Transcriptional Profiling of Whisker Follicles and of the Striatum

International Journal of Molecular Sciences

Article Transcriptional Proﬁling of Whisker Follicles and of the Striatum in Methamphetamine Self-Administered Rats

1, 2, 1 3 1, Won-Jun Jang † , Taekwon Son †, Sang-Hoon Song , In Soo Ryu , Sooyeun Lee * and Chul-Ho Jeong 1,* 1 College of Pharmacy, Keimyung University, Daegu 42601, Korea; [email protected] (W.-J.J.); [email protected] (S.-H.S.) 2 Research Institute of Pharmaceutical Sciences, College of Pharmacy, Seoul National University, Seoul 08826, Korea; [email protected] 3 Substance Abuse Pharmacology Group, Korea Institute of Toxicology, Daejeon 34114, Korea; [email protected] * Correspondence: [email protected] (S.L.); [email protected] (C.-H.J.) These authors contributed equally to this work. † Received: 28 August 2020; Accepted: 20 November 2020; Published: 23 November 2020

Abstract: Methamphetamine (MA) use disorder is a chronic neuropsychiatric disease characterized by recurrent binge episodes, intervals of abstinence, and relapses to MA use. Therefore, identification of the key genes and pathways involved is important for improving the diagnosis and treatment of this disorder. In this study, high-throughput RNA sequencing was performed to find the key genes and examine the comparability of gene expression between whisker follicles and the striatum of rats following MA self-administration. A total of 253 and 87 differentially expressed genes (DEGs) were identified in whisker follicles and the striatum, respectively. Multivariate and network analyses were performed on these DEGs to find hub genes and key pathways within the constructed network. A total of 129 and 49 genes were finally selected from the DEG sets of whisker follicles and of the striatum. Statistically significant DEGs were found to belong to the classes of genes involved in nicotine addiction, cocaine addiction, and amphetamine addiction in the striatum as well as in Parkinson’s, Huntington’s, and Alzheimer’s diseases in whisker follicles. Of note, several genes and pathways including retrograde endocannabinoid signaling and the synaptic vesicle cycle pathway were common between the two tissues. Therefore, this study provides the first data on gene expression levels in whisker follicles and in the striatum in relation to MA reward and thereby may accelerate the research on the whisker follicle as an alternative source of biomarkers for the diagnosis of MA use disorder.

Keywords: methamphetamine; drug reward; RNA sequencing; drug addiction

1. Introduction Methamphetamine (MA) is a highly addictive psychostimulant whose abuse frequently leads to physical and psychological dependence. In recent years, there was a dramatic increase in the use of MA in North America [1] and in East Asian countries, including Korea [2]. Preclinical and clinical studies indicate that MA exposure results in substantial physical and psychiatric impairments such as depression, cognitive deﬁcits, and psychotic behavior [3]. Repeated consumption of drugs such as cocaine and MA can cause the transition from casual to compulsive drug use, and this transition is thought to be initiated by neuroadaptive changes in brain circuits involved in drug rewards and cognitive processes that regulate habitual behaviors [4,5]. These neuroadaptive changes

Int. J. Mol. Sci. 2020, 21, 8856; doi:10.3390/ijms21228856 www.mdpi.com/journal/ijms Int. J. Mol. Sci. 2020, 21, 8856 2 of 20 are commonly caused by complicated alterations of gene expression implicated in transcriptional regulation, synaptic plasticity, and intracellular signaling pathways in brain regions related to drug addiction [6,7]. Therefore, it may be worthwhile to investigate the change of gene expression patterns in the striatum during MA self-administration for speeding up the diagnosis and treatment of MA use disorder. There are major obstacles to the search for novel relevant biomarkers in the brain: primarily the difficulty with obtaining brain tissue from living donors and the lack of reliable experimental animal models. The brain is an ectodermal tissue and shares its developmental origin with scalp hair follicles, which are readily accessible mini-organs in the skin. Recently, Yoshikawa and colleagues proposed hair follicles as a beneficial genetic resource for the identification of potential biomarkers of autism and chronic psychosis (as in schizophrenia) [8]. After co-examining the gene expression in hair follicles and post-mortem brain tissue samples from patients with autism, they suggested that hair follicles may serve as an alternative apparatus reflecting the state of the central nervous system [8]. Consistently with this finding, our previous data have revealed that alterations of gene expression patterns in the whisker follicles of MA self-administered rats may serve as useful indicators of MA rewards [9]. However, to date, large-scale evaluation and comparison of the molecular changes have not been conducted between hair follicles and the striatum. High-throughput RNA sequencing (RNA-seq) is a high-productivity transcriptome profiling method designed to quantify changes in RNA expression, thereby revealing the functional output of a disease state such as a drug use disorder. To understand how to diagnose and treat MA addiction, it is crucial to find hub genes and to delineate their neurobiological functions related to MA rewards and addiction. Numerous relevant biological events point to the need for large-scale analyses to uncover the diverse pathways and mechanisms that are most prominently involved in MA addiction. In this study, a rat MA self-administration model [10] was used to study potential molecular phenomena underlying MA rewards, and the collected gene information was analyzed by means of several bioinformatics tools, such as multivariate analysis and network-based topological analysis. This study provides data on overall gene expression in whisker follicles and in the striatum of the same rats following MA self-administration. Using genome-wide quantification of transcripts, we aimed to find and clarify the role of concordantly and differentially expressed genes (DEGs) in the two tissues and to reveal the gene–gene interaction mechanisms behind the rewarding effect of MA.

2. Results

2.1. Methamphetamine Self-Administration This procedure was performed on rats according to the experimental schedule illustrated in Figure1A. Figure1B,C show the number of active and inactive lever responses during saline (SA) or MA self-administration (2 h/day, 16 days) under the fixed-rate 1 (FR1) schedule. The MA group showed clear goal-oriented behavior with a significantly higher number of active lever presses than that of the SA group (self-condition: p < 0.0001; subjects (matching): p < 0.001). No significant differences in the numbers of inactive lever presses were observed between both groups (self-condition: p = 0.3177; subjects (matching): p < 0.0001). The number of drug infusions was significantly higher in MA self-administering rats than in SA self-administering rats (self-condition: p < 0.0001; subjects (matching): p < 0.0001) and showed <10% variation during the last 3 days of the experiment (Figure1D). The number of MA infusions on the final day was 23 4.2 (mean SEM (standard error of mean)). ± ± Two-way repeated measure analysis of variance (ANOVA) uncovered a significant difference in the number of infusions between SA and MA self-administered rats. In addition, we assessed the expression of several marker proteins in the dopamine signaling pathway in the striatum of rats exposed to MA self-administration. Our data revealed that the expression levels of a D1 dopamine receptor (D1DR) and brain-derived neurotropic factor (BDNF) that play a role in the regulation of synaptic plasticity were significantly increased in the striatum of the methamphetamine self-administered rats compared Int. J. Mol. Sci. 2020, 21, 8856 3 of 20 with those of the saline control. In addition, the phosphorylation of tyrosine hydroxylase (p-TH, ser40) was significantlyInt. J. Mol. Sci. 2020 increased,, 21, x FOR but PEER the REVIEW total level was not changed in the striatum of the methamphetamine 3 of 20 self-administered rats (Figure S1). These results are consistent with previous reports indicating that consistent with previous reports indicating that these proteins could be regulated by exposure to MA these proteins could be regulated by exposure to MA or other drugs of abuse such as cocaine [11,12]. or other drugs of abuse such as cocaine [11,12].

(A)

Self-administration: Sampling 2h/day, FR1, 20s time-out, 0.05 mg/inf MA or SA SA (n=6) Whisker follicle • SA: n=6 MA (n=6) • MA: n=6 Striatum

-10 Food training -7 Surgery -6 Recovery 016Self- • SA: n=6 (3 days) (1 day) (7 days) administration • MA: n=6 (16 days) (B) (C) **** **** **** **** **** **** **** *** ** **** *** ** ** #### Number of active lever press lever active of Number

(D)

Figure 1. Methamphetamine (MA) self-administration. (A) The schematic diagram of the experimental timelineFigure of saline 1. Methamphetamine (SA) or methamphetamine (MA) self-administration. (MA) self-administration (A) The schematic (n = 6 per diagram group). of (B ,Cthe) The experimental timeline of saline (SA) or methamphetamine (MA) self-administration (n = 6 per group). numbers of active and inactive lever presses by SA or MA self-administration. (D) The number of (B,C) The numbers of active and inactive lever presses by SA or MA self-administration. (D) The infusions during MA self-administration at 0.05 mg/kg (2 h/day, ﬁxed-rate 1 (FR1), 20 s time-out). number of infusions during MA self-administration at 0.05 mg/kg (2 h/day, fixed-rate 1 (FR1), 20 s Data were displayed as mean SEM (n = 6/group) and were subjected to two-way repeated measure time-out). Data were displayed± as mean ± SEM (n = 6/group) and were subjected to two-way repeated ANOVA,measure followed ANOVA, by Bonferroni’sfollowed by Bonferroni’s multiple comparisonmultiple comparison post hoc post test: hoc ** test:p < **0.01, p < 0.01, *** ***p

2.2. RNA-Seq2.2. RNA-Seq Analysis Analysis and and Gene Gene Expression Expression Proﬁling Profiling SamplesSamples were were sequenced sequenced on on the the Illumina Illumina HiSeq 2000 2000 platform, platform, with with paired-end paired-end 101 bp 101 reads bp for reads for mRNA-seqmRNA-seq usingusing the the TruSeq TruSeq SBS SBS Kit Kitv3 (Illumina). v3 (Illumina). The raw The imaging raw imagingdata were datatransformed were transformed by base- by base-callingcalling into sequence into sequence data and data stored and in storedFASTQ format. in FASTQ After format. removing After the adapters, removing approximately the adapters, approximately39.22 million 39.22 reads million could reads be detected could in be each detected sample. in eachAlignment sample. results Alignment from TopHat2 results showed from TopHat2 that showedthe that unique the mapping unique mapping rate was 91.52% rate was on 91.52%average, on and average, 17,711 genes and 17,711 could be genes found could when be compared found when

