Increased Toll‐like Receptor‐MyD88‐NFκB‐Proinflammatory neuroimmune signaling in the orbitofrontal cortex of humans with alcohol use disorder

Abstract Background Many brain disorders, including alcohol use disorder (AUD), are associated with induction of multiple proinflammatory genes. One aspect of proinflammatory signaling is progressive increases in expression across cells and induction of other innate immune genes. High‐mobility group box 1 (HMGB1) heteromers contribute to amplification by potentiating multiple proinflammatory responses, including Toll‐like receptors (TLRs). TLR signaling recruits coupling proteins linked to nuclear transcription factors that induce proinflammatory cytokines and chemokines and their respective receptors. We tested the hypothesis that AUD induction of TLR expression increases levels of proinflammatory genes and cellular signaling cascades in association with neurodegeneration in the orbitofrontal cortex (OFC). Methods Postmortem human OFC tissue samples (n = 10) from males diagnosed with AUD were compared to age‐matched moderate drinking controls (CON). Neuroimmune signaling molecules were assessed using immunohistochemistry for protein and reverse transcription polymerase chain reaction for messenger RNA (mRNA). Results In the AUD OFC, we report induction of the endogenous TLR agonist HMGB1 as well as all TLRs assessed (i.e., TLR2‐TLR9) except TLR1. This was accompanied by increased expression of the TLR adaptor protein myeloid differentiation primary response 88 (MyD88), activation of the proinflammatory nuclear transcription factor nuclear factor kappa‐light‐chain‐enhancer of activated B cells (NFκB), and downstream induction of proinflammatory cytokines, chemokines, and their corresponding receptors. Several of these proinflammatory signaling markers are expressed in glia and neurons. The induction of HMGB1‐TLR‐MyD88‐NFκB proinflammatory signaling pathways correlates with neurodegeneration (i.e., Fluoro‐Jade B), lifetime alcohol consumption, and age of drinking onset. Conclusion These data implicate the induction of HMGB1‐TLR‐MyD88‐NFκB cascades through coordinated glial and neuronal signaling as contributors to the neurodegeneration seen in the postmortem human OFC of individuals with AUD.


