Alcohol Consumption Modulates Host Defense in Rhesus Macaques by Altering Gene Expression in Circulating Leukocytes Free

Alcohol use disorder (AUD) results in a significant increase in the incidence and severity of infections, such as bacterial pneumonia, tuberculosis, hepatitis C virus, and HIV (1–3). Similarly, chronic ethanol consumption in rodents results in increased pathogen burden and impaired ability to clear Listeria monocytogenes (4), Mycobacterium tuberculosis (5), and influenza virus (6). Likewise, rhesus macaques given ethanol via intragastric cannula show increased SIV replication compared with controls (7). Increased vulnerability to infection in individuals with AUD is due to changes in barrier function, as well as innate and adaptive immunity (8). Dysregulation of tight junction proteins in the lungs and gut increases permeability, leading to bacterial translocation into the alveolar space and circulation, respectively (2, 9). In addition, AUD results in the inhibition of phagocytic functions, reduction of chemotaxis and aberrant cytokine production, and diminished lymphocyte numbers and Ag-specific responses (10).

In contrast, data from several studies support a beneficial role for moderate alcohol consumption in immunity. Moderate alcohol consumption is associated with a decreased incidence of the common cold in humans (11–13), as well as improved bacterial clearance and increased delayed cutaneous hypersensitivity response following infection with Mycobacteria bovis in rats (14). Recently, we showed, using a macaque model of ethanol self-administration (15), that moderate consumption resulted in a more robust T cell and Ab vaccine response to modified vaccinia Ankara (MVA), whereas heavy drinkers generated blunted T cell and Ab responses compared with controls (16). Moreover, we showed that the dose-dependent effects of ethanol on the immune response to the MVA vaccine were independent of changes in the frequency of major immune cell subsets. Specifically, numbers of circulating lymphocytes, monocytes, and neutrophils, as well as the frequency of CD4 T cells, CD8 T cells, and CD20 B cells (and their naive and memory subsets), did not differ between control and ethanol-consuming animals (16). Instead, we detected changes in the expression of several microRNAs (miRNAs) associated with the development and function of the immune system, suggesting that ethanol dose–dependent modulation of immunity is mediated by changes in gene expression. Therefore, in this study, we compared the transcriptomes of PBMCs isolated from controls and moderate and heavy drinkers on day 7 post-MVA vaccination.

Our results revealed that chronic heavy ethanol consumption was associated with the significant downregulation of genes involved in immune response to infection and wound healing, as well as upregulation of genes associated with the development of obstructive lung disease and cancer. In contrast, chronic moderate alcohol consumption was associated with the reduced expression of genes involved in neoplasia and the upregulation of genes involved in host defense. To uncover mechanisms underlying the alterations in PBMC transcriptomes, we also examined changes in miRNA expression. Our analysis showed that chronic heavy ethanol consumption altered the expression of several miRNAs whose targets were differentially expressed in our data set and are involved in cancer progression and immune function. Overall, data presented in this article provide novel insights into the mechanisms by which excessive alcohol consumption interferes with immune responses and exacerbates comorbidities, such as poor wound healing, lung disease, and cancer, and moderate consumption improves immunity.

This study was performed in strict accordance with the recommendations detailed in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, the Office of Animal Welfare, and the U.S. Department of Agriculture. All animal work was approved by the Oregon National Primate Research Center Institutional Animal Care and Use Committee.

The animal model and vaccination strategy were described previously (16). Briefly, we used schedule-induced polydipsia to establish reliable self-administration of 4% (w/v) ethanol in eight male rhesus macaques (15). Four animals served as controls, for a total of 12 animals. Following a 4-mo induction period, animals were allowed a choice of 4% ethanol or water for 22 h/d every day for 12 mo. In this nonhuman primate model of voluntary self-administration, animals segregate naturally into heavy and moderate drinkers within 2–3 mo, and these patterns remain stable for ≥12 mo (17). In this specific cohort, ethanol-drinking animals segregated into two cohorts (n = 4 each), based on average blood ethanol concentration (BEC) values: moderate drinkers with average BEC of 22.3–48.8 mg/dl and heavy drinkers with average BEC of 90–126 mg/dl (16). All 12 animals were vaccinated with MVA prior to the induction of ethanol and again after 7 mo of open access to ethanol. We used PBMCs isolated 7 d after booster vaccination for RNA and miRNA expression analysis. Only three animals from each group had sufficient numbers of PBMCs for RNA sequencing.

