Chronic ethanol consumption: role of TLR3/TRIF-dependent signaling

Abstract

Chronic ethanol consumption stimulates neuroimmune signaling in the brain, and Toll-like receptor (TLR) activation plays a key role in ethanol-induced inflammation. However, it is unknown which of the TLR signaling pathways, the myeloid differentiation primary response gene 88 (MyD88) dependent or the TIR-domain-containing adapter-inducing interferon-β (TRIF) dependent, is activated in response to chronic ethanol. We used voluntary (every-other-day) chronic ethanol consumption in adult C57BL/6J mice and measured expression of TLRs and their signaling molecules immediately following consumption and 24 hours after removing alcohol. We focused on the prefrontal cortex where neuroimmune changes are the most robust and also investigated the nucleus accumbens and amygdala. Tlr mRNA and components of the TRIF-dependent pathway (mRNA and protein) were increased in the prefrontal cortex 24 hours after ethanol and Cxcl10 expression increased 0 hour after ethanol. Expression of Tlr3 and TRIF-related components increased in the nucleus accumbens, but slightly decreased in the amygdala. In addition, we demonstrate that the IKKε/TBK1 inhibitor Amlexanox decreases immune activation of TRIF-dependent pathway in the brain and reduces ethanol consumption, suggesting the TRIF-dependent pathway regulates drinking. Our results support the importance of TLR3 and the TRIF-dependent pathway in ethanol-induced neuroimmune signaling and suggest that this pathway could be a target in the treatment of alcohol use disorders.

INTRODUCTION

Chronic alcohol consumption leads to adaptive changes in the central nervous system (CNS) that contribute to dependence and the development of alcohol use disorders (AUDs) (Diamond & Gordon 1997; Harper 2009). The prefrontal cortex (PFC), a brain region central to cognitive behavior, is important in the development of AUDs (Mukherjee et al. 2008; Abernathy et al. 2010). Chronic alcohol use produces molecular adaptations in the innate immune system, and the PFC is particularly sensitive to these changes (Goral et al. 2008; Crews 2012; Stolyarova et al. 2015). Although the PFC shows the strongest immune response to ethanol, the nucleus accumbens (NAc) and amygdala (AMY) are also affected (He & Crews 2008; Bajo et al. 2014; Osterndorff-Kahanek et al. 2015). There is evidence that neuroimmune mechanisms cause CNS damage and also promote alcohol dependence, suggesting that these pathways are potential therapeutic targets for the deleterious effects of alcohol (Alfonso-Loeches et al. 2010; Zou & Crews 2010; Blednov et al. 2011; Blednov et al. 2012).

Toll-like receptors (TLRs) are innate immune proteins and key components of the neuroimmune response to ethanol, particularly TLR2, TLR3 and TLR4 (Alfonso-Loeches et al. 2010; Crews et al. 2013). TLRs are pattern recognition receptors, and their signaling is triggered through a variety of pathogen-derived ligands. TLR2 and TLR4 are found on the cell surface and respond to bacterial ligands while TLR3 and TLR7 are in endosomes and respond to viral ligands (Lehnardt 2010). In addition, TLRs can respond to endogenous ligands like high mobility group box 1 (HMGB1) (Crews et al. 2013; Anggayasti et al. 2017). TLRs signal through two pathways, the myeloid differentiation primary response (MyD88) and the TIR-domain-containing adapter protein (TRIF/TICAM-1) pathways. In the MyD88-dependent pathway, the adapter protein MYD88 binds to the Toll/IL-1 receptor (TIR) domain and recruits interleukin-1 (IL-1) receptor-associated kinase 4 (IRAK4), IRAK1 and TNF receptor-associated factor 6 (TRAF6). IRAK4 then phosphorylates IRAK1, which leads to the release of TRAF6 and the formation of the TRAF6 complex. Members of the TRAF6 complex phosphorylate the inhibitor of nuclear factor kappa-B kinase (IKK) complex, allowing release and activation of nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) (Takeda & Akira 2005). NF-κB activation leads to the transcription of proinflammatory cytokines, such as IL-1β, tumor necrosis factor alpha (TNF-α) and interIeukin-6 (IL-6). The TRIF-dependent pathway uses the adapter protein TRIF, which signals through IKKε (also referred to as IKKi) and TANK-binding kinase 1 (TBK1), leading to the phosphorylation of interferon regulator factor 3 (IRF3) and the transcription of type I interferons. Activation of the TRIF-dependent pathway also leads to increased transcription of interferon inducible genes like C-X-C motif chemokine 10 (Cxcl10/IP-10) and chemokine (C-C motif) ligand 5 (Ccl5/RANTES) (Weighardt et al. 2004; Hirotani et al. 2005). All TLRs (except TLR3) signal through the MyD88-dependent pathway while TLR3, TLR4 and possibly TLR2 signal through the TRIF-dependent pathway (Takeda & Akira 2005; Nilsen et al. 2015).

Although studies have shown increased TLR mRNA and protein levels with high dose ethanol exposure, few have examined the effects of voluntary ethanol consumption on the MyD88 versus TRIF pathway components in the brain (Crews et al. 2013; Lippai et al. 2013; Whitman et al. 2013). It is important to understand the downstream signaling events that mediate ethanol-induced changes in order to target the relevant pathways and simultaneously alter the activity of multiple TLRs. The goal of this study was to examine TLR signaling components after voluntary chronic ethanol consumption.

We investigated both mRNA and protein changes in TLRs and components of the MyD88-dependent and TRIF-dependent pathways using C57BL/6J mice undergoing an every-other-day two-bottle choice (EOD-2BC) paradigm, which is a voluntary drinking test that leads to escalation in drinking and changes in gene expression (Crabbe et al. 2012; Osterndorff-Kahanek et al. 2013). We examined three different brain regions and two different time points to compare changes that occur immediately after consumption, while alcohol is still present in low levels, with those that occur when the mice are anticipating reinstatement of alcohol. We present novel evidence for TRIF-dependent and brain region-dependent signaling during alcohol withdrawal that may influence craving. To determine the functional importance of TRIF signaling in regulating ethanol consumption, we show that the TRIF pathway inhibitor Amlexanox blunts the inflammatory response in the brain and reduces ethanol consumption. Our results suggest that TRIF-dependent signaling may have a role in the development of AUDs and is a potential therapeutic target.