INTRODUC TI ON
Neurodegeneration and neuroimmune signaling are associated with several disorders of the CNS, including Alzheimer's disease (AD) and related dementias, Parkinson's disease (PD), and alcohol use disorder (AUD) (Cheng et al., 2017;Crews et al., 2017;Van Hoesen et al., 2000;Hornberger et al., 2011;Kamal et al., 2020;Qin & Crews, 2012b). Neuroimmune signaling molecules of the innate immune system are expressed in the human brain, but their role in neurodegeneration is poorly understood. The orbitofrontal cortex (OFC), which is critically involved in regulating executive function (Schoenbaum et al., 2006), is vulnerable to damage in human AUD. Significant neuronal loss, volumetric reductions, and decreased connectivity are reported in the OFC of individuals with AUD and are accompanied by deficits in executive functioning (Crews & Boettiger, 2009;Miguel-Hidalgo et al., 2006;Moorman, 2018;Qin & Crews, 2012b). Studies in alcoholtreated rodents and postmortem humans diagnosed with AUD report decreased cortical neuron populations (Harper & Kril, 1989;Miguel-Hidalgo et al., 2006) and increased markers of neuronal degeneration (Pascual et al., 2007;Q in & Crews, 2012b). Induction of proinflammatory neuroimmune signaling is thought to contribute to neurodegeneration (Crews & Vetreno, 2014;Glass et al., 2010;Kamal et al., 2020). Increased expression of neuroimmune signaling molecules is observed in the brains of alcohol-treated rodents and postmortem humans with AUD (Crews et al., 2013;Erickson et al., 2019b;Guerri & Pascual, 2019;, and ex vivo studies report that ethanol induces Toll-like receptor (TLR) 7 signaling that can cause neuronal death . However, the broader relationship between proinflammatory neuroimmune signals and neurodegeneration in the human AUD brain remains unclear.
Accumulating evidence implicates endogenous dangerassociated molecular patterns (DAMPs) and TLRs as factors contributing to neuroinflammation and neural degeneration across neurodegenerative disease states Okun et al., 2009;Paudel et al., 2020). High-mobility group box 1 (HMGB1), which is an endogenous cytokine-like DAMP that signals across and regulates several TLRs, activates TLRs that signal through kinases to induce the proinflammatory nuclear transcription factor nuclear factor kappa-light-chain-enhancer of activated B cells (NFκB). NFκB, in turn, induces transcription of proinflammatory cytokines, chemokines, and other genes associated with innate immune responses (Crews et al., 2011;Liu et al., 2017;Pahl, 1999).
In rodent models, alcohol increases brain expression of HMGB1, TLRs, and proinflammatory genes through NFκB activation, which is accompanied by increased markers of neurodegeneration (Crews et al., 2006;McCarthy et al., 2018;Montesinos et al., 2016;Qin & Crews, 2012a;Vetreno & Crews, 2018;Vetreno et al., 2018). Further, treatment of mice with the inflammagen lipopolysaccharide persistently induces proinflammatory cytokines (e.g., tumor necrosis factor alpha [TNFα], interleukin-1 beta ) and chemokines (e.g., chemokine [C-C motif] ligand 2 [CCL2]), which are accompanied by a delayed, progressive degeneration of substantia nigral dopaminergic neurons, similar to human PD (Qin et al., 2008;Qin et al., 2007). We find similarly increased expression of HMGB1, TLRs 2-4, and CCL2 as well as markers of neurodegeneration in the postmortem human AUD brain (Crews et al., 2013;Qin & Crews, 2012b). However, the association between neuroimmune signaling and neurodegeneration in human AUD is unknown. These findings led us to test the hypothesis that AUD-associated induction of HMGB1 signaling through TLRs induces neuroimmune NFκB-cytokine and chemokine signaling cascades that contribute to neurodegeneration in the human OFC.

Human tissue
Postmortem human OFC paraffin-embedded and frozen tissue samples from moderate drinking control (CON) and AUD subjects Subject information was collected through personal interviews and next-of-kin interviews as well as medical records and is presented in Table 1. The NSW BTRC donor program uses a premortem consent program inviting members of the community to donate their postmortem brain tissue to neuroscience research, allowing collation and study of individuals with a wide variety of clinical histories of AUD and alcohol drinking. Since establishing an accurate alcohol drinking history and age of drinking onset is critical, trained clinical nurses and psychologists from the NSW BTRC performed extensive interviews with the human volunteers and their families. Alcohol drinking history and age of drinking onset information were derived from personal interviews with the volunteers as well as medical records and next-of-kin interviews. In cases where the age of drinking onset was unclear, an age of 25 was recorded (Sheedy et al., 2008). AUD subjects reported an average age of drinking onset of 16.6 (±0.5) years of age, which was compared to age-matched CONs whose average age of drinking onset was 24.5 (±0.5). It is noteworthy that 100% of AUD subjects were able to recall their age of drinking onset, whereas only 10% of CONs were able to provide accurate information. Only individuals with AUD uncomplicated by liver cirrhosis and/or nutritional deficiencies were included in this study. All psychiatric and AUD diagnoses were confirmed using the Diagnostic Instrument for Brain Studies that complies with the Diagnostic and Statistical Manual of Mental Disorders (Dedova et al., 2009

