two-way ANOVA tests demonstrated significant effects of AUD, trophic biomarker, and AUD x trophic biomarker interactive effects on trophic factor expression (Table 2). Post hoc multiple comparison tests demonstrated a significant inhibitory effect of AUD on HGF (Figure 5B), but not on NGF (Figure 5A), VEGF (Figure 5C), SCF (Figure 5D), FGF (Figure 5E), or SDF (Figure 5F).

For these studies, we compared the expression levels of insulin/IGF growth factors, their receptors, and IRS molecules using multiplex RNA hybridization assays of INSULIN, IGF1, IGF2, INSULINR, IGF1R, IGF2R, IRS1, IRS2, and IRS4. The rationale was that oligodendrocyte/myelin cellular functions are regulated by insulin/IGF pathways. The relative abundance of the mRNA transcripts was determined by calculating the ratios of target genes to RPL13a. The two-way ANOVA test revealed significant effects of insulin/IGF pathway mRNA transcript, but not AUD or AUD × insulin/IGF pathway mRNA transcript interactions (Table 3). Post hoc tests revealed that none of the AUD-associated changes in expression of INSULIN/IGF/IRS gene pathway molecules were statistically significant (Figure 6). Nonetheless, the corresponding heatmap showed that, except for IGF1, which was relatively reduced, AUD was associated with consistently higher levels of all INSULIN/IGF/IRS pathway mRNA transcripts (Figure 3B).

To examine the effects of AUD on Insulin/IGF/IRS1 signaling through Akt pathways that lead to mTOR, we measured total and phosphorylated proteins using 7-plex bead-based ELISA platforms. In addition, to assess the effects on relative levels of phosphorylation, we calculated the ratios of phosphorylated to total protein immunoreactivity. Results were analyzed by two-way ANOVA with post hoc multiple comparison tests. The Two-way ANOVA tests detected significant AUD effects on the total protein and relative levels of phosphorylation, and significant Akt pathway biomarker effects on the total, phosphorylated, and relative phosphorylation levels of the signaling molecules. Additionally, a significant AUD × Akt pathway biomarker interaction was detected for total signaling protein expression (Table 2). The corresponding graphs display the within-group and between-group differences in total protein (Figure 7A), phosphoprotein (Figure 7B), and relative levels of protein phosphorylation (Figure 7C) with significant post hoc multiple comparisons test results. The main findings were significantly reduced Akt (Figure 7A), GSK-3β (Figure 7A), and pS-GSK-3β (Figure 7B), and increased pY/total-IGF1-R and pT/total-PRAS40 (Figure 7C) in AUD relative to control samples.

Notch pathway genes were evaluated because preclinical studies linked chronic heavy ethanol consumption and inhibition of ASPH and Notch pathway mRNA expression to white matter atrophy and cerebellar pathology [70]. Multiplex mRNA hybridization assays measured ASPH, NOTCH1, JAGGED1, HES1, HEY1, HIF1α, ABCG2, CASR and POLR2a, with results normalized to RPL13. Two-way ANOVA tests detected significant variance related to the Notch-related mRNA biomarkers, but not to AUD or the AUD × Notch-related mRNA biomarker interactive effects (Table 3). Violin plots with significant post hoc multiple comparison test results demonstrated AUD-associated reductions in ASPH (Figure 8A) and HES1 (Figure 8D), but none of the other mRNAs measured in the panel (Figures 8B,C,E–I). The corresponding heatmap, which includes the HPRT1 control gene, illustrates the within- and between-group differences in gene expression, along with the significant effects of AUD on ASPH and HES1 (Figure 3C).

Despite its novelty in the analysis of alcohol-related cerebellar white matter degeneration in humans, the study has limitations that should be addressed in future research. The subject numbers were modest (just 6 brains per group) due to limited samples available from cases with uncomplicated ARBD without co-existing problems caused by exposure to other drugs of abuse. The inclusion of males and not females is a weakness that could be explained by the male predominance of heavy drinkers, including in Australia. ¹ The study included only postmortem brains corresponding to ARBD endpoints. The research findings could be strengthened by examining brain tissue or other relevant specimens at earlier timepoints in ARBD. Finally, the inclusion of a subset of cases in which the participants ceased to drink heavily for a significant period would help assess abstinence-related reversibility of ARBD cerebellar WM pathology in humans.