Int. J. Mol. Sci. 2020, 21, 8856 4 of 20

comparedInt. J. Mol. against Sci. 2020,the 21, x ratFORreference PEER REVIEW genome rn6. Integrative RNA-seq profiling was performed 4 of 20 on whisker follicle samples and striatum samples collected from six SA self-administered and six MA self-administeredagainst the rat rats.reference A total genome of 87 genesrn6. Integrat (falseive discovery RNA-seq rate profiling (FDR) was<0.05, performed|fold change on whisker (FC)| 1.5) follicle samples and striatum samples collected from six SA self-administered and six MA self-≥ were found to be differentially expressed in the striatum in the MA group relative to the SA group administered rats. A total of 87 genes (false discovery rate (FDR) <0.05, |fold change (FC)| ≥ 1.5) were (Table S1), and 253 DEGs with FDR< 0.05 and |FC| 1.5 were identified in whisker follicles (Table S2). found to be differentially expressed in the striatum≥ in the MA group relative to the SA group (Table MA self-administeredS1), and 253 DEGs with rats FDR< showed 0.05 and a significant |FC| ≥ 1.5 down-regulationwere identified in whisker of 14 genesfollicles in (Table thestriatum S2). MA and 182 genesself-administered in the whisker rats showed follicle, a whereas significant 73 down-reg genes inulation the striatum of 14 genes and in 71 the genes striatum in the and whisker 182 genes follicle werein significantly the whisker up-regulated follicle, whereas ascompared 73 genes in with the SAstriatum self-administered and 71 genes rats.in the The whisker hierarchical follicle clusteringwere revealedsignificantly that the up-regulated identified genes as compared were highly with consistentSA self-administered between tworats. groupsThe hierarchical of each whisker clustering follicle and therevealed striatum that tissuethe identified (Figure genes S2). Next,were highly we conducted consistent a between series of two bioinformatics groups of each analyses whisker tofollicle study the RNA-seqand the data striatum according tissue to (Figure the procedure S2). Next, presentedwe conducted in Figurea series2 .of We bioinformatics analyzed the analyses enriched to study function the RNA-seq data according to the procedure presented in Figure 2. We analyzed the enriched and pathway of up-regulated and down-regulated DEGs in the striatum (Figure3A–D) and the whisker function and pathway of up-regulated and down-regulated DEGs in the striatum (Figure 3 A–D) and follicle (Figure3E–H), respectively, using the Database for Annotation, Visualization, and Integrated the whisker follicle (Figure 3E–H), respectively, using the Database for Annotation, Visualization, Discoveryand Integrated (DAVID) Discovery online tool (DAVID) based online on the tool two ba complementarysed on the two complementary databases of databases Gene Ontology of Gene (GO) and theOntology Kyoto (GO) Encyclopedia and the Kyoto of Genes Encyclopedia and Genomes of Genes (KEGG). and Genomes The p -value(KEGG). and The gene p-value count and were gene set to <0.05count and were2 as set significant to <0.05 and thresholds ≥2 as significant for enrichment thresholds analysis for enrichment in DAVID analysis tool. in DAVID tool. ≥

Raw data

Trimming

Mapping to reference genome (UCSC rn6)

Transcript assembly

Abundance estimation

Nomalization

Differential expression test

Differential expressed genes Functional Enrichment Analysis (FDR<0.05, |FC|≥1.5) • Up-regulated DEGs (S:73, WF:71) • Down-regulated DEGs (S:14, WF:182) • Striatum: 87 genes • GO (BP, CC,MF), KEGG • whisker follicle: 253 genes

PPI Network Analysis Multivariate Analysis Network Analysis • Total DEGs • Up-regulated DEGs • Down-regulated DEGs PCA PPI Network • Functional Analysis

Correlation Analysis Cluster Analysis (MCODE) Centrality Analysis • Pearson correlation •Degreecutoff:10 • Degree • Correlation Coefficient •Haircut:on (cutoff: S=2, WF=5) (cutoff: 0.6) •Fluff:on • Betweenness centrality • Node density cutoff: 0.1 (Top20) • Node score cutoff: 0.2 • K-Core: 2 • Max. Depth: 100

• Striatum • Striatum • Striatum (30 genes) (3 clusters) (20 genes) • Whisker follicle • Whisker follicle • Whisker follicle (72 genes) (6 clusters) (20 genes)

• Striatum Functional (15 genes) Analysis • Whisker follicle of Clusters (62 genes)

• Striatum Functional Analysis between tissues (49 genes) •KEGGpathways • Whisker follicle • p-value cutoff: 0.05 (129 genes) • Kappa score: 0.3

Figure 2. The workflow of computational analysis of the RNA sequencing (RNA-seq) data. A workflow diagram for the quality control and differential gene expression analysis of the RNA-seq data from the striatum or from the whisker follicle of MA self-administered rats. Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 5 of 20

Figure 2. The workflow of computational analysis of the RNA sequencing (RNA-seq) data. A workflow diagram for the quality control and differential gene expression analysis of the RNA-seq data from the striatum or from the whisker follicle of MA self-administered rats.

2.3. Multivariate Analysis We analyzed 87 DEGs in the striatum and 253 DEGs in the whisker follicle from the RNA-seq data on the two tissues. To check the quality of the data and evaluate the magnitude of separation from the control, principal component analysis (PCA) was applied to the DEGs from the SA and MA groups of the two tissues. Expectedly, the PCA score plot of DEGs showed a much more clear separation between two groups in each tissue than that of total genes (Figure S3A–D). As shown in Figure S3C,D, 2D score scatter plots of PCA mostly occupied the space of principal component 1 (PC1) and principal component 2 (PC2). In the striatum, contributions of PC1 and PC2 to the maximum variance in the dataset were calculated and found to be 48.1% and 20.5%, respectively (Figure S3C). In addition, PC1 and PC2 in the whisker follicle covered 81.9% and 4.5% of the variance in the dataset, respectively (Figure S3D). In both tissues, most samples in the MA groups were clearly separated from the SA groups (Figure S3C,D). Next, to identify the critical genes that contribute to the separation between the groups, we performed correlation analysis and correlation pattern searching (Figure S3E,F). Pearson’s correlation coefficients for each gene were calculated. To screen out the critical genes, the correlation coefficient for each gene was limited to 0.6 or higher (relative coefficient r ≥ 0.6 is thought to denote a strong correlation). Thirty critical genes in the striatum (Table S3) and 72 critical genes in the whisker follicle (Table S4) were identified, and the top 25 were Int.represented J. Mol. Sci. 2020 by, 21means, 8856 of a correlation pattern searching tool (Figure S3G,H). The positive 5and of 20 negative correlations of the genes are indicated by the red and blue color, respectively.

(A) GO: Biological Process 10.0 25.0 Up-regulated p-value 8.0 20.0 Ratio 6.0 15.0

4.0 10.0 Ratio(%)

-Log10(p-value) 2.0 5.0

0.0 0.0 10.0 25.0 Down-regulated 8.0 20.0

6.0 15.0

4.0 10.0 Ratio(%)

-Log10(p-value) 2.0 5.0

0.0 0.0 aging ossification wound healing oxygen transport response to drug cell-cell signaling female pregnancy response to cAMP epnetoresponse hypoxia peptide cross-linking fat cell differentiation response to estradiol response to radiation inner ear development promoter proliferation response to calcium ion regulation of cell growth osteoblast differentiation response to light stimulus stimulus response to progesterone neural development retina collagen fibril organization response to glucocorticoid epnetoresponse hormone peptide collagen biosynthetic process hormone stimulus epnetoresponse substance organic chemical synaptic transmission epnetoresponse peroxide hydrogen cellular response to calcium ion response to mechanical stimulus polymerase II promoter chloride transmembrane transport regulation of type B pancreatic cell cellular response to glucose stimulus transcription from RNA polymerase II Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW cellular response to hormone stimulus 6 of 20 positive regulation of synapse assembly cellular response to corticotropin-releasing positive regulation of synaptic transmission positive regulation of fibroblast proliferation cellular response to fibroblast growth factor growth to response cellular fibroblast positive regulation of transcription from RNA

(B) GO: Cellular Component (C) GO: Molecular Function (D) KEGG 10.0 40.0 10 25 10 30 Up-regulated Up-regulated p-value Up-regulated 8.0 8 20 8 30.0 Ratio 20 6.0 6 15 6 20.0 4.0 4 10 4 Ratio(%) Ratio(%) Ratio(%) 10 p-value 10.0 -Log10(p-value) -Log10(p-value) 2.0 -Log10(p-value) 2 5 2 p-value Ratio Ratio 0.0 0.0 0 0 0 0 10.0 25.0 10 30 Down-regulated Down-regulated 8.0 20.0 8

heme binding 20 protein binding xgnbinding oxygen 6.0 15.0 iron ion binding 6 haptoglobin binding 4

4.0 10.0 Ratio(%) Ratio(%)

oxygen transporter activity 10 -Log10(p-value)

synaptic vesicle Nicotine addiction excitatory synapse xrclua space extracellular extracellular matrix blood microparticle extracellular region hemoglobin complex basement membrane yatcvscecycle Synaptic vesicle extracellular exosome endoplasmic reticulum Glutamatergic synapse African trypanosomiasis postsynaptic membrane PI3K-Akt signaling pathway synaptic vesicle membrane haptoglobin-hemoglobin complex proteinaceous extracellular matrix

Retrograde endocannabinoid signaling

(E) GO: Biological Process 10 25 Up-regulated p-value 8 Ratio 20 6 15

4 10 Ratio(%) 2 5 -Log10(p-value)

0 0 10 25 Down-regulated 8 20

6 15

4 10 Ratio(%) 2 5 -Log10(p-value)

0 0 translation vasculogenesis cell proliferation trachea formation aerobic respiration response to ethanol kidney development etd cross-linking peptide neuron differentiation response to hormone forebrain development neutrophil aggregation astrocyte development ATP metabolic process regulation of translation osteoblast differentiation mitochondrial translation erythrocyte homeostasis blood vessel development renal system development metanephros development vitamin A metabolic process eerlcortexcerebral development cellular oxidant detoxification regulation of cell proliferation skeletal system development substantia nigra development regulation of gene expression glutathione metabolic process smoothened signaling pathway embryonic digit morphogenesis positive regulation of translation peptidyl-cysteine S-nitrosylation ribosomal small subunit assembly positive regulation of epithelial cell… positive regulation of organ growth type B pancreatic cell development cellular phosphate ion homeostasis positive regulation of endothelial cell… response to reactive oxygen species nRAgie rRNA guided pseudouridine… snoRNA positive regulation of catalytic activity positive regulation of cell proliferation epithelial-mesenchymal cell signaling ebaosspu morphogenesis septum membranous etdlcsen S-trans-nitrosylation peptidyl-cysteine positive regulation of gene expression negative regulation of oligodendrocyte… positive regulation of peptide secretion positive regulation of peptidase activity positive regulation of transcription from… negative regulation of gene expression oiierglto fitiscapoptotic… of intrinsic regulation positive hydrogen ion transmembrane transport cellular response to nerve growth factor… ATP synthesis coupled proton transport ATP coupled proton synthesis eaierglto ftasrpinfrom… transcription of regulation negative ATP hydrolysis coupled proton transport positive regulation of transcription, DNA-… odontogenesis of dentin-containing tooth negative DNA-… regulationtranscription, of response to endoplasmic reticulum stress mitochondrial electron transport, ubiquinol… cellular response to growth factor stimulus positive regulation of branching involved in…

regulation of smooth muscle cell proliferation

(F) GO: Cellular Component Figure(G) 3. GO:Cont Molecular. Function (H) KEGG 20 50 20 25 25 15 Up-regulated Up-regulated p-value Up-regulated p-value 16 40 20 20 12 15 Ratio Ratio 12 30 15 15 9 10 8 20 10 10 6 Ratio(%) Ratio(%) Ratio(%) 5 -Log10(p-value) -Log10(p-value) -Log10(p-value) 4 p-value 10 5 5 3 Ratio 0 0 0 0 0 0 20 50 20 25 25 15 Down-regulated Down-regulated Down-regulated 16 40 20 20 12 15 12 30 15 15 9 10 8 20 10 10 6 Ratio(%) Ratio(%) Ratio(%) 5 -Log10(p-value) -Log10(p-value) 4 10 -Log10(p-value) 5 5 3