Total RNA was isolated from PBMCs using the miRNeasy kit (QIAGEN, Valencia, CA). One microgram of RNA was used to generate libraries using the NEBNext Ultra Directional RNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, MA). Poly(A)-enriched mRNA was fragmented, followed by cDNA synthesis with random hexamers. This product underwent end-repair, adapter ligation, and size selection using AMPure XP beads (Beckman Coulter, Brea, CA) to isolate cDNA templates of 320 nt that were amplified by PCR. Each library was prepared with unique index primers for multiplexing and subjected to single-end 100-bp sequencing on the HiSeq 2500 platform (Illumina, San Diego, CA).

One microgram of total RNA, extracted as described above, underwent adapter ligation and primer hybridization prior to cDNA synthesis and PCR amplification using the NEBNext Small RNA Library Prep Set for Illumina Kit (New England Biolabs). Size selection was performed with AMPure XP beads (Beckman Coulter) to isolate cDNA templates of 140 nt. Each library was prepared with unique index primers for multiplexing and subjected to single-end 50-bp sequencing on the HiSeq 2500 platform (Illumina). We were unsuccessful in generating one library from one of the heavy drinkers.

Data analysis was performed with the RNA-sequencing (RNA-Seq) workflow module of the systemPiperR package available on Bioconductor (18, 19). Quality reports were generated with the seeFastq function. RNA-Seq reads were mapped with the splice junction aware short read alignment suite Bowtie2/Tophat2 (20, 21) against the Macaca mulatta genome sequence from Ensembl (22). For the alignments, we used default parameters of Tophat2 optimized for mammalian genomes. Raw expression values in the form of gene-level read counts were generated with the summarizeOverlaps function (23). We counted only reads overlapping exonic regions of genes, discarding reads mapping to ambiguous regions of exons from overlapping genes. Given the nonstranded nature of RNA-Seq libraries, the read counting was performed in a nonstrand-specific manner. The RNA-Seq data were deposited in the National Center for Biotechnology Information Sequence Read Archive under accession number SRP064253 (http://www.ncbi.nlm.nih.gov/sra). Analysis of differentially expressed genes (DEGs) was performed with the generalized linear model method from the edgeR package (24, 25). DEGs were defined as those with a fold change (FC) ≥ 2 and a false discovery rate ≤ 0.05. Enrichment analysis of functional annotations was performed to identify significant biological pathways, including gene ontology (GO) terms and disease biomarkers using MetaCore software (GeneGo, Philadelphia, PA).

Adaptor contaminations were removed (trimmed) from the reads using the preprocessReads function from the systemPipeR package. The preprocessed reads were aligned with Bowtie2 (20, 21) against the M. mulatta genome sequence, with settings optimized for miRNA alignments, including tolerance of multiple mappings. Reads overlapping with miRNA gene ranges were counted with the summarizeOverlaps function, as described above, but in a strand-specific manner. The miRNA gene coordinates, required for this step, were downloaded from miRBase (release version 19). The small RNA-Seq data were deposited in the National Center for Biotechnology Information Sequence Read Archive under accession number SRP064540 (http://www.ncbi.nlm.nih.gov/sra). Differentially expressed miRNA genes were identified with edgeR, as described above. TargetScan was used to predict genes for each differentially expressed miRNA with a high context ratio of 0.95. These targets were compared with our list of DEGs among the three groups of rhesus macaques. These combinations of differentially expressed mRNA and miRNA were then segregated based on the directions of FCs.

cDNA was synthesized from RNA, isolated as above, using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). mRNA expression was determined by quantitative RT-PCR using TaqMan primer and probe kits specific for M. mulatta cDNA sequences and a StepOnePlus instrument (Life Technologies, Grand Island, NY). mRNA expression levels of LYZ (Rh02902590), PTGS2 (Rh02787804), THBS1 (Rh00962902), Kruppel-like factor (KLF)4 (Rh02847953), TLR4 (Rh01060206), CD14 (Rh03648680), CD163 (Hs00174705), and FN1 (Rh02621780) for each sample were calculated relative to control RPL32 mRNA expression using ΔCt calculations.

There were 514 DEGs between controls on day 7 post-MVA vaccination (C7) and heavy drinkers on day 7 post-MVA vaccination (H7; C7–H7), 479 of which were annotated, with 356 downregulated and 123 upregulated genes with heavy drinking (Supplemental Table I). We identified 368 DEGs between moderate drinkers on day 7 post-MVA vaccination (M7) and heavy drinkers (M7–H7), 347 of which were annotated, with 290 downregulated and 57 upregulated genes with heavy drinking (Supplemental Table II). Finally, of the 60 DEGs between controls and moderate drinkers (C7–M7), 47 were annotated, with 29 downregulated and 18 upregulated with moderate ethanol consumption (Fig. 1A, Supplemental Table III).