RNA extraction and reverse transcription PCR (RT-PCR)
Total RNA was extracted from frozen human OFC tissue samples from CON and AUD subjects by homogenization in TRI reagent (Sigma-Aldrich) following the single-step method of RNA isolation (Chomczynski & Sacchi, 2006). RNA quality and concentration was determined using a NanoDrop 1000 (Thermo Fisher Scientific).
Total mRNA was reverse transcribed as previously described (Vetreno et al., 2018). RT-PCRs were run on a Bio-Rad CFX system (Bio-Rad). SYBER Green PCR Master Mix (Life Technologies) was used for RT-PCR. RT-PCR was run with an initial activation for 10 min at 95°C, followed by 40 cycles of denaturation (95°C, 15 s), annealing/extension (57-58°C, 1 min), and melt curve. The primer sequences are presented in Table S1. Differences in primer expression between groups are expressed as cycle time (Ct) values normalized with β-actin, and relative differences between groups calculated and expressed as the percent difference relative to CONs.
binding was conducted employing the above-mentioned procedures with omission of the antibody.

Microscopic quantification
BioQuant Nova Advanced Image Analysis software (R&M Biometric) was used for image capture and quantification of immunohistochemistry. Representative images were captured using an Olympus BX50 microscope and Sony DXC-390 video camera linked to a computer. For each measure, the microscope, camera, and software were background corrected and normalized to preset light levels to ensure fidelity of data acquisition. A modified unbiased stereological quantification method was used to quantify TLR9-, MyD88-, IKKβ-, cleaved IL-1β-, CCL2-, and CXCL8-immunoreactive (+IR) cells in the postmortem human OFC. We previously reported that comparison of traditional unbiased stereological methodology with our modified unbiased stereological approach yielded nearly identical values relative to control subjects (Crews et al., 2004). The outlined regions of interest were determined and data expressed as cells/mm 2 .

Fluoro-Jade B immunofluorescent staining
Fluoro-Jade B immunofluorescent data were used from a previously published study (Qin & Crews, 2012b

Fluorescent immunohistochemistry and microscopy
Paraffin-embedded human OFC sections were deparaffinized, washed in PBS, and antigen retrieval performed by incubation in

Statistical analysis
Statistical analysis was performed using SPSS. Sample size determinations were based on previously published studies (Crews et al., 2013;Liu et al., 2020). Two-tailed Student's t tests were used to assess human demographics, RT-PCR, and immunohistochemistry data unless otherwise reported. Levene's Test for

Induction of TLRs and HMGB1 signaling genes in the postmortem OFC of individuals with AUD
Toll-like receptors, originally characterized as receptors for exogenous pathogens, have emerged as neuroimmune signaling receptors in the sterile brain that respond to endogenous agonists, thereby inducing self-expression and amplifying signals across cells through autocrine and paracrine mechanisms. In previous studies, we reported significantly increased protein levels of TLR2, TLR3, TLR4, and TLR7 as well as the endogenous TLR agonist HMGB1 in the AUD brain Crews et al., 2013;Qin et al., 2021). In the present investigation, we assessed induction of the universal TLR agonist HMGB1 and its association with levels of TLRs 1-9 in the postmortem human OFC of individuals diagnosed with AUD (n = 10) relative to age-matched CONs (n = 10).

Increased expression of NFκB family and signaling genes in the postmortem OFC of individuals with AUD
Preclinical studies find alcohol induces and releases HMGB1 that activates TLRs and other neuroimmune receptors, increasing transcription of NFκB proinflammatory genes, including TLRs, within and across cells (Coleman & Crews, 2018;Crews et al., 2017). We  (Table S4).

Increased expression of chemokine and chemokine receptor genes in the postmortem OFC of individuals with AUD
We previously reported that AUD increased CCL2 protein expression in the postmortem human hippocampus, amygdala, VTA, and substantia nigra , and CXCL8 protein expression in the hippocampus . In the present study, we investigated expression of multiple chemokines, including the C-C motif chemokine (β-chemokine) and C-X-C motif chemokine and CCR2 (t(18) = 2.9, p = 0.010) was increased approximately 1.5fold and 2.5-fold, respectively, in the AUD OFC relative to CONs ( Figure 4D), consistent with AUD induction of C-C motif chemokines and their receptors ( Figure 4E).