0 0 0 0 0 0 cytosol ribosome nucleolus cytoplasm Ribosome RNA binding Spliceosome myelin sheath mitochondrion GAIT complex focal adhesion mRNA binding keratin filament protein binding protein complex enzyme binding Prostate cancer onfe envelope cornified eaoi pathways Metabolic Parkinson's disease Alzheimer's disease nemdaefilament intermediate poly(A) RNA binding Basalcellcarcinoma Huntington's disease Lsm1-7-Pat1 complex extracellular exosome small ribosomal subunit precatalytic spliceosome rtolcn inProteoglycans cancer Hippo signaling pathway xdtv phosphorylation Oxidative Toll-like receptor 4 binding structural molecule activity peptidase activator activity protein C-terminus binding box H/ACA snoRNA binding box H/ACA snoRNP complex mitochondrial inner membrane ubiquitin protein ligase binding 1-yeslcooa complex spliceosomal U12-type cytosolic large ribosomal subunit cytosolic small ribosomal subunit structural constituent of ribosome translation initiation factor binding iohnra nemmrn space intermembrane mitochondrial box H/ACA telomerase RNP complex iohnra large subunit ribosomal mitochondrial mitochondrial small ribosomal subunit mitochondrial proton-transporting ATP… mitochondrial proton-transporting ATP… intracellular ribonucleoprotein complex iohnra eprtr chain I complex respiratory mitochondrial hydrogen ion transmembrane transporter… ubiquinol-cytochrome-c reductase activity o-looi fatty liverNon-alcoholic disease (NAFLD) iohnra eprtr chain III complex respiratory mitochondrial proton-transporting ATP synthase activity,… inln ahasrgltn pluripotency… pathways regulating Signaling AHdhdoeae(bqioe activity (ubiquinone) NADH dehydrogenase

Figure 3. Functional enrichment analysis of up- and down-regulated differentially expressed genes (DEGs) in the striatum (A–D) and whisker follicle (E–H). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway were analyzed in up- and down-regulated differentially expressed genes (DEGs) using Database for Annotation, Visualization, and Integrated

Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 6 of 20

heme binding 20 protein binding xgnbinding oxygen 6.0 15.0 iron ion binding 6 haptoglobin binding 4

4.0 10.0 Ratio(%) Ratio(%)

oxygen transporter activity 10 -Log10(p-value)

-Log10(p-value) 2.0 5.0 2 nui-iegot atrbinding factor growth insulin-like transcriptional activator activity, RNA… 0.0 0.0 platelet-derived growth factor binding 0 0 extracellular matrix structural constituent RNA polymerase II core promoter proximal… Malaria synapse cell junction collagen trimer synaptic vesicle Nicotine addiction excitatory synapse xrclua space extracellular blood microparticle extracellular matrix extracellular region hemoglobin complex basement membrane yatcvscecycle Synaptic vesicle extracellular exosome endoplasmic reticulum Glutamatergic synapse African trypanosomiasis postsynaptic membrane PI3K-Akt signaling pathway synaptic vesicle membrane haptoglobin-hemoglobin complex proteinaceous extracellular matrix

Retrograde endocannabinoid signaling

(E) GO: Biological Process 10 25 Up-regulated p-value 8 Ratio 20 6 15

4 10 Ratio(%) 2 5 -Log10(p-value)

0 0 10 25 Down-regulated 8 20

6 15

4 10 Ratio(%) 2 5 -Log10(p-value)

0 0 translation vasculogenesis cell proliferation trachea formation aerobic respiration response to ethanol kidney development etd cross-linking peptide neuron differentiation response to hormone forebrain development neutrophil aggregation astrocyte development ATP metabolic process regulation of translation osteoblast differentiation mitochondrial translation erythrocyte homeostasis blood vessel development renal system development metanephros development vitamin A metabolic process eerlcortexcerebral development cellular oxidant detoxification regulation of cell proliferation skeletal system development substantia nigra development regulation of gene expression glutathione metabolic process smoothened signaling pathway embryonic digit morphogenesis peptidyl-cysteine S-nitrosylation positive regulation of translation ribosomal small subunit assembly positive regulation of epithelial cell… positive regulation of organ growth type B pancreatic cell development cellular phosphate ion homeostasis positive regulation of endothelial cell… Int. J. Mol. Sci. 21 response to reactive oxygen species 2020, , 8856 rRNA guided pseudouridine… snoRNA 6 of 20 positive regulation of catalytic activity epithelial-mesenchymal cell signaling positive regulation of cell proliferation ebaosspu morphogenesis septum membranous etdlcsen S-trans-nitrosylation peptidyl-cysteine positive regulation of gene expression negative regulation of oligodendrocyte… positive regulation of peptide secretion positive regulation of peptidase activity positive regulation of transcription from… negative regulation of gene expression oiierglto fitiscapoptotic… of intrinsic regulation positive hydrogen ion transmembrane transport cellular response to nerve growth factor… ATP synthesis coupled proton transport ATP coupled proton synthesis eaierglto ftasrpinfrom… transcription of regulation negative ATP hydrolysis coupled proton transport positive regulation of transcription, DNA-… odontogenesis of dentin-containing tooth negative DNA-… regulationtranscription, of response to endoplasmic reticulum stress mitochondrial electron transport, ubiquinol… cellular response to growth factor stimulus positive regulation of branching involved in…

regulation of smooth muscle cell proliferation

(F) GO: Cellular Component (G) GO: Molecular Function (H) KEGG 20 50 20 25 25 15 Up-regulated Up-regulated p-value Up-regulated p-value 16 40 20 20 12 15 Ratio Ratio 12 30 15 15 9 10 8 20 10 10 6 Ratio(%) Ratio(%) Ratio(%) 5 -Log10(p-value) -Log10(p-value) -Log10(p-value) 4 p-value 10 5 5 3 Ratio 0 0 0 0 0 0 20 50 20 25 25 15 Down-regulated Down-regulated Down-regulated 16 40 20 20 12 15 12 30 15 15 9 10 8 20 10 10 6 Ratio(%) Ratio(%) Ratio(%) 5 -Log10(p-value) -Log10(p-value) 4 10 -Log10(p-value) 5 5 3

0 0 0 0 0 0 cytosol ribosome nucleolus cytoplasm Ribosome RNA binding Spliceosome myelin sheath mitochondrion GAIT complex focal adhesion mRNA binding keratin filament protein binding protein complex enzyme binding Prostate cancer onfe envelope cornified eaoi pathways Metabolic Parkinson's disease Alzheimer's disease nemdaefilament intermediate poly(A) RNA binding Basalcellcarcinoma Huntington's disease Lsm1-7-Pat1 complex extracellular exosome small ribosomal subunit precatalytic spliceosome rtolcn inProteoglycans cancer Hippo signaling pathway xdtv phosphorylation Oxidative Toll-like receptor 4 binding peptidase activator activity protein C-terminus binding structural molecule activity box H/ACA snoRNA binding box H/ACA snoRNP complex mitochondrial inner membrane ubiquitin protein ligase binding 1-yeslcooa complex spliceosomal U12-type cytosolic large ribosomal subunit cytosolic small ribosomal subunit structural constituent of ribosome translation initiation factor binding iohnra nemmrn space intermembrane mitochondrial box H/ACA telomerase RNP complex iohnra large subunit ribosomal mitochondrial mitochondrial small ribosomal subunit mitochondrial proton-transporting ATP… mitochondrial proton-transporting ATP… intracellular ribonucleoprotein complex iohnra eprtr chain I complex respiratory mitochondrial hydrogen ion transmembrane transporter… ubiquinol-cytochrome-c reductase activity o-looi fatty liverNon-alcoholic disease (NAFLD) iohnra eprtr chain III complex respiratory mitochondrial proton-transporting ATP synthase activity,… inln ahasrgltn pluripotency… pathways regulating Signaling AHdhdoeae(bqioe activity (ubiquinone) NADH dehydrogenase

FigureFigure 3. 3.Functional Functional enrichmentenrichment analysisanalysis of up- and down-regulated differentially diﬀerentially expressed expressed genes genes (DEGs)(DEGs) in in the the striatum striatum (A– D()A and–D) whiskerand whisker follicle (follicleE–H). Gene(E–H Ontology). Gene (GO)Ontology and Kyoto(GO) Encyclopediaand Kyoto ofEncyclopedia Genes and Genomesof Genes and (KEGG) Genomes pathway (KEGG) were path analyzedway were in up- analyzed and down-regulated in up- and down-regulated diﬀerentially expresseddifferentially genes expressed (DEGs) genes using (DEGs) Database using for Database Annotation, for Annotation, Visualization, Visualization, and Integrated and Integrated Discovery (DAVID) database. The horizontal axis represents various functional terms, including biological process

(BC) (A,E), cellular component (CC) (B,F), molecular function (MF) (C,G), and KEGG pathway (D,H). The vertical axis represents the p-value ( Log10) and ratio (%) of the number genes enriched in GO − and KEGG terms.

2.3. Multivariate Analysis We analyzed 87 DEGs in the striatum and 253 DEGs in the whisker follicle from the RNA-seq data on the two tissues. To check the quality of the data and evaluate the magnitude of separation from the control, principal component analysis (PCA) was applied to the DEGs from the SA and MA groups of the two tissues. Expectedly, the PCA score plot of DEGs showed a much more clear separation between two groups in each tissue than that of total genes (Figure S3A–D). As shown in Figure S3C,D, 2D score scatter plots of PCA mostly occupied the space of principal component 1 (PC1) and principal component 2 (PC2). In the striatum, contributions of PC1 and PC2 to the maximum variance in the dataset were calculated and found to be 48.1% and 20.5%, respectively (Figure S3C). In addition, PC1 and PC2 in the whisker follicle covered 81.9% and 4.5% of the variance in the dataset, respectively (Figure S3D). In both tissues, most samples in the MA groups were clearly separated from the SA groups (Figure S3C,D). Next, to identify the critical genes that contribute to the separation between the groups, we performed correlation analysis and correlation pattern searching (Figure S3E,F). Pearson’s correlation coefficients for each gene were calculated. To screen out the critical genes, the correlation coefficient for each gene was limited to 0.6 or higher (relative coefficient r 0.6 is thought to denote a ≥ strong correlation). Thirty critical genes in the striatum (Table S3) and 72 critical genes in the whisker follicle (Table S4) were identified, and the top 25 were represented by means of a correlation pattern searching tool (Figure S3G,H). The positive and negative correlations of the genes are indicated by the red and blue color, respectively.

2.4. Construction of Protein–Protein Interaction (PPI) Networks Next, we constructed protein–protein interaction (PPI) networks from total, up-regulated, and down-regulated DEGs of the two tissues, respectively, based on the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database [13] (Figure S4). The PPI networks were built using the Cytoscape software. The fold change of each node was expressed as color intensity. The up- and down-regulated networks of each tissue showed a separate form from the total network, but Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 7 of 20

Discovery (DAVID) database. The horizontal axis represents various functional terms, including biological process (BC) (A,E), cellular component (CC) (B,F), molecular function (MF) (C,G), and KEGG pathway (D,H). The vertical axis represents the p-value (−Log10) and ratio (%) of the number genes enriched in GO and KEGG terms.