Heavy ethanol consumption was associated with the largest changes in gene expression compared with controls and moderate consumption (Fig. 1A, 1B), with 171 downregulated DEGs (Fig. 1C) and 26 upregulated DEGs (Fig. 1D) compared with controls and moderate drinkers. The overwhelming majority of the downregulated (Fig. 1E) and upregulated (Fig. 1F) DEGs showed a 2–4 FC in expression. To confirm the RNA-Seq results, eight genes differentially expressed with chronic heavy ethanol consumption (LYZ, PTGS2, THBS1, KLF4, TLR4, CD14, CD163, and FN1) were selected for confirmation using quantitative RT-PCR. Changes in the expression level of all eight DEGs were confirmed (Supplemental Fig. 1). To better understand the biological relevance of these gene-expression changes, we conducted functional enrichment analysis using the MetaCore pathway-mapping tool.

Of the 171 genes repressed in H7 compared with C7 and M7, 114 mapped to the following GO terms: response to stress, response to wounding, inflammatory response, response to lipid, and positive regulation of response to external stimulus (Fig. 2A). Several DEGs mapping to “response to stress” encode microbial sensors, notably formyl peptide receptor (FPR)2 (FC = 128.2), TLR4 (FC = 5.6), CD14 (FC = 3.8), pyrin domain–containing-3 (FC = 3.3), TLR8 (FC = 2.9), and nucleotide-binding oligomerization domain–containing-2 (NOD2; FC = 2.4). Other DEGs encode immune receptors, such as cadherin EGF LAG seven-pass G-type receptor-1 (CELSR1; FC = 30.6), syndecan-2 (SDC2; FC = 7.8), plasminogen activator receptor (FC = 6.5), macrophage scavenger receptor-1 (FC = 5.7), IL-1R type-1 (FC = 5.7), IL-13Rα-1 (FC = 4.8), CCR1 (FC = 4.5), and neuropilin-1 (NRP1; FC = 3.4). Additional DEGs encode chemokines, cytokines, growth factors, and antimicrobial peptides, including matrix metalloproteinase-1 (FC = 8.6), CXCL8 (FC = 7.5), oncostatin-M (OSM; FC = 4.4), IL-1β (FC = 3.7), vascular endothelial growth factor (VEGF)-A (FC = 5.2), heparin-binding EGF-like growth factor (HBEGF; FC = 3.4), and S100 calcium binding protein-8/9 (FC = 3.6/2.9).

Several genes listed above also mapped to “response to wounding” and “inflammatory process” and play a role in wound healing (Fig. 2B). For instance, CELSR1, SDC2, VEGFA, HBEGF, and NRP1 promote wound closure (26, 27), whereas NOD2, pyrin domain–containing-3, CD14, coagulation factor plasminogen activator inhibitor-2 (FC = 17.6), complement component 5aR receptor-1 (FC = 4.0), complement component 3a receptor-1 (C3AR1; FC = 2.9), and IL-1β (which activates CXCL8, FC = 7.5) promote chemotaxis and leukocyte extravasation into injury sites (28–31).

Additional analysis showed that 89 genes map to these disease categories: obstructive lung diseases, pathologic processes, hypersensitivity, bacterial infection and mycoses, and inflammation (Fig. 2C). Of the 52 genes mapping to “obstructive lung diseases” (Fig. 2D), 24 interact with each other (Fig. 2E) and are important for lung homeostasis, notably matrix metalloproteinase-1 [FC = 8.6, involved in lung alveolar epithelial cell migration (32)], VEGFA [important for alveolar structure (33)], cathelicidin antimicrobial peptide (LL37, FC = 2.8), and aryl hydrocarbon receptor [AHR; FC = 2.7, regulates apoptosis of lung epithelial cells (34)]. DEGs mapping to “pathological processes” include annexin-2 receptor (FC = 17.1), cysteine-rich secretory protein LCCL domain–containing-2 (FC = 9.7), epiregulin (FC = 9.2), triggering receptor expressed on myeloid cells-1 (TREM1; FC = 4.7), and HBEGF, which are involved in protection against cancer, sepsis, endotoxin shock, and necrotizing enterocolitis (35–39).

Of the 185 DEGs repressed in H7 compared with C7, 128 mapped to these GO terms: response to wounding, regulation of response to stimulus, positive regulation of response to stimulus, response to stimulus, and system development (Fig. 3A). The 37 genes mapping to “response to wounding” play an important role in wound healing (Fig. 3B), including diacylglycerol kinase [FC = 5.8, regulates fibroblast migration (40)], catenin α-1 [FC = 3.4, promotes wound repair in bronchial epithelial cells (41)], metallopeptidase inhibitor-2 [FC = 3.4, involved in wound closure (42)], solute carrier (SLC) family-11 [FC = 2.2, regulates macrophage activation in cutaneous wounds (43)], CSF1 [FC = 2.1, involved in neoangiogenesis (44)], and hepatocyte growth factor [HGF; FC = 2.0, accelerates wound re-epithelialization (27)]. One highly downregulated gene mapping to “response to stimulus” is eosinophil peroxidase (FC = 22.6), a potent toxin for bacteria and parasites (45).

