Abstract

   Alcohol consumption induces hepatocyte damage through complex processes
   involving oxidative stress and disrupted metabolism. These factors
   alter proteomic and epigenetic marks, including alcohol-induced protein
   acetylation, which is a key post-translational modification (PTM) that
   regulates hepatic metabolism and is associated with the pathogenesis of
   alcohol-associated liver disease (ALD). Recent evidence suggests lysine
   acetylation occurs when a proximal cysteine residue is within ∼15 Å of
   a lysine residue, referred to as a cysteine-lysine (Cys-Lys) pair.
   Here, acetylation can occur through the transfer of an acetyl moiety
   via an S → N transfer reaction. Alcohol-mediated redox stress is known
   to occur coincidentally with lysine acetylation, yet the biochemical
   mechanisms related to cysteine and lysine crosstalk within ALD remain
   unexplored. A murine model of ALD was employed to quantify hepatic
   cysteine redox changes and lysine acetylation, revealing that alcohol
   metabolism significantly reduced the cysteine thiol proteome and
   increased protein acetylation. Interrogating both cysteine redox and
   lysine acetylation datasets, 1280 protein structures generated by
   AlphaFold2 represented by a 3D spatial matrix were used to quantify the
   distances between 557,815 cysteine and lysine residues. Our analysis
   revealed that alcohol metabolism induces redox changes and acetylation
   selectively on proximal Cys-Lys pairs with an odds ratio of 1.88
   (p < 0.0001). Key Cys-Lys redox signaling hubs were impacted in
   metabolic pathways associated with ALD, including lipid metabolism and
   the electron transport chain. Proximal Cys-Lys pairs exist as sets with
   four major motifs represented by the number of Cys and Lys residues
   that are pairing (Cys[1]:Lys[1], Cys[x]:Lys[1], Cys[1]:Lys[x] and
   Cys[x]:Lys[x]) each with a unique microenvironment. The motifs are
   composed of functionally relevant Cys-Ly altered within ALD,
   identifying potential therapeutic targets. Furthermore, these unique
   Cys-Lys redox signatures are translationally relevant as revealed by
   orthologous comparison with severe alcohol-associated hepatitis (SAH)
   explants, revealing numerous pathogenic thiol redox signals in these
   patients.

   Keywords: Cysteine proteomics, Redox, Acetylation, Protein modeling,
   Alcohol-associated liver disease, Mass spectrometry

Graphical abstract

   [39]Image 1
   [40]Open in a new tab

1. Introduction

   Alcohol is a toxic and habit-forming substance that is consumed by over
   62 % of people in the United States [[41]1]. Worldwide, chronic
   consumption of alcohol contributes to disease etiology in approximately
   46 % of patients with liver disease and alcohol use is among the
   leading causes of preventable death [[42]2]. Unfortunately, aside from
   abstinence and liver transplantation there are no approved therapeutic
   interventions [[43]3]. The challenges impeding the development of
   therapeutics for alcohol-associated liver disease (ALD) are
   multifaceted, encompassing disease heterogeneity among individuals with
   ALD adding to the complexity of clinical trials, late stage diagnosis,
   and limited biochemical understanding of early stages of ALD
   progression [[44]4]. Further characterization of biochemical changes
   driven by alcohol metabolism will provide therapeutic advances
   ameliorating hepatic alcohol toxicity.

   The pathological onset and development of ALD includes the disruption
   of many biochemical processes resulting in metabolic, genetic, and
   proteomic alterations [[45]3]. One such result attributed to metabolic
   disruption, is the robust increase in the post-translational
   modification (PTM), lysine (Lys) acetylation, driven by alcohol
   metabolism [[46][5], [47][6], [48][7]]. Like many other PTMs, Lys
   acetylation is critical for regulating a host of metabolic and cellular
   functions, such as cellular respiration, fatty acid handling and
   signaling [[49]8]. The increase of Lys acetylation has been well
   characterized within ALD models and is associated with metabolic
   disruption [[50]9,[51]10]. Furthermore, a recent thiol-centric
   proteomic analysis in ALD revealed the global increase of hepatic
   protein cysteine (Cys) residues in their thiol/thiolate form (reduced)
   [[52]11]. This study mapped fundamental protein thiols and identified
   locations of redox perturbation attributed to alcohol metabolism which
   can be used to extrapolate the response of redox regulated enzymes. The
   observed Cys reduction confirms that the hepatic proteome responds to
   alcohol metabolism through multiple routes of PTM signaling. One unique
   mechanism of modification is the non-enzymatic transfer of an acetyl
   moiety from a Cys thiol to a Lys ε-amine [[53]12]. Our recent findings
   of increased Lys acetylation and a reduced thiol proteome suggests
   cysteine-lysine (Cys-Lys) proximity may be a key metabolically derived
   feedback loop regarding Cys redox changes and Lys acetylation in
   response to alcohol toxicity [[54]10,[55]11].

   Emerging research suggests Lys acetylation can be facilitated by the
   proximity of an acetylated Cys thiol [[56]13]. More specifically, Lys
   acetylation was found to be significantly increased when a Lys is
   within 15 Å of a Cys, termed a proximal Cys [[57]12,[58]13]. Here, the
   mechanism of acetylation is proposed to occur due to the Cys thiol
   attacking the thiol ester carbonyl of the acetyl moiety of acetyl
   coenzyme A (Ac-CoA), resulting in an S-acetylation intermediate
   followed by S → N transfer of the acetyl moiety to a nearby Lys ε-amine
   [[59]12]. Due to pK[a] ∼8.3–8.5 Cys thiol residues more readily react
   with Ac-CoA molecules; thus, stabilizing the acetyl moiety and enabling
   the Lys ε-amine to react with the thiol ester carbonyl rather than
   Ac-CoA(12). This concept has been further supported by the in vitro
   identification of the proximity of nucleotide binding sites increasing
   Lys acetylation [[60]14]. Further, the structural similarities between
   co-factors enables association with similar binding sites resulting in
   non-discriminate CoA binding and enabling acyl modifications on Lys
   proximal to that binding pocket [[61]12,[62]13,[63]15]. Here, we are
   structurally focused on alcohol-related changes to Cys and Lys residues
   by evaluating the distance between the thiol sulfur atom (S^−) and the
   ε-amine nitrogen (NH[3]^+) atom across three-dimensional (3D) protein
   structures, while noting co-factor binding affiliation. Consequently,
   uncovering Cys-Lys residues significantly impacted by alcohol
   metabolism expands the scope of enzymes critical to ALD pathology.

   Here, we present the first analysis integrating hepatic quantitative
   Cys redox proteomics and Lys acetylation to experimentally determine
   proximal Cys-Lys pairs. Evaluating Cys-Lys relationships within a model
   of ethanol consumption is critical to further understanding hepatic
   alcohol-induced biochemical changes and the interplay between PTMs and
   protein function. The implementation of a 6-week Lieber-DeCarli model
   recapitulates early stages of ALD pathology and enables the examination
   of proteomic changes occurring during alcohol exposure and the onset of
   ALD(7). The integrated evaluation of Cys redox and acetyl-Lys proteomic
   datasets was coupled with cutting-edge predictive protein structure
   modeling which revealed novel spatial interactions of Cys-Lys amino
   acids and their modifications. StructureMap enabled the application of
   structural data from the AlphaFold2 artificial intelligence program
   [[64]16,[65]17]. The integration of quantitative proteomic data and
   structural information revealed an increase in the likelihood of a
   change in Cys redox status when Cys and Lys residues are within 15 Å.
   Furthermore, this study identified a trend among these proximal Cys-Lys
   pairs where cofactor binding, such as metal or NAD^+, is associated
   with modifications [[66]14,[67]15]. Finally, database mining across a
   cohort of severe alcohol-associated hepatitis (SAH) patients revealed
   numerous orthologous proteins containing regulatory Cys-Lys pairs.
   Supporting that specific translational Cys-Lys pairs may be key
   therapeutic targets for the treatment of end-stage liver disease
   [[68]18].