2.4. Construction of Protein–Protein Interaction (PPI) Networks Next, we constructed protein–protein interaction (PPI) networks from total, up-regulated, and Int. J. Mol. Sci. 2020, 21, 8856 7 of 20 down-regulated DEGs of the two tissues, respectively, based on the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database [13] (Figure S4). The PPI networks were built using no networksthe Cytoscape were lost software. or newly The formed. fold change We of analyzed each node thewas enriched expressed function as color intensity. and pathway The up- of and the six networksdown-regulated using the DAVIDnetworks tool of each and summarizedtissue showed thea separate results form in Tables from S5the and total S6 network, with biological but no networks were lost or newly formed. We analyzed the enriched function and pathway of the six process (BP), cellular component (CC), molecular function (MF) functional tools as well as KEGG. networks using the DAVID tool and summarized the results in Table S5-S6 with biological process The total(BP), network cellular of thecomponent striatum (CC), consisted molecular of 69 function nodes (MF) and 72functional edges and tools appeared as well as to KEGG. contain The several total fragmentednetwork modules of the (Figure striatum4A). consisted Most of theof genes69 nodes that and make 72 upedges the and striatum appeared network to contain were found several to be up-regulatedfragmented by MA.modules The (Figure total network 4A). Most of of whisker the genes follicle that make consisted up the ofstriatum 223 nodes network and were 1191 found edges and lookedto be like up-regulated a single network by MA. thatThe total was network densely of packed whisker around folliclethe consisted center of (Figure 223 nodes4B). and Genes 1191 located edges near the centerand looked of the like whisker a single follicle network network that was tended densel toy be packed down-regulated around the bycenter MA, (Figure while some4B). Genes genes located outsidelocated thenear center the center were of found the whisker to be up-regulatedfollicle network by tended MA to be down-regulated by MA, while some genes located outside the center were found to be up-regulated by MA

(A)

Hbb Striatum Alas2 Tnnt2 Hba-a2 Alox15 Tnnt2 (Nodes: 69, Edges: 72) Alpl Vim Vim Bgn Bgn Col1a1 LOC1001348 71 Serpinf1 Col3a1 Fmod Slc22a6 Apln Serpinh1 Hbb Cdc42e p4 Col1 a1 Tagln Selenow Fxyd5 Alas2 Ier5 Hba-a2 Lrg1 Serpinf1 Alox15 Fmod Fzd2 Rbm3 Col3 a1 Mgp Pnisr Len g8

Serpinh1 Dusp1 LOC1001348 71 Htra1 Sult1a1 Igfbp2 Junb Cbln1 Tagln Aifm3 Sod3 Slc17a6 Lrg1 Slc17a7 Igf2 Fosb Arc Ccn5 Cxcl14 Fosl2 Nr4a1 Gria2 Egr2 Irs2 Rem2 Per1 Nr4a3 Mgp Egr4 Htra3 Module 2 Tmem132e Gjb6 Slc25a 25 Wfdc1 Gabra1 C1qa Ddit4 Rnase4Slc13a4Snhg11 C1qa Cd74 Klf10 C1qb Metrn Hapln4 Oxt Tmem119

Cd74 Module 3 C1qb

Dusp1

Junb Cbln1 Slc17a6 Slc17a7 Fosb Arc Fosl2 Nr4a1 Gria2 Egr2

Nr4a3 Egr4 Fold change Module 1 Gabra1 Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW -303 8 of 20

(B) Whisker follicle Ostc (Nodes: 223, Edges: 1191) Psmg4 Eif3hRpl34 Sp6 Polr1d Jpt1 Sec61bRps15a Mrps21 RT1-A2 Sec61g Rps24 Arl4a Mrpl52 Rpl22Rps29Rpl10aRpl27a Cnih 4 Ostc Rsrp1 Dnl z Ap2a1 Psm g 4 Rps17Rps9 Eif3kRps23 Mrpl41 Eif3hRpl3 4 Rps27aRpl37aRps3a Prcp Vamp8 Mrpl23 Cd k2ap 2 Sec61bRp s15 a Mr ps21 Cltb Rps10Rpl36al Rp s24 Lu c 7l 3 Sec61g Ar l4 a Eif3el1 Use1 Atl2 Sm im 2 2 Ar l4 d Rplp1 Mrpl54 Mrps35 Rpl2 2 RpRpl10 s29Rpl27 a a Mr pl5 2 Ap2a 1 Rps27l Mrps33Mrps18c Rp s1s9 7 Eif3kRp s23 Mr pl4 1 Rpl22l1 Chchd1 Rp s27Rpl37 a Rp a s3a Mr pl2 3 Ubc Polr2g Nop10 Ro r2 Cltb Vamp8 Rp s10Rpl36a l Plekhm2 R G D156394 1Tuft1 Sdf2l1 Eif3el1 Tmem14c Polr2k Magoh Rpl13Timm13 Creld2 Rplp 1 Mr pl5 4 Mr ps35 Tars Rps27 Mrps16 T s pan1 4 Rp s27 l MrMr ps33 ps18 c Sf3b5 Ar glu 1 Ub c Polr2g Rpl22lNop1 1 0 ElobCh chd 1 Polr2k Magoh Rpl1Timm 3 13 R G D131058 7 Serpina1 1Acy3 Snrpd2 UqcrhTmem256 Hn rnph 2 Sf3b5 Rp s27Bloc1s1Mr ps16 Mr pl3 0 Tra2b SrsSnrpd2 f3 Uq crhTmem256 Ciao2 b Psmb6 Ar l6ip 5 Tra2b Psm b 6 Gmfg S100a 1 Rbp 1 Gar1 Atp5f1eMrpl55 Uchl1 Tardbp Gar1 Atp5f1eMr pl5 5 R G D156564 1 Uchl1 Tardbp Lsm7 Lsm3Naa3T omm7 8 Pfdn5Tmem160 Bola1 Tomm7Pfdn5 Smo Atp5mgPfdn1 Pfdn1 Nme1 Uq cc2 March7 Atp5mg Elavl1 Anap c11Tbca ScSm and im 1 2 6 Lce1m Lce1f Nme1 Nrtn Ntrk2 Atp5pAtp5pd f Atp5mc1 Tbca Uqcc2 Uq crq Ndu fa7 Clp x Anapc11 Scand1 Ctnnb 1 Ube2 s Atp5mc3 Co x7 b Glrx5 Ar p c 5 l Atp5pd Rn f19 b UqNdu crbNdu fb2 fa2 Sub1 Atp5me Per1 Serpinb1Nr2c2ap a Xbp1 Sod1 Ndu fb5 Ptprs Ex o sc7 Fbxl15 M G C10564 9 Ube2s Uqcrq Plp1 App Nud t14 Ndu fa5 S100a 8 Ctnnb1 Atp5mc3 Alas1 Mbp Chu rc1 Fxyd4 Lce1l Rnf19b Gapdh Ps ene n Hin t2 Trir Sod1 Atp5me Nfkbia Co x17 Fbxl15 Cad m4 Tnfrsf11b S 100a 9 Dh cr7 Pgp Nudt14 Co x2 0 App Np w Ak t1 Mif Micos10 Gpcpd1 Sy ng r2 S lc 34a 2 Mbtps1 Il18 Gapdh Pr d x1 Ogt Abracl CstbSerpinb1 3 Nfkbia Module 1 Oga Il36a Fis1 Gpx4 Cebp b Tagln Id2 Csde1 Endou Cd9 9 Tmsb10 Ddi t4 Pdcd4

Seleno w Stfa3 Tm4sf4 Lrrc 8e De fb1 Foxc2 Fkbp2 Ivl Gstm5 Stmn4 Dpp 3 Ar pp1 9 Cmss1 Ggt6 Tpm4 L O C68042 8Sept7 Rbp 2 Irf2bp l Kr t3 6 Il17 b Hb b Maff

S 100a1 0 Kr t7 5 Em l2 If i27l2 bKr t ap19 -5

Dnlz Polr1d Jpt1

Mrps21

Mrpl41 Cnih4 Dnlz Mrpl23 Rsrp1 Mrpl54 Mrps33Mrps18cElobChchd1 Mrps21 Timm13Bloc1s1Mrps16 Mrpl30 Sf3b5 Luc7l3 Mrpl52 Snrpd2 UqcrhTmem256 Arl6ip5 Psmb6 Mrpl41 Lsm3 Atp5f1ePfdn5Mrpl55 RGD1565641 Mrpl23 Naa38Tomm7 Tmem160Atp5mgPfdn1 Anapc11Tbca Uqcc2 Smim26 Mrpl54 Mrps35 Tars Atp5pfAtp5pdAtp5mc1 Mrps33Mrps18cElobChchd1 Uqcrq Ndufa7 Arglu1 Polr2kPolr2gMagohNop10Timm13 Atp5mc3UqcrbNdufb2Cox7bNdufa2 Glrx5 Hnrnph2 Sf3b5 Bloc1s1Mrps16 Mrpl30 Sub1Sod1 Atp5me Srsf3Snrpd2 Tmem256 Ciao2b Ndufb5MGC105649 Tra2b Psmb6 Ndufa5 TardbpLsm7 Gar1 Lsm3 Pfdn5Mrpl55 RGD1565641Bola1 Churc1 Naa38Tomm7 Tmem160Pfdn1 Hint2 Nme1 PsenenCox17 Trir Elavl1 Anapc11Tbca Uqcc2 Scand1Smim26 Ctnnb1 Sub1Sod1 Micos10 Exosc7 Nudt14 Prdx1 Churc1 Module 2 Gapdh Hint2

Akt1

Fis1

Pdcd4 Module 3

FigureFigure 4. Construction 4. Construction of protein–protein of protein–protein interaction interaction networks networks (PPI) (PPI) fromfrom DEGs. (A (A) )Construction Construction of of a PPI networka PPI network in the striatum.in the striatum. (B) Construction (B) Construction of a of PPI a PPI network network in thein the whisker whisker follicle. follicle. The The putativeputative PPI networksPPI ofnetworks DEGs wereof DEGs visualized were viavisualized the Search via Toolthe Search for the Tool Retrieval for the of InteractingRetrieval of GenesInteracting/Proteins Genes/Proteins (STRING) application (confidence (score) cut-off = 0.4) in the Cytoscape software (STRING) application (confidence (score) cut-off = 0.4) in the Cytoscape software (v3.7.1). Red nodes (v3.7.1). Red nodes denote up-regulated genes, and blue nodes indicate down-regulated genes. The denote up-regulated genes, and blue nodes indicate down-regulated genes. The intensity of each node intensity of each node color means the fold change. The top three modules of each tissue were color meansidentified the by fold the MCODE change. plug-i The topn in threethe Cytoscape modules software. of each tissue were identified by the MCODE plug-in in the Cytoscape software. 2.5. Cluster Analysis We performed a cluster analysis on the PPI networks of the two tissues to elucidate the network structure and the relations among its components [14,15]. The cluster analysis was conducted using the Molecular Complex Detection (MCODE) plug-in of the Cytoscape software, which identifies highly interactive nodes with high scores in terms of the function and localization similarity of proteins [15]. Using the cluster analysis by MCODE, we found top three statistically significant gene modules in the network of the striatum and of the whisker follicle, respectively (Figure 4). To verify the biological functions of each module in the networks, we performed pathway enrichment functional analysis by means of the ClueGO application based on the KEGG database (Tables S7 and S8). Our data revealed that module 1 of the striatum network was significantly enriched with addiction-related functions such as nicotine addiction, cocaine addiction, amphetamine addiction, glutamatergic synapse, retrograde endocannabinoid signaling, and osteoclast differentiation signaling (Figure 5A). Notably, FosB, Fosl2, Arc, and Junb were identified as statistically significant genes that were up-regulated after MA self-administration, whereas Gabra1, Slc17a6, Slc17a7, and Gria2 were identified as down-regulated genes (Figure 5A). Next, we performed pathway enrichment functional analysis on two representative modules of whisker follicles. As a result, functionally relevant pathways were analyzed in module 1 and module 2 (Figure 5B,C). Module 1 of whisker follicles turned out to be enriched with pathways associated with neurodegenerative diseases (such as Huntington’s disease, Alzheimer’s disease (AD), and Parkinson’s disease) and oxidative phosphorylation as well as a pathway related to ribosomes (Figure 5B). Of note, module 2 of whisker follicles was also mainly enriched with pathways associated with neurodegenerative diseases,