Further analysis showed 106 genes mapped to these disease categories: obstructive lung diseases, pathological processes, immune system diseases, bronchial diseases, and immediate hypersensitivity (Fig. 3C). Genes in “obstructive lung diseases” play an important role in lung function (Fig. 3D). For instance, IL-1R–like-1 (FC = 9.2), TLR5 (FC = 6.0), arachidonate 15-lipoxygenase (ALOX15, FC = 5.5), TREM2 (FC = 2.6), and TNFR superfamily-1A (TNFRSF1A, FC = 2.0) promote host defense against bacterial infection in the lungs and regulate lung inflammation, whereas HGF (FC = 2.0) promotes lung regeneration after injury (27). Polymorphisms in myeloperoxidase (FC = 7.4) and A Disintegrin and Metalloproteinase domain-12 (ADAM12, FC = 4.5) modulate the development of lung cancer (46, 47).

Several DEGs that mapped to “immune system diseases” also mapped to “obstructive lung diseases” and “pathological processes,” including C-type lectin domain family-10A (CLEC10A, FC = 4.8) and ATP-binding cassette subfamily-C-2/3 (FC = 2.9/3.2), which play a role in Ag recognition and presentation (48, 49), as well as EGF-like module receptor-1 (FC = 3.7) and IL-12Rβ-2 (FC = 2.8), which are critical for host defense (50). Genes unique to this category (Fig. 3E) regulate inflammation, such as leucine-rich repeat-containing-18 (FC = 104.5) and hemoglobin subunit γ-2 [FC = 8.9 (51, 52)], as well as lymphocyte proliferation and differentiation, including SH2B adaptor-3 [SH2B3, FC = 2.0 (53)] and killer cell lectin-like receptor subfamily-G1 [FC = 2.4 (54)].

Of the 119 genes downregulated in H7 compared with M7, 66 mapped to the following GO terms: immune system processes, response to stress, immune response, defense response, and regulation of immune system processes (Fig. 4A). In total, 47 mapped to immune system–related GO terms, 20 of which interact with each other (Fig. 4B). DEGs mapping to these GO terms are involved in lymphocyte activation and recruitment [CXCL10, FC = 17.3; SLC16A1, FC = 3.2 (55); sterile α-motif-domain Src homology-domain nuclear localization signals-1, FC = 3.7 (56); ICAM1, FC = 5.5; CD83, FC = 3.0 (57)], antimicrobial response [TNFα, FC = 7.9; ficolin-2, FC = 6.5 (58); MD2, FC = 2.8], and regulation of gene expression [v-ets avian erythroblastosis virus oncogene homolog-2 (ETS2), FC = 2.5; B cell lymphoma 2-related protein-A1, FC = 5.4 (59); B cell CLL/lymphoma-3/6, FC = 2.5/2.0 (60, 61); KLF10, FC = 3.0 (62); and NF of κ light polypeptide gene enhancer in B cells inhibitor-α, FC = 2.9 (63)].

Several genes mapping to “immune system process” (Fig. 4C) also map to “response to stress,” notably hypoxia-inducible factor-1α (FC = 3.0), which regulates expression of genes that counter oxidative stress, and ETS2, which is induced by shear stress to preserve the integrity of microvascular walls (64). Additional notable DEGs that only mapped to “response to stress” included Gadd45-γ [FC = 8.4, important in antitumor immune responses (65)] and BMX nonreceptor tyrosine kinase [FC = 11.5, promotes tight junction formation in epithelial cells during chronic hypoxia (66)].

Within the 26 genes upregulated in H7 compared with C7 and M7 (Fig. 5A), several encode transcription factors associated with skin, colorectal, breast, and lymphoma cancers, notably IFN-α-inducible protein 27-like-1 [FC = 68.0 (67)], AXIN2 [FC = 2.2 (68)], lymphoid enhancer-binding factor-1 [LEF1; FC = 2.0 (69)], Meis homeobox-1 [FC = 2.4 (70)], and four-and-a-half LIM domains-1 [FC = 2.1 (71)]. Interestingly, increased expression of retinoid X receptor-γ (FC = 5.2), which is associated with sensation seeking (72), a behavioral trait common among alcoholics, was detected. Another overexpressed gene was serum deprivation protein response (FC = 2.8), which induces deformation of plasma membrane invaginations, impairing endocytosis and, potentially, Ag presentation (73). Finally, expression of regulator of G-protein signaling (RGS)18 (FC = 3.3), growth factor-independent 1B transcription repressor (FC = 2.3), and rho guanine nucleotide exchange factor-4 (FC = 6.1), which play a role in megakaryocyte differentiation (74, 75), were also increased.

