Effects of chronic intermittent ethanol exposure and withdrawal on neuroblastoma cell transcriptome

Abstract

Cycles of heavy drinking and abstinence can lead to ethanol abuse disorder. We studied the effects of chronic intermittent ethanol exposure (CIE) over three weeks on neuroblastoma cells, using an ethanol concentration frequently attained in binge drinking (40 mM, 184 mg/dl). There were many changes in gene expression but most were small. CIE affected pathways instrumental in the development or plasticity of neurons, including axonal guidance, reelin signaling and synaptogenesis. Genes involved in dopamine and serotonin signaling were also affected. Changes in transporters and receptors could dampen both NMDA and norepinephrine transmissions. Decreased expression of the GABA transporter SLC6A11 could increase GABA transmission and has been associated with a switch from sweet drinking to ethanol consumption in rats. Ethanol increased stress responses such as unfolded protein response. TGF-β and NFκB signaling were increased. Most of the genes involved in cholesterol biosynthesis were decreased in expression. Withdrawal for 24 h after CIE caused most of the CIE-induced expression changes to move back toward unexposed levels.

Introduction

In the United States, 8.5% of adults age 18 and older have met criteria for alcohol use disorders in the last 12 months, and 30.3% have met them at some point in their lifetime (Hasin & Grant, 2015). The economic burden of alcohol misuse in the US was recently estimated at $249 billion per year, with three quarters of the total cost related to binge drinking (Sacks, Gonzales, Bouchery, Tomedi, & Brewer, 2015). In 2012–2013, 33% of US adults reported past year drinking at binge levels (five drinks or more for men and four drinks or more for women, over 2 hours). Many drank at binge levels at least weekly (Hingson, Zha, & White, 2017) and 13% reporting drinking beyond twice binge level, defined by NIAAA as “high intensity drinking” (Hingson et al., 2017).

We modeled the effects of repeated binge drinking on a neuroblastoma cell line, using chronic intermittent exposure to ethanol (CIE). This model will be useful to determine whether the effects of repeated ethanol exposures differ from those seen after single acute exposure (e.g. McClintick et al., 2014; McClintick et al., 2019). CIE has been modeled in animals in several ways (Becker & Lopez, 2004; Bell, Rodd, Engleman, Toalston, & McBride, 2014; Jury et al., 2017; McClintick et al., 2015). The Becker protocol in many mice studies uses four 16 h cycles of ethanol vapor separated by 8 h periods of withdrawal (Becker & Lopez, 2004; Jury et al., 2017). In P (alcohol preferring) and HAD (high alcohol drinking) rat strains, which voluntarily consume ethanol, several different multiple scheduled access protocols have been used, including a single 2 h or 4 h bout, or four 1 h bouts during the dark cycle, usually for 5 days a week for 3–4 weeks (Bell et al., 2014; McBride et al., 2013; McClintick et al., 2015). These protocols can achieve high blood alcohol levels (BEC); using voluntary consumption methods, P rats reached BEC of 40–90 mg/dl (8.7 – 19.5 mM) (Bell et al., 2014); with ethanol vapor chambers BEC of 175–325 mg/dl (38 – 70 mM) can be achieved (Jury et al., 2017; Melendez, 2011).

We chose the neuroblastoma cell line SH-SY5Y as a model, because it expresses more of the genes found in brain tissue than other readily available cells. Among these are dopamine receptor D2 (DRD2), two serotonin receptors, three opioid receptors (OPRM1, OPRD1, OPRL1) several GABA(A) receptor subunits, three metabotropic glutamate receptors (GRM7, GRM2 & GRM8), subunits of AMPA, NMDA and kainate iontropic receptors, nicotinic and muscarinic cholinergic receptor subunits, and two adrenergic receptors (ADRA2A and ADRA2C).

We repeatedly exposed the SH-SY5Y cells to 40 mM ethanol for 4 h per day, 4 days per week over 3 weeks. This concentration is equivalent to a blood alcohol concentration of 0.18%, which can be reached by a 180-pound man drinking about 7 beers over 2 h, and is below the level defined as “high intensity drinking” (Hingson et al., 2017). We found that the expression of many genes was affected. We also examined the effect of a 24 h withdrawal after the last exposure, and found most of those genes returned toward baseline expression.

