To obtain RNA for TRAP, we used a mouse strain in a C57BL/6J background that expresses a fluorescently tagged ribosomal protein (EGFP-Rpl10a) in the presence of tetracycline-controlled transactivator (tTA). TRAP mice for our study were derived by crossing C57BL/6J-Tg(tetO-EGFP/Rpl10a)5aReij/J with B6.Cg-Tg(CaMKIIα-tTA)1Mmay/DboJ(CaMKIIα-tTA) mice. Both mouse lines were purchased from Jackson Laboratories. The resulting offspring have ribosomes tagged with EGFP expressed only in CaMKIIα expressing cells (Heiman et al., 2008). Genotyping of mice was performed according to standard procedures.

All animal studies were conducted in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. Mice were group housed on a standard 12 light/12 dark cycle and allowed ad libitum access to food and water. Male transgenic mice 8-12 weeks old were randomly assigned to three treatment groups: control, ethanol (EtOH), and ethanol + ABT-888 (EtOH + ABT-888). The animals were administered i.p. twice a day with normal saline (CTL) or ethanol (EtOH, to a final daily dose of 2 g/kg) for four consecutive days (2 h between the first and second injection of the day). The dosing strategy was the same as the involuntary ethanol administration protocol reported earlier (Vallerini et al., 2020). We simulated the drinking-in-the-dark binge drinking paradigm (this is a 4d drinking paradigm) in which, PARP inhibitor reduced drinking. The time point of ABT-888 administration was chosen since in our earlier study (Vallerini et al., 2020), we observed a change in drinking behavior after a single dose of ABT-888 was administered after a period of 4 days of EtOH administration. ABT-888 (25 mg/kg) was co-administered with ethanol on the fourth day in a sub-group of mice that received ethanol in the previous three days (EtOH + ABT). Two hours after the second injection on day 4, mice were sacrificed via rapid CO₂ asphyxiation and decapitation (Supplementary Figure 4). Blood was rapidly collected from the trunk, centrifuged at 2500 RPM for 15 minutes to separate the serum, and kept at –80°C until analyzed; the brain was dissected on a brain block to isolate the PFC from the two hemispheres. This was based on coordinates from our earlier publication (Vallerini et al., 2020) (anterior/posterior: +1.9 mm; medial/lateral: ±0.5 mm; dorsal/ventral: –2.5 mm from Bregma). PFC was immediately homogenized and subject to immunoprecipitation-driven RNA extraction according to the TRAP protocol as well as total-RNA extraction for the INPUTs.

An enzymatic assay calibrated against a standard curve of ethanol known concentrations was utilized to determine BECs (Jung and Férard, 1978). Briefly, ethanol standard samples were prepared by mixing absolute ethanol and ultrapure water to yield a 1000 mg/dL ethanol stock solution. This solution was serially diluted to yield a set of eight standard samples in the range of 7.8–1,000 mg/dl ethanol. Perchloric acid (40 μl of 3.75%) was added to 10 μl of samples (serum) and standards and centrifuged for 6 minutes at 2000 RPM. The supernatant was retained. β-Nicotinamide adenine dinucleotide lithium salt from Saccharomyces cerevisiae (Sigma #N7132) was added at a final concentration of 2.5 mM, and Alcohol Dehydrogenase from Saccharomyces cerevisiae (Sigma #A7011) was added at a final concentration of 5 μM to samples and standards and incubated at 35°C for 40 minutes. BEC was calculated in milligrams per deciliter and noted as follows: control group (13.2 ± 18.9 mg/dl; n = 7), EtOH group (285 ± 84.2 mg/dl; n = 5) and EtOH + ABT-888 group (221 ± 113 mg/dl; n = 5).

The technology was developed in the Heintz lab at Rockefeller University (Heiman et al., 2008). Briefly, PFC samples were homogenized in the appropriate buffer (containing NP-40, KCl, Tris, MgCl2, cycloheximide, protease inhibitor cocktail, RNAse inhibitors and DTT) and centrifuged. Supernatant (40 μl) was collected for total-RNA (INPUT), and the remaining supernatant (approximately 400 μl) was transferred to a separate tube and treated with anti-EGFP antibodies and Agarose Plus A/G beads (Santa Cruz_sc-2003*) overnight. After centrifugation at 1200 RPM for 1 min, the supernatant was removed and the beads, containing ribosome-bound RNA (TRAP samples), were washed 4 times with a high-salt buffer solution (containing NP-40, KCl, Tris, MgCl2, cycloheximide, and DTT) and eluted with QIAZOL (Qiagen # 79306). TRAP-RNA and total-RNA samples were extracted using the miRNeasy Mini Kit (Qiagen # 217004), according to manufacturer’s instructions.