Of the 97 genes upregulated in H7 compared with only C7, 26 mapped to the following GO terms: response to wounding, regulation of body fluids, platelet activation, wound healing, and platelet degranulation. Genes that mapped to “response to wounding” (Fig. 5B) include connexin-43 (GJA1, FC = 11.0), a gap junction associated with impaired wound healing (76); IL-17F (FC = 6.7), which can delay wound closure (77); and P-selectin (FC = 2.3), a glycoprotein that is highly expressed in wounds (78). Genes with roles in cardiovascular disease mapped to the disease category “infarction,” including carbonic anhydrase III (FC = 5.2), glycoprotein VI (FC = 2.9), phosphodiesterase-6H (FC = 2.6), and integrin-α-2b [ITGA2B; FC = 2.6 (79–82)]. Furthermore, heavy ethanol consumption upregulated genes associated with cancer, notably transient receptor potential cation channel-M1 (FC = 14.9), tripartite motif containing-31 (FC = 9.8), RGS6 (FC = 7.6), and CLDN5 [FC = 2.2 (83–85)].

A total of 21 of the 31 genes upregulated in H7 compared with only M7 mapped to “neuro-ectodermal tumors” (Fig. 5C). These genes are either expressed at high levels in cancer, including δ-like-1 [DLK1, FC = 7.3 (86)] and insulin receptor substrate-1 [IRS1; FC = 4.8 (87)] or are involved in progression of cancer, such as phosphatidylinositol-3,4,5-trisphosphate-dependent Rac exchange factor-2 [FC = 11.7 (88)], latent TGF-β–binding protein-2/3 [FC = 3.6/2.6 (89, 90)], and stromal Ag-3 [FC = 2.3 (91)].

Of the 29 annotated genes upregulated in M7 compared with C7 (Fig. 6A), 4 play a role in chemotaxis: CXCL3 (FC = 50.3, critical for leukocyte chemotaxis), IL-1α (FC = 23.4, recruitment of neutrophils), CCL3 [FC = 17.1, recruits T cells (92)], and CCL4L1/2 [FC = 8.1/5.5, trafficking of NK cells (93)]. Other genes significantly upregulated in M7 include acute-phase protein pentraxin-3 [PTX3, FC = 15.3 (94)], ceruloplasmin (FC = 3.7), and granzyme A (FC = 2.2), which are expressed primarily by NK cells and play a role in host defense (95, 96), as well as GJA1 (FC = 9.6), important for barrier function (97).

Several of the 18 genes downregulated in M7 compared with C7 (Fig. 6B) are involved in cancer progression, including transmembrane protein-98 (FC = 286.9), serine-protease temperature requirement-A1 (FC = 24.1), DNA nucleotidylexotransferase (FC = 5.7), MMP9 (FC = 3.5), and ten-eleven translocation-1 [FC = 3.0 (98–102)]. Interestingly, ABCA9 (FC = 6.9), retinoic acid-binding receptor-related orphan receptor-C (FC = 3.6), 2′-5′-oligoadenylate synthetase-2 (FC = 2.6), CD84 (FC = 2.2), and myxovirus resistance protein-1 (FC = 2.2), which play a role in innate immunity, were downregulated in M7.

To begin uncovering the mechanisms underlying changes in gene expression regulation with moderate and heavy ethanol consumption, we compared the miRNA expression profiles of the same PBMCs isolated from controls, moderate, and heavy drinkers on day 7 post-MVA vaccination. miRNAs are ∼22-nt-long endogenous RNAs that target mRNAs for translational repression or degradation (103), and several reports indicate that ethanol can modulate miRNA expression (104). As described for mRNA expression, the largest differences in miRNA expression were observed between controls and heavy drinkers; only a few miRNAs were differentially expressed between controls and moderate drinkers. Interestingly, no differentially expressed miRNAs were detected between heavy and moderate drinkers. There were 79 differentially expressed miRNAs between controls and heavy drinkers: 37 were upregulated (Fig. 7A), and 42 were downregulated (Fig. 7B).

Importantly, 53 of these miRNAs have mRNA targets within our data set. Heavy drinking led to the upregulation of 29 miRNAs known to regulate 25 target mRNAs that were downregulated in our data set. A subset of these mRNA–miRNA pairs is shown in Fig. 7C; for a complete list, please refer to Table I. Some of the downregulated mRNAs were targeted by several differentially expressed miRNAs. For instance, miR-16, miR-15b, miR-195, and miR-374b target VEGFA, which was downregulated >5-fold. Similarly, miR-30b and miR-30c target SH2B3, which was downregulated 2-fold, whereas miR-125a and miR-125b both target SCARB1, which was downregulated >3-fold. Of the 24 miRNAs downregulated with heavy drinking, 7 were associated with an increase in their mRNA targets (Fig. 7C, Table I). For example, the downregulated miR-101 targets GJA1, which was upregulated 11-fold with heavy drinking. miR-144 targets AXIN2, which was upregulated 2-fold. miR-183 and miR-202 both target ROBO2, which was upregulated >3-fold. Finally, miR-29a, miR-29b, and miR-29c all target SH3PXD2A, which was upregulated 4-fold in our data set.