Methods

Cell culture and ethanol treatment of SH-SY5Y cells

SH-SY5Y cells (CRL-2266, Lot 63724189, ATCC, Manassas, VA) were cultured in a 1:1 mixture of EMEM (ATCC) and F12K medium (10025-CV, Thermo Fisher Scientific) with 10% (vol/vol) fetal bovine serum (ATCC) and 1% penicillin and streptomycin. Cells were plated at a density of 1.5 × 10⁶ in 25 cm² canted neck flasks (Corning, Corning, NY). We modeled chronic intermittent exposure to ethanol (CIE) by using a pattern of 4 h exposure to 40 mM ethanol on 4 successive days, followed by 3 days without, repeated for 3 weeks (see Supplementary Figure 1). One hour after seeding, the cells were treated for 4 h with a “binge level” concentration of ethanol (40 mM) or left untreated, after which both sets of cells were washed with 4 mL 1X phosphate-buffered saline (PBS), and medium without ethanol was added. The 4 h ethanol exposure, wash and fresh medium was repeated Monday through Thursday for one pair of samples and Tuesday through Friday for a second. After washing the cells on Thursday (or Friday for the second pair), all dishes were split and plated in medium without ethanol until the following Monday. This pattern was repeated in week 2. In week 3, the treated group was split into two flasks on Monday (or Tuesday) and both were treated with ethanol as above for 3 days but on Thursday (or Friday), one group was treated with 40 mM ethanol for 4 h, the other received medium without ethanol. All cells were collected after 4 h. RNA was extracted using a Direct-zol DNA/RNA miniprep kit (R2080, Zymo Research, www.zymoresearch.com) per manufacturer’s instructions. The repeated washings and media changes removed most non-adherent cells, so the cells being studied were primarily adherent. Six independent biological replicates were conducted for each condition, with each pair of samples starting in on a new day (Monday or Tuesday of 3 successive weeks). The experiment thus spanned 5 weeks, during which cells were from passages 6 to 14 after receipt from ATCC.

RNA sequencing

Total RNA was evaluated for its quantity and quality using an Agilent Bioanalyzer 2100; RNA integrity numbers were all greater than 9. Fifty nanograms of total RNA was used for each sequencing library. cDNA library preparation was carried out at the Center for Medical Genomics, and included mRNA enrichment using oligo dT-beads to capture polyA RNA, RNA fragmentation, cDNA synthesis, ligation of index adaptors, and amplification, following the KAPA mRNA Hyper Prep Kit Technical Data Sheet, KR1352 – v4.17 (Roche Corporate). Each resulting indexed library was quantified and its quality accessed by Qubit and Agilent Bioanalyzer, and all libraries were pooled at equal molarity. Five microliter aliquots of 2 nM pooled libraries per lane were denatured, neutralized and applied to flow cell deposition on both surfaces of the flow cell; library cluster amplification was performed on a cBot 2 System (Illumina, Inc.). The resulting flow cell was sequenced on an Illumina HiSeq 4000, 75 base paired-end sequencing (Illumina, Inc.). Approximately 30 M reads per library were generated (minimum 26.5 M, maximum 35.4 M). A Phred quality score (Q score) was used to measure the quality of sequencing. More than 90% of the sequencing reads reached Q30 (99.9% base call accuracy).

Sequence alignment and gene counts

The sequencing data were first assessed for quality control using FastQC (v.0.11.5, Babraham Bioinformatics, Cambridge, UK). Then all sequenced libraries were mapped to the human genome (GRCh38/hg38) using the STAR RNA-seq aligner (v.2.5) (Dobin et al., 2013) with the following parameter: “--outSAMmapqUnique 60”. The reads distribution across the genome was assessed using bamutils (from ngsutils v.0.5.9) (Breese & Liu, 2013). Uniquely mapped sequencing reads were assigned to GRCh38/hg38 refGene genes using featureCounts (subread v.1.5.1) (Liao, Smyth, Shi, 2014) with the following parameters: “−s 2 –p –Q 10”.

Analyses

Reads aligned to known genes were analyzed by edgeR (Robinson, McCarthy, & Smyth, 2010); median alignments per sample was 20.7 M (minimum 17.7 M, maximum 24.1 M). Counts were normalized to counts per million reads (CPM) for each sample. Data were examined by Multidimentional Scaling in the edgeR package (Robinson et al., 2010) to detect outliers. Most samples clustered by experimental replicate, but one outlier was detected (Supplementary Figure 2), so all 3 samples from the replicate containing the outlier were removed from analyses, leaving 5 sets of 3 samples each: untreated, repeatedly treated (CIE), and repeatedly treated but withdrawn for 24 h (CIE+withdrawn). The MDS plot without the omitted samples (Supplementary Figure 3) indicates that there is a batch effect by replicate therefore analysis included replicate as a factor. To avoid analyzing genes that were not expressed or expressed at near background levels, only genes that had 2 or more CPM in at least 5 samples were retained for analysis (cf. (McClintick & Edenberg, 2006). Paired analysis was done based on experimental replicate: ethanol vs. controls, ethanol vs. ethanol withdrawal, and ethanol withdrawal vs. controls. False discovery rates (FDR) were calculated using the method of Benjamini & Hochberg (Benjamini & Hochberg, 1995) within edgeR.

GTEx data (GTEx Consortium, 2013; Lonsdale et al., 2013) were used to determine whether the gene was expressed in brain tissue. We considered any gene expressed at ≥ 1 Transcripts Per kilobase Million (TPM; read counts per million reads divided by length of transcript in kb) in any brain region to be significantly expressed in brain.