MATERIAL AND METHODS

Ethics statement

All procedures were approved by the University of Texas at Austin Institutional Animal Care and Use Committee (animal protocol number AUP-2013-00061) and adhered to the NIH Guidelines. The University of Texas at Austin animal facility is accredited by the Association for Assessment and Accreditation of Laboratory animal Care.

Animals and voluntary ethanol consumption

Studies were conducted in adult (6–8 weeks old) drug-naïve C57BL/J (B6) male mice (Jackson Laboratories, Bar Harbor, ME, USA). Mice were individually housed and allowed to acclimate to upright bottles 1 week before the start of the experiment. The experimental rooms were maintained at an ambient temperature of 21 ± 1°C, 40– 60 percent humidity and a regular light/dark schedule (7 AM–7 PM). Food and water were available ad libitum.

An EOD-2BC paradigm was used (Crabbe et al. 2012; Osterndorff-Kahanek et al. 2013), and mice were randomly assigned to the control or treatment group. Treatment and control groups each contained 15 mice per time point (60 total). Control groups had access to water every day. Treatment groups had EOD access to water and a 15 percent (v/v) ethanol solution, and water only on off days. On days that ethanol was available, bottle positions were alternated to control for potential side preferences. Ethanol and water bottles were weighed after drinking days, and animals were weighed once per week to calculate consumption (Supporting Information Fig. S1). The study concluded after 60 days (30 drinking days).

Blood alcohol measurements and tissue harvest

Mice were sacrificed either immediately after ethanol was removed (0-hour group) or 24 hours after ethanol was removed (24-hour group). For the 0-hour group, retro-orbital bleeds were performed before sacrifice to determine blood alcohol concentration. Blood alcohol concentration values, expressed as milligram ethanol per deciliter of blood (Supporting Information Table S1), were determined spectrophotometrically by an enzyme assay (Lundquist et al. 1959).

Five mice per group were used for immunohistochemistry. These mice were anesthetized using isofluorane and given intraperitoneal injections of Euthanasia III Solution (TW Medical, Lago Vista, TX, USA). Once death was confirmed, transcardial perfusion was begun with phosphate-buffered saline (PBS) and followed by 4 percent paraformaldehyde (PFA). The whole brain was removed and post-fixed in 4 percent PFA. The remaining 10 mice per group were used for RNA and protein isolation. Following rapid cervical dislocation, brains were removed and placed on ice. The PFC was dissected as previously described (Osterndorff-Kahanek et al. 2013), cut in half and flash frozen in liquid nitrogen. The remainder of the brain was flash frozen separately.

Tissue punches

Frozen brains were mounted in optimum cutting temperature compound (VWR International, Randor, PA, USA) and placed in isopentane on dry ice. Micropunches of the NAc and AMY were taken as previously described (Osterndorff-Kahanek et al. 2015). The following coordinates, anterior-posterior distance from bregma, were utilized: NAc (+1.8 mm to +0.6 mm) and AMY (−0.9 mm to −1.8 mm).

RNA isolation and quantitative reverse transcription polymerase chain reaction

Half of each frozen PFC and the tissue punches were used for RNA isolation using the MagMax-96 Total RNA Isolation Kit (Thermo Fisher Scientific Inc., Rockford, IL, USA). The RNA yield was quantified on a NanoDrop 1000 spectrophotometer and assessed for quality on an Agilent 2200 TapeStation (Agilent Technologies, Santa Clara, CA, USA). RNA was reverse transcribed into cDNA using the Applied Biosystems High-capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific Inc.). cDNA was tested for genomic DNA contamination and showed at least a 10 Cq difference between the reverse transcription (+RT and −RT) samples (Bustin et al. 2009). Applied Biosystems TaqMan^® Gene Expression Assay (Thermo Fisher Scientific Inc.) primers were used, and specific assay IDs are shown in Supporting Information Table S2. Quantitative polymerase chain reactions (qPCRs) were performed using SsoAdvanced™ Universal Probes Supermix (Bio-Rad, Hercules, CA, USA) in 10-μl reactions containing 18 ng of cDNA. All reactions were performed in triplicate and included a negative control. qPCR reactions were carried out using the CFX384 Real-time System (Thermo Fisher Scientific Inc.). For the PFC, three reference genes (Gusb, Gapdh and Hprt) were tested for each time point, and the best two genes were selected based on normalization factors in the qbase+ software (Biogazelle, Gent, Belgium). The 0-hour samples were normalized to Gapdh and Hprt; the 24-hour samples were normalized to Gapdh and Gusb. The NAc and AMY micropunches were normalized only to Gapdh due to smaller quantities of cDNA. Relative quantification of mRNA levels was determined using the qbase+ software.

Protein isolation and western blot analysis

Half of each PFC was used for protein isolation. Tissue was homogenized in 200 μl of lysis buffer [150 mM NaCl, 50 mM Tris-HCl pH 7.4, 1 mM ethylenediaminetetraacetic acid, 1 percent Triton-X-100, 1 percent sodium deoxycholic acid, 0.1 percent sodium dodecyl sulfate and 1X Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific Inc.)], centrifuged for 10 minutes at 10 000 × g, aliquotted and frozen at −80°C. Protein concentrations were determined using the DC Protein Assay (Bio-Rad). Cell lysates (40 μg) were boiled for 5 minutes, run on 10 percent Mini-protean TGX Precast Gels (Bio-Rad) and transferred to polyvinylidene difluoride membranes using semi-dry transfer. Membranes were blocked with 5 percent dried milk in Tris-buffered saline with 0.5 percent Tween-20 (TBST) and incubated overnight at 4°C with primary antibody (Supporting Information Table S3a). Membranes were washed with TBST and incubated with horseradish peroxidase-conjugated secondary antibodies in 5 percent dried milk in TBST (Supporting Information Table S3b). Bands were visualized using enhanced chemiluminescence (Pierce) and imaged on film using G:BOX Chemi XX6 (Syngene, Cambridge, UK). Bands were quantified using ImageJ and normalized to GAPDH and β-actin. All blots were repeated at least once.