Using a macaque model of ethanol self-administration, we recently showed that heavy ethanol consumption suppresses, whereas moderate ethanol consumption enhances, T and B cell responses to MVA (15). The goal of this study was to uncover the molecular mechanisms underlying this dose-dependent effect. We used RNA-Seq to identify changes in gene expression on day 7 following vaccination with MVA, the earliest time point at which we detected differences in Ab and T cell responses (15). We also sequenced small RNA molecules to gain insight into mechanisms underlying the changes in gene regulation. Overall, our study revealed robust changes in gene and miRNA expression among controls and moderate and heavy drinkers, with fewer changes between moderate drinkers and controls compared with either group versus heavy drinking status.

The changes in gene expression reported in this article provide novel insights into the reduced immune response to vaccination and the increased vulnerability to infection seen in humans with AUD. Specifically, we detected large decreases in the expression of microbial sensors that are critical in the detection of bacterial peptides [FPR2/3 (105)], LPS (MD2, CD14, TLR4), bacterial flagellin (TLR5), and certain helminths and filoviruses [CLEC10A (48)]. There was also decreased expression of genes important for Ag presentation [ATP-binding cassette subfamily-C-2/3, CLEC10A (48, 49)], recruitment of immune cells [C3AR1, complement component 5aR receptor-1, CCR2, CXCL3, CXCL8, CXCL10, CCL3, CCL4L1/2, ICAM1 (30, 92, 106–108)], and soluble mediators that play a role in response to infection [TNF-α, TNFRSF1A, IFN-γ, IFNGR1, IL-1R, IL-1R1, IL-1RL1, IL-1β, S100 calcium binding protein-8/9, LL37, eosinophil peroxidase, and granzyme A (45, 109)]. As previously described (110), expression of IFN signaling IFN receptor-2 was downregulated in heavy drinkers, which would contribute to deficits in both innate and adaptive immunity.

We also found decreased expression of lymphocyte activation markers, including CD83 and killer cell lectin-like receptor subfamily-G1 (54). In our previous study, excessive ethanol consumption suppressed, whereas moderate ethanol consumption enhanced, MVA-specific IgG responses (16). These defects in Ab production could be explained, in part, by the decreased expression of transcription factors BCL3 and BCL6, which are important in germinal center formation, isotype class switching, and hypermutation (60, 61); SH2B3, which regulates B cell development (53); and sterile α-motif-domain Src homology-domain nuclear localization signals-1, an adapter protein involved in immune cell signaling (56). Reduced expression of KLF10, which suppresses regulatory T cells by upregulating TGF-β (62), could further explain the reduction in T and B cell responses.

Our study also revealed insights into the mechanisms by which moderate alcohol consumption stimulates immunity. Animals that drank moderate amounts of ethanol showed increased expression of chemokines CCL3 and CCL4L1, which signal through CCR5 to recruit memory T cells (92), and IL-1α and CXCL3, which recruit neutrophils. Although a vigorous innate immune response is critical, it is equally important for the host to minimize damaging inflammation. PTX3, which plays a role in the resolution of inflammation (94), and the NF-κB inhibitor, NF of κ light polypeptide gene enhancer in B-cells inhibitor-α, were significantly upregulated with moderate drinking. These observations are in line with previous studies that showed that moderate alcohol consumption in humans significantly alters genes involved in B cell, T cell, and IL-15 signaling pathways, as well as attenuates NF-κB signaling pathways in leukocytes (111).

Several of the genes that were differentially expressed with ethanol consumption were previously described as being important for mounting immune responses to vaccination. For instance, studies that investigated yellow fever vaccine (YF-17D)-induced signatures in blood of healthy adults reported significant increases in the expression of proinflammatory mediators CXCL10 and IL-1α and complement gene C3AR1, which were found to be predictive of robust vaccine responses (112). Another study identified ETS2 as an additional key regulator of the early innate immune response to YF-17D (113). In our study, we observed a 17-fold downregulation of CXCL10, a 3-fold downregulation of C3AR1, and a 2.5-fold downregulation of ETS2 with excessive ethanol consumption, which could explain the suppression of vaccine responses in this cohort. In contrast, we detected a 23-fold upregulation of IL-1α in moderate drinkers compared with controls, reinforcing the association of this marker with successful immune responses.