2. Materials and methods

2.1. Animals

   All animal-related experiments were approved by the Institutional
   Animal Care and Use Committee of the University of Colorado and were
   performed in accordance with published National Institutes of Health
   guidelines. Healthy male C57BL/6J mice were purchased from The Jackson
   Laboratory (Bar Harbor, ME) and housed in a temperature-controlled
   (20–22 °C) room on a 6am to 6pm light/dark cycle. 12-week-old mice were
   fed a modified Lieber-DeCarli liquid based-diet (Bio-Serv, Frenchtown,
   NJ) for 6 weeks. The diet consisted of 44 % fat-derived calories, 16 %
   protein-derived calories and the remaining balance was comprised of
   either carbohydrate or alcohol-derived calories. Control (n = 5) and
   alcohol (n = 5) mice were pair-fed where alcohol-derived calories were
   replaced with calories from the carbohydrate source (maltose-dextrin).
   Mice in the alcohol group began the study at 2 % alcohol (v/v) and the
   concentration of alcohol was increased every week until 6 % alcohol
   (v/v) for the final week as previously described [[69]11].

2.2. Cysteine proteomic analysis

   The click chemistry Cys redox assay used was adapted from Yang et al.
   [[70]19] and Pertova et al. [[71]20] to quantify protein Cys residues
   that underwent a redox change due to alcohol metabolism by nHPLC-MS/MS.
   This assay utilized an isotopically labeled biotin-azide to “click” the
   free protein thiols of reduced Cys modified by alkynyl iodoacetamide.
   The pair-fed control and alcohol-fed murine hepatic peptides were
   combined 1:1 and were used to generate an accurate mass and retention
   time (AMRT) library [[72]11]. General protein quantitation (GPQ) was
   performed on the individual samples and used to normalize Cys redox
   proteomic abundance changes. MS quantitation data was extracted and
   aligned for both the protein quantitation and the Cys proteomic
   analysis as described elsewhere [[73]11].

2.3. Acetylomic sample preparation

   A different piece of hepatic tissue from the same mice assayed in the
   click chemistry Cys redox experiment (n = 5, per group) was homogenized
   by bead milling using a Qiagen TissueLyser LT set to 50Hz for 6 min.
   Samples were homogenized in a 10:1 diluted urea lysis buffer as
   outlined in the PTMScan HS Acetyl-Lysine Motif kit (Ac–K kit) (Cell
   Signaling Technology #46784) at a ratio of 10 μL/mg of tissue (Danvers,
   MA). Samples were spun down in a microcentrifuge at room temperature to
   pellet any remaining tissue debris not solubilized and then a BCA
   protein assay was performed on the supernatant. Equivalent 4 mg
   aliquots of protein were taken from each sample. One hundred nanograms
   of Acetyl BSA internal standard was added to each sample followed by
   reduction, alkylation, and digestion with trypsin at 1:60
   enzyme:protein for 18 h. After digestion, samples were desalted on
   C[18]SPE columns as previously described [[74]7,[75]10]. The peptide
   eluent was divided to yield four 1 mg aliquots and dried down in 2 mL
   microcentrifuge tubes using a Savant SPD1010 SpeedVac Vacuum
   Concentrator under reduced pressure. One aliquot was used for general
   protein quantitation and another aliquot was used for
   Immunoprecipitation (IP) of acetyl-Lys peptides. The IP and subsequent
   desalting of Immunoprecipitated peptides using Piece™ C18 Spin Tips
   (Thermo Scientific, PN 84850) was performed exactly as stated in steps
   III and IV of the protocol provided with the Ac–K kit. Acetyl peptides
   were dried down via SpeedVac and stored at −80 °C for LC-MS/MS analysis
   (Waltham, MA).

2.4. General protein quantitation

   GPQ sample aliquots from the acetylomic sample preparation were
   suspended in a 3 % acetonitrile with 0.1 % formic acid solution
   containing 15 fmol/μL of Procal Spike Mix Retention Time Standards
   (TRK-1-10pmol) to control for instrument variability [[76]21]. Samples
   were then loaded onto a 2 cm PepMap 100, nanoviper trapping column and
   chromatographically resolved on-line using a 0.075 × 150 mm, 3.0 μm
   EASY-Spray PepMap RSLC C[18] reverse phase nano column and an Ultimate
   3000 RSCLnano LC system (Thermo Scientific, Waltham, MA). Mobile phases
   consisted of water +0.1 % formic acid (A) and 100 %
   acetonitrile + 0.1 % formic acid (B). Samples were loaded onto the
   trapping column at 5.0 μL/min for 4.5 min at initial condition before
   being chromatographically separated at a flow rate of 300 nL/min using
   a gradient of 3–7% B over 2.5 min, 7–33 % B over 93 min for a total
   95.5 min gradient at 40 °C. The gradient method was followed by a
   column wash at 70 % B for 5 min. Data was collected on an Orbitrap
   Eclipse (Thermo Scientific) equipped with an EASY-Spray Source operated
   using intensity dependent Collision induced dissociation (CID) MS/MS to
   generate peptide ID's. MS^1 spectra were acquired in the Orbitrap
   (Resolution = 120k; AGQ target = 100 %; MaxIT = Auto; RF Lens = 30 %;
   mass range = 350–1500; Profile data) (Waltham, MA). Precursors selected
   for MS/MS were filtered by mono isotopic precursor selection (MIPS)
   model set to peptide with an intensity threshold of 15000 and only
   charge states 2–6 were allowed. Dynamic exclusion was employed for 10 s
   and the dependent scan for a single charge state per precursor only was
   set to true. MS^2 spectra were collected using CID in the linear ion
   trap with a cycle time of 1 s between MS Orbitrap scans (Isolation
   window = 1.2 m/z [quadrupole], rate = rapid; AGQ target = Standard;
   MaxIT = Dynamic; CID = 35 %).

   Data were searched and extracted using SEQUEST HT and the label-free
   quantitation workflow in Proteome Discover (PD) software version
   2.5.0.400 that includes the minora feature detector, feature mapper,
   and precursor ions quantifier algorithms. Spectra were searched against
   a combined Procal Retention Time (RT) standard peptide database and the
   SwissProt Mus musculus database downloaded July 17th, 2023, allowing up
   to 2 missed tryptic cleavages with fixed carbamidomethylation
   (cysteine) and dynamic deamidation (asparagine or glutamine) and
   oxidation (methionine) modifications. The monoisotopic peptide mass
   tolerance allowed was ±10.0 ppm and the MS/MS tolerance was ±0.5Da.
   P-values related to peptide-spectrum matches were adjusted to a 1 %
   false discovery rate (FDR) using the percolator algorithm.
   Additionally, only high confidence peptides were allowed for protein
   scoring and a protein FDR cut-off of 1 % was also used. A signal to
   noise (S/N) threshold of 1.5 was set for minora feature detector and at
   least 6 points had to be found across a peak. Peptide spectrum matches
   (PSMs) confidence levels were set to high for feature ID linking. A
   coarse retention time (RT) alignment of data was performed using vendor
   suggested default values. Following coarse RT alignment, the retention
   time tolerance was automatically determined by PD software based on
   coarse aligned data files, mass tolerance of 10 ppm, minimum S/N
   threshold of 5 were set for feature linking and mapping. Precursor
   quantification was based on area and protein abundances were estimated
   by summing the abundances of all peptides matched to a specific protein
   using the Unique + Razor setting in PD software. The total Procal
   summed abundance for each sample was used to adjust protein abundances
   of all other identified and quantified proteins in that sample to
   control for instrument variability. The GPQ Procal corrected protein
   abundances were then used to adjust acetyl peptide abundance on a per
   protein and sample basis.

2.5. Acetylomic analysis by LC-MS/MS

   Dried immunoprecipitated peptides from the acetylomic sample
   preparation were re-suspended in 3 % acetonitrile 0.1 % formic acid
   solution with 15 fmol/μL of Procal Spike Mix. LC-MS/MS data was
   acquired on one third of each acetyl sample using the same method
   parameters as the GPQ data acquisition above with the following
   exceptions: a shorter gradient was used for peptide separation, the ion
   trap scan rate was set to Normal, and maximum injection time type was
   set to Auto. The shorter gradient used is described as follows: Samples
   were loaded onto the trapping column at 5.0 μL/min for 5 min at initial
   condition before being chromatographically separated at a flow rate of
   300 nL/min using a gradient of 3–7% B over 3 min, 7–26 % B over 40 min,
   and 26–33 % B over 4 min for a total 47 min gradient at 40 °C. The
   gradient method was followed by a column wash at 70 % B for 4 min.