Immunohistochemistry

Brains were post-fixed for 24 hours in 4 percent PFA at 4°C, cryoprotected for 24 hours in 20 percent sucrose and mounted in molds with optimal temperature compound. Frozen brains were sectioned coronally (20 μm thick) and placed free floating into PBS. Sections were permeabilized in optimized detergent (0.1 percent Triton-X-100 or 0.1 percent sodium dodecyl sulfate) and blocked in 10 percent goat or donkey serum for 1 hour at room temperature. Sections were incubated with primary antibody overnight at 4°C (Supporting Information Table S3c). The following day, sections were washed with PBS and incubated with secondary antibody (Supporting Information Table S3d) for 2 hours at room temperature. Appropriate quality control staining was performed for labeling and secondary antibodies (Supporting Information Fig. S2). Sections were mounted on slides using Vectashield containing 4′,6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI) (Vector Labs, Burlingame, CA, USA).

Microscopy

Quantification of immunopositive cells was performed using a Zeiss Axiovert 200M fluorescent light microscope (Zeiss, Thornwood, NY, USA) equipped with an Axiocam b/w camera. Brain regions were identified using a mouse brain atlas as previously described (Zuloaga et al. 2014). Bilateral images of the PFC (distance from bregma: +2.8 to +2.24 mm), NAc (+1.10 to +0.8 mm) and AMY (−1.20 to −1.60 mm) were captured using a 20× objective. All immunohistochemistry was quantified bilaterally within fixed area frames: PFC (box, 645 μm × 645 μm), NAc (circle, 575 μm diameter) and AMY (circle, 675 μm diameter), using the ImageJ plug-in ITCN (http://rsb.info.nih.gov/ij/pIugins/itcn.htmI). Areas used for quantification are shown in Supporting Information Fig. S2. Any auto-fluorescence around the perimeter of the tissue was not included. We quantified IRF3 and IKKε immunoreactivity as the number of immunopositive cells per section divided by total cell count (i.e. DAPI). For all cell quantifications, cells were counted in both hemispheres for a given region and summed. Total cell counts for each animal were then averaged and are presented as percent immunopositive cells per section. For IBA1 staining, images were thresholded and then each microglial cell was counted (particle count), average particle size was determined and mean pixel intensity was quantified within the selected area (methods described in https://www.unige.ch/medecine/bioimaging/files/1914/1208/6000/Quantification.pdf). Selected area was calculated as sum area of particles/area fraction; density was calculated as particle count/selected area. Because of the diffuse staining pattern of GFAP, it was difficult to count individual cells, so the analysis was performed as previously described (Alfonso-Loeches et al. 2010). GFAP immunoreactivity was calculated as the percentage of thresholded area occupied by the specific staining in relation to the whole area. The results are expressed as a fold change over control values. Approximately three medial PFC slices from coronal brain sections of five animals per group were analyzed.

Amlexanox treatment

Mice consumed 15 percent ethanol for at least 8 weeks using the EOD-2BC paradigm described previously. After this period, ethanol consumption was measured for at least 4 days to ensure stable consumption. Ethanol intake was then measured after saline administration for 2 days, and mice were grouped to provide similar levels of ethanol intake and preference based on the consumption during these 2 days. On day 4, mice were administered saline or Amlexanox once daily, and results are presented as the average from 2-day periods of consecutive drinking using different bottle positions. Amlexanox (100 mg/kg) was administered orally by gavage (p.o.). Amlexanox was purchased from Abcam (ab142825). Drug was prepared as suspensions in saline with five to four drops of Tween-80 and administered once daily in a volume of 0.05 ml/10 g of body weight 30 minutes before drinking experiments. Saline containing four to five drops of Tween-80 was administered to control groups.

Drug effect on ethanol consumption and preference

Ethanol consumption, preference and total fluid consumption were measured as previously described (Blednov et al. 2003). Briefly, bottles containing 15 percent ethanol and water were weighed before they were placed in the cage and again after 5 hours. Ethanol consumption was measured based on g/kg body weight/5 hours. Preference for ethanol was calculated as the amount of ethanol consumed divided by the total amount of fluids consumed per day (a value >50 percent indicates preference for ethanol).

Amlexanox-Polyinosinic:polycytidylic acid treatment

Mice were divided into four groups (n = 10 per group) and were given an oral treatment (saline or Amlexanox 100 mg/kg) followed by an i.p. injection 30 minutes later [saline or polyinosinic:polycytidylic acid (Poly I:C) 5 mg/kg]. Mice were sacrificed 3 hours after the i.p. injection, and the PFC was fresh harvested as described previously. RNA isolations and qPCR analysis were performed as described previously.

Statistical analysis

Results are reported as the mean ± standard error of the mean. Data were analyzed using GRAPHPAD software (GraphPad Software Inc., San Diego, CA, USA) and two-tailed Student’s t-tests unless otherwise noted (unpaired parametric test assumes populations have the same standard deviation). Statistical outliers were identified using Grubb’s test and removed from analysis. Consumption following Amlexanox or saline treatment was analyzed using a two-way analysis of variance and Tukey’s multiple comparisons test. Gene expression changes following Poly I:C and Amlexanox treatments were analyzed using a one-way analysis of variance and Tukey’s multiple comparisons test.

RESULTS

Chronic ethanol consumption increases expression of Tlrs

To determine whether chronic ethanol consumption changes mRNA levels in the TLR signaling pathways in the PFC, qPCR analysis was performed 0 and 24 hours after ethanol removal. Several TLRs are altered in other studies using different alcohol paradigms (Alfonso-Loeches et al. 2010; Crews et al. 2013; Lippai et al. 2013; Whitman et al. 2013). Consistent with previous studies, qPCR analysis confirmed that Tlr2, Tlr3, Tlr4 and Cd14 (TLR co-receptor) are increased following voluntary ethanol consumption at the 24-hour time point (Fig. 1a). Although Tlr2 and Tlr4 have been implicated in alcohol action in more studies than Tlr3 (Robinson et al. 2014), the largest fold-change (2.12) was observed for Tlr3. Like Tlr3, Tlr7 is another endocytic TLR that recognizes viral RNA and leads to the production of type I interferons, despite signaling through the MyD88-dependent pathway. Tlr7 has been implicated in alcohol liver disease (Cha et al. 2012) and in the suppression of cytokines by acute ethanol exposure (Pruett et al. 2004). To determine whether Tlr7 expression changed with chronic ethanol consumption, qPCR was performed at both time points. In contrast to the other TLRS, Tlr7 expression increased 0 hour after ethanol removal.