Our gene-expression analysis is also in line with clinical observations linking alcohol abuse with impaired wound healing (114), increased susceptibility to wound infections (115), and delay of wound closure (116). These defects have significant clinical ramifications because half of emergency room trauma cases involve alcohol exposure (117). Previous studies showed that ethanol exposure at the time of traumatic injury impairs wound closure via decreased proinflammatory cytokine release, neutrophil recruitment, and phagocytic function (114). Our gene-expression data support and extend these earlier observations. We detected significantly decreased expression of multiple components of the innate immune system that play a critical role in the prevention of wound infections, including pattern recognition receptors (FPR2/3, NALP3, NOD2, MD2, CD14, TLR4, TLR5, TLR8, CLEC10A, SCARB1), proinflammatory cytokines and their receptors (TNFRSF1A, IFN-γ, IFNgγ1, IL-1R, IL-1R1, IL-1RL1, IL-1β, IL-13Rα-1), and chemokines and their receptors (CXCL2, CXCL8, CXCL10, CCR1, CCR2, OSM, CSF3R, and CSF1).

Earlier studies suggested that the disruption of VEGF signaling and reduced expression of hypoxia-inducible factor-1α in endothelial cells with chronic alcohol consumption interferes with wound closure (116). Our gene-expression data also show a significant decrease in the expression of both of those genes with excessive ethanol consumption. Moreover, we detected fewer transcripts of NRP1, which is expressed by endothelial cells and associates with the VEGF receptor to promote angiogenesis and wound repair (118). We also report decreased expression of CELSR1 and SDC2, which were shown to promote effective wound repair (26, 119), as well as secreted proteins, such as fibronectin and MMP-1, which promote platelet aggregation, angiogenesis, and tissue remodeling (120, 121). The expression of additional growth factors, HGF and HB-EGF, which promote angiogenesis and tissue regeneration, was also significantly decreased (27). Furthermore, heavy alcohol abuse increased expression of additional genes known to interfere with wound healing [PSEL, VWF, CX43, IL-17 (77, 78, 122)].

Additionally, our data are in line with clinical observations that heavy alcohol consumption is associated with increased incidence of chronic obstructive pulmonary disease (123, 124), lung injury in response to inflammatory insults (125), acute respiratory distress syndrome (126), and risk for mortality in acute lung injury patients (127). Impaired immunity and increased oxidative stress, both consequences of AUD, are considered risk factors for chronic obstructive pulmonary disease and acute respiratory distress syndrome (128). Our transcriptome analysis provides new insights into the mechanisms that underlie increased susceptibility and severity of lung injury and chronic lung inflammatory diseases. Heavy ethanol consumption was associated with downregulation of several genes important for maintaining lung homeostasis that can be categorized as transcription factors (AHR, P21, RORg, ATF/CREB), receptors (TREM1, FCgR, NOR1), immune signaling molecules (IL-1R, IL-1β, CXCL8, CD14, CD163, G-CSF, TNFR1, TLR5, ALOX15, ADAM12), transporters (SLC11A1/16A1), and growth factors (VEGF, HBEGF, HGF). AHR and RORg play a role in suppressing lung inflammation (34, 129), whereas decreased p21 expression is associated with hypoxia-induced lung disease (130). In addition, transcripts associated with immune genes that promote host defense against pulmonary infection (ALOX15, CD14, G-CSF, TREM1, TLR5, NRAMP1, TNFR1) were reduced with heavy drinking (131–134). Critical growth factors important for repairing lung injury were also downregulated. Decreased levels of VEGF correlate with loss of alveolar structure in emphysema patients (33) and a compromised integrity of the alveolar–capillary barrier (135). Finally, HGF is important for lung development and promotes regeneration after lung injury in animals (27).

Chronic alcohol consumption is associated with an increased risk for cardiovascular disease (136) and stroke (137). Our analysis revealed increased expression of genes implicated in heart disease, including CLDN5, VWF, phosphodiesterase-6H, ITGA2B, and carbonic anhydrase III (79, 81, 82, 138, 139), as well as megakaryocyte differentiation [RGS18, growth factor-independent 1B transcription repressor, and rho guanine nucleotide exchange factor-4 (74, 75)]. ITGA2B is used as a biomarker for myocardial infarction risk, and therapeutic modulation of phosphodiesterases is one strategy for treating cardiovascular disease. Interestingly, claudin-5 levels are reduced in human and mice models of cardiomyopathy; therefore, its increased expression in this study might be a compensatory mechanism.