Qiagen’s Ingenuity Pathway Analysis Spring 2019 release (IPA; QIAGEN Inc., www.qiagenbioinformatics.com/products/ingenuitypathway-analysis) was used to analyze genes that were differentially expressed when treated with ethanol (at FDR < 0.20; using “Base Genes only” as background). Pathways were limited to those with FDR < 0.20 and 5 or more differentially expressed genes. The results for the upstream analysis were limited to those entries with absolute z-score ≥ 1.9.

The genes in Table 1 (FDR< 0.20, absolute fold ≥ 1.2) were imported into FUMA Version: 1.3.5e (27 May 2019) (Watanabe, Taskesen, van Bochoven, & Posthuma, 2017) at 2019-11-17 20:17:42 for GENE2FUNC analysis vs. GTEx version 8 all genes using standard parameters with minimum overlap set to 5 differentially expressed genes and FDR < 0.2.

Table 1.

Expression changes in SH-SY5Y neuroblastoma cells due to chronic intermittent ethanol treatment (40mM). Genes with FDR < 0.20, absolute fold changes > 1.2. (subset of Supplementary Table S1). EvC: ethanol vs. control.

Gene Symbol	Gene Name	Fold Ethanol vs. Control	FDR
GPRC5C	G protein-coupled receptor class C group 5 member C	1.68	2.86E-04
RHBDF1	rhomboid 5 homolog 1	1.54	4.70E-02
LRRC29	leucine rich repeat containing 29	1.54	1.42E-02
ZNF433	zinc finger protein 433	1.51	3.07E-04
SLC16A14	solute carrier family 16 member 14	1.50	9.97E-13
ARMC12	armadillo repeat containing 12	1.50	2.27E-04
WTIP	WT1 interacting protein	1.48	7.07E-03
FZD5	frizzled class receptor 5	1.47	9.50E-04
FAM129A	family with sequence similarity 129 member A	1.46	1.40E-01
CDHR3	cadherin related family member 3	1.46	8.31E-03
MYLK-AS1	MYLK antisense RNA 1	−1.44	1.65E-02
GPRIN2	G protein regulated inducer of neurite outgrowth 2	1.42	1.72E-02
CELSR1	cadherin EGF LAG seven-pass G-type receptor 1	1.40	7.70E-02
HGFAC	HGF activator	1.40	1.02E-03
LOC100272217	uncharacterized LOC100272217	−1.40	1.18E-02
SLC2A10	solute carrier family 2 member 10; class III facilitative glucose transporter	1.39	2.15E-03
PRNP	prion protein	1.38	2.39E-04
LINC00473	long intergenic non-protein coding RNA 473	1.37	2.98E-03
LOC100130992	uncharacterized LOC100130992	−1.36	8.50E-02
ZNF844	zinc finger protein 844	1.36	2.65E-07
KLHL35	kelch like family member 35	1.36	1.33E-01
PTH1R	parathyroid hormone 1 receptor	1.36	2.16E-02
TMEM200C	transmembrane protein 200C	1.35	3.85E-02
GAS1	growth arrest specific 1	1.34	1.45E-04
WNT5A	Wnt family member 5A	1.34	1.06E-01
GUSBP5	glucuronidase, beta pseudogene 5	−1.34	1.96E-01
CMAHP	cytidine monophospho-N-acetylneuraminic acid hydroxylase, pseudogene	−1.33	4.00E-03
LBX2-AS1	LBX2 antisense RNA 1	1.33	1.17E-02
DLK2	delta like non-canonical Notch ligand 2	1.33	9.22E-02
LTB4R	leukotriene B4 receptor	1.32	4.34E-07
RPL21P28	ribosomal protein L21 pseudogene 28	−1.32	1.96E-02
SERTM2	serine rich and transmembrane domain containing 2	−1.32	3.32E-14
TMC7	transmembrane channel like 7	1.32	7.54E-02
CCM2L	CCM2 like scaffold protein	1.32	5.03E-02
KLF3-AS1	KLF3 antisense RNA 1	−1.32	1.02E-01
ICAM3	intercellular adhesion molecule 3	1.32	1.55E-01
LOC100996419	uncharacterized LOC100996419	−1.31	1.92E-01
TBXAS1	thromboxane A synthase 1	−1.31	2.29E-02
STARD8	StAR related lipid transfer domain containing 8	1.31	1.30E-01
NRTN	neurturin	1.31	5.47E-03
PACRG	parkin coregulated	1.30	1.09E-01
LOC105373156	uncharacterized LOC105373156	−1.30	9.10E-02
TMEM266	transmembrane protein 266	1.29	1.09E-01
MIR4458HG	MIR4458 host gene	−1.29	2.57E-03
RN7SL832P	RNA, 7SL, cytoplasmic 832, pseudogene	−1.29	7.10E-02
DENND1B	DENN domain containing 1B	1.29	1.49E-01
EXOC3-AS1	EXOC3 antisense RNA 1	−1.29	1.95E-03
ABHD11	abhydrolase domain containing 11	−1.29	9.16E-04
ITPKB	inositol-trisphosphate 3-kinase B	1.29	1.73E-04
JCAD	junctional cadherin 5 associated	1.28	1.93E-01
PCF11-AS1	PCF11 antisense RNA 1	−1.28	5.48E-02
USP44	ubiquitin specific peptidase 44	1.28	1.99E-02
ROR1-AS1	ROR1 antisense RNA 1	−1.28	1.39E-02
MKLN1-AS	MKLN1 antisense RNA	1.28	1.65E-01
KCNH6	potassium voltage-gated channel subfamily H member 6	1.28	8.39E-02
UBTD1	ubiquitin domain containing 1	1.28	2.82E-02
SUSD5	sushi domain containing 5	1.27	1.79E-01