Figure 1.

RT-qPCR analysis of the PFC 0 and 24 hours after ethanol removal (n = 10/group). (a) The 0-hour group showed increased Tlr7 expression while the 24-hour ethanol group showed increased Tlr2, Tlr3, Tlr4 and Cdl4 mRNA expression. (b) The 0-hour ethanol group showed increased Myd88 and Irak4 expression. (c) Increased Ikki expression was observed in the 0-hour group, while increased Trif and Irf3 expression was found in the 24-hour ethanol group. (d) The 0-hour group showed increased Irf7 expression with ethanol. (e) The 0-hour ethanol group showed increased Il1b and Cxcl10 expression while the 24-hour ethanol group showed increased Il6 expression. (f) The 24-hour ethanol group showed increased expression of Cd11b, a microglial marker. All values are expressed as fold change over control ± SEM. P < 0.05 compared with control, one-tailed t-test, *P < 0.05 compared with control, two-tailed t-test; **P < 0.01 compared with control, two-tailed t-test; ***P < 0.001 compared with control, two-tailed t-test. PFC, prefrontal cortex; RT-qPCR, reverse transcription-quantitative polymerase chain reaction; SEM, standard error of the mean; TLRs, Toll-like receptors; TRIF, TIR-domain-containing adapter protein

Chronic ethanol consumption increases MyD88-related and TRIF-related mRNA levels in a time-dependent manner

The qPCR was then used to evaluate changes in components of the MyD88-dependent pathway, which is common to all TLRs except TLR3, and hypothesized to be affected by chronic drinking. Only modest increases were observed for Myd88 and Irak4 at 0 hour, and all MyD88 pathway components were unchanged at 24 hours (Fig. 1b). Investigation of the TRIF-dependent pathway revealed an increase in Ikke at 0 hour and an increase in Trif and 1rf3 24 hours after ethanol removal (Fig. 1c). Tbk1 mRNA expression was unchanged. These results suggest that chronic ethanol consumption mainly changes expression of TRIF-dependent pathway mRNA. There are also molecules that are common to both the MyD88-dependent and TRIF-dependent pathways. Traf3 and Irf7 are involved in both TRIF-dependent signaling and TLR7 signaling through the MyD88-dependent pathway while Traf6 is involved in all MyD88-dependent and TRIF-dependent signaling. Measurement of these mRNAs using qPCR revealed that Traf3 and Traf6 were unchanged at both time points, but Irf7 expression increased at 0 hour (Fig. 1d).

Chronic ethanol consumption increases Il1b and Cxcl10 mRNA immediately after ethanol removal and Il6 24 hours after ethanol removal

The TLR pathway activation leads to transcription of cytokines, chemokines and interferons, and several cytokine mRNAs have been previously linked to alcohol exposure (Il1b, Tnfa and Il6) (Alfonso-Loeches et al. 2010; Zou & Crews 2010; Lippai et al. 2013; Whitman et al. 2013). Expression of these cytokines was measured to evaluate activation of the MyD88-dependent pathway. In addition, transcriptional outputs of the TRIF-dependent pathway (1fnb, Cc15 and Cxcl10) were measured; however, only Cxcl10 was expressed at high enough levels to be quantified. Expression of Il1b and Cxcl10 mRNA increased at 0 hour while Il6 expression increased 24 hours after ethanol removal (Fig. 1e). Il1b transcription is primarily induced by MyD88-dependent signaling, while Cxcl10 is mostly induced by TRIF-dependent signaling and Il6 requires both pathways (Hirotani et al. 2005). These data suggest that both pathways may be activated in response to chronic ethanol exposure.

Chronic ethanol consumption increases a microglial, but not an astrocyte, marker at 24 hours

Several studies have shown that glial activation is an important component of ethanol-induced neuroimmune signaling (He & Crews 2008; Fernandez-Lizarbe et al. 2009; Lippai et al. 2013). Microglia and astrocytes respond to pathogens and danger signals (i.e. HMGB1) by releasing cytokines, which can then further activate glia (Crews & Vetreno 2015). To determine the impact of chronic voluntary ethanol consumption on glial activation, we measured the astrocyte marker Gfap, the microglial marker Cd11b and the activated microglial marker Cd68. Cd68 expression decreased at 0 hour while Cd11b mRNA increased 24 hours after ethanol removal (Fig. 1f). Immunohistochemistry was also used to evaluate any morphological or protein level changes in microglia and astrocytes in the PFC. Immunohistochemistry revealed few changes, with only a decrease in microglial density at the 0-hour time point (Supporting Information Fig. S4).

Chronic ethanol consumption did not change protein levels in the prefrontal cortex

After gene expression changes were observed for TLRs and TRIF-dependent pathway transcripts, protein levels were measured from the same PFC tissue using western blots. No significant differences were seen in any of the proteins measured (Supporting Information Fig. S5). Due to the small fold-changes, it is possible that western blots are not sensitive enough to detect protein level changes or that protein expression is changing at different time points.

Chronic ethanol consumption leads to increased IKKε and IRF3 immunopositive cells in the prefrontal cortex and nucleus accumbens

Immunohistochemistry, which may be a more sensitive measurement compared with western blots, was used to determine the number of IKKε and IRF3 immunopositive cells. In the PFC, there was a significant increase in the percentage of IKKε immunopositive cells at 0 hour and a significant increase in the percentage of both IKKε and IRF3 immunopositive cells 24 hours after ethanol removal (Fig. 2a). The increases were larger at the 24-hour time point, which is consistent with the qPCR data. It is interesting to note that changes in both mRNA and IKKε protein are seen at 0 hour. In addition to the PFC, immunopositive cells were measured in the NAc and AMY. The NAc showed similar changes to the PFC, with a small increase in IKKε and IRF3 at 0 hour and larger increases in both IKKε and IRF3 at 24 hours (Fig. 2b). No changes were found in the AMY at 0 hour, but there was a decrease in IKKε and an increase in IRF3 at 24 hours (Fig. 2c). These results suggest that components of the TRIF signaling pathway are increased on the protein level by chronic ethanol in a brain region-specific manner.