RNA samples were quantified using the Quantus fluorimeter (Promega) with dual RNA/DNA quantification. Levels of remaining DNA did not exceed 10% of the total amount of Nucleic Acid. RNA Integrity Numbers (RIN) were assessed using Agilent 4200 TapeStation (Agilent) (Mean:7.33; SD: 0.46; Min: 6.4; Max: 7.8). The samples were subjected to rRNA depletion and sequencing libraries prepared for Illumina sequencing (CORALL Total RNA-seq Library Prep Kit with Unique Dual Indices [Lexogen, PN M11696] with RiboCop HMR rRNA Depletion Kit V1.3 [Lexogen, PN K03796]). Final amplified libraries were purified, quantified and average fragment sizes confirmed to be 254 –323 and then subjected to test sequencing on a MiniSeq instrument in order to check sequencing efficiencies. Sequencing was carried out on NovaSeq 6000 (Illumina), SP flowcell, 2 × 50 bp PE reads.

The resulting FASTQ files were checked for adapters and following QC, raw reads were aligned to the reference genome, mm10, in a splice-aware manner using the STAR aligner (Dobin et al., 2013). The genome reference utilized ENSEMBL gene and transcript annotations, which included non-coding RNAs in addition to mRNAs. The expression level of the genes was quantified using FeatureCounts (Liao et al., 2014). Differential expression statistics (Log2 fold-change and p-value) were computed using edgeR (Robinson et al., 2010; McCarthy et al., 2012) on raw expression counts obtained from quantification. Prior to analysis, the counts were filtered to exclude any gene that either had less than a total of 500 counts across all samples or was detected in fewer than 3 samples. The data were normalized using the trimmed mean of m-values (TMM) normalization. The normalized data were modeled across treatment status, i.e., Control, EtOH or EtOH + ABT-888, as well as by sample type, i.e., Input or TRAP. The terms of the model were tested using the likelihood ratio test, i.e., glmLRT function, within edgeR. Pairwise tests of the expression data were computed using the exactTest function within edgeR. In all cases, adjusted p-values were computed using false discovery rate (FDR) correction (Benjamini and Hochberg, 1995). Significant genes were determined based on an FDR threshold of 0.01 or 0.05 in the comparison.

Six experimental samples were obtained from the three experimental conditions: Control_TRAP (n = 3); EtOH_TRAP (n = 4); EtOH + ABT-888_TRAP (n = 4); Control_total-RNA (n = 3); EtOH_total-RNA (n = 4); EtOH + ABT-888_total-RNA (n = 4). We further performed the following comparisons indexed in Table 1: (A1) Control_TRAP vs. Control_total-RNA; (A2) EtOH_TRAP vs. EtOH_total-RNA; (A3) EtOH + ABT-888_TRAP vs. EtOH + ABT-888_total-RNA; (B1) EtOH_TRAP vs. Control_TRAP; (B2) EtOH + ABT-888_TRAP vs. Control_TRAP; (C1) EtOH_total-RNA vs. Control_total-RNA; (C2) EtOH + ABT-888_total-RNA vs. Control_total-RNA. The first three comparisons in Table 1 (A1, A2, and A3) utilized total-RNA measurements for scaling Ribosome engagement (RE), similar to the published RE method (Ouwenga et al., 2017; Rodrigues et al., 2020). We also performed comparisons entirely within the cortical samples used for both TRAP RNA (B1 and B2) and total-RNA (C1 and C2). Gene lists were generated from these comparisons using an FDR < 0.01 (A1, A2, A3) or 0.05 (B1, B2; C1, C2) and subjected to an overrepresentation analysis for biological pathways using Ingenuity Pathway Analysis (IPA^®; QIAGEN).

IPA Pathway lists were prioritized by computing −log p-values and Z scores to identify known biological pathways that appeared most significantly affected by the genes in the data set. The significance values (−log p-value of overlap) for the canonical pathways indicate the probability of association of molecules from our dataset with the canonical pathway by random chance alone. IPA assigns the negative −log p-value to pathways based on a Fisher’s exact test of the probability of the number of molecules from the user-created dataset included in the given pathways vs. being included based on chance alone. Z-scores measure the inclusion/exclusion of genes within networks according to experimental conditions and indicate the activation/inhibition states of the molecule. Gene lists were uploaded to IPA with Ensemble ID, Log2 FC and FDR values and the resulting Canonical Pathway information along with –log p-values obtained.

The RNA used for sequencing was utilized for cDNA preparation and target gene validations by qRT-PCR using PikoReal Real-Time PCR (Thermo Fisher, Waltham, MA, USA) and Maxima^® SYBR Green/ROX qPCR Master Mix (Thermo Fisher, Waltham, MA, USA). To confirm amplification specificity, the PCR products were subject to a melting curve analysis, in which only one peak was observed. Crossing point values were measured with the PikoReal analysis software. Primers (Supplementary Table 1) were designed to span at least one intron exon boundary. The following cycling conditions were used: 10 min at 95°C then 40 cycles at 95°C for 30 s, 60°C for 1 min, and 72°C for 30 s. For normalization of mRNA expression, the geometric mean of Gapdh and β-Actin was used. Relative fold changes were determined using the ΔΔCt method.