MIR3064	microRNA 3064	−1.27	9.74E-04
KCNC4	potassium voltage-gated channel subfamily C member 4	1.27	3.24E-02
LDLRAD4	low density lipoprotein receptor class A domain containing 4	1.27	8.43E-07
FAT1	FAT atypical cadherin 1	1.27	1.32E-03
FAM78A	family with sequence similarity 78 member A	1.27	1.14E-04
LOC107984035	uncharacterized LOC107984035	−1.27	1.41E-01
AJM1	apical junction component 1 homolog	1.27	1.14E-01
FBXO32	F-box protein 32	1.26	5.19E-04
BAG3	BCL2 associated athanogene 3	1.26	6.11E-03
LRIG1	leucine rich repeats and immunoglobulin like domains 1	1.26	1. 16E-01
SLC16A9	solute carrier family 16 member 9	1.26	1.52E-09
MAN1C1	mannosidase alpha class 1C member 1	1.26	2.75E-02
INHBA	inhibin subunit beta A	−1.26	9.69E-02
HTRA1	HtrA serine peptidase 1	1.26	3.09E-02
SLC6A9	solute carrier family 6 member 9, glycine	1.26	6.10E-02
SWSAP1	SWIM-type zinc finger 7 associated protein 1	−1.26	1.85E-01
KCNB1	potassium voltage-gated channel subfamily B member 1	1.25	9.93E-07
HIST2H2BE	histone cluster 2 H2B family member e	−1.25	5.58E-03
SMIM18	small integral membrane protein 18	−1.25	1.05E-01
JAK3	Janus kinase 3	−1.25	1.69E-01
OLFML2B	olfactomedin like 2B	1.25	7.65E-02
PTPN3	protein tyrosine phosphatase, non-receptor type 3	1.25	1.73E-01
MEF2C	myocyte enhancer factor 2C	1.25	4.11E-02
SPRY1	sprouty RTK signaling antagonist 1	−1.25	4.16E-03
MGEA5	new symbol OGA O-GlcNAcase	1.25	8.39E-14
GRASP	general receptor for phosphoinositides 1 associated scaffold protein	1.25	1.29E-01
TRIM9	tripartite motif containing 9	1.24	3.34E-02
ANKRD1	ankyrin repeat domain 1	−1.24	1.09E-01
LOC100506282	uncharacterized LOC100506282	−1.24	1.17E-01
NBAT1	neuroblastoma associated transcript 1	−1.24	1.41E-01
NIPAL2	NIPA like domain containing 2	1.24	1.28E-01
LINC00638	long intergenic non-protein coding RNA 638	1.24	1.73E-01
CEBPB	CCAAT enhancer binding protein beta	1.23	1.98E-02
EMILIN2	elastin microfibril interfacer 2	1.23	1.40E-01
ZNF441	zinc finger protein 441	1.23	1.54E-04
C17orf51	chromosome 17 open reading frame 51	1.23	5.46E-03
ERVMER34–1	endogenous retrovirus group MER34 member 1, envelope	1.23	9.78E-02
SYBU	syntabulin	1.23	2.48E-02
ABHD15	abhydrolase domain containing 15	1.23	6.63E-02
QRICH2	glutamine rich 2	1.23	3.25E-03
LOC100287015	uncharacterized LOC100287015	−1.23	8.50E-02
JDP2	Jun dimerization protein 2	1.22	1.09E-01
HLX	H2.0 like homeobox	1.22	1.70E-01
CHST8	carbohydrate sulfotransferase 8	1.22	3.09E-05
LINC01003	long intergenic non-protein coding RNA 1003	−1.22	5.58E-03
WSCD2	WSC domain containing 2	1.22	1.91E-01
LINC00526	long intergenic non-protein coding RNA 526	−1.22	1.79E-01
SMAD6	SMAD family member 6	1.22	5.67E-07
DGAT2	diacylglycerol O-acyltransferase 2	1.22	1.69E-01
SESN2	sestrin 2	1.22	3.83E-03
MIR17HG	miR-17–92a-1 cluster host gene	−1.22	1.44E-02
HAPLN3	hyaluronan and proteoglycan link protein 3	1.22	1.51E-01
CCDC3	coiled-coil domain containing 3	1.22	2.65E-02
CCDC61	coiled-coil domain containing 61	−1.21	7.70E-02
TGFBR2	transforming growth factor beta receptor 2	1.21	1.09E-01
KCTD16	potassium channel tetramerization domain containing 16	1.21	1.30E-01
SLCO1A2	solute carrier organic anion transporter family member 1A2	−1.21	1.92E-01
TUFT1	tuftelin 1	1.21	3.27E-03
RWDD2B	RWD domain containing 2B	−1.21	5.38E-07
CLYBL	citrate lyase beta like	−1.21	1.36E-01
PXDC1	PX domain containing 1	1.21	1.60E-01
WNT6	Wnt family member 6	1.21	9.10E-02
SWI5	SWI5 homologous recombination repair protein	1.21	1.01E-01
ADAMTS9-AS2	ADAMTS9 antisense RNA 2	−1.21	1.59E-01
LOC150051	uncharacterized LOC150051	−1.21	1.41E-01
TSPEAR-AS2	TSPEAR antisense RNA 2	−1.21	3.27E-09
FZD4	frizzled class receptor 4	1.21	3.15E-02
GABPB1-IT1	GABPB1 intronic transcript	−1.21	8.43E-07
CCEPR	cervical carcinoma expressed PCNA regulatory lncRNA	−1.21	2.07E-04
TYSND1	trypsin domain containing 1	1.20	7.20E-04
NOTCH1	notch 1	1.20	9.73E-02
FUT10	fucosyltransferase 10	1.20	1.44E-02
CHST15	carbohydrate sulfotransferase 15	1.20	4.07E-02
ONECUT2	one cut homeobox 2	1.20	1.76E-01
ADCK2	aarF domain containing kinase 2	1.20	4.03E-03
ADRA2C	adrenoceptor alpha 2C	1.20	4.83E-03

Results

To determine the effects of repeated ethanol exposure on SH-SY5Y neuroblastoma cells, we used a 3-week chronic intermittent exposure to ethanol (CIE) paradigm. Cells were exposed to 40 mM ethanol (or control) for 4 h per day on 4 successive days, followed by 3 days without ethanol, for 3 weeks. The repeated exposures were akin to three weeks of binge drinking on 4 successive days with 3 abstinent days in between. The 40 mM ethanol dose is equivalent to a blood alcohol concentration of 0.18%, which can be surpassed in a bout of high intensity drinking. On the last week, after 3 days of exposure, half of the ethanol-treated cells were withdrawn for 24 h while the other half received the fourth 4 h exposure (Supplementary Figure 1), to determine how much recovery occurred in that time. The only ADH detectably expressed in these cells is ADH5, which has a very high K_m for ethanol and is not likely to carry out significant metabolism; mRNA levels are the same under all 3 conditions and very similar to the levels seen in various brain regions in GTEx.

Ethanol Exposure vs. Control

In these SH-SY5Y cells, 12,599 genes were expressed significantly over background, 97% of which were also expressed in at least one human brain region. CIE led to the differential expression of 1,498 genes at FDR < 0.20 (Supplementary Table S1; all data have been deposited into GEO, accession number GSE139408); 744 (50%) were decreased in expression. Most of the changes were small, with 91% (1365) having an absolute fold change ≤ 1.2 (full list in Supplementary Table S1). The 133 genes with absolute fold change ≥ 1.2 are listed in Table 1. Among the differentially expressed genes were those related to neuron function or development, including AMPA glutamate receptor subunit (GRIA2), ephrin B2 (EFNB2) and two ephrin receptors (EPHA2, EPHB4), genes encoding proteins that degrade neurotransmitters (MAOB and COMT), and dopa decarboxylase which produces dopamine and serotonin. Norepinephrine transporter SLC6A2 and glycine transporter SLC6A9 were increased in expression, while GABA transporter SLC6A11 was decreased; all three remove their respective neurotransmitter from the synaptic cleft. Several voltage-gated potassium channel subunits were increased in expression as were three calcium channel subunits. Calcium subunit CACNA2D2 had decreased expression. Several genes that are important for signaling were found in multiple affected pathways; these include MAP/ERK kinases (MAPK1, MAPK12, MAP2K1, MAP2K12, MAP3K1, MAP3K3), GAB1 and IRS1 that activate ERK signaling by binding to GRB2, and adenylate cyclases (ADCY1, ADCY3, ADCY6 and ADCY9; all with increased expression). Genes involved in neural development, axon growth, synaptogenesis, etc. were affected by ethanol, including growth factor receptors FGFR1 and FGFR2, WNT5A, and four frizzled receptors, FZD2, FZD4, FZD5 and FZD7 (all increased).