Figure 2.

Immunohistochemistry for IRF3 and IKKε 0 and 24 hours after ethanol removal. Representative images are shown from one control and one ethanol sample. Graphs show quantification from all samples (n = 5/group), and values represent the average ± SEM percent of cells positive for either IKKε or IRF3 over the total number of DAPI positive cells. Scale bar for prefrontal cortex and nucleus accumbens = 50 μM and for amygdala = 100 μM. (a) In the prefrontal cortex, there is an increase in the percentage of IKKε positive cells at 0 hour and of both IKKε and IRF3 positive cells at 24 hours. (b) In the nucleus accumbens, there is an increase in the percentage both IKKε and IRF3 positive cells at 0 and 24 hours. (c) In the amygdala, there are no changes at 0 hour, and at 24 hours, there is a decrease in IKKε positive cells and an increase in IRF3 positive cells. ^#P < 0.05, one-tailed t-test, *P < 0.05 compared with control, two-tailed t-test; **P < 0.01 compared with control, two-tailed t-test; ***P < 0.001 compared with control, two-tailed t-test [Colour figure can be viewed at wileyonlinelibrary.com]

Chronic ethanol consumption leads to increased Tlr3 mRNA in the nucleus accumbens and decreased Tlr mRNA in the amygdala

Because changes in the percentage of immunopositive cells in the NAc and AMY were observed, mRNA levels in these brain regions were measured from micropunch control, two-tailed t-test, ***P < 0.001 compared with control, two-tailed t-test. AMY, amygdala; NAc, nucleus accumbens; RT-qPCR, reverse transcription-quantitative polymerase chain reaction; SEM, standard error of the mean; TLRs, Toll-like receptors; TRIF, TIR-domain-containing adapter protein samples. Because of limited sample, only transcripts that changed in the PFC were measured. In the NAc, there was an increase in Tlr3 mRNA at the 24-hour time point while there was a decrease in Tlr3 and Tlr4 at 24 hours in the AMY (Fig. 3 a). Despite changes in immunopositive cells, there were no changes in expression in the TRIF-dependent pathway components at either time point in either brain region (Fig. 3b). The AMY showed a decrease in Cxcl10 at the 24-hour time point, consistent with the decreased expression of Tlr3/4 and IKKε immunopositive cells seen in that brain region (Fig. 3c). Like the PFC, both the NAc and AMY showed an increase in Il1b expression at the 0-hour time point (Fig. 3c). This early increase in Il1b has also been observed in the hippocampus and VTA (data not shown), suggesting a brain-wide increase in Il1b immediately after ethanol removal.

Figure 3.

RT-qPCR analysis from the NAc and AMY 0 and 24 hours after ethanol removal (for the 0-hour time point n = 10/group and for the 24-hour time point n =10 for the control group and n = 9 ethanol group). (a) The NAc shows increased Tlr3 mRNA levels while the AMY shows decreased Tlr3 and Tlr4 at 24 hours. (b) No changes were seen in the TRIF pathway genes. (c) Both the NAc and AMY showed increased Il1b expression at 0 hour, and the AMY showed decreased Cxd10 expression at 24 hours. All values are expressed as fold change over control ± SEM. ^#P < 0.05 compared with control, one-tailed t-test; *P < 0.05 compared with control, two-tailed t-test; ***P < 0.01 compared with control, two-tailed t-test, ***P < 0.001 compared with control, two-tailed t-test. AMY, amygdala; NAc, nucleus accumbens; RT-qPCR, reverse transcription-quantitative polymerase chain reaction; SEM, standard error of the mean; TLRs, Toll-like receptors; TRIF, TIR-domain-containing adapter protein

Inhibitor of TRIF-dependent pathway decreases ethanol consumption

To evaluate whether changes in TRIF-dependent signaling may regulate ethanol consumption, mice were treated with the drug Amlexanox. Amlexanox is considered a specific inhibitor of IKKε and TBK1 (Reilly et al. 2013), although it does inhibit GRK5 (Homan et al. 2014) and the S100A13-FGF1 complex (Rani et al. 2010) at higher concentrations. Previous work showed that Amlexanox decreases kinase activity of IKKε and TBK1 as well as phosphorylation of IRF3 in response to the TLR3 ligand Poly I:C (Reilly et al. 2013). At a dose of 100 mg/kg, Amlexanox reduced ethanol consumption and preference in mice undergoing a chronic EOD-2BC paradigm but did not significantly change total fluid consumption (Fig. 4). These data suggest a role for activation of the TRIF-dependent pathway in the regulation of ethanol consumption.

Figure 4.

Measures for saline and Amlexanox (100 mg/kg) treatment groups under baseline (both groups received saline injections) and drug conditions. All measurements are taken 5 hours following treatment. (a) The Amlexanox group shows decreased ethanol consumption. (b) The Amlexanox group shows decreased preference for ethanol. (c) The Amlexanox group shows no change in total fluid intake after treatment compared to baseline. Bars labeled with the same letter are not statistically different while bars with different letters are (two-way analysis of variance with Tukey’s multiple comparisons test, n =11 per group)

Amlexanox reduces poIyinosinic:poIycytidyIic acid-induced increase in Cxcl10 expression

Amlexanox decreases the Poly I:C response in vitro and decreases high-fat diet-induced expression of Cxcl10 in liver (Reilly et al. 2013), but there is no evidence that it inhibits the TRIF-dependent pathway in the brain. To address this, we measured Cxcl10 expression in the PFC of mice that were treated with saline only, Amlexanox only, the TLR3 ligand Poly I:C and both Poly I:C and Amlexanox. Administration of Poly I:C resulted in a ~2000-fold increase in Cxcl10 expression; however, administration of Poly I:C with Amlexanox reduces the Poly I:C-induced increase by 40 percent (Supporting Information Fig. S7).