   Data were searched and extracted using the same parameters as GPQ data
   with the following exceptions: Up to 4 missed tryptic cleavages were
   allowed, acetylation (lysine) was an allowed dynamic modification, and
   spectra were searched against a combined Procal RT standard peptide
   database, SwissProt Mus musculus database downloaded July 17th, 2023,
   and bovine serum albumin (BSA) database. Protein level summed area
   abundances of Procal peptides and spiked in acetylated BSA peptides
   were used to correct all identified and quantified acetyl peptide area
   abundances on a sample per sample basis. First, the total sum of Procal
   area abundance within each sample was used to correct the total sum of
   spiked acetylated BSA peptide abundances within each sample. A
   normalization factor was then generated for each sample using the total
   Procal adjusted acetylated BSA signal found within each sample and
   these factors were used to normalize acetyl peptide levels within each
   sample. The Procal adjusted acetylated BSA normalized acetyl peptide
   abundance data was then normalized on a protein/sample specific basis
   based on Procal adjusted general protein quantitation data abundances.
   GPQ sample E5 was used as the reference data file for GPQ factors for
   each protein. To summarize, Procal standards were used to control for
   run-to-run variability, acetyl BSA was used to control for sample
   preparation variation, and general protein quantitative data was used
   to adjust for protein abundance differences between each sample.

   Statistical analysis was then performed in Mass Profiler Professional
   software V.15.1 (Agilent Technologies) on adjusted acetyl peptide data
   using an unpaired t-test with Benjamini-Hochberg multiple testing
   correction (Santa Clara, CA). A corrected p-value (p (Corr)) ≤ 0.05 and
   log[2] fold change (FC) cut-off of |0.58| were used to determine
   statistical significance of acetylated peptides [[77]22].

2.6. Cysteine-lysine proximity and structural analysis

   1280 unique UniProt IDs were used to extract CIF (crystallographic
   information files) and PAE (predicted aligned error) files from the
   AlphaFold Protein Structure Database version 4 (AlphaFold DB accessed
   03.09.2024; [78]http://alphafold.ebi.ac.uk) [[79]17]. The site-directed
   PTM proximity analysis evaluated the distance between amino acid
   residues experimentally determined to be acetylated or redox changed
   with predicted proteins structures using Python Program StructureMap
   described in Bludau et al.
   ([80]http://github.com/MannLabs/structuremap) [[81]16]. Peptides with
   multiple modified sites were stratified into individual sites, and the
   FC values were averaged. The lowest corrected p-value associated with
   each individual site was used for subsequent analysis. The Lys amine
   and the Cys thiol were assigned coordinates on experimentally pertinent
   proteins then the proximity of all corresponding Lys and Cys residues
   was quantified to identify all possible Cys-Lys pairs
   ([82]https://github.com/FritzLaboratory). Amino acid residues with a
   quality score below 70 % were excluded according to AlphaFold's
   confidence metrics. Additionally, amino acid residues with a PAE
   greater than 15 Å were excluded from this analysis because 15 Å is
   predicted to be necessary for biological interactions [[83]13]. As a
   novel technology AlphaFold2 has several limitations due to current
   modeling capacity. For example, in this analysis, 20 proteins from our
   datasets were unable to be predicted and/or represented by a distance
   matrix by AlphaFold2. Additionally, only 70 % of residues had locations
   predicted with sufficient accuracy to be included in this analysis. The
   ability to accurately predict protein structure by AlphaFold2 is
   dependent on reliable homology protein sequences and preexisting
   structures suggesting that less studied proteins are more likely to be
   excluded from this analysis. Furthermore, the structures present in the
   AlphaFold DB are limited to their monomeric form therefore we were
   unable to analyze potential Cys-Lys pairs between multimeric
   proteoforms [[84]17,[85]23].

   The 3D structure visualization of AlphaFold2 DB predicted structures
   was conducted in Discovery Studio 2021 (BIOVIA). AlphaFold2 DB protein
   structures visualized in this manuscript include: glutaredoxin-related
   protein 5 (Glrx5) (AF-[86]Q80Y14–F1), enoyl-CoA hydratase 1 (Echs1)
   (AF-[87]Q8BH95–F1), peroxisomal 2,4-dienoyl-CoA reductase (Decr2)
   (AF-[88]Q9WV68–F1) and alcohol dehydrogenase (Adh1) (AF-[89]P00329-F1).

2.7. Pathway analysis of acetylated and redox-modified protein

   Peptides identified by the acetylomic analysis were condensed to the
   protein level by combining peptide abundance values and the inverse
   fold change was used due to the common association of acetylation as an
   inhibitory modification [[90]24]. Additionally, peptides from the Cys
   redox proteomic analysis were condensed to the protein level by
   combining modified peptide abundance values. All proteomic analyses
   were performed on the same murine liver samples. Ingenuity Pathway
   Analysis (IPA) Software (QIAGEN) and protein IDs were used to identify
   associated canonical pathways and toxicity phenotypes. Expression
   analysis with a right-tailed Fisher's exact test with
   Benjamini-Hochberg multiple testing correction was used to identify
   significantly enriched pathways. A corrected p-value of <0.05 was
   considered statistically significant. Activation and inhibition
   z-scores were calculated by IPA's z-score algorithm to predict
   significant canonical pathway activation (z-score>2) or inhibition
   (z-score < −2). The UniProt application programming interface (API)
   (version 2024_03) was queried to identify the subcellular location and
   molecular function annotation for all proteins found in both the
   acetylomic and redox proteomic datasets (see [91]Supplemental 1 Cys Lys
   StructureMap). Proteins associated with one or more subcellular
   locations or functions were summarized for all possibilities. The motif
   enrichment analysis displayed in [92]Table 1 was conducted with
   Database for Annotation, Visualization and Integrated Discovery (DAVID)
   (see [93]Supplemental 3 CysLys Motif). CysLys orthologous subcellular
   location, canonical pathway enrichment, and metabolic pathways were
   identified by DAVID and substrate usage were determined by Reactome
   ([94]https://reactome.org/) [[95]25,[96]26].

Table 1.

   Description of Cys-Lys pair motif and associated subcellular
   compartments, biological processes, enzymatic class, and the number of
   Cys-Lys pairs in each motif.
   Cys[1]:Lys[1] Cys[x]:Lys[1] Cys[1]:Lys[x] Cys[x]:Lys[x]
   Subcellular location Mitochondria Mitochondria Cytoplasm Mitochondria
   Biological process Metabolic pathways Metabolic pathways Coronavirus
   disease - COVID-19 Carbon metabolism
   Enzymatic class Oxidoreductase Oxidoreductase, Metal Ion Binding
   Protein binding, nucleotide binding Metal Ion Binding
   # of Cys-Lys Pairs 345 178 419 23
   [97]Open in a new tab

2.8. Human ortholog identification

   Murine proteins containing Cys-Lys pairs with significantly modified
   residues were cross-referenced with a previously published human SAH
   explant acetylomic dataset [[98]27]. The BioMart R package was
   implemented to access Ensembl 111: Jan 2024
   ([99]https://www.ensembl.org) [[100]28]. UniProt version 2024_03 was
   used to align ADH1 amino acid sequences ([101]https://www.uniprot.org/)
   [[102]29].