Int. J. Mol. Sci. 2020, 21, 8856 8 of 20

2.5. Cluster Analysis We performed a cluster analysis on the PPI networks of the two tissues to elucidate the network structure and the relations among its components [14,15]. The cluster analysis was conducted using the Molecular Complex Detection (MCODE) plug-in of the Cytoscape software, which identifies highly interactive nodes with high scores in terms of the function and localization similarity of proteins [15]. Using the cluster analysis by MCODE, we found top three statistically significant gene modules in the network of the striatum and of the whisker follicle, respectively (Figure4). To verify the biological functions of each module in the networks, we performed pathway enrichment functional analysis by means of the ClueGO application based on the KEGG database (Tables S7 and S8). Our data revealed that module 1 of the striatum network was significantly enriched with addiction-related functions such as nicotine addiction, cocaine addiction, amphetamine addiction, glutamatergic synapse, retrograde endocannabinoid signaling, and osteoclast differentiation signaling (Figure5A). Notably, FosB, Fosl2, Arc, and Junb were identified as statistically significant genes that were up-regulated after MA self-administration, whereas Gabra1, Slc17a6, Slc17a7, and Gria2 were identified as down-regulated genes (Figure5A). Next, we performed pathway enrichment functional analysis on two representative modules of whisker follicles. As a result, functionally relevant pathways were analyzed in module 1 and module 2 (Figure5B,C). Module 1 of whisker follicles turned out to be enriched with pathways associated with neurodegenerative diseases (such as Huntington’s disease, Alzheimer’s disease (AD), and Parkinson’s disease) and oxidative phosphorylation as well as a pathway related to ribosomes (Figure5B). Of note, module 2 of whisker follicles was also mainly enriched with pathways associated Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 9 of 20 with neurodegenerative diseases, although all genes of module 2 are different from those of module 1 (Figurealthough5C). Moreover,all genes ofApp module, Uchl1 2 ,are and differentAp2a1 were from identifiedthose of module as statistically 1 (Figure significant 5C). Moreover, genes whoseApp, expressionUchl1, and increased Ap2a1 were after identified MA self-administration, as statistically significant whereas genes all the whose genes expression that make increased up these after two modulesMA self-administration, featured a down-regulation whereas all pattern.the genes Overall, that make 15 genesup these in thetwo striatummodules andfeatured 62 genes a down- in the whiskerregulation follicle pattern. were Overall, selected 15 by genes the cluster in the striatum analysis. and 62 genes in the whisker follicle were selected by the cluster analysis. (A)

Fosl2 Dusp1 Gabra1 Junb Cbln1 Slc17a6 Cocaineoc addictionon Slc17a7 Nicotine adddiction Fosb Arc Fosb Osteoclast Junb Fosl2 Nr4a1 Gria2 Egr2 Gria2 differentiation Slc17a7 Nr4a3 Slc17a6 Egr4 Retrograde endocannabinoid signaling Gabra1 Amphetaminne Glutamatergic synapse addiction Striatum: Module 1 Arc

(B)

Ostc Psmg4 Atp5me Atp5mg Eif3hRpl34 Rplp1 Sec61bRps15a Mrps21 Rps29 Sec61g Rps24 Arl4a Rps23 Rpl22Rps29Rpl10aRpl27a Mrpl52 Mrps18c Ap2a1 Polr2k Rpl10a Rps17Rps9 Eif3kRps23 Mrpl41 Mrps21 Rps27aRpl37aRps3a Mrpl23 Cltb Vamp8 Rps10Rpl36al Oxidative phoph sphohorylation Rpl13 Rps15a Eif3el1 App Rplp1 Mrpl54 Mrps35 Cltb Mrps16 Rps27l Mrps33Mrps18c Uqcrq Ubc Polr2g Rpl22l1Nop10 Chchd1 Atp5f1e Rps17 Polr2k Magoh Rpl13Timm13 Mrps16 Uqcrh Rpl22 Sf3b5 Rps27 Atp5mc3 Snrpd2 UqcrhTmem256 Alzheimer disease Ap2a1 Rps27l Rps27a Tra2b Psmb6 Atp5pd Uchl1 Tardbp Gar1 Atp5f1ePfdn5Mrpl55 Tomm7 Pfdn1 Rpl36al Nme1 Atp5mg Gapdh Rps3a Tbca Uqcc2 Hunntington disease Ribobosome Anapc11 Scand1 Mrpl23 Atp5pd Parkinsono disease Uqcrq Rpl37a Ctnnb1 Ube2s Atp5mc3 Polr2g Rps9 Rnf19b Rpl34 Sod1 Atp5me Sod1 Rps10 Fbxl15 Rpl27a App Nudt14 Uchl1 Rps27 Gapdh Rpl22l1 Nfkbia Rps24 Whisker follicle: Module 1

Dnlz

Mrps21 Oxidative Psenen phosphophorylation Atp5mc1 Mrpl41 Atp5mc3 Mrpl23 Mrps18c Mrpl54 Atp5pd Mrpl30 Mrps33Mrps18cElobChchd1 Cox17 Timm13 Parkinson diseeaeas Sf3b5 Bloc1s1Mrps16 Mrpl30 Atp5mg Snrpd2 Atp5me Atp5f1e UqcrhTmem256 Alzheh imer diseas Mrps21 Arl6ip5 Psmb6 Atp5f1eMrpl55 Atp5pf Lsm3Naa38Tomm7Pfdn5Tmem160 RGD1565641 Uqcrb Ribosome Atp5mgPfdn1 Uqcrq Tbca Uqcc2 Mrpl23 Anapc11 Atp5pfAtp5pdAtp5mc1Smim26 Uqcrh Uqcrq Ndufa7 Mrps16 Atp5mc3UqcrbNdufb2Cox7bNdufa2 Glrx5 Cardiac musclecontractio Cox7b Sub1 Atp5me Ndufb5 Ndufa7 Sod1 Ndufb5MGC105649 Ndufa5 Ndufa5 Churc1 Ndufa2 Hint2 Ndufb2 PsenenCox17 Trir Huntington diseaeas Thermogene sis Micos10 Prdx1 Non-alcoholic fatty liver diseas (NAFLD)

Sod1 Whisker follicle: Module 2 Retrogradeendocannab ino signaling

Figure 5. Cluster analysis of DEGs in whisker follicles or in the striatum. (A–C) Gene pathway networks that represent significantly enriched KEGG terms and related genes in designated modules. The sub-networks were analyzed and visualized using the ClueGO plug-in in the Cytoscape software.

2.6. Centrality Analysis For the topological analysis of the network, the degrees and betweenness centrality values were calculated with the NetworkAnalyzer tool [15] of the Cytoscape software. To identify hub genes that are important for network cross-talk and biological essentiality [16,17], two topological features, the degree and betweenness centrality, were considered. In each network (the striatum and the whisker

Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 9 of 20

although all genes of module 2 are different from those of module 1 (Figure 5C). Moreover, App, Uchl1, and Ap2a1 were identified as statistically significant genes whose expression increased after MA self-administration, whereas all the genes that make up these two modules featured a down- regulation pattern. Overall, 15 genes in the striatum and 62 genes in the whisker follicle were selected by the cluster analysis. (A)

(B)

Ostc Psmg4 Atp5me Atp5mg Eif3hRpl34 Rplp1 Sec61bRps15a Mrps21 Rps29 Sec61g Rps24 Arl4a Rps23 Rpl22Rps29Rpl10aRpl27a Mrpl52 Mrps18c Ap2a1 Polr2k Rpl10a Rps17Rps9 Eif3kRps23 Mrpl41 Mrps21 Rps27aRpl37aRps3a Mrpl23 Cltb Vamp8 Rps10Rpl36al Oxidative phoph sphohorylation Rpl13 Rps15a Eif3el1 App Rplp1 Mrpl54 Mrps35 Cltb Mrps16 Rps27l Mrps33Mrps18c Uqcrq Ubc Polr2g Rpl22l1Nop10 Chchd1 Atp5f1e Rps17 Polr2k Magoh Rpl13Timm13 Mrps16 Uqcrh Rpl22 Sf3b5 Rps27 Atp5mc3 Snrpd2 UqcrhTmem256 Alzheimer disease Ap2a1 Rps27l Rps27a Tra2b Psmb6 Atp5pd Uchl1 Tardbp Gar1 Atp5f1ePfdn5Mrpl55 Tomm7 Pfdn1 Rpl36al Nme1 Atp5mg Gapdh Rps3a Tbca Uqcc2 Hunntington disease Ribobosome Anapc11 Scand1 Mrpl23 Atp5pd Parkinsono disease Uqcrq Rpl37a Ctnnb1 Ube2s Atp5mc3 Polr2g Rps9 Rnf19b Rpl34 Sod1 Atp5me Sod1 Rps10 Fbxl15 Rpl27a App Nudt14 Uchl1 Rps27 Int. J. Mol. Sci.Gapdh 2020, 21, 8856 Rpl22l1 9 of 20 Nfkbia Rps24 Whisker follicle: Module 1

(C)

Dnlz

Sod1 Whisker follicle: Module 2 Retrogradeendocannab ino signaling

FigureFigure 5. 5.Cluster Cluster analysis analysis ofof DEGsDEGs inin whiskerwhisker folliclesfollicles or in the striatum. (A (A–C–C) )Gene Gene pathway pathway networksnetworks that that represent represent significantly significantly enriched enriched KEGGKEGG terms and and related related genes genes in in designated designated modules. modules. TheThe sub-networks sub-networks were were analyzed analyzed and and visualized visualized usingusing the ClueGO plug-in plug-in in in the the Cytoscape Cytoscape software. software. 2.6. Centrality Analysis 2.6. Centrality Analysis ForFor the the topological topological analysis analysis of of the the network,network, thethe degrees and and betweenness betweenness centrality centrality values values were were calculatedcalculated with with the the NetworkAnalyzer NetworkAnalyzer tool tool [[15]15] ofof thethe Cytoscape software. software. To To identify identify hub hub genes genes that that areare important important for for network network cross-talk cross-talk and and biologicalbiological essentiality [16,17], [16,17], two two topological topological features, features, the the degreedegree and and betweenness betweenness centrality, centrality, were were considered. considered. In each network network (the (the striatum striatum and and the the whisker whisker follicle), the degrees were set to cut-off levels of 2 and 5, respectively. The genes filtered by degree values are listed in the order of betweenness centrality values, and each of the top 20 genes was chosen as a hub gene. The hub genes of each tissue were visualized in scatter plots with the degrees and betweenness centrality values as axes (Figure6A,B, left panel). A list of hub genes is shown at the right of Figure6A,B. KEGG pathway analysis shown in Supplementary Table S9 revealed that the hub genes of the striatum were mainly enriched with addiction-related functions and nervous system-related functions. The hub genes of the whisker follicle were enriched with neurological disorder-related functions.