DISCUSSION

Emerging evidence supports the role of neuroinflammatory mechanisms in alcohol dependence and alcohol-induced brain damage. Our group and others have shown that ethanol upregulates inflammatory mediators in the brain, which activate neuroimmune pathways (Blednov et al. 2012; Crews et al. 2013; Lippai et al. 2013). Studies have focused on TLR4 because of its role in producing proinflammatory cytokines that can promote neuroinflammatory-dependent brain damage (Alfonso-Loeches et al. 2010; Robinson et al. 2014). Given that MyD88 is the adapter protein for TLR4 (and all other TLRs except TLR3), we hypothesized that chronic ethanol drinking would lead to increased expression of the MyD88-dependent pathway. However, our findings indicate that EOD ethanol consumption primarily increases expression of Tlr3 and components of the TRIF-dependent pathway (Fig. 5). We observed upregulation of TRIF signaling components at both the mRNA and protein levels as well as increased expression of Cxcl10, a transcriptional output of TRIF pathway activation. Although it is not clear how endosomal TLRs (e.g. TLR3 and TLR7) or the TRIF-dependent pathway is targeted by alcohol, recent studies suggest that alcohol increases expression of endogenous ligands (HMGB1 and miRNAs) that may preferentially target specific TLRs (Crews et al. 2013; Coleman et al. 2017). The TRIF-dependent pathway changes appear to be both brain region-specific and time-specific, with the greatest increase in signaling components observed in the PFC 24 hours after the last ethanol exposure and an increase in transcriptional outputs at 0 hour (Table 1). Moreover, we provide evidence that the TRIF-dependent pathway may regulate drinking by showing that a IKKε/TBK1 inhibitor decreases ethanol consumption. A potential role for the TRIF-dependent pathway in ethanol consumption points to the utility of drugs that target this system for treating AUDs.

Figure 5.

Schematic of the Toll-like receptor (TLR) signaling pathways in the prefrontal cortex. Dark blue represents RNA changes at 24 hours, purple represents RNA changes at 0 hour, black dots indicate that polyinosinic:polycytidylic acid-induced mRNA was decreased with Amlexanox, checkered hatching represents protein increases at 0 and 24 hours while diagonal hatching represents protein increases at 24 hours. Gray represents no change, and white indicates that the pathway component was not measured in this study. Light blue means that expression was too low to measure in this study [Colour figure can be viewed at wileyonlinelibrary.com]

Table 1.

Effects of chronic ethanol on mRNA and protein levels in (a) the PFC, (b) the NAc and (c) the AMY were examined 0 and 24 hours after ethanol removal.

(a) mRNA and protein changes in PFC
	PFC
	mRNA		protein-IHC
	0 hour	24 hours	0 hour	24 hours
Toll-like receptors and co-receptors
Tlr2	–	↑(1.26)
Tlr3	–	↑(2.12)
Tlr4	–	↑(1.49)
Tlr7	↑(1.15)	—
Cd14	–	↑(1.19)
MyD88-dependent pathway signaling molecules
Myd88	↑(1.07)	—
Irak1	–	—
plrak1
Irak4	↑(1.19)	—
Ikkb	–	—
TRIF-dependent pathway signaling molecules
Trif		↑(1.23)
Ikki	↑(1.17)		↑(1.08)	↑(1.24)
Irf3	–	↑(1.56)	—	↑(1.68)
Molecules involved in both pathways
Irf7*	↑(1.18)	—
Traf3*	–	—
Traf6	–	—
Transcriptional outputs of MyD88-dependent pathway
Il1b	↑(4.45)	—
Tnfa	–	—
Il6	–	↑(1.58)
Transcriptional outputs of the TRIF-dependent pathway
Cxcl10	↑(2.11)	—
Ccl5	–	—
Ifnb	–	—
Glial markers
Cd11b/Iba1		↑(1.26)
Gfap	–	—
Cd68	↓(0.89)	—

(b) mRNA and protein changes in NAc
	NAc
	mRNA		protein-IHC
	0 hour	24 hours	0 hour	24 hours
Toll-like receptors
Tlr2	–	—
Tlr3	–	↑(1.15)
Tlr4	–	—
MyD88-dependent pathway signaling molecules
Myd88	–	—
TRIF-dependent pathway signaling molecules
Trif	–	—
Ikki	–	—	↑(1.05)	↑(1.36)
Irf3	–	—	↑(1.10)	↑(1.31)
Cytokines and chemokines
Il1b	↑(4.24)	—
Cxcl10	–	—

(c) mRNA and protein changes in AMY
	AMY
	mRNA		protein-IHC
	0 hour	24 hours	0 hour	24 hours
Toll-like receptors
Tlr2	–	—
Tlr3	–	↓(0.87)
Tlr4	–	↓(0.87)
MyD88-dependent pathway signaling molecules
Myd88	–	—
TRIF-dependent pathway signaling molecules
Trif	–	—
Ikki	–	—	—	↓(0.88)
Irf3	–	–	–	↑(1.15)
Cytokines and chemokines
Il1b	↑(3.25)	—
Cxcl10		↓(0.37)

Effects are shown as increases, decreases or no change (—); blank cells indicate that mRNA or protein was not measured. Gene names are indicated in the left column, and the numbers in parentheses indicate fold change compared with control group. AMY = amygdala; IHC = immunohistochemistry; NAc = nucleus accumbens; PFC = prefrontal cortex. Genes with * are only used in the non-canonical MyD88-dependent pathway that leads to the transcription of type I interferons.