2.9. Immunoblotting

   30 μg of murine hepatic whole tissue lysate was prepared by adding 4x
   Laemmli loading buffer (Bio Rad) then denatured by boiling sample for
   5 min at 92 °C. Protein was separated by sodium dodecyl sulfate
   polyacrylamide gel electrophoresis (Waltham, MA). Total protein
   abundance was determined using 2,2,2-trichloroethanol (Sigma-Aldrich)
   and used to normalize the signal from specific antibodies (St. Louis,
   MO). The anti-acetylated-lysine antibody was purchased from Cell
   Signaling Technology (#9441) and was used at a 1:1000 dilution
   (Danvers, MA). Blots were visualized using a Bio-Rad ChemiDoc MP
   Imaging Systems and analyzed using Bio-Rad ImageLab 6.0.1 software. For
   immunoblot densitometry, graphs represent the average within each group
   (n = 5, per group) with error bars representing the within group
   standard deviation.

2.10. Statistical analysis

   Proteins were divided into four mutually exclusive groups 1) proteins
   with at least one peptide that contained significantly (p (Corr)≤0.05,
   log[2]FC≥|0.58|) acetylated/deacetylated Lys and at least one peptide
   having Cys redox significantly (p (Corr)≤0.05, log[2]FC≥|0.58|)
   changed, 2) proteins containing a peptide with a significantly
   acetylated/deacetylated Lys and no peptides with a significantly
   changed Cys, 3) proteins with a peptide containing a significantly
   changed Cys, and 4) proteins which have no significantly modified
   peptides. At the protein level, a Fisher's Exact test was used to
   determine if more co-occurring alcohol-induced peptide modifications on
   the same protein ([103]Supplemental Tables 1 and 2) were observed than
   were expected by chance. For all Cys-Lys pairs within 15 Å of each
   other, a Fisher's Exact test and odds ratio (OR) were used to compare
   the odds of Lys acetylation between pairs with a modified Cys and pairs
   without a modified Cys ([104]Supplemental Table 3). For all the
   significant PTM (acetylated Lys and modified Cys), distances to the
   closest proximal Cys (for Acetyl-Lys) and Lys (for Cys Mod) residue
   were also examined ([105]Fig. 4C). For immunoblot densitometry, graphs
   represent the within treatment group average (n = 5, per group) with
   error bars representing the standard error of the mean (SEM).
   Statistical analyses and graphical representations were completed and
   generated by GraphPad Prism (10.1.0) and R Statistical Software
   (9.0)/RStudio 4.3.3 (2023.12.1 + 402).

Fig. 4.

   [106]Fig. 4
   [107]Open in a new tab

   StructureMap was used to determine Cys and Lys proximity of AlphaFold2
   predicted protein structures found to be modified by alcohol. A)
   Proteins found to be modified by alcohol metabolism were used in the
   Python program StructureMap to evaluate the distance between Cys and
   Lys residues on predicted protein structures. The program mapped 1281
   of the 1300 proteins input and quantified the distance between 557,815
   Cys-Lys pairs. Adh1 (AF-[108]P00329-F1) demonstrated the use of
   AphaFold2 predicted structure to map amino acid residues. B) Breakdown
   of Cys residues and Lys residues within the proteins mapped by
   StructureMap either modified or unmodified containing circles
   representative of approximately 450 amino acids. C) Bar plot depicting
   number of Cys residues with a redox change and distance to nearest Lys
   residue and acetylated Lys residue and the distance to nearest Cys
   residue. Acetyl-Lys: Acetylated Lysine; Adh1: Alcohol dehydrogenase;
   Cys-Redox: Cysteine redox proteome.

3. Results

3.1. Characterization of alcohol-induced hepatic acetylome modifications

   The analysis of whole liver acetylomic profiles revealed significant
   differences in murine hepatic protein acetylation due to alcohol
   metabolism. A total of 2266 peptides from 1006 proteins were found to
   be acetylated at one or more Lys sites. Of those, 1373 peptides were
   significantly acetylated in response to chronic alcohol consumption and
   contained an acetylated Lys with a log[2] difference greater than 0.58
   (i.e., 50 % higher acetylation in the alcohol group compared to the
   control group) (dark green [109]Fig. 1A). In contrast, 73 peptides were
   found to be significantly deacetylated with chronic alcohol consumption
   (dark yellow [110]Fig. 1A). Additionally, 120 peptides were found to
   have increased acetylation by 50 % or more (log[2](FC)≥0.58 difference)
   and 38 were found to have decreased acetylation (log[2](FC) ≤ -0.58
   difference) but were not significantly changed by the multiple
   comparison-corrected threshold (p (Corr)≤0.05) (light green and light
   yellow, respectively [111]Fig. 1A). The phenomenon of more acetylation
   than deacetylation due to alcohol in the acetylomics was validated by
   immunoblot, observing a 1.92-fold increase in whole cell liver lysates
   ([112]Supplemental Fig. 1). The IPA Toxicity Function analysis of
   proteins found to be acetylated or deacetylated examined associations
   with pathological endpoints and identified diseases linked with chronic
   alcohol overconsumption, such as liver steatosis and hepatocellular
   carcinoma ([113]Fig. 1B) ([114]Supplemental 2 IPA Acetylome and
   CysLys). The canonical pathway with the highest enrichment score (i.e.,
   largest -log[10](p (Corr)) was mitochondrial dysfunction which is
   associated with the development of hepatic lipid accumulation
   ([115]Fig. 1C) [[116]6]. Proteins driving this association were
   primarily components of the electron transport chain (ETC) and are
   involved in ATP synthesis, the second most enriched pathway. Proteins
   driving the enrichment included multiple subunits from Complex I,
   Complex II, Complex III, Cytochrome C, Complex IV, and ATP synthase
   ([117]Fig. 2). Eighteen subunits of Complex I were found to be
   acetylated with NDUFS2 having the largest fold change, indicating key
   acetyl-mitochondrial disruptions in energy homeostasis through the
   electron transport chain during alcohol toxicity ([118]Fig. 2).

Fig. 1.

   [119]Fig. 1
   [120]Open in a new tab

   Characterization of alcohol-induced hepatic acetylome modifications. A)
   This volcano plot depicts acetylated (dark green log[2](FC)≥|0.58|,
   -log[10](p (Corr))>1.30) or deacetylated (dark yellow -log[2]FC≤|0.58|,
   -log[10](p (Corr))≥1.30) peptides due to alcohol exposure.
   Non-significant peptides that are below the FC threshold are shown in
   grey. The dashed horizontal line marks the significance threshold (p
   (Corr) = -log[10](0.05)) and vertical marks indicate log[2](FC) of
   |1.5| and -|1.5|. B) Bar plot of whole liver acetylomic pathway
   enrichment of toxicity endpoints with –log[10](p (Corr)) representing
   enrichment score and values within each bar indicating the number of
   molecules identified within each pathway. C) Bubble plot of acetylomic
   canonical pathway enrichment with the -log[10](p (Corr)) displaying
   enrichment score, number of molecules denoted by size, and color
   representing pathway activation (orange) and inhibition (purple). ALT:
   Alanine aminotransferase; ATP: Adenosine triphosphate; EIEF2:
   Eukaryotic initiation factor 2; GCN2: Serine/threonine-protein kinase
   general control nonderepressible 2; TCA: Tricarboxylic acid cycle.

Fig. 2.

   [121]Fig. 2
   [122]Open in a new tab

   Alcohol induced acetylation of proteins within the electron transport
   chain. Illustration of proteins containing acetylated lysines
   significantly changed by multiple comparison-corrected test within the
   electron transport chain. The data is presented as log[2] fold-change
   of ethanol over control. Figure made in BioRender.com. Atp5: Adenosine
   triphosphate synthase 5; Cox; Cyclooxygenase, Cyc: Cytochrome subunit;
   Cyt C; Cytochrome complex, ETF; Electron-transfer flavoprotein, Nduf;
   NADH:ubiquinone oxidoreductase, SDH; Succinate dehydrogenase complex.