2.7. Identification of Potential Diagnostic Markers In the multivariate analysis and network analysis, 49 and 129 statistically significant genes were eventually selected from the DEGs of the striatum and of the whisker follicle, respectively, and these chosen genes are presented in a Venn diagram according to the statistical analysis performed (Figure7). Of the 49 genes selected in the striatum, the activity-regulated cytoskeletal gene (Arc) and Jun B proto-oncogene (Junb) were found to be crucial genes selected by each of the three statistical analyses (Figure7A). In the whisker follicle, the amyloid precursor protein ( App) gene was also found to be a crucial gene selected by all three statistical analyses (Figure7B). Notably, Per1, Ddit4, and Tagln were identified as common genes between whisker follicles and the striatum (Figure7A,B). We also verified the expression of these key genes by Western blot and RT-PCR analyses (Figure S5). Next, to examine the comparability of gene expression between whisker follicles and the striatum, we conducted functional analysis of relations of the two tissues using the ClueGO application in the Cytoscape software. In Figure8, tissue-specific functions in the striatum (blue) and in the whisker follicle (green) and common functions between the two tissues (yellow) are presented. Notably, several pathways, including synaptic vesicle cycle and retrograde endocannabinoid signaling pathways, were identified as common functions that link the whisker follicle tissues and striatal tissues, indicating the cross-talk between both tissues. Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 10 of 20

follicle), the degrees were set to cut-off levels of 2 and 5, respectively. The genes filtered by degree values are listed in the order of betweenness centrality values, and each of the top 20 genes was chosen as a hub gene. The hub genes of each tissue were visualized in scatter plots with the degrees and betweenness centrality values as axes (Figure 6A,B, left panel). A list of hub genes is shown at the right of Figure 6A,B. KEGG pathway analysis shown in Supplementary Table S9 revealed that the hub genes of the striatum were mainly enriched with addiction-related functions and nervous Int.system-related J. Mol. Sci. 2020, 21 functions., 8856 The hub genes of the whisker follicle were enriched with neurological10 of 20 disorder-related functions.

(A) List of top 20 hub genes from PPI network in striatum Striatum Symbol EntrezID FC Degree BC 0.8 LOC100134871 100134871 3.02 3 0.571 Lrg1 367455 2.30 2 0.536 0.7 Lrg1 LOC100134871 Arc 54323 2.43 3 0.476 Gria2 29627 -1.57 4 0.448 0.6 Acr Col1a1 29393 2.36 9 0.437 Gria2 C1qa 298566 1.57 3 0.429 0.5 Col1a1 Nr4a1 79240 3.39 9 0.402 Alas2 25748 3.50 4 0.339 0.4 C1qa Nr4a1 Fosb 100360880 2.07 8 0.145 Egr2 114090 3.29 6 0.145 Alas2 0.3 Tagln 25123 1.79 5 0.115 Col3a1 84032 2.33 6 0.100 Tagln Betweenness centrality Fosb 0.2 Col3a1 Egr2 Hba-a2 360504 3.07 3 0.089 Vim Vim 81818 1.55 4 0.089 Hba-a2 Bgn 25181 1.67 6 0.081 0.1 Bgn Slc17a6 84487 -4.71 3 0.062 Slc17a7 116638 -3.13 3 0.062 0.0 012345678910 Dusp1 114856 1.62 7 0.012 Nr4a3 58853 5.60 6 0.007 Degree Junb 24517 2.01 5 0.002

(B) List of top 20 hub genes from PPI network in whisker follicle Whisker follicle Symbol EntrezID FC Degree BC 0.16 Sod1 24786 -1.65 20 0.125 Ubc 50522 -1.77 33 0.109 0.14 Sod1 Akt1 24185 1.72 20 0.084 Ubc Gapdh 24383 -1.66 15 0.065 0.12 Atp5f1e 245958 -2.13 43 0.056 Akt1 App 54226 1.62 14 0.049 0.10 ATP5f1e Glrx5 362776 -1.52 8 0.047 Gapdh Pfdn5 300257 -1.74 39 0.046 Pfdn5 0.08 Snrpd2 Snrpd2 680309 -1.66 34 0.045 Glrx5 Mrps16 Elavl1 363854 -1.89 16 0.044 Hint2 App 0.06 Elavl1 Rps23 Hint2 313491 -1.56 14 0.043 Gpx4 Psmb6 Mrps16 688912 -1.50 38 0.041

Betweenness centrality Mrps33 Gpx4 29328 -1.68 9 0.039 0.04 Tagln Ntrk2 Rps23 124323 -1.59 42 0.037 Psmb6 29666 -1.65 22 0.033 0.02 Ndufa2 Tagln 25123 1.63 7 0.033 Mrps33 296995 -1.89 34 0.032 0.00 Ndufa2 291660 -1.67 40 0.031 0 5 10 15 20 25 30 35 40 45 Ntrk2 25054 2.12 8 0.029 Degree Vamp8 83730 -1.62 8 0.026

FigureFigure 6. 6.Centrality Centrality analysis analysis and and lists lists of of the the top top 20 20 hub hub genes. genes. (A ,(BA),B Betweenness) Betweenness centrality centrality values values andand degrees degrees of of the the entire entire PPI PPI networks networks werewere obtainedobtained byby means of the the NetworkAnalyzer NetworkAnalyzer tool tool in in the theCytoscape Cytoscape software software (v3.7.1). (v3.7.1). Cut-offs Cut-oﬀ sof of the the degree for PPIPPI networksnetworks inin eacheach tissue tissue were were set set to to 2 2 (striatum)(striatum) and and 5 (whisker 5 (whisker follicle). follicle). Hub Hub genes genes were were arranged arranged in in the the order order of of decreasing decreasing betweenness betweenness centralitycentrality values. values. A A scatter scatter plot plot of of betweenness betweenness centrality centrality versus versus the the degree degree for for nodes nodes from from PPI PPI networksnetworks (left (left panel). panel). The The top top 20 20 hub hub genes genes of of PPI PPI networks networks are are listed liste ind in the the lower lower panel. panel. FC, FC, fold fold change;change; BC, BC, betweenness betweenness centrality. centrality.

Int. J. Mol. Sci. 2020, 21, 8856 11 of 20 Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 12 of 20

(A)

Multivariate Analysis Network Analysis Rbm3 (30) Centrality analysis (20) Nr4a3 Miat Nr4a1 Lrg1 Snhg11 Alas2 Pnisr Egr2 Leng8 Tagln 7 LOC10013487 Ier5 Vim Gria2 Egr4 Bgn Col1a1 RGD1564664 2 Dusp1 C1qa Per1 Fosb Slc25a25 Col3a1 Irs2 25 9 Hba-a2 Klf10 2 Slc17a6 Htra3 Slc17a7 Fam107a Ddit4 Selenow 1 Serpinh1 Cxcl14 3 Aifm3 C1qb Apln Hbb Hsd11b1 Gabra1 Tmem119 Fosl2 Network Analysis Sod3 Arc Cluster analysis (15) Htra1 Junb Pygm Striatum (49 genes)

(B)

Rps9 Rps27a Network Analysis Anapc11 Psenen Rpl27a Cluster analysis (62) Ap2a1 Rpl10a Atp5mc3 Atp5mc1 Rpl13 Atp5mg Atp5me Rpl22 Polr2k Atp5pd Rpl34 Rpl22l1 Ogt Luc7l3 Atp5pf Rpl36al Rps17 Ciao2b Rsrp1 Cltb Rpl37a Rps27 Cnih4 Creld2 Cox17 Rplp1 Rps24 MGC105649 Xbp1 Cox7b Rps10 Tra2b Arl4a RGD1310587 Elob Rps15a Magoh Tardbp Ivl 41 Lsm3 Rps27l Cox20 Syngr2 Lsm7 Rps29 Lce1m Eml2 Mrpl23 Rps3a LOC100912041 Wfs1 Mrpl30 Sf3b5 Mrpl41 Ggt6 Mrps18c Srsf3 Tbca Nub1 12 Mrps21 Uchl1 Pgp Arl6ip5 Ndufa5 Uqcrb Tmem256 Cadm4 8 Ndufa7 Uqcrh RGD1565641 Lrrc8e 1 Ndufb2 Uqcrq Rbp2 Tmem65 Ndufb5 Use1 Ifi27l2b Gpcpd1 Polr2g Il36a Krt75 56 Gmfg Dhcr7 3 8 Smim26 Eif3h Atp5f1e Rbp1 Per1 Akt1 Gapdh Lce1f Dmkn Network Analysis Glrx5 Mrps16 Arglu1 Serpinb13 Centrality analysis (20) Gpx4 Ndufa2 Defb1 Nfkbia Hint2 Rps23 Serpinb1a Ubb Ntrk2 Snrpd2 Sdf2l1 Acy3 App Psmb6 Sod1 Cstb LOC680428 Tagln Vamp8 Arl4d Ddit4 Mrps33 Ubc Sat1 Tm4sf4 Multivariate Analysis Pfdn5 (72) Elavl1 Whisker follicle (129 genes)

FigureFigure 7. Potential7. Potential diagnostic diagnostic markers markers identified identified in thein the striatum striatum (A )(A and) and in whiskerin whisker follicles follicles (B )( ofB) of methamphetaminemethamphetamine (MA) (MA) self-administered self-administered rats. rats. The The Venn Venn diagram diagram shows shows the the potential potential markers markers identifiedidentified by by the the multivariate multivariate and and network network analyses. analyses. Orange Orange color: color: genes genes increased increased by by MA, MA, Blue Blue color: color:genes genes decreased decreased by MA, by MA,Red Redboldfaced: boldfaced: The genes The genes identified identified both in both the in striatum the striatum and in and whisker in whiskerfollicles. follicles.

Int. J. Mol. Sci. 2020, 21, 8856 12 of 20 Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 13 of 20

Junb

Mrps21 Mrps16 Fosl2 Nfkbia Rpl10a Rps27l Osteoclast Rps23Rps15a Rpl13 differentiation Rplp1 Rps27a Mrpl23 Fosb Ogt Rps29 Pygm Rpl22 Rpl37a Rpl34 Insulin resistance Akt1 Arc Mrps18c Ribosome Mrpl30 Rpl36al Irs2 Rps17 Rpl27a Rpl22l1 Cocaine addiction Xbp1 Rps3a Rps9 Per1 Rps27 Rps10 Rps24 Amphetamine Non-alcoholic fatty Circadian entrainment addiction liver disease Longe vity regulating Oxidative (NAFLD) Cox20 pathway Gria2 phosphorylation Atp5i Cox17 Ndufa5 Atp5l Gabra1 Ndufa7Ndufb2 Retrograde Thermogene sis Nicotine addiction Ndufb5Ndufa2 Uqcrb Slc17a7 endocannab inoid Uqcrh Atp5j signaling Cox7b Uqcrq Atp5g1 Slc17a6 Sod1 Atp5e Atp5g3 Ubb Staphylococcus Atp5h aureus infection C1qb Cardiac muscle Synaptic vesicle cycle Parkinson disease Prion diseases contraction Chagas disease Cltb Uchl1 (American C1qa App Ap2a1 Alzheimer disease trypanosomiasis) Vamp8 Huntington disease Gapdh Psenen Protein digestion and Polr2g Use1 Complement and absorption Polr2k SNARE interactions coagulation cascades Col1a1 Pertussis Serpinb1a in vesicular transport Amoebiasis RNA polymerase Tra2b Col3a1 Lsm3 Hba2 African Sf3b5 Hbb Serpinb13 trypanosomiasis Magoh LOC100134871 AGE-RAGE signaling SpliceosomeSnrpd2 Malaria pathway in diabetic Srsf3 Lsm7 complications