Chronic alcohol abuse encompasses a relapsing cycle of intoxication, withdrawal and craving resulting in aberrant neuroplasticity in corticolimbic structures (e.g. PFC), mesolimbic (e.g. NAc) and the extended AMY circuit. Expression changes in immune-related genes in the PFC, NAc and AMY following chronic intermittent ethanol treatment showed little overlap between brain regions, suggesting that there are region-specific differences in immune signaling (Osterndorff-Kahanek et al. 2015). Our results also support brain region-specific differences in neuroinflammatory signaling. The PFC showed the largest increase in Tlr mRNA expression and in TRIF-related expression of mRNA and protein. Similar changes in protein were found in the NAc, but only Tlr3 mRNA increased in this region. This could be due to differences in temporal signaling events in these brain regions, suggesting a dominance of corticolimbic structures during craving. It is also possible that TLR3 is the predominant subtype altered in the NAc, while TLR2, TLR3 and TLR4 all increase in the PFC. Interestingly, the AMY showed decreased Tlr3, Tlr4 and Cxcl10 mRNA levels and decreased IKKε protein at 24 hours. These results are consistent with data that showed psychological stress produced divergent cytokine gene expression in the medial PFC and the AMY (Vecchiarelli et al. 2016). However, increased IRF3 protein expression was observed in the AMY at 24 hours, potentially due to decreased TLR signaling and compensatory increased IRF3 signaling. Although Tlr and TRIF-dependent changes in the AMY were the opposite of those seen in the PFC and NAc, all three brain regions showed an increase in Il1b at the 0-hour time point suggesting that there may be distinct signaling events occurring. This increase in Il1b is noteworthy because of the involvement of the IL-1 system with GABAergic transmission and evidence that IL-1 is also involved in disorders, like depression, that have high comorbidity with AUDs (Bajo et al. 2015a; Bajo et al. 2015b; Barnes et al. 2016). Further studies are necessary to elucidate the differences in TLR expression and signaling in various brain regions.

The EOD drinking paradigm produces escalation of intake and maintains high alcohol consumption on drinking days (Crabbe et al. 2012). The 0-and 24-hour time points were selected to compare neuroimmune responses at different stages during chronic alcohol consumption (i.e. craving versus bingeing). The 0-hour time point assesses immune activation during the last ethanol exposure while ethanol is still present, albeit in low levels, whereas the 24-hour time point represents the postwithdrawal stage when the mice are anticipating the availability of alcohol. We observed effects of alcohol consumption that are dependent on the time after withdrawal with more MyD88-dependent pathway changes at the 0-hour time point while most of the TLR and TRIF-dependent pathway changes are at the 24-hour time point, consistent with the idea that the MyD88-dependent pathway is activated earlier than the TRIF-dependent pathway (Kawai & Akira 2007). In addition, we saw increased expression of Cxcl10, a transcriptional output of TRIF signaling, at the 0-hour time point in the PFC. These results suggest that the pathway components are increasing expression at 24 hours, leading to increased activation of the pathway that results in transcription induction later, at the 0-hour time point. However, few studies have looked at TRIF pathway components and outputs at multiple time points, and future work would be needed to verify this time course. Because we observed the largest increase in TRIF-dependent components after 24 hours, we suggest that this pathway may be involved in the post-withdrawal craving stage. Amlexanox reduced ethanol consumption when administered at this 24-hour time point (right before drinking was reinstated), further suggesting that TRIF signaling is involved in craving. Interestingly, Cxcl10 expression decreased at the 0-hour time point in the AMY, possibly due to differences in temporal regulation between brain regions or differences in the timeline for transcription versus degradation of mRNA. There are some timing inconsistencies, such as Ikke increasing at the 0-hour time point and not the 24-hour time point or Il6 and Il1b changing at different time points. TLR signaling is complex and involves many levels of regulation, including negative and positive feedback as well as components that are common to multiple pathways (Medzhitov & Horng 2009). It is likely that any perturbation of this pathway could impact expression of other signaling molecules involved and that the timing of these changes would depend on the speed and level of regulation.

In addition to identifying time-specific and brain region-specific changes in the TRIF-dependent pathway, our results suggest a causal role for TRIF signaling in the regulation of consumption. We utilized Amlexanox, an immunomodulatory drug that was recently found to be a potent inhibitor of IKKε/TBK1. Although it has not been demonstrated that Amlexanox can cross the blood brain barrier, Lipinski’s rule of five (Lipinski et al. 2001; Lipinski 2004) suggests it does and we show that it inhibits PFC induction of Cxcl10 in response to a TLR3 agonist. Our Amlexanox results show that global inhibition of the TRIF pathway reduces ethanol consumption. Previous work has suggested that TLR signaling regulates ethanol consumption, with TLR2 and CD14 null mice showing decreased ethanol consumption in several tests (Blednov et al. 2017). However, no previous studies have investigated components of the TRIF-dependent pathway. Interestingly, MyD88 null mice increase ethanol consumption, possibly because of upregulation of TRIF-dependent signaling (Blednov et al. 2017). It is important to note that inhibition of TLR4 signaling does not reduce alcohol consumption emphasizing the importance of the TLR3/TRIF pathway (Harris et al. 2016; Blednov et al. 2017).

Many of the TLR pathway components measured here have been studied following other methods of alcohol exposure. Il1b increased in mouse cortex after an ethanol-based diet, and Il6 increased in mouse cerebellum after binge ethanol exposure (Whitman et al. 2013; Kane et al. 2014). Tlr2 and Tlr4 have been shown to increase after various drinking paradigms, and Tlr3 was increased following chronic intragastric ethanol exposure (Crews et al. 2013; Lippai et al. 2013; Whitman et al. 2013). In vitro studies in microglia and astrocytes have shown that ethanol can activate both the MyD88-dependent and TRIF-dependent pathway, consistent with our detection of changes in components of these pathways in vivo (Fernandez-Lizarbe et al. 2009; Pascual-Lucas et al. 2013). In contrast to our results, some studies reported increases in MyD88 and TLR protein, Gfap mRNA, TNF-α (mRNA and protein) and IL-1p protein (Alfonso-Loeches et al. 2010; Crews et al. 2013; Lippai et al. 2013; Whitman et al. 2013). These differences are likely due to the differing alcohol consumption paradigms, which are known to produce distinct effects on gene expression (Osterndorff-Kahanek et al. 2013; Osterndorff-Kahanek et al. 2015). Interestingly, previous rodent ethanol studies have shown marked changes in glial morphology and marker expression (Alfonso-Loeches et al. 2010; Lippai et al. 2013), while this study showed few. Thus, it is worth noting that even though it did not induce dramatic glial changes, the voluntary ethanol paradigm still led to changes in TLR and cytokine expression. These results suggest that gene expression changes can occur in the absence of obvious glial morphological changes. Although controversial, it is also possible that CNS cell types other than microglia and astrocytes are contributing the gene expression changes.