DISCUSS ION
To our knowledge, this is the first study to report upregulation of a broad number of frontal cortical genes involved in HMGB1-TLR-MyD88-NFκB cytokine and chemokine neuroimmune cascades in the postmortem human OFC of individuals with AUD that we link to neurodegeneration. The RT-PCR and IHC data presented here, which allows for resolution of AUD-induced changes in the postmortem human brain that are observed in preclinical models, complement, and extend prefrontal cortex AUD transcriptome studies (Brenner et al., 2020) as well as molecular and histochemical assessments of proinflammatory neuroimmune induction in the AUD brain (Crews et al., 2013;Qin & Crews, 2012b). We report that AUD induces gene expression of HMGB1 as well as multiple TLRs (i.e., TLRs 2-9), but not TLR1. Toll-like receptors are well-characterized pattern recognition receptors in brain implicated in numerous disease states, including AUD and AD (Coleman & Crews, 2018;Crews et al., 2017;Paudel et al., 2020). HMGB1 is a highly conserved, ubiquitously expressed nuclear protein released from cells in response to injury whereupon it acts as a universal TLR agonist through either direct activation (i.e., TLR4) or indirect activation through formation of heteromers with TLR agonists (e.g., TLR7 and TLR9; Yanai et al., 2009;Andersson & Tracey, 2011;Ivanov et al., 2007). In preclinical studies, alcohol releases HMGB1-let7 miRNA heteromers in microglial vesicles, activating TLR7-induced gene expression and neurodegeneration Qin et al., 2021). Studies in bone marrow-derived DCs find that HMGB1 release with CpG-DNA, which is a TLR9 ligand, enhances TLR9 proinflammatory gene induction (Ivanov et al., 2007).
The ability of TLRs to recognize and respond to endogenous HMGB1 and similar DAMPs results in a sterile inflammatory state wherein innate immune signaling is induced in brain in the absence of invading pathogens (Coleman and Crews, 2018). Indeed, alcohol causes nuclear release and circulation of HMGB1 (Crews et al., 2013). Ligation of HMGB1 to TLRs recruits intracellular adaptor signaling molecules, including MyD88, TRIF, and the viral-mediated TLR adaptor IRF3.
We report that AUD induces MyD88, the downstream NFκBactivating IKK complex, and the NFκB transcription factors NFKB1 and RELA in the OFC, consistent with TLR coupling to MyD88 activating the proinflammatory nuclear factor NFκB and downstream transcription of proinflammatory signaling molecules (Kawai & Akira, 2007;Pahl, 1999;Schmitz & Baeuerle, 1991). Indeed, we previously reported that AUD increases activated pRELA+IR in the human OFC (Qin et al., 2021), consistent with activation of NFκB.
Further, we report that AUD induces proinflammatory cytokines as well as C-C motif and C-X-C motif chemokines and their corresponding receptors, consistent with NFκB transcription of proinflammatory signaling (Pahl, 1999;Schmitz & Baeuerle, 1991). and neurons. Microglia, which are the resident monocyte-like glia of the brain, express many of these proinflammatory signaling genes, and studies in postmortem human AUD brain  and preclinical rodent models  find alcohol exposure increases ramified microglia populations, consistent with microglial activation and induction of proinflammatory cytokines and chemokines. While microglial depletion in mice blocks acute alcohol induction of TNFA and TLR7, it does not block HMGB1, TLR2, TLR3, TLR4, IL1B, IL6, or CCL2, suggesting that these latter genes may be independent of microglia . However, in a recent study, media transfer from ethanol-treated organotypic hippocampal slice cultures (OHSCs) containing microglial microvesicles to naïve OHSCs induced TNFA and other proinflammatory cytokines that was blunted in microglialdepleted ethanol-treated media consistent with microglial vesicular signaling contributing to some aspects of alcohol-induced neuroimmune activation . Further, while microglia depletion prevents escalations voluntary alcohol intake and concomitant anxiety-like behavior in alcohol dependent mice, it does not affect voluntary ethanol intake in nondependent mice supporting a role for microglia in alcohol dependence (Warden et al., 2021;Warden et al., 2020). The observation that numerous proinflammatory neuroimmune genes are unaltered following microglial depletion in vivo suggests that astrocytes and neurons express higher levels of neuroimmune mediators than previously known. Indeed, astrocytes and neurons are emerging as important contributors to neuroimmune responses (Erickson et al., 2021;Lawrimore et al., 2019;Lawrimore & Crews, 2017;Moffat & Ron, 2021) and a recent cell type-specific transcriptome study in the postmortem human brain revealed unique neuroimmune gene signatures across glia and neurons with robust changes in astrocytes (Brenner et al., 2020;Erickson et al., 2019a;Soreq et al., 2017). While HMGB1 is enriched across all cell types, TLR2, TLR7, and TLR9 are particularly enriched in microglia at baseline, whereas TLR3 is enriched in astrocytes and the remaining TLRs are modestly expressed across glia and neurons (Soreq et al., 2017).
However, alcohol exposure can cause induction in other cell types.
In the present study, we report TLR9 co-expression with microglia as well as astrocytes and neurons, consistent with previous studies reporting colocalization of TLR2, TLR3, and TLR4 with neurons in the postmortem human brain (Casula et al., 2011;Crews et al., 2013;Maroso et al., 2010;Zurolo et al., 2011). All TLRs, except TLR3, signal through the TLR adaptor protein MyD88, which is expressed across cell types but is particularly enriched in microglia and astrocytes (Brenner et al., 2020;Soreq et al., 2017). Alcohol releases HMGB1 from neurons and microglia (Crews et al., 2013;Lawrimore & Crews, 2017), and we find HMGB1 is positively correlated with all induced TNFα, but not IL-1β (Horng et al., 2002;Kawai et al., 2001). Members of the NFκB-activating IKK complex are expressed in neurons, astrocytes, and microglia in the human brain, whereas NFKB1 and RELA are similarly expressed across cell types, but particularly enriched in microglia (Soreq et al., 2017). and TNFA did not correlate with TLR expression, suggestive of different signaling mechanisms (Horng et al., 2002;Qin et al., 2007).
Taken together, these data suggest that induction of HMGB1-TLR-MyD88-NFκB cytokine and chemokine signaling in the human OFC involves a coordinated cascade of neuroimmune signaling involving microglia as well as astrocytes and neurons.
Proinflammatory neuroimmune induction and neurodegeneration are hallmark features of neurodegenerative disorders and addiction Kempuraj et al., 2016;Kohno et al., 2019). Human neuroimaging studies report decreased OFC volume and brain connectivity (Moorman, 2018) (Qin & Crews, 2012b) with neuroimmune signaling molecules colored for significance in the postmortem human OFC. Red-labeled correlations indicate statistically significant correlation p value < 0.001; green-labeled correlations indicate statistically significant correlation p value < 0.01; blue-labeled correlations indicate statistically significant correlation p value < 0.05 after correction for multiple comparisons using Benjamini-Hochberg procedure (B-H Critical) for controlling false positives. Black-labeled HMGB1, IKKβ+IR, TNFA, and CCL8 correlated with Fluoro-Jade B+IR but did not achieve significance following correction for multiple comparisons using B-H Critical. Alcohol causes nuclear release of the ubiquitously expressed universal TLR agonist high-mobility group box 1 (HMGB1) (Crews et al., 2013) that binds to and activates TLRs located on glia and neurons (Crews et al., 2013). Microglia, the resident monocyte-like cell of the CNS, express all Toll-like receptors (TLRs) under basal conditions and contribute to initiation of proinflammatory neuroimmune cascades Soreq et al., 2017). Ligation of HMGB1 to all TLRs, except TLR3, recruits the intracellular TLR coupling protein MyD88, activating downstream activation and nuclear translocation of the proinflammatory mediator nuclear factor kappa B (NFκB) and transcription of proinflammatory cytokines and chemokines. These data