IPA identified 114 pathways that met FDR < 0.20 and contained at least 5 differentially expressed genes (Supplementary Table S2). [Several of the pathways were redundant, sharing all or most of their genes with another pathway.] The most significant pathway that was affected by CIE was cholesterol biosynthesis; the rate-limiting gene for cholesterol production, HMGCR, and 10 other genes in that pathway showed decreased expression after CIE. Pathways that showed overall activation (i.e., had positive z-scores) include dopamine receptor signaling, TGF-β signaling, and synaptogenesis signaling. Other affected pathways important in neurons were reelin signaling in neurons, axonal guidance, Ephrin A signaling, and GABA receptor signaling.

The Unfolded Protein Response pathway had many genes increased in expression. Many of these pathways are driven by a core set of genes: the adenylate cyclases, fibroblast growth factor receptors, WNTs, MAP kinases and frizzled receptors.

Upstream regulator analysis identified putative proteins and chemicals that may be responsible for the pattern of changes in expression (Supplementary Tables S3 & S4, respectively). This analysis indicated increased activity for TP53, NFκB, INSIG1, cAMP and ATF4, genes that regulate large numbers of other genes, as well as dexamethasone, which indicates changes similar to cortisol treatment. Regulators that appear to be less active include HIF1A, RARA, insulin receptor, estrogen receptor and SREBF1/2. SREBF1/2 regulate the expression of genes in the cholesterol pathway.

Effects of Withdrawal

CIE-treated samples that were withdrawn from ethanol for the 24 h before harvest were very similar to the controls that had not been exposed to ethanol at all during the 3 weeks. Only 9 genes differed from the unexposed cells at FDR < 0.20. Four of these had changes similar to the effects of ethanol treatment without withdrawal, while the other 5 did not have significant changes after ethanol alone (Supplementary Table S1).

The comparison of ethanol exposed cells to cells withdrawn for 24 h reinforced this: it was very similar to the comparison of exposed cells to the control cells. The correlation of expression levels was quite good (Figure 1; the coefficient for the regression line was 1.1, R²=0.83), suggesting that the recovery was mostly but not quite complete after the 24 h withdrawal.

Figure 1.

Scatter plot of log₂ ratios comparing ethanol treatment to control (y-axis) and ethanol treatment to ethanol plus 24h withdrawal (x-axis). Genes shown differ in expression between ethanol vs. control at FDR < 0.20.

We analyzed the genes in Table 1 using FUMA GENE2FUNC analysis (Watanabe et al., 2017). This highlighted GO cellular components neuron, synapse, post-synapse, and synapse part (Supplementary Table S5). Other functional entries include epithelial to mesenchymal transition, pluripotency, neurogenesis, neuron cell death and neuron differentiation

Discussion

A series of repeated exposures of SH-SY5Y neuroblastoma cells to binge-drinking levels of ethanol over 3 weeks (CIE) led to changes in the expression of many genes, although most changes were of small magnitude (Supplementary Table S1). These fell into a number of pathways relevant to neuronal cells, many of which have been shown to be affected in the brains of animals exposed to CIE. Perhaps surprisingly, nearly all of the changes returned toward control levels after only 24 h without ethanol.

The CIE treatment led to lower expression of most of the genes in the cholesterol biosynthesis pathway. Three brain regions, dorsal raphe nucleus, periaqueductal gray, and ventral hippocampus, from CIE treated adolescent P rats also had lower expression of cholesterol synthesis genes (McClintick et al., 2015, 2016, 2018). Lower expression of the genes in cholesterol biosynthesis pathway was also detected in lymphoblastoid cell lines (LCLs) from alcoholics, compared to cells from controls (McClintick et al., 2019), although 48 h treatment with 75mM ethanol to LCLs did not result in decreased expression of these genes (McClintick et al., 2019). Lower expression was also found in post-mortem frontal cortex, basolateral amygdala and temporal cortex of alcoholics compared to controls (Mayfield et al., 2002; Ponomarev, Wang, Zhang, Harris, & Mayfield, 2012; Sokolov, Jiang, Trivedi, & Aston, 2003).

Genes affected by CIE fell into several pathways associated with neuronal development and/or plasticity (Supplementary Table S2). Fifty-eight genes in the axonal guidance pathway were affected, among which 38 were increased in expression (including frizzled and ephrin receptors, semaphorins and plexins), but the tubulins were decreased in expression. Tubulins make up microtubules necessary for development and movement of neurons (Breuss, Leca, Gstrein, Hansen, & Keays, 2017). Related pathways that share some of the same genes include reelin signaling in neurons, Ephrin A signaling, synaptogenesis signaling, and Wnt/β-catenin signaling. Reelin signaling is important in neuronal migration and modulation of synaptic functions (Lee & D’Arcangelo, 2016). TGF-β signaling, involved in late-stage adult neurogenesis (He et al., 2014), showed an overall increase in activity. All of these pathways except synaptogenesis were affected in at least one brain region of the CIE treated P rats (Supplementary Table S2) (McClintick et al., 2015, 2016, 2018). TGFβ signaling was also increased in the LCLs of alcoholics (McClintick et al., 2019). There is substantial overlap of genes within these pathways, including multiple Wnts and their Frizzled receptors, FGF receptors, ephrin receptors, semaphorins and adenylate cyclases.