3.2. Integration of alcohol-induced cysteine redox and lysine acetylation
proteomic datasets

   Alcohol-induced changes in murine hepatic thiol redox proteome were
   evaluated previously, noting that the majority of Cys thiols were
   significantly reduced [[123]11,[124]15]. The whole-liver acetylomic
   analysis described above ([125]Fig. 1, [126]Fig. 2) was conducted on
   the same biological replicates used in the Cys proteomic analysis for
   PTM proteomic integration ([127]Fig. 3A). Proteins that underwent an
   alcohol-induced Lys acetylation were more likely to have a change in
   Cys redox status than proteins that did not have alcohol-induced Lys
   acetylation (OR = 1.47, p-value = 0.034, [128]Supplemental 1 Cys Lys
   StructureMap). Protein with an alcohol-induced Cys redox change had a
   higher probability of having alcohol-induced Lys acetylation than
   proteins without an alcohol-induced redox change (OR = 1.33,
   p-value = 0.054, [129]Supplemental 2 IPA Acetylome and CysLys).
   Proteins identified within both datasets were characterized by querying
   UniProt API to identify the molecular functions enriched.
   Oxidoreductase enzymes or redox reactive proteins contained the
   greatest number of proteins ([130]Fig. 3B). Furthermore, UniProt API
   was queried to categorize the subcellular localization of these
   proteins ([131]Fig. 3C). Mitochondrial and cytosolic proteins were the
   first and second most common locations of enzymes found in both the Cys
   redox and acetyl-Lys integration analysis. Next, Tox Functions were
   identified using IPA for proteins containing both Cys redox changes and
   Lys acetylation. The greatest enrichment for hepatic disease
   pathologies were liver hyperplasia, hepatocellular carcinoma, liver
   steatosis, and glutathione depletion ([132]Fig. 3D). IPA canonical
   pathway analysis identified enrichment of pathways related to the
   development of liver disease, including sirtuin (Sirt) signaling,
   mitochondrial dysfunction, and xenobiotic metabolism ([133]Fig. 3E).
   Furthermore, the pathway with the highest enrichment was
   lipopolysaccharide/Interleukin 1 (LPS/IL-1) mediated inhibition of RXR
   function with direct implications for fatty acid metabolism and hepatic
   steatosis ([134]Fig. 3F).

Fig. 3.

   [135]Fig. 3
   [136]Open in a new tab

   Pathway enrichment analysis of proteins identified by the cysteine
   proteomic and acetylomic analysis. A) Schematic for the development and
   integration of a cysteine proteomic and acetylomic analysis on 6-week
   Lieber-DeCarli (LD) mouse tissue. B) Bar plot depicts the number of
   acetylated and redox changed hepatic proteins within each enzymatic
   class. C) Pie chart displaying the percentage of acetylated and redox
   changed hepatic proteins within each subcellular compartment. D) Bar
   plot of selected toxicity functions identified through of analysis of
   acetylated and redox modifies hepatic proteins with –log[10](p (Corr))
   representing enrichment score and values within each bar indicating the
   number of molecules identified within each pathway. E) Canonical
   pathway enrichment of acetylated and redox modified hepatic proteins
   with the -log[10](p (Corr)) representing enrichment score and the
   number of molecules represented by size. F) IL-1 mediated inhibition of
   RXR function pop-out depicts the fatty acid metabolizing enzymes
   regulated by RXRA (enrichment by IPA). Figure made in BioRender.com.
   ACS: Acetyl-CoA synthetase; aIAM: Alkynyl iodoacetamide; Az-UVB: UV
   cleavable biotin-azide; CPT: Carnitine palmitoyltransferase; ER:
   Endoplasmic reticulum; HMG: Hydroxymethylglutaryl-CoA synthase; IL-1:
   Interleukin 1; IP: Immunoprecipitation; LC-MS/MS: Liquid Chromatography
   tandem mass spectrometry; LD: Lieber-DeCarli; RXRA: Retinoid X receptor
   alpha; SLC: Solute carrier; SA: Streptavidin.

3.3. Chronic alcohol consumption modifies cysteine-lysine pairs

   StructureMap enabled the integration of Cys proteomic and Lys
   acetylomic data with AlphaFold2's structural information. AlphaFold2 is
   an artificial intelligence system able to predict 3D protein structure
   [[137]17]. A total of 1300 unique proteins identified within our Cys
   proteome and acetylome datasets were uploaded to StructureMap for the
   quantification of distance between Cys and Lys residues on fully folded
   structures plotted by distance matrix ([138]Fig. 4A–[139]Supplemental 1
   Cys Lys StructureMap). Of the 1300 proteins, StructureMap was able to
   confidently map 1280 proteins. In total, the distances between 557,815
   Cys and Lys residues (Cys-Lys pairs) were analyzed. These Cys-Lys pairs
   include 9895 unique Cys residues and 33,849 unique Lys residues
   ([140]Fig. 4B). Following the implementation of AlphaFold2's confidence
   metrics with a cut-off of 70 % quality score and 15 Å predicted aligned
   error (PAE), we identified 388,698 Cys-Lys pairs that we used to
   evaluate the relationship between the physical distance between Cys-Lys
   residues and the presence of alcohol-induced PTMs of these residues. Of
   these Cys-Lys pairs, 39,696 have at least one modified residue within
   the pair, 25,886 pairs have at least one redox changed Cys, and 15,724
   pairs have an acetylated or deacetylated Lys residue. When only
   considering Cys-Lys pairs within 15 Å (28,238 pairs), the probability
   that the Lys is acetylated is higher among pairs with a modified Cys
   than the pairs with no modified Cys (OR = 1.89, p-value<0.001)
   ([141]Supplemental 3 CysLys Motif). Here in [142]Fig. 4C the number of
   modified Cys residues to the nearest Lys residue and acetylated Lys
   residues to the nearest Cys residue was stratified by distance to
   inform on Cys and Lys microenvironments. These results implicate Lys
   residue proximity as a factor when considering alcohol metabolism's
   impact on Cys redox status, specifically when a Lys is within 5–7.5 Å
   ([143]Fig. 4C). Indeed, 95 % of redox-modified Cys have at least one
   proximal or within 15 Å Lys residue. Interestingly, this trend did not
   apply to Lys acetylation, given that 47 % of acetyl-Lys are within 15 Å
   of a Cys residue and 53 % are farther than 15 Å from a Cys residue
   ([144]Fig. 4C).

3.4. Cys-Lys proximity motifs

   Our integrated proteome analyses revealed four different motifs of
   Cys-Lys pairs modified by alcohol consumption. These include proteins
   containing: 1) one Lys proximal to one Cys (Cys[1]:Lys[1]); 2) one Lys
   proximal to multiple Cys (Cys[x]:Lys[1]); 3) one Cys proximal to
   multiple Lys (Cys[1]:Lys[x]); and 4) multiple Lys proximal to multiple
   Cys (Cys[x]:Lys[x]). The pairs evaluated contained at least one residue
   with an alcohol-induced PTM (i.e., redox-modified Cys and/or acetylated
   Lys residue) to ensure a mechanistic interpretation of alcohol-induced
   biochemical pathologies ([145]Supplemental 3 CysLys Motif).
   Additionally, enrichment analysis was conducted on the proteins
   containing different Cys-Lys motifs. The different motifs are enriched
   for varying subcellular locations, biological processes, and enzymatic
   classes ([146]Table 1). The Cys[1]:Lys[1] motif predominately contains
   acetylated Lys with 99 % unmodified Cys within 15 Å ([147]Fig. 5A). In
   total, 345 Cys[1]:Lys[1] pairs were identified and Glrx5 was selected
   to demonstrate the 1Cys:1Lys relationship ([148]Fig. 5B). The most
   common of these motifs was Cys[1]:Lys[x] with 419 Cys-Lys pairs
   quantified in total. With the Cys[1]:Lys[x] motif, the majority of Cys
   were found to have undergone a significant change in redox status
   ([149]Fig. 5C) and this is illustrated with Echs1 ([150]Fig. 5D).
   Evaluation of Cys[x]:Lys[1] motifs showed that all Lys but 1 within
   this motif were acetylated/deacetylated out of 178 Cys-Lys pairs
   ([151]Fig. 5E). Decr2 was used to illustrate the Cys[x]:Lys[1] motif
   ([152]Fig. 5F). Finally, an example of the Cys[x]:Lys[x] motif is
   detailed in the next section using the alcohol metabolizing enzyme,
   Adh1.

Fig. 5.