A caveat of this study is that we were unable to show increased phosphorylation of IRF3 as a measure of pathway activation. We attempted to measure phosphorylated IRF3 and interferon expression, but were unable to detect them at high enough levels in brain (Supporting Information Fig. S6). However, we were able to measure the interferon inducible gene Cxcl10, which is transcribed in response to TRIF pathway activation (Weighardt et al. 2004; Hirotani et al. 2005). Our data showing that the TLR3 ligand Poly I:C leads to increased Cxcl10 expression in the PFC, further supports this chemokine as a measure of pathway activation. Because of the transient nature of phosphorylation states, changes in transcriptional outputs are a more reliable measure of pathway activation. The increased Cxcl10 expression in the PFC at the 0-hour time point supports the idea that the TRIF-dependent pathway is being activated in response to chronic ethanol. CXCL10 is a chemokine that is responsible for recruiting cells that express CXCR3 (microglia, dendritic cells and T lymphocytes) and has roles in regulating apoptosis, cell growth and proliferation, and synaptic activity (Liu et al. 2011; Gruol 2016). Increased CXCL10 therefore could increase the immune response to ethanol as well as contribute to the alcohol-induced damage or changes in neuronal activity. In addition, the fact that an inhibitor of this pathway decreases drinking and partially rescues Poly I:C-induced Cxcl10 increases, further supports the idea that TRIF pathway activation is involved in the ethanol response as well as regulation of consumption.

The fold changes reported here are relatively small, and it is important to consider that there may be a dilution effect when examining heterogeneous cell populations. If immune signaling molecules are preferentially expressed in glial cells, studying the entire brain region would minimize any cell type-specific changes that may be present (Mayfield et al. 2013). We have observed greater differences in gene expression in more homogenous cell fractions, which can reveal discrete, localized changes (Most et al. 2014). In addition, the drinking paradigm we used here is voluntary and does not produce as extreme changes as those seen with high dose paradigms. For example, 5 g/kg intragastric ethanol daily for 10 days (a binge model) produced increases in Tlr mRNA that ranged from 1.5 to 2.5 fold (Crews et al. 2013).

In summary, voluntary chronic ethanol consumption increases expression of brain Tlrs, particularly Tlr3 and components of the TRIF-dependent pathway, 24 hours after ethanol exposure. These changes could promote ethanol consumption and dependence, supported by the fact that a TRIF pathway inhibitor decreased ethanol consumption.

Supplementary Material

FigS1

Figure S1. Average consumption shown as mean (g/kg/day) ± SEM for all animals over the course of the 15 percent ethanol EOD-2bc drinking experiment (n =15 for each time point). The x-axis represents days where ethanol was present.

FigS2

Figure S2. Brain regions used for immunohistochemistry counts: (a) PFC, (b) NAc and (c) AMY.

FigS3

Figure S3. Quality Controls for Immunohistochemical Quantification. (a) Labeling staining controls for all staining conditionings in nucleus accumbens of a C57BI6/J male mouse-primary antibody and secondary antibody omitted, 10 percent donkey serum only. Image taken on fluorescent microscope (20×).(b) Secondary controls for all secondary antibodies in nucleus accumbens of a C57BI6/J male mouse-primary antibody omitted, 10 percent donkey serum and appropriate secondary. Image taken on a fluorescent microscope (20×).

FigS4

Figure S4. Immunohistochemistry for microglia and astrocyte markers. (a) Microglia staining using the microglial marker IBA1. At both time points, control and ethanol PFC were compared (n = 5/condition, 3 slices per animal) and representative images are shown (20×) with an inset at 300 percent magnification. Graphs show quantification of microglial density (number of cells/μM²), mean pixel intensity, and average size (μM). Values represent the mean ± SEM. Only microglial density was statistically different, with an increase at the 0-hour time point. * p < 0.05, 2- tailed T-test. (b) Representative images of GFAP immunoreactivity in the PFC at the 24-hour time point (20×) with an inset at 300 percent magnification. Graphs show GFAP immunoreactivity quantification in PFC of mice 24 hours after ethanol relative to control. Immunoreactivity is quantified as the percentage of thresholded area occupied by GFAP in relation to the whole area. Values represent the mean ± SEM of five animals per group and three slices analyzed per animal.

FigS5

Figure S5. Western blot analysis of the PFC 0 and 24 hours after ethanol removal. Representative blots are shown from one control (C) and one ethanol (E) sample. Graphs show quantification from a complete blot (n = 10/group). Protein levels were normalized to β-actin and GAPDH. No differences were seen in (a) TLR protein levels, (b) MYD88 pathway protein levels, (c) TRIF pathway protein levels, (d) Cytokines protein levels, or (e) Glial marker protein levels. All values are expressed as relative protein expression (normalized to β-actin) ± SEM. Statistics were performed using a 2-tailed t-test.

FigS6

Figure S6. Western blots using 3 commercially available pIRF3 antibodies showing that signal is not able to be detected in the PFC despite being detected in the spleen.

FigS7

Figure S7. Cxcl10 qPCR analysis of the PFC after saline oral and saline I.P injection (S/S), saline oral and Poly I: C (5 mg/kg) I.P. injection (S/P), Amlexanox oral (100 mg/kg) and saline I.P. injection (A/S), and Amlexanox oral (100 mg/kg) with Poly I:C (5 mg/kg) I. P. injection. Poly I:C significantly increases expression of Cxcl10 mRNA while Amlexanox significantly decreases Poly I:C induced expression. Different letters indicate significantly different means (ANOVA with Tukey’s multiple comparisons test, p < .05).

TabS1

Table S1. Blood alcohol concentration (mg/dL) at the 0-hour time point.

TabS2

Table S2. TaqMan^® Gene Expression Assay IDs used for RT-qPCR.

TabS3

Table S3. Antibody names, manufacturers, and dilution information for (a) western blot primary antibodies, (b) western blot secondary antibodies, (c) immunohistochemistry primary antibodies, and (d) immunohistochemistry secondary antibodies.