Chronic ethanol treatment also affected genes that are associated with neurotransmission. Dopamine signaling was increased, including DDC which produces dopamine from l-dopa, but DBH, which converts dopamine to norepinephrine, is also increased in expression. Increased expression of ADRA2C, one of the α2-adrenergic autoreceptors, could attenuate norepinephrine release, an effect that could be amplified by the increased expression of the norepinephrine transporter, SLC6A2. Increases in the glycine transporter SLC6A9 and serine transporter SLC1A4 could dampen glutamate signaling via the NMDA receptor, because glycine and or D-serine act as co-agonists of that receptor; the dampening could be amplified because the glutamate transporter SLC1A2 is also increased. Glycine can also act as a neurotransmitter. GRIA2, the glutamate ionotropic receptor AMPA type subunit 2, was decreased. The GABA pathway was affected; although the analysis did not predict a direction, the GABA transporter SLC6A11 was decreased, which could mean less uptake of GABA and more activity. Decreased expression of SLC6A11/GAT3 in sweet drinking rats caused them to increase their intake of ethanol, and SLC6A11 was decreased in expression in the amygdala of post-mortem brains (Augier et al., 2018). In CIE treated P rats, the GABA, serotonin, and glutamate pathways were strongly affected, especially in the dorsal raphe nucleus (McClintick et al., 2015, 2016, 2018). In the DRN there were large decreases in most of the genes of the serotonin pathway, and more subtle changes in glutamate and GABA subunits.

Alcohol has a known proinflammatory effect in multiple organs including the brain (Crews & Vetreno, 2016; J. Mayfield, Ferguson, & Harris, 2013; McClintick et al., 2013) and lymphoblastoid cells (McClintick et al., 2014; McClintick et al., 2019). A few pathways and upstream regulators related to inflammation were increased in the CIE cells, including the key regulators NFκB and cytokines C5, CCL5 and IL11. LTB4R, a receptor for the proinflammatory leukotriene LTB4, had a large increase in expression. Although no Toll-like receptors (TLRs) were detectably expressed in these cells, genes downstream of TLR4 were affected, possibly via effects of related activators such as C5 and NFκB. Both WNT5A and its receptor FZD5 increased in expression; this combination regulates inflammatory responses, linking innate and adaptive immune responses (Blumenthal et al., 2006). The pair can also activate the Wnt5a/JNK pathway which is important in neural differentiation from stem cells (Jang, Park, & Jeong, 2015). The significant FUMA gene sets include 42 sets from Immunologic signatures (Supplementary Table S5).

Several stress-responsive pathways were significantly affected, with most affected genes having increased expression. These include the unfolded protein response pathway, endoplasmic stress response and the NRF2 mediated oxidative stress pathway. These were also significantly affected in the LCLs treated with ethanol and in LCLs of alcoholics compared to controls (McClintick et al., 2019). The Hif1α pathway, which responds to low oxygen stress, is decreased, which might increase the impact of the other stress-responsive pathways. Nearly 200 genes downstream of TP53 were affected. These include cyclins and several genes in the minichromosome maintenance complex (all decreased) which are involved in cell cycle checkpoint and control. Also downstream of TP53 are SREBF1 and cholesterol biosynthesis genes.

Even after 3 weeks of repeated exposure to 40 mM ethanol, a single 24 h withdrawal largely reversed the effects of ethanol. There were nearly no genes that significantly differed between the withdrawn cells and the untreated cells (Supplementary Table S1), and the comparison of CIE-treated to the withdrawn cells is very similar to the comparison of CIE-treated to control (unexposed) cells (Supplementary Figure 2). The latter pair of comparisons shows that the return to baseline is not complete, but clearly substantial. There have been several studies of the timecourse of the effects of CIE in C57BL/6J mice (exposed via vapor chamber). There were many changes in gene expression detected immediately after 4 weeks of CIE exposure, with a slight drop in differentially expressed mRNAs and a more pronounced drop in differentially expressed miRNAs 8 h later and a further drop after 120 h (Osterndorff-Kahanek et al., 2015, 2018). Those experiments provided evidence for overrepresentation of genes characteristic of astrocytes and microglia in the amygdala and to a lesser extent in the nucleus accumbens at the early times, but no evidence of overexpression of genes characteristic of neurons. In the pre-frontal cortex, in contrast, neuronal genes were overrepresented along with astrocytes and, to a lesser degree, microglia (Osterndorff-Kahanek et al., 2015). A similar experiment looked only at 0 and 8 h after CIE exposure, and found many differences in gene expression in pre-frontal cortex but few in hippocampus or nucleus accumbens; the number of differentially expressed genes in pre-frontal cortex dropped markedly by 8 h (Melendez et al., 2012). Another experiment showed that pre-frontal cortex had the most differences at time 0, dropping by half at 8 h and to about 10% by 72 h (Smith et al., 2016). These experiments reveal much complexity in the timecourse and nature of genes affected, and differences among brain regions. They also suggest changes in the cellular composition of different regions at different times. Our study of a single cell type removes these complexities and focuses on understanding how a single cell type is affected in a single environment. The recovery toward baseline is a common feature of all of these experiments.