Finally, heavy alcohol use is a major risk factor for liver, head and neck, and colorectal cancers (140–142). Our gene-expression analysis showed increased expression of several genes that promote cancer progression with chronic heavy alcohol consumption (latent TGF-β–binding protein-2/3, IRS1, SFRP5, LEF1, DLK1, stromal Ag-3, and H2B). Higher levels of IRS1 are found in hepatocellular carcinoma and breast, ovarian, and colorectal cancers (143–146). Increased expression of LEF1 is associated with human endometrial tumors and prostate and colon cancer (147–149). DLK1 is also expressed at higher levels in colon adenocarcinomas, pancreatic islet carcinomas, and small cell lung carcinomas (86). In contrast, moderate alcohol consumption is associated with a reduced risk for developing kidney cancer (150), Hodgkin’s lymphoma (151), and thyroid cancer (152). Our study revealed that moderate drinking repressed genes associated with reduced cancer incidence, including transmembrane protein-98 (101), serine-protease temperature requirement-A1 (100), DNA nucleotidylexotransferase (98), ten-eleven translocation-1 (99), and MMP9 (102).

We also investigated differences in miRNA expression levels among the three experimental groups. miRNAs can modulate gene expression through translational repression or degradation of target mRNAs and play a critical role in regulating immune function (153). We identified several differentially expressed miRNAs with validated mRNA targets present in our RNA-Seq data set. Interestingly, and as described for mRNA, several of these upregulated miRNAs have also been implicated in the development and progression of cancer. For instance, miR-494 was shown to be upregulated in hepatocellular carcinomas and promote proliferation in tumor cells (154), whereas miR-106a is upregulated in gastric, colorectal, and pancreatic cancers in humans (155). As described previously in human hepatocytes and cholangiocytes treated with ethanol (156), miR-34a was upregulated >5-fold in our data set. Furthermore, ethanol-induced hypomethylation of the miR-34a promoter, which results in increased expression of this miRNA, plays a role in the development of alcoholic liver disease (156).

Moreover, several upregulated miRNAs in our data set are involved in modulating immune responses. For example, and as we recently reported (16), miR-221 was upregulated in PBMCs of heavy drinkers. We previously showed that increased levels of miR-221 resulted in decreased expression of transcription factors STAT3 and ARNT, which, in turn, regulate expression of VEGF, G-CSF, and HGF (16, 157). Indeed, VEGFA and HGF were downregulated in this study by 5- and 2-fold, respectively. Another target of miR-221, RGS6, was also downregulated in this study. In addition, upregulation of miR-125b interferes with the innate immune response following LPS stimulation or microbial infection (158). Finally, miR-223, upregulated 19-fold in our data set, inhibits NF-κB activation, angiogenesis, and endothelial cell proliferation, thereby impairing wound healing (159) and inflammation (160).

Many of the downregulated miRNAs are also involved in cancer. For instance, miR-203 was identified as a tumor suppressor and inhibits proliferation in colorectal cancer cell lines (161). Similarly, expression of miR-144 is significantly decreased in human lung cancer and inhibits proliferation in lung cancer cell lines (162). Finally, chronic ethanol feeding in a mouse model of alcoholic steatohepatitis also led to downregulation of miR-183 (163). Additionally, most of the downregulated miRNAs modulate immunity. miR-183 levels were shown to be positively associated with phagocytosis by macrophages (164). As we reported previously (16, 157), miR-29a expression was modulated by heavy drinking. Specifically, miR-29a was downregulated, and its target, SH3PXD2A, was upregulated. Finally, in addition to its role as a tumor suppressor, increased expression of miR-203 was shown to be important for anti-inflammatory responses (165).

In summary, our studies revealed that heavy ethanol consumption results in the downregulation of genes that promote resolution of infection and wound healing and protect against obstructive lung diseases and cancer, whereas moderate drinkers showed increased expression of genes associated with enhancing immune responses. Heavy drinking status also resulted in upregulation of genes involved in impaired wound healing and cancer progression compared with controls and moderate drinkers, whereas moderate ethanol consumption lowered the expression of genes associated with cancer. Moreover, heavy ethanol consumption altered the expression of several miRNAs whose targets were differentially expressed in our data set and are involved in cancer progression and immune function. One of the strengths of this study is that we used an outbred animal model of voluntary self-administration that faithfully recapitulates human behavior and physiology. However, a caveat of the current study is that only three animals/group were analyzed. Future studies are needed to extend these observations, using a larger cohort of animals, to other infectious agents, such as influenza. Future studies will also investigate the mechanisms underlying dose-dependent changes in gene expression by uncovering factors regulating gene expression, such as epigenetic changes within specific immune cells that an influence expression of both mRNA and miRNA. Using the nonhuman primate model of alcohol self-administration, longitudinal gene regulation, and epigenetic changes in immune cells and target organs offers the promise for understanding the complicated and dose-dependent impact of alcohol on immunity and health.

We thank Sumana Pasala for help with RNA extraction and library construction.