   [153]Fig. 5
   [154]Open in a new tab

   CysLys pair motifs. A) Cys[1]:Lys[1]. Number of pairs with redox
   changed or unchanged Cys paired with an acetylated or unmodified Lys.
   B) AlphaFold2 predicted structure of Glrx5 which contains one Cys-Lys
   pair with a redox modified cysteine and acetylated Lys. Cys-Lys pairs:
   Cys63 – Lys55. C) Cys[1]:Lys[x]. Bar plot showing the distribution of
   pairs comprised of a Cys residue proximal to multiple Lys residues,
   with x representing the number of Lys residues. D) AlphaFold2 predicted
   structure of Echs1 contains a singular cysteine residue proximal to
   three lysine residues. Cys-Lys pairs: Cys62 – Lys101, Lys56. E)
   Cys[x]:Lys[1]. Bar plot showing the distribution of pairs comprised of
   a Lys residue proximal to multiple Cys residues, with x representing
   the number of Cys residues. F) AlphaFold2 predicted structure of Decr2
   which contains two cysteine residues proximal to a singular lysine
   residue. Cys-Lys pairs: Lys62 – Cys13, Cys80. Cys: Cysteine; Lys:
   Lysine.

3.5. Evaluation of Adh1 binding sites containing Cys-Lys pairs

   Adh1 is a primary ethanol metabolizing enzyme in the liver which drives
   the reduction in the NAD+/NADH ratio and acetaldehyde (AcH) generation.
   This enzyme included both a Cys[1]:Lys[1] pair and the Cys[x]:Lys[x]
   motif in different locations of the protein. At the NAD ^+ binding
   pocket a Cys[1]:Lys[1] pair was identified and the Lys229 residue was
   found to be a part of NAD ^+ binding [[155]30]. Furthermore, Lys229
   (FC = 1.66, p (Corr) = 0.01) was found to be significantly acetylated
   proximal to Cys241. In contrast, the zinc binding pocket was found to
   have a Cys[x]:Lys[x] motif with three of the zinc coordinating Cys to
   be significantly reduced by alcohol (Cys98, FC = 4.95, p (Corr) = 0.01;
   Cys101 FC = 8.27, p (Corr) = 0.01) and are proximal to two acetylated
   Lys (Lys324 FC = 1.99, p (Corr) = 0.03; Lys105 FC = 1.27, p
   (Corr) = 0.05) ([156]Fig. 6) [[157]30,[158]31]. Alignment of human ADH1
   isoforms and mouse Adh1 amino acid sequences revealed that Cys-Lys
   pairs within zinc and NAD^+ binding sites were 100 % conserved between
   species and isoforms ([159]Fig. 6) [[160]30].

Fig. 6.

   [161]Fig. 6
   [162]Open in a new tab

   CysLys pairs are conserved within the zinc and NAD^+ binding sites of
   Adh1. The panel to the left depicts cysteines that coordinate Adh1 zinc
   and the proximal lysine residues. The right panel depicts the Lys that
   binds to NAD^+ and its proximal cysteine. Adh1 structure was generated
   by AlphaFold2. Amino acid sequences between mouse and human Adh1 were
   aligned. Bolded within the Zn^2+ and NAD^+ binding region represent the
   amino acids involved in Zn^2+ binding and NAD^+ binding. NAD^+:
   Nicotinamide adenine dinucleotide; Zn^2+: Zinc.

3.6. Identification of human proteins orthologous to murine proteins with
Cys-Lys pairs detected within an acetylomic profile of severe
alcohol-associated liver hepatitis explants

   Characterizing how alcohol metabolism impacts protein biochemistry
   through undiscovered mechanisms is key to understanding the
   pathological progression of alcohol-associated end-stage liver disease
   and therapeutic development. Therefore, translating our murine findings
   to the human condition is paramount. Thus, the Ensembl genome browser
   was used to identify orthologous proteins containing one or more
   Cys-Lys pair modified by alcohol metabolism (see [163]Supplemental 4
   SAH CysLys Orthology and Enrichment). The human orthologous proteins
   identified were cross referenced with an acetylomic analysis conducted
   on hepatic explants from SAH patients [[164]27]. The human protein
   orthologs with murine counterparts containing Cys-Lys pairs modified by
   alcohol metabolism were all found to have significantly decreased
   acetylation in the patients with SAH. Thus, all hepatic proteins with
   altered acetylation in patients with SAH were found to have Cys-Lys
   pairs in our murine model of ALD. To assess how Cys-Lys orthologs
   translate to disease pathology, the list of proteins was submitted to
   DAVID, identifying significantly enriched biological processes. Then
   Reactome was used to determine enzyme substrate use and connection
   within biological processes ([165]Fig. 7). The processes enriched
   included ethanol metabolism, glycolysis/gluconeogenesis, β-oxidation,
   TCA and ketogenesis (see [166]Supplemental 4 SAH CysLys Orthology and
   Enrichment). Key enzymes within these pathways such as ADH1A/B,
   aldolase B (ALDOB), ECHS1, succinate-CoA ligase (SUCLG1), and
   hydroxyacyl-CoA dehydrogenase trifunctional protein (HADHA) were
   identified, offering avenues for exploring how these PTMs impact
   protein structure and function in the context of disease pathology
   ([167]Fig. 7).

Fig. 7.

   [168]Fig. 7
   [169]Open in a new tab

   Proteins identified to be deacetylated within severe alcohol-associated
   hepatitis (SAH) explant tissue containing Cys-Lys pairs separated by
   subcellular compartments and associated pathways. Depicted are
   orthologous enzymes of murine proteins containing Cys-Lys pairs and
   human proteins found deacetylated within SAH explant tissues. All
   enzymes are represented by red ovals and the substrates they interact
   with. The pathways that were enriched are placed within their
   respective subcellular location and include glycolysis/gluconeogenesis,
   alcohol metabolism, fatty acid metabolism, citric acid cycle and
   ketogenesis. Figure made in [170]BioRender.com. ACADVL: Very long-chain
   specific acyl-CoA dehydrogenase; ACA-CoA: Acetoacetyl-CoA; Ac-CoA:
   Acetyl-CoA; ACAT1: Acetyl-CoA acetyltransferase 1; ACSL1: Acetyl-CoA
   synthetase 1; SAH: severe alcohol-associated hepatitis; ALDOB: Aldolase
   fructose-bisphosphate B; CBR1: Carbonyl reductase 1; EHHADH: Enoyl-CoA
   hydratase and 3-hydroxyacyl CoA dehydrogenase; ENO1: Enolase 1; FAs:
   Fatty acids, FASN: Fatty acid synthase; HADHA: Hydroxyacyl-CoA
   dehydrogenase trifunctional protein; HMGCL:
   3-hydroxy-3-methylglutaryl-CoA lyase; HMGCS2: Hydroxymethylglutaryl-CoA
   synthase; GOT2: Glutamic-oxaloacetic transaminase 2; IDH1: Isocitrate
   dehydrogenase 1; LD: Lieber-DeCarli; MDH1: Malate dehydrogenase 1;
   MDH2: Malate dehydrogenase 1; PALM-CoA: Palmitoyl-CoA; PEP:
   Phosphoenolpyruvate; PGK1: Phosphoglycerate kinase 1; SUCLG1:
   Succinate-CoA ligase.

4. Discussion

   Alcohol metabolism alters the hepatic proteome through both a reduction
   in the Cys proteome and Lys hyperacetylation. Here, we show for the
   first time that these alcohol-induced protein modifications may
   regulate protein structure and redox function due to Cys-Lys proximity.
   Indeed, both Cys and Lys are independently regulated through a host of
   well-characterized PTMs such as Cys-glutathionylation or
   Lys-ubiquitination [[171]32]. However, this newly discovered feedback
   mechanism plays a key role across a host of metabolic processes by
   uniquely sensing 1) the hepatic metabolic state by acetylation and 2)
   oxidative stress via thiol redox signaling. Indeed, this coordinated
   crosstalk among Cys-Lys pairs highlights key redox signaling hubs
   responsible for hepatic responses to alcohol toxicity.

   Our whole liver acetylomic analysis of murine tissue revealed that many
   acetylated proteins were within the ETC ([172]Fig. 2). These findings
   suggest that the acetylation of the ETC complexes results in diminished
   respiratory output which is supported by a decrease in ATP and NAD^+
   consequent to alcohol metabolism [[173]33,[174]34]. Additionally,
   several of the sites found to be acetylated by alcohol consumption have
   been previously identified in Sirt3-deficient mice including
   NADH:ubiquinone oxidoreductase (Complex I) and ATP synthase (Complex V)
   [[175]35,[176]36]. This directly correlates with the pathway analysis
   enrichment of the sirtuin signaling pathway noted above. Sirt3 is the
   major deacetylase in the mitochondria, which is the subcellular
   compartment with the most acetylated proteins, making Sirt3 vital to
   regulating acetylation in ALD [[177]37]. Sirt3-deficient murine models
   have decreased ATP concentrations, supporting the concept that Lys
   hyperacetylation results in ETC inhibition [[178]10]. These findings
   have been validated within ATP synthase and succinate dehydrogenase
   (SDHA, Complex II). For example, ATP synthase acetylation decreases
   protein activity and is regulated by Sirt3 [[179]36]. Also,
   hyperacetylation of Complex II blocks substrate entry, inhibits
   activity, and is also regulated by Sirt3 [[180]24,[181]38].
   Furthermore, Complex I acetylation has been found to wholly control
   respiratory output within other models and is likely a crucial
   component to NAD^+ depletion in ALD [[182]39,[183]40]. Overall, the
   evaluation of alcohol-induced whole liver acetylation and the
   interrogation of the cross talk between the hepatic acetylomic and
   cysteine proteomic alterations revealed a critical interplay of carbon
   and redox stress.

   The integration of proteins found by the acetylomic and redox proteomic
   analyses revealed an alcohol-related association between redox active
   Cys residues and Lys acetylation. The proteins identified across these
   datasets were primarily comprised of oxidoreductase enzymes or enzymes
   that catalyze redox reactions ([184]Fig. 3B), and the mitochondria and
   cytosol are the most common subcellular locations of these proteins
   ([185]Fig. 3C). Among the Toxicity Functions identified by IPA, liver
   steatosis was one of the most enriched, validating our murine model of
   early-stage ALD ([186]Fig. 3D). Proteins associated with liver
   steatosis are central to β-oxidation, for example, acyl-CoA synthetase
   long chain family member 1 (ACSL1) breaks down long chain fatty acids.
   Similarly, carnitine palmitoyltransferase I (CPT1A) transports fatty
   acids into the mitochondria. Two other proteins were also found related
   to β-oxidation, HADHA and acyl-CoA dehydrogenase (ACADS)
   [[187]41,[188]42]. Each of these enzymes are critical in the transport,
   oxidation, and metabolism of fatty acids, revealing that Cys redox
   signaling and Lys acetylation play a role in alcohol-induced redox
   signaling. Canonical pathway analysis found both modifications to be
   central to the regulation of lipid and xenobiotic metabolism through
   enzymes regulated by the RXRA transcription factor within the liver
   ([189]Fig. 3E). Changes in RXRA concentrations have been connected to
   IL-1, which are proinflammatory cytokines well documented to be
   elevated within ALD patients and contribute to hepatocyte disfunction
   leading to fibrosis/cirrhosis [[190]43]. Subsequently, IL-1 and alcohol
   consumption diminish RXRA concentrations resulting in the disruption of
   lipid handling [[191]44]. Studies have identified an association
   between depleted RXR protein levels and the inhibition of fatty acid
   oxidation leading to hepatic lipid accumulation in ALD [[192]45].
   Integrating our datasets to identify key protein signatures through Cys
   and Lys modification enabled the identification of new mechanisms of
   metabolic control essential to lipid handling and the pathogenesis of
   fat deposition in the early stages of ALD.

   The evaluation of 3D protein structures and quantifying the proximity
   between redox-modified Cys and Lys residues revealed specific paired
   Cys and Lys sites involved in enzymatic regulation. Furthermore, this
   analysis demonstrated that Cys redox status was more responsive to Lys
   proximity. In our model of ALD, the optimal distance for a Lys to
   impact Cys reduction appears to be 5-7.5 Å ([193]Fig. 4C). The strong
   structural link between alcohol metabolism and the effect it had on the
   Cys proteome is exemplified by the finding that 95 % of redox-modified
   Cys residues were paired with a proximal Lys ([194]Fig. 4C). The
   mechanism driving this interaction is yet to be elucidated but is
   likely due, in part, to the reductive stress generated during the
   metabolism of alcohol to AcH and acetate. Interestingly, the proximity
   of a Cys only resulted in Lys acetylation 50.5 % of the time. This
   suggests that Lys acetylation is occurring through additional
   mechanisms such as other structural protein features or environmental
   factors. The alkaline pH of the mitochondria and elevated
   concentrations of Ac-CoA are both noted as an evolutionary mechanism
   resulting in fewer Cys-Lys pairs conserved versus the cytosol,
   specifically in longer living species, such as non-human primates and
   humans [[195]46]. Contrary to longer lived species, mice have conserved
   Cys-Lys pairs within their mitochondrial matrix [[196]13]. The
   metabolic feedback loop driven by the generation of Ac-CoA and
   increased likelihood of enzymatic inhibition has not evolutionarily
   impacted Cys-Lys pair conservation in mice. This necessitates cross
   referencing murine Cys-Lys pairs and conservation within the human
   proteome. An important evolution of our dataset is the finding that
   proximal Cys-Lys pairs regulate Cys reactivity with the potential for
   alcohol-induced Lys acetylation to impact protein function outside the
   canonical functions of Lys modification.

   Embedded within the alcohol-induced proximal Cys-Lys findings is the
   potential for protein regulation. Our analysis revealed that unique
   Cys-Lys structural motifs may define enzymatic processes. These motifs
   describe shared amino acid spatial interactions directed through
   electrostatics to facilitate the transfer of an acetyl moiety from a
   Cys thiol to a Lys ε-amine, thus impacting Cys reactivity. The four
   motifs identified are separated by the number of Cys and Lys residues
   within a defined 15 Å sphere and are noted in [197]Table 1 above and
   illustrated in [198]Fig. 5, [199]Fig. 6. One of the proteins containing
   a Cys[1]:Lys[1] pair was Glrx5, found to have been reduced at Cys63
   (FC = 3.27, p (Corr) = 0.026) and paired with acetyl Lys55 ([200]Fig.
   5B). This impacts a regulatory hotspot as Lys55 binds to glutathione
   and Cys63 coordinates the 2Fe–2S cluster [[201]47]. The reduction of a
   redox active Cys indicates diminished enzymatic activity resulting from
   alcohol metabolism. Another example of this motif is glutamate
   oxaloacetate transaminase 2 (Got2), which is acetylated at Lys404
   (FC = 2.66, p (Corr) = 0.0002) and is proximal to Cys382. The
   acetylation of Got2 at Lys159, Lys185 and Lys404 has been associated
   with an increased affinity for MDH2 (Malate dehydrogenase 2) and is
   regulated by Sirt3 [[202]48]. This provides a clear example of a
   proximal Cys-Lys pair involved in the redox regulation that has been
   conserved as a regulatory mechanism to control enzyme activity during
   sustained oxidative stress. Illustrating the second motif, Echs1
   contains a singular Cys proximal to multiple Lys (Cys[1]:Lys[x]) and
   was found to be modified at Lys101 and Cys62 (FC = 3.94, p
   (Corr) = 0.002 and FC = 5.01, p (Corr) = 0.01, respectively) ([203]Fig.
   5D). Cys62 is additionally proximal to Lys101, Lys56, and Lys115.
   Lys101 contributes to enoyl-CoA substrate binding and acetylation at
   Lys101 is proposed to inhibit enzymatic activity [[204]49]. ECHS1
   deficiency is associated with alterations in metabolic activity
   including diminished oxidative phosphorylation and dysregulation of
   fatty acid oxidation, both commonly observed in models of ALD
   [[205]50,[206]51]. A third motif is characterized by having multiple
   Cys paired with one proximal Lys. Decr2 contains this Cys[x]:Lys[1]
   motif where Lys64 is proximal to Cys13 and Cys80. Lys64 aids in
   NADP ^+ binding and was significantly deacetylated by alcohol
   metabolism (FC = −1.53, p (Corr) = 0.01) [[207]5,[208]52]. Decr2
   carries out the degradation of peroxisomal unsaturated fatty enoyl-CoA
   esters and this proximal Cys-Lys pair likely plays an important role in
   lipid homeostasis. Importantly, this site was previously found to be
   acetylated in a model of ALD [[209]5,[210]53]. The proximity of
   multiple Cys residues to a Lys in this motif could signify that these
   conserved Cys-Lys pairs induce S → N transfers resulting in the
   acetylation of these sites and regulating protein activity. The final
   motif identified is Cys[x]:Lys[x] which was found within Adh1, a key
   enzyme in alcohol pathogenesis. Adh1 was further evaluated due to the
   presence of Cys-Lys motifs (Cys[x]:Lys[x] and Cys[1]:Lys[1]) within the
   zinc and NAD^+ binding regions ([211]Fig. 6). Adh1 catalyzes the
   reduction of alcohol to AcH using zinc to transfer electrons, reducing
   NAD^+ to NADH [[212]30]. Our redox proteomic analysis revealed that Cys
   residues involved in the coordination of zinc were reduced, suggesting
   that zinc was no longer bound. This is further supported by the
   identification of acetylated proximal Lys which is likely acetylated
   via S → N acetyl moiety transfer. As stated above, Lys229 (FC = 1.66, p
   (Corr) = 0.01) a component of the NAD^+ binding site was found to be
   acetylated while also being proximal to Cys241 [[213]30]. The
   inhibition of Adh1 could be one mechanism to limit the amount of AcH
   generated and to conserve hepatic NAD^+ concentrations. Supporting this
   idea, previous researchers have shown that Adh1 inhibition is
   protective, decreasing lipogenic genes and triglyceride accumulation in
   mice exposed to alcohol [[214]9]. Indeed, AcH inhibits the transport of
   NADH into the mitochondria, limiting cofactor concentrations.
   Understanding how these four Cys-Lys motifs regulate activity must be
   carried out on an individual protein basis and will be key to revealing
   potential therapeutic targets focused on the early intervention of ALD.
   Future studies may also require the analysis of protein quaternary
   structures, such as dimer and tetramer multimers, since there is a high
   chance that Cys-Lys pairs exist throughout these proteoforms.

   Our analysis reveals that alcohol-induced metabolic disturbances leave
   their mark on Cys-Lys pairs in a unique manner. Our murine hepatic
   Cys-Lys analysis aligned translationally when compared with acetylation
   targets identified previously in SAH human explants ([215]Fig. 7). SAH
   is a complex disease that damages hepatocytes and impairs metabolic
   activity resulting in decreased acetylation across protein lysine
   residues and provides a unique signature that inversely aligns with
   murine acetylation in an early model of ALD, where fully functioning
   hepatocytes are actively metabolizing alcohol [[216]27]. This alignment
   suggests that Cys-Lys motifs correlate in a regulatory nature across
   species and murine Cys-Lys pairs are comparable to the human condition.
   Several alcohol metabolizing orthologs were identified by our spatial
   Cys-Lys analysis. Two of the human ADH1 isoforms, ADH1A and ADH1B, as
   well as Catalase (CAT) and Aldehyde dehydrogenase (ALDH2) were found to
   have decreased acetylation in the SAH hepatic explants and align with
   our murine Cys-Lys pairs [[217]54]. In addition to ethanol metabolizing
   enzymes, the TCA cycle and glycolysis/gluconeogenesis were found to
   contain similar proximal Cys-Lys pairs in SAH tissue, with altered
   acetylation profiles [[218]55]. Glycolytic enzymes found to be
   deacetylated in SAH explants with orthologous murine Cys-Lys pair(s)
   include ALDOB, Phosphoglycerate kinase 1 (PGK1) and Enolase 1 (ENO1).
   Inhibition of ETC, TCA, and glycolysis contribute to a metabolic shift
   toward fatty acid synthesis and steatosis. Consequently, there would be
   an increase in lipid uptake by fatty acid transporters of the solute
   carrier family (SLC), SLC27A5 which was found to be significantly
   reduced and SLC27A2 which was found to be significantly reduced and
   acetylated in mice. Also, Acetyl-CoA synthetase 1 (ACSL1), Very
   long-chain specific acyl-CoA dehydrogenase (ACADVL), and HADHA were
   found to be significantly reduced and acetylated likely contributing to
   impaired lipid catabolism due to our murine model of ALD [[219]56].

   Notably, numerous ketogenic enzymes were found to contain proximal
   Cys-Lys pairs within our murine and SAH acetylomic analysis, including
   CPT1, Acetyl-CoA acetyltransferase 1 (ACAT1), MDH2,
   Hydroxymethylglutaryl-CoA synthase (HMGCS2) and
   3-hydroxy-3-methylglutaryl-CoA lyase (HMGCL). CPT1 is critical to
   initiating ketogenesis by transporting long-chain fatty acids into the
   mitochondria and converting them into fatty acyl-carnitines. This
   results in increased Ac-CoA concentrations via β-oxidation and elevated
   acetoacetyl-CoA (ACA-CoA) and thus initiating the ketogenic pathway
   [[220]8,[221]57]. Additionally, this pathway is induced when
   oxaloacetate levels are low as with the activation of MDH2 via
   acetylation and reducing oxaloacetate to malate while oxidizing NADH to
   NAD^+ [[222]48]. HMGCS2 will then use ACA-CoA to generate β-hydroxy
   β-methylglutaryl-CoA (HMG-CoA). Furthermore, HMGCS2 was found to be
   acetylated at regulatory Lys310 in the absence of SIRT3, potentially
   impacting the generation of ketone bodies [[223]58]. Further evaluation
   of these pathways is critical to understanding protein specific changes
   that could contribute to the metabolic disruption induced by alcohol
   consumption. Vital targets bridge models of ALD and SAH pathology
   solidifying the necessity of multiple approach analysis for therapeutic
   target identification.

   Overall, our analysis reveals that alcohol metabolism specifically
   modifies Cys-Lys redox-centric protein motifs and supports the
   hypothesis that these unique metabolic-sensing thiol signaling hubs
   play a key role in the pathogenesis of ALD. Future work will continue
   to characterize spatially resolved Cys-Lys biochemical mechanisms in
   protein pathways associated with the onset and progression of ALD with
   a focus on lysine acetylation, redox modifications, and impact on
   enzyme function. These results begin to define the complex
   spatiotemporal relationships protein modifications play in directing
   adaptive responses to altered hepatic redox signaling resulting from
   alcohol consumption.

CRediT authorship contribution statement

   Courtney D. McGinnis: Writing – review & editing, Writing – original
   draft, Visualization, Validation, Software, Methodology, Investigation,
   Formal analysis, Data curation. Peter S. Harris: Writing – review &
   editing, Visualization, Validation, Methodology, Investigation, Formal
   analysis, Data curation. Brenton I.M. Graham: Validation, Software,
   Formal analysis. John O. Marentette: Methodology, Investigation, Data
   curation. Cole R. Michel: Writing – review & editing, Validation,
   Methodology, Investigation, Formal analysis, Data curation. Laura M.
   Saba: Writing – review & editing, Visualization, Validation, Software,
   Investigation, Funding acquisition, Formal analysis. Richard Reisdorph:
   Writing – review & editing, Validation, Methodology, Formal analysis,
   Conceptualization. James R. Roede: Writing – review & editing, Writing
   – original draft, Validation, Methodology, Investigation,
   Conceptualization. Kristofer S. Fritz: Writing – review & editing,
   Writing – original draft, Visualization, Validation, Supervision,
   Software, Resources, Project administration, Methodology,
   Investigation, Funding acquisition, Formal analysis, Data curation,
   Conceptualization.

Declaration of competing interest

   The authors declare no conflicts of interest.

Acknowledgements