Examining the effects of CIE exposure on neuroblastoma cells allowed us to measure changes in neurotransmitter systems that cannot be studied in non-neural cell lines. There are limitations, however. The lack of other cell types found in brain could affect their response to ethanol, as could the lack of the spatial differentiation in brain.

What may be surprising is that pathways related to neuronal development and connectivity appear to be activated, with most genes in axonal guidance and synaptogenesis pathways increased in expression. The cholesterol biosynthesis pathway was decreased. TP53-controlled genes included some affecting the cell cycle, which may indicate that cell cycle were being slowed by checkpoint mechanisms. Stress systems like the unfolded protein response were activated; NFκB1 expression is increased and genes downstream of NFκB were affected. Another surprising finding was that the many changes evoked by the CIE exposure over 3 weeks returned nearly to baseline after only 24 h of withdrawal. It is possible that more lasting changes could result from exposures over a much longer period, or that chronic heavy exposure of individuals might lead to changes in connectivity within the brain, or might predispose to stronger effects of ethanol on reexposure.

Supplementary Material

1

Supplementary Figures & Tables

Supplementary Figure 1: Schematic of method for ethanol exposure.

2

Supplementary Figure 4. Scatter plot of the log2 ratios for all 3 comparisons using genes that were significantly changed ethanol vs. control, FDR< 0.20. Ethanol/control (x-axis), Ethanol/WD (Y-axis) orange, WD/control (Y-axis) blue

3

Supplementary Figure 2: Multidimensional scaling plot of all samples showing outlier in replicate 3 samples.

4

Supplementary Figure 3: Multidimensional scaling plot of all samples after replicate 3 samples were removed. Dimensions 1 and 4 are shown. Demonstrates differences between replicates and treatment groups.

5

Supplementary Table S1. Genes affected by ethanol only treatment vs. control or ethanol only vs ethanol + WD, FDR<0.20 in at least one comparison. Results shown for ethanol only vs. control, ethanol + withdrawal vs. control, ethanol only vs. ethanol + withdrawal, fold, pvalue and false discovery rate (FDR) shown for each. This table contains all genes which met FDR< 0.20 for any of the 3 comparisons. Bolded results for each comparison show those meeting FDR< 0.20 for that comparison. FPKM is fragments per kilobase per million mapped reads, average for each group. GTEX, largest average TPM among all brain regions listed in GTEx (https://gtexportal.org/home/). LCL 48h: results for 48 h 70 mM ethanol treatment of LCLs, PMID 30639126, McClintick et al., 2019. Bold indicates significant at FDR< 0.05. PM Brain: From a list of genes differentially expressed in at least one of 13 Post-mortem studies of brain tissue from alcoholics vs. controls.

Supplementary Table S2. Pathways associated with ethanol only treatment, FDR< 0.20. Ingenuity pathway analysis results for genes affected by ethanol only treatment vs. control with FDR< 0.020. Results are displayed for pathways significant at FDR< 0.20. Overlapping pathways are displayed after the most significant pathway and shaded with pale gray.

Supplementary Table S3. Upstream regulators - genes, proteins, protein complexes Ingenuity pathway analysis upstream results for genes affected by ethanol only treatment vs. control with FDR< 0.20. Fold changes, regardless of significance, are given for genes expressed in these cells. Blanks for fold indicates that this gene did not meet the threshold for expression.

Supplementary Table S4. Upstream regulators - chemicals, drugs Ingenuity pathway analysis upstream results for genes affected by ethanol only treatment vs. control with FDR< 0.20. S4 lists results for exogenous chemicals, drugs or proteins; fold change is irrelevant.

Supplementary Table S5. FUMA gene sets FUMA gene sets that were significant (FDR<0.20, 5 or more genes in the set) based upon the genes in Table 1, FDR<0.20 absolute fold ≥ 1.2.

Highlights.

• Chronic intermittent ethanol exposure of neuroblastoma cells affected expression of 1,498 genes.

• Gene expression returns to near baseline after 24 hour withdrawal from CIE.

• Expression changes could affect glutamate transmission via AMPA and NMDA receptors.

• Genes for cholesterol biosynthesis are decreased in expression.

• Axonal guidance, reelin signaling and synaptogenesis pathways were affected.