implicate induction of HMGB1-TLR-MyD88-NFκB cascades through coordinated glial and neuronal signaling as contributing to neurodegeneration in the postmortem human OFC of individuals with AUD [Color figure can be viewed at wileyonlinelibrary.com] . The AUD-induced increases of HMGB1-TLR-MyD88-NFκB cytokine and chemokine neuroimmune signaling are likely the result of both alcohol drinking and alcohol-induced neurodegeneration. These findings are consistent with chronic alcohol exposure progressively inducing persistent HMGB1-TLR-MyD88-NFκB cytokine and chemokine neuroimmune signaling that may contribute to the modest and diffuse neurodegeneration associated with AUD relative to AD and other neurodegenerative disorders that may sensitize the brain to age-related neurodegeneration and disease.
While neurodegeneration is an important endpoint in this study and neuroimmune signaling molecules are expressed by glia and neurons, accumulating evidence implicates neuroimmune induction in regulating brain function. Neuroimmune signaling is thought to contribute to increased alcohol drinking and drug addiction as knockout of many key neuroimmune genes, including TLR4, IL6, and CCL2, decreases alcohol consumption in mice (Blednov et al., 2005;Blednov et al., 2012). Interestingly, we report that an earlier age of drinking onset negatively correlates with HMGB1-TLR-MyD88-NF-κB cytokine and chemokine signaling, likely contributing to the increased risk for adult AUD development in individuals that start drinking at a younger age (Dawson et al., 2008;Grant, 1998).
Neuroimmune induction also appears to contribute to the withdrawal and negative affect associated with AUD as brain levels of several proinflammatory signaling molecules, including TNFα and CCL2, are increased during alcohol withdrawal in mice (Qin et al., 2008), and ventricular injection of TNFα or CCL2 potentiates alcohol-induced withdrawal-associated negative affect in rats (Breese et al., 2008). Further, in humans with AUD, plasma levels of several circulating cytokines (e.g., TNFα, IL-1β, IL-6) and chemokines (e.g., CXCL8) correlate with alcohol craving (Heberlein et al., 2014;Leclercq et al., 2014). Inhibition of amygdalar IL-1β reduces bingelike alcohol drinking (Marshall et al., 2016). Similarly, inhibition of IL-1β in the ventral tegmental area prevents cocaine-induced dopamine release in the nucleus accumbens (Northcutt et al., 2015) whereas neuronal NFκB is essential for amplification of cocaine addiction (Russo et al., 2009), further implicating neuroimmune signaling in addictive behaviors. Recently, AUD and alcohol abuse in humans has emerged as a potential etiological factor contributing to the onset of dementia later in life (Kamal et al., 2020). Thus, our finding of AUD-induced increases of HMGB1-TLR-MyD88-NFκB cytokine and chemokine neuroimmune signaling suggests this signaling cascade should be considered in studies linking innate immune signaling to mood, cognition, and addiction.
In conclusion, we report induction of the universal endogenous TLR agonist HMGB1 in the postmortem human OFC of individuals with AUD that parallels increases of all TLRs (i.e., TLR2-TLR9) except TLR1. Induction of HMGB1-TLR signaling is accompanied by increased expression of the TLR adaptor protein MyD88, activation of the proinflammatory nuclear transcription factor NFκB, and downstream induction of proinflammatory cytokines, chemokines, and their corresponding receptors. Expression of several of these proinflammatory signaling markers is observed in glia and neurons. Induction of the HMGB1-TLR-MyD88-NFκB proinflammatory signaling pathways correlates with neurodegeneration (i.e., Fluoro-Jade B) as well as lifetime alcohol consumption and age of drinking onset. These data implicate induction of HMGB1-TLR-MyD88-NFκB cascades through coordinated glial and neuronal signaling as contributing to neurodegeneration in the postmortem human OFC of individuals with AUD.

ACK N OWLED G EM ENT
We thank Jennie Vaughn for assistance editing the manuscript.

CO N FLI C T S O F I NTE R E S T
None.