Figure 8. TheThe interaction interaction network between the striatum and the whisker follicle of methamphetamine self-administered rats. Blue Blue color: color: A A striatum-specific striatum-specific pathway pathway (big (big circles) circles) or or genes (small circles), green color: a whisker follicle-specific follicle-specific pathway (big circles) or genes (small circles), and yellow color: the pathways or genes common forfor thethe twotwo tissues.tissues. 3. Discussion Multivariate analysis is a useful statistical technique for understanding the structure and relationsDrug of addictiona dataset and is afor complicated evaluating a chronic biomarker disorder candidate that with is conceptualized respect to various as experimental a cycle that includesmeasurements a reward [20,21]. effect, As withdrawal multivariate e ffanalyses,ect, craving, PCA and and a correlation relapse to drug-seekinganalysis were behavior conducted [18 here,19]. Theusing short-access the MetaboAnalyst self-administration v4.0 (http://www.metaboanalyst.ca) model used in the present online study istool. designed Both PCA to study and thecorrelation reward process,analysis whichcan visualize is an initial variation step of in MA a dataset addiction. and identify To find potential technical diagnostic or biological genes outliers in this that pathological were not process,excluded the at the transcriptional alignment step. changes Topological caused byanalysis MA self-administration based on a PPI network were investigatedplays an important in whisker role folliclesin the prediction and in the of the striatum. function Previously, of genes or we protei havens exploredconnected the to a changes network in and gene provides expression a view in into the whiskerthe functional follicle relations of MA self-administered of the interactions rats among to identify the genes novel or indicators proteins in of a the biological MA reward system and [15,22]. found thatPPI genenetworks expression generally in the have whisker a variety follicle of could topolo begical used tofeatures understand such theas rewardingthe degree, e ffbetweennessect of MA in thecentrality, striatum closeness [9]. Nonetheless, centrality, nobody shortest has path directly length, examined and clustering the concomitant coefficient gene that expressionhelp to clarify during the MAarchitecture self-administration and function both of inthe the network striatum [15,23]. and in whiskerAlthough follicles each ofof thethese same analyses rat at the has same its timeown points,advantages, nor have there any is researchersalso a risk that placed the their data results may be in misinterpreted the context of systems or missed biology. if only Therefore, one type the of examinationanalysis is applied. of differential Therefore, gene we expression employed in a mult discreteiple brainbioinformatics region, the tool striatum,s to analyze and at anthe alternative RNA-seq samplingdata, and sitethis (whiskerstrategy follicles)might be willmore reveal reliable an additionalthan a single-analysis transcriptional approach landscape in that for MAthe multiple rewards. analyses can reduce error and improve the authenticity of the data.

Int. J. Mol. Sci. 2020, 21, 8856 13 of 20

Furthermore, it might be interesting to investigate whether the non-invasive analysis of whisker follicles is comparable to gene expression analysis in the striatum and yields similar results. Moreover, our study provides convergent systems biological evidence of genetic networks common between whisker follicles and the striatum in rats, and this evidence could further elucidate MA addiction in humans. Multivariate analysis is a useful statistical technique for understanding the structure and relations of a dataset and for evaluating a biomarker candidate with respect to various experimental measurements [20,21]. As multivariate analyses, PCA and correlation analysis were conducted here using the MetaboAnalyst v4.0 (http://www.metaboanalyst.ca) online tool. Both PCA and correlation analysis can visualize variation in a dataset and identify technical or biological outliers that were not excluded at the alignment step. Topological analysis based on a PPI network plays an important role in the prediction of the function of genes or proteins connected to a network and provides a view into the functional relations of the interactions among the genes or proteins in a biological system [15,22]. PPI networks generally have a variety of topological features such as the degree, betweenness centrality, closeness centrality, shortest path length, and clustering coefficient that help to clarify the architecture and function of the network [15,23]. Although each of these analyses has its own advantages, there is also a risk that the data may be misinterpreted or missed if only one type of analysis is applied. Therefore, we employed multiple bioinformatics tools to analyze the RNA-seq data, and this strategy might be more reliable than a single-analysis approach in that the multiple analyses can reduce error and improve the authenticity of the data. Independent multivariate and network analyses detected 49 and 129 genes (including Arc, Junb, and App, which are expected to play a major role in MA reward and addiction) in the striatum and whisker follicle, respectively. Some reports indicate that an acute injection of MA significantly up-regulates many immediate early genes such as members of the AP-1 family [24,25], and that acute and chronic administration of MA leads to increased Arc expression in the rat cerebral cortex, striatum, and hippocampus [26]. Our data revealed that several AP-1 family members such as Junb, Fosl2, and Fosb were up-regulated by MA self-administration in the striatum. The MA self-administered rats showed increased expression of a synaptic-plasticity-related gene, Arc, in the striatum. Arc is an effector immediate early gene participating in synaptic plasticity and memory consolidation [27,28]. An increased expression of Arc in the visual cortex has been reported to decrease synaptic strength by inducing α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor endocytosis [29,30], which is consistent with our data showing decreased expression of Gria2, which is a gene encoding a subunit of the AMPA receptor. Our results revealed that highly connected gene modules are associated with drug addiction in the striatum. By contrast, in whisker follicles, DEGs known (through annotation) to be associated with neurodegenerative disorders were prominent. These findings imply that MA self-administration induces differential gene expression in both tissues, thus activating distinct functions. However, in this context, it is noteworthy that Arc is required for postsynaptic trafficking and processing of Appand amyloid β formation, which appear to be relevant to AD pathogenesis [31]. In whisker follicles, the App gene has been identified as a potential marker gene, which is in agreement with our previous data suggesting that App is an addiction-related gene in the whisker follicle in MA self-administered rats [9]. Although much remains to be studied, we cannot ignore the possibility that MA-induced genetic changes disturbing homeostatic synaptic plasticity in the brain are somehow reflected in whisker follicles, thereby linking the functions between striatal tissues and whisker follicle tissues. In fact, a recent report supports this hypothesis in that the pathological changes associated with AD can be reflected in peripheral tissues, fluids, and cells such as fibroblasts, the olfactory epithelium, whole blood, and plasma, which are considered useful sources for potential biomarkers of AD [32]. Moreover, in our study, several DEGs including the circadian clock gene period 1 (Per1) were identified in both the striatum and whisker follicles, i.e., in the overlap between their gene sets. Drug addiction has long been linked to disruptions in diurnal and circadian rhythms. Of note, it has been Int. J. Mol. Sci. 2020, 21, 8856 14 of 20 demonstrated that drugs of abuse alter the expression of circadian-clock genes in reward-related regions of the brain, thereby directly affecting ongoing circadian rhythms [33]. In addition, people with insomnia are more prone to addiction [34], suggesting that the disruption of circadian rhythms may be strongly related to drug reward and addiction. In agreement with this notion, acute MA administration leads to a rapid induction of Per1 expression in the dorsal striatum of mice [35]; this change may increase drug seeking and craving when drugs are anticipated. Additionally, a recent report suggests that Per1 expression in peripheral cells such as leukocytes and hair follicles may be used as an appropriate biomarker for the assessment of biological clock traits [36]. Therefore, our findings strongly indicate that Per1 expression in peripheral hair follicles is an indicator of MA reward in the brain. Collectively, our findings provide overall gene expression profiles and identify key genes and pathways related to MA reward in whisker follicles and in the striatum as a result of a series of bioinformatics analyses. The gene expression patterns were found to be mostly different between whisker follicles and the brain, pointing to the activation of independent functions in the two tissues in response to the MA. The gene expression patterns were found to be mostly different between whisker follicles and the brain, pointing to the activation of independent functions in the two tissues in response to the MA. Nevertheless, our findings suggest that common alterations of expression of several genes and pathways in the brain and peripheral whisker follicles may serve as diagnostic markers of the rewarding effect of MA. Future research directions may also be highlighted.

4. Materials and Methods

4.1. Animals Male Sprague–Dawley rats (Daehan Animal, Seoul, Korea) weighing between 260 and 280 g at the start of the study were housed individually on a 12 h light-and-dark cycle, with all the experiments conducted during the light cycle. The rats had ad libitum access to food and water throughout the experiment. All the animals were handled (for acclimation) for at least 5 days prior to the surgical procedure and experiments. All experimental procedures were approved by the Institutional Animal Care and Use Committee at the Korea Institute of Toxicology (Approval No. KIT-1605-0710, B116061, 12-09-2016) and met the Association for Assessment and Accreditation of Laboratory Animal Care International guidelines for the care and use of laboratory animals.

4.2. Methamphetamine Self-Administration and Sample Collection Before the MA self-administration experiment, rats were trained to press the active (drug-paired) lever (30 min/day for 3 days) for a food pellet reward. Only the rats that earned at least 100 pellets in the last session of the training were chosen for the further experiment. Surgical and post-surgical techniques have been described in detail in other studies [37,38]. After surgery, the rats were allowed to recover for 7 days. The recovered rats were moved to a special chamber, and either MA dissolved in SA or SA alone was infused into the rat when the lever was pressed. The MA self-administration was performed in daily 2 h sessions (0.05 mg/kg per intravenous infusion) for 16 days. The numbers of lever presses and infusions were analyzed by two-way repeated measure (RM) analysis of variance (ANOVA) with modality of self-administration (MA, SA) and days as between- and within-subject factors, respectively, using GraphPad Prism 8.0.2. After two-way RM ANOVA, Bonferroni multiple comparison tests were used for post-hoc analysis. After the end of the MA self-administration protocol, whisker follicles and the striatum were collected as described elsewhere [9,39]. The collected tissue samples were stored at 80 C until analysis. − ◦ 4.3. RNA Extraction Total RNA was isolated from whisker follicles and the striatum by the conventional method (with the TRIzol® reagent; Invitrogen, Carlsbad, CA, USA). The quality and concentration of RNA were Int. J. Mol. Sci. 2020, 21, 8856 15 of 20 assessed on a ﬁlter-based multi-mode microplate reader (FLUOstar Omega, BMG Labtech, Ortenberg, Germany). Residual genomic DNA was removed by incubation with RNase-free DNase I for 30 min at 37 ◦C.

4.4. Construction of Transcriptome Libraries The isolated total RNA was processed for preparation of an mRNA sequencing library using a TruSeq RNA Sample Preparation Kit v2 (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions. In brief, mRNAs were isolated from 1 µg of total RNA on RNA purification beads by polyA capture, which was followed by enzymatic shearing. After first- and second-strand cDNA synthesis, A-tailing and end repair were performed for the ligation of proprietary primers that incorporate unique sequencing adaptors with an index for tracking Illumina reads from multiplexed samples run on a single sequencing lane. For each library, an insert size of approximately 200-500 bp was confirmed by an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA), and quantification of the library was carried out by real-time PCR. Samples were sequenced on the Illumina HiSeq 2000 platform, with paired-end 101 bp reads for mRNA-seq using the TruSeq SBS Kit v3 (Illumina). The raw imaging data were transformed by base-calling into sequence data and stored in FASTQ format.

4.5. Processing of the RNA-Seq Data Paired-end reads of the 24 independent samples were trimmed with Cutadapt to remove both PCR and sequencing adapters [40]. The trimmed reads were aligned to the rn6 rat reference genome in the TopHat2 software [41]. The quantiﬁcation of gene expression was performed with Cuﬄinks [42] to calculate the fragments per kilobase of transcript per million mapped reads (FPKM) values.

4.6. Statistical Analysis of Differentially Expressed Genes (DEGs) The presence of statistically significant differential expression was determined in Cuffdiff [43] at the gene level. Then, we chose gene expression differences based on their average FDRs and average log2-fold expression changes between the SA self-administration group (control) and the MA self-administration group in whisker follicles and in the striatum. For these comparisons, we assumed that genes are differentially expressed if the FDR was below 0.05 and the absolute value of fold change was >1.5.

4.7. Principal Component Analysis (PCA) and Hierarchical Clustering These analyses were performed by means of MetaboAnalyst v4.0 (https://www.metaboanalyst. ca)[44]. The PCA was intended to discriminate between MA and SA samples of each tissue on the basis of their RNA-seq proﬁles. The values of DEGs were transformed by the autoscaling method (mean-centered and divided by the standard deviation of each variable) in MetaboAnalyst; 2D score plots of PCA between principal component 1 (PC1) and 2 (PC2) were constructed next. Red and green areas denote 95% conﬁdence regions of MA and SA groups, respectively. The hierarchical clustering was also performed on MA and SA samples of each tissue. Euclidean distances were used as a distance measured between samples.