Statistical significance was determined using one-way analysis of variance (ANOVA) followed by post-hoc Bonferroni or Tukey tests or with Student’s t-test as appropriate. All statistical analyses were performed using GraphPad Prism for Windows (GraphPad Software, San Diego, CA, USA)¹ or SPSS (IBM Corp. released 2016, IBM SPSS Statistics for Windows, version 24.0, Armonk, NY, USA).

In preliminary TRAP experiments we measured several markers that would confirm the successful isolation of RNA from CaMKIIα-expressing cells. Specifically, we measured Egfp, CamKIIα, S100b, and Ef101557 expression in TRAP samples. We expected isolated ribosomes to have high green-fluorescent protein mRNA expression (Egfp) and CamKIIα expression and low expression of the astrocyte specific-S100b and microglial-specific Ef101557 transcripts. In support of the ability of the TRAP technique to isolate CaMKIIα-expressing cells we found Egfp (2.654 vs. 0.790, t₆ = 2.547, p = 0.04, n = 4 per group) and CamKIIα (1.416 vs. 0.265, t₄ = 6.022, p = 0.004, n = 3 per group) to be highly elevated in the TRAP samples compared to total-RNA, while astrocyte marker S100b (2.624 vs. 1.715, t₆ = 3.357, p = 0.02, n = 4 per group) and microglial marker Ef101557 (29.985 vs. 25.121, t₆ = 6.063, p = 0.0009, n = 4 per group) were reduced.

With the initial hypothesis that EtOH will induce a shift in RE transcripts which ABT-888 may reverse, we proceeded to identify shifts in IPA pathways between the experimental conditions. From Figure 1, we identified the top 20 canonical pathways based on IPA generated −log p-value scores in each computed ratio, and further identified a subset of pathways that were common (Figure 2) across the indexed conditions (A1 vs. A2 vs. A3; and B1 vs. B2). In Figure 2A where pathways are scaled to total-RNA, we note “Molecular Mechanisms of Cancer” and “Estrogen Receptor Signaling” pathways show a strong induction with EtOH following a more suppressive/normalization state with ABT-888. Likewise, IL-8 Signaling and CXCR4 Signaling show a strong induction which appears to reverse with ABT-888. We next examined modulation of these pathways in only TRAP samples across the experimental conditions by comparing lines B1 and B2 from Table 1 and are presented in Figure 1B, of which “Epithelial Adherens Junction Signaling,” “Insulin Receptor Signaling,” and “Regulation of elF4 and p70S6K Signaling,” are the top 3 (Figure 2B). Comparing the two conditions (B1 and B2), we observed that most of the top 20 enriched pathways had an identifiable downward shift in RE after ABT-888 treatment (Figure 2B).

We analyzed the data set from TRAP (translating) and total-RNA (transcribed) to detect RNA transcripts that are significantly enriched or reduced in both ribosomes and the total-RNA pool. We identified 785 transcripts significantly altered (FDR < 0.05) in both the RE and total-RNA pools, suggesting coordinated changes in both transcription and translation regardless of cell type. These changes were in the same direction in all but three instances. Expression of Fbln5 was increased with EtOH in the total-RNA pool and decreased in the RE pool, while expression of Pltp6 and Ptpn6 was decreased with EtOH in the total-RNA pool and increased in the RE pool. We next used IPA to identify enriched canonical pathways in this data set of 785 transcripts. The top 20 pathways ranked according to their negative log p-value are shown in Figure 3 and included “3-Phosphoinositide Degradation,” “Coronavirus Replication Pathway,” and “Breast Cancer Regulation by Stathmin 1.”

We wanted to identify the number of transcripts whose EtOH-induced expression changes were nullified or “reversed” by the addition of ABT-888. We identified 1115 (1365-250) transcripts or 82% (1115/1365) of all EtOH affected genes to be “reversed” by ABT-888 co-administration (shown as a volcano plot; Supplementary Figures 2, 3). A schematic representation is shown in Figure 4. These genes were analyzed by IPA to identify enriched pathways and the 15 highest-ranking canonical pathways are indicated in Table 2 and Figure 5. We observed “Insulin Receptor Signaling” emerging as the top ranked canonical pathway (Table 2), suggesting that ethanol’s effects on this pathway are regulated by PARP signaling.

We elected to validate six genes present in the 15 highest-ranking enriched canonical pathways emerging from the ethanol-regulated and PARP-reverted RE pool (Table 2 and Figure 5). The genes Irs2, Irs4, Scnn1a, Stxbp4, Rasd1 (Insulin Receptor Signaling), and Mapk11 (Chemokine Signaling) were chosen based on the differences in logFC between the EtOH_TRAP/Control_TRAP and the EtOH + ABT-888_TRAP/Control_TRAP comparisons. Genes showing a ΔlogFC value higher than 0.50 were selected for qRT-PCR validations (Table 3). Validation results confirmed our hypothesis of ABT-888 induced reversal of EtOH dysregulation of Irs2, Irs4, Rasd1, and Scnn1a (Figure 6). In the case of Rasd1, although ABT-888 was able to reverse the effects of EtOH, it was not reduced to control levels. All these validated targets, except Mapk11, lie within the Insulin Receptor Signaling pathway as noted in Table 2.