4.8. Correlation Analysis Correlation matrix and pattern analyses of DEGs from the striatum and from the whisker follicle were performed using MetaboAnalyst. The correlation analysis was intended to evaluate the strength of a linear relation between diﬀerent DEGs of each tissue. The raw data on DEGs were converted by the autoscaling method of the MetaboAnalyst online tool. For correlation matrix and pattern analyses, Pearson’s correlation coeﬃcient (Pearson’s r) was computed as a correlation parameter (distance measure). Pearson’s r adopts a value between +1 and 1, where +1 is a total positive linear correlation, − Int. J. Mol. Sci. 2020, 21, 8856 16 of 20 and 1 is a total negative linear correlation. The values of Pearson’s r for the entire gene set were − obtained using PatternHunter in MetaboAnalyst, and the top 25 genes were visualized in a bar chart.

4.9. Centrality Analysis PPI networks were constructed using the STRING application in Cytoscape (https://cytoscape.org). The DEGs of whisker follicles and of the striatum were employed for a protein query in the STRING application for analysis of the PPI networks. The confidence cut-offs and maximum additional interactors of the network visualization were set to 0.4 and 0.0, respectively. The color intensity of each node represented a fold change. To clarify the functional and biological importance of the genes, we analyzed topological structures of the PPI networks. For the determination of hub genes in the PPI networks of each tissue, the degrees and betweenness centrality values based on topological features were calculated via the NetworkAnalyzer tool in Cytoscape. The cut-offs for the degree in the networks of the striatum and of the whisker follicle were set to 2 and 5, respectively. The genes in the networks were ranked in the order of increasing betweenness centrality values, and the top 20 genes of the lists were designated as hub genes.

4.10. Functional and Pathway Enrichment Analyses For functional and pathway enrichment analysis of the DEGs or PPI networks, two complementary analyses of Gene Ontology (GO, http://geneontology.org/), including biological process (BP), cellular component (CC) and molecular function (MF), and Kyoto Encylopedia of Genes and Genomes (KEGG, https://www.genome.jp/kegg/) were performed. The Database for Annotation, Visualization, and Integrated Discovery (DAVID, web-based online tool (https://david.ncifcrf.gov)) was used to perform GO functional and the KEGG pathway enrichment analysis. The significant thresholds of p-value and gene count for enrichment analysis were set to <0.05 and 2, respectively. ≥ 4.11. Cluster and Functional Annotation Analyses of PPI Networks To extract a highly connected cluster from the above-mentioned constructed PPI networks, we investigated each statistically significant cluster by the MCODE [45] clustering algorithm. The clusters of highly connected nodes were identified at the following parameters: degree cut-off = 10, node score cut-off = 0.2, K-core = 2, and max depth up to 100. Bader and Hogue have described the principle and operating procedures of the MCODE clustering algorithm and have defined each parameter in detail [45]. Then, we investigated the functional pathways of each statistically significant cluster to identify the major functions of the clusters using the ClueGO [46] application in Cytoscape. KEGG was employed as a database for pathway analysis, and two parameters of the ClueGO application, the p value and kappa score, were set to 0.05 and 0.3, respectively. The gene pathway functional networks of each cluster were visualized by the ClueGO plug-in in the Cytoscape software.

4.12. Identiﬁcation of Potential Markers among the DEGs of Whisker Follicles or of the Striatum To identify the potential markers among the DEGs of each tissue, we constructed a Venn diagram from the genes identiﬁed by three procedures: correlation analysis, centrality analysis, and cluster analysis. The blue and orange color of the gene name represented genes that were down-regulated or up-regulated in each tissue, and the red color represented the up-regulated genes common for the two tissues. To investigate functional linkages between the two tissues, we performed KEGG pathway enrichment analysis on potential markers of the striatum (49 genes) and of the whisker follicle (129 genes) and visualized the gene pathway network using the ClueGO application in Cytoscape.

4.13. Data Availability The sequencing data analyzed during the current study are available in the Sequence Read Archive (SRA) database (accession number PRJNA633547). Int. J. Mol. Sci. 2020, 21, 8856 17 of 20

Supplementary Materials: Supplementary materials can be found at http://www.mdpi.com/1422-0067/21/22/ 8856/s1. Supplementary Materials and Methods; Table S1. List of differentially expressed genes (87 genes) in striatum; Table S2. List of differentially expressed genes (253 genes) in whisker follicle; Table S3. Correlation coefficient list of DEGs in striatum; Table S4. Correlation coefficient list of DEGs in whisker follicle; Table S5. Functional enrichment analysis in PPI networks of the striatum; Table S6. Functional enrichment analysis in PPI networks of the whisker follicle; Table S7. KEGG pathway enrichment analysis of modules in the striatum PPI network; Table S8. KEGG pathway enrichment analysis of modules in the whisker follicle PPI network; Table S9. KEGG pathway enrichment analysis of 20 hub genes in the striatum and whisker follicle; Figure S1. Protein expression level of dopamine signaling in the striatum after MA self-administration; Figure S2. The hierarchical cluster analysis of DEGs in the striatum and whisker follicle; Figure S3. PCA and correlation pattern analysis of DEGs in the striatum and whisker follicle; Figure S4. Construction of protein–protein interaction (PPI) networks from total, up-, and down- regulated DEGs in the striatum and whisker follicle; Figure S5. Validation of several key genes in the striatum and whisker follicle. Author Contributions: Conceptualization, S.L. and C.-H.J.; animal experiment, S.-H.S. and I.S.R.; software, W.-J.J. and T.S.; writing—original draft preparation, W.-J.J. and T.S.; writing—review and editing, S.L. and C.-H.J.; visualization, W.-J.J.; funding acquisition, S.L. and C.-H.J. All authors have read and agreed to the published version of the manuscript. Funding: This research was supported and funded by the Basic Science Research Program of the National Research Foundation of Korea (NRF), the Ministry of Education (NRF-2016R1A6A1A03011325), and the Bio & Medical Technology Development Program of the NRF, the Ministry of Science, ICT & Future Planning (NRF 2015M3A9E1028327). Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations

DEGs Diﬀerentially expressed genes DAVID Database for Annotation, Visualization, and Integrated Discovery GO Gene Ontology BP Biological process CC Cellular component MF Molecular function KEGG Kyoto Encyclopedia of Genes and Genomes PCA Principal component analysis PPI Protein–protein interaction STRING Search Tool for the Retrieval of Interacting Genes/Proteins MCODE Molecular complex detection BC Betweenness centrality MA Methamphetamine SA Saline FR1 Fixed ratio 1 FDR False discovery rate AD Alzheimer’s disease PD Parkinson’s disease FC Fold change FPKM Fragments per kilobase of transcript per million mapped reads

References

1. Centers for Disease Control and Prevention. Acute public health consequences of methamphetamine laboratories—16 states, January 2000-June 2004. MMWR Morb. Mortal. Wkly. Rep. 2005, 54, 356–359. Available online: https://www.cdc.gov/mmwr/preview/mmwrhtml/mm5414a3.htm (accessed on 14 April 2005). 2. Kwon, N.J.; Han, E. A commentary on the eﬀects of methamphetamine and the status of methamphetamine abuse among youths in South Korea, Japan, and China. Forensic. Sci. Int. 2018, 286, 81–85. [CrossRef] 3. Kalechstein, A.D.; Newton, T.F.; Green, M. Methamphetamine dependence is associated with neurocognitive impairment in the initial phases of abstinence. J. Neuropsychiatry Clin. Neurosci. 2003, 15, 215–220. [CrossRef] Int. J. Mol. Sci. 2020, 21, 8856 18 of 20

4. Jentsch, J.D.; Taylor, J.R. Impulsivity resulting from frontostriatal dysfunction in drug abuse: Implications for the control of behavior by reward-related stimuli. Psychopharmacology 1999, 146, 373–390. [CrossRef] [PubMed] 5. Koob, G.F.; Volkow, N.D. Neurocircuitry of addiction. Neuropsychopharmacology 2010, 35, 217–238. [CrossRef] 6. Luscher, C.; Malenka, R.C. Drug-evoked synaptic plasticity in addiction: From molecular changes to circuit remodeling. Neuron 2011, 69, 650–663. [CrossRef] 7. Robison, A.J.; Nestler, E.J. Transcriptional and epigenetic mechanisms of addiction. Nat. Rev. Neurosci. 2011, 12, 623–637. [CrossRef] 8. Maekawa, M.; Yamada, K.; Toyoshima, M.; Ohnishi, T.; Iwayama, Y.; Shimamoto, C.; Toyota, T.; Nozaki, Y.; Balan, S.; Matsuzaki, H.; et al. Utility of Scalp Hair Follicles as a Novel Source of Biomarker Genes for Psychiatric Illnesses. Biol. Psychiatry 2015, 78, 116–125. [CrossRef][PubMed] 9. Song, S.H.; Jang, W.J.; Hwang, J.; Park, B.; Jang, J.H.; Seo, Y.H.; Yang, C.H.; Lee, S.; Jeong, C.H. Transcriptome profiling of whisker follicles in methamphetamine self-administered rats. Sci. Rep. 2018, 8, 11420. [CrossRef] 10. O’Connor, E.C.; Chapman, K.; Butler, P.; Mead, A.N. The predictive validity of the rat self-administration model for abuse liability. Neurosci. Biobehav. Rev. 2011, 35, 912–938. [CrossRef] 11. Yao, L.; Fan, P.; Arolfo, M.; Jiang, Z.; Olive, M.F.; Zablocki, J.; Sun, H.L.; Chu, N.; Lee, J.; Kim, H.Y.; et al. Inhibition of aldehyde dehydrogenase-2 suppresses cocaine seeking by generating THP, a cocaine use-dependent inhibitor of dopamine synthesis. Nat. Med. 2010, 16, 1024–1028. [CrossRef][PubMed] 12. Krasnova, I.N.; Chiflikyan, M.; Justinova, Z.; McCoy, M.T.; Ladenheim, B.; Jayanthi, S.; Quintero, C.; Brannock, C.; Barnes, C.; Adair, J.E.; et al. CREB phosphorylation regulates striatal transcriptional responses in the self-administration model of methamphetamine addiction in the rat. Neurobiol. Dis. 2013, 58, 132–143. [CrossRef][PubMed] 13. von Mering, C.; Jensen, L.J.; Snel, B.; Hooper, S.D.; Krupp, M.; Foglierini, M.; Jouffre, N.; Huynen, M.A.; Bork, P. STRING: Known and predicted protein-protein associations, integrated and transferred across organisms. Nucleic Acids Res. 2005, 33, D433–D437. [CrossRef][PubMed] 14. Barabasi, A.L.; Oltvai, Z.N. Network biology: Understanding the cell’s functional organization. Nat. Rev. Genet. 2004, 5, 101–113. [CrossRef] 15. Miryala, S.K.; Anbarasu, A.; Ramaiah, S. Discerning molecular interactions: A comprehensive review on biomolecular interaction databases and network analysis tools. Gene 2018, 642, 84–94. [CrossRef] 16. Gursoy, A.; Keskin, O.; Nussinov, R. Topological properties of protein interaction networks from a structural perspective. Biochem. Soc. Trans. 2008, 36, 1398–1403. [CrossRef] 17. Yu, H.; Kim, P.M.; Sprecher, E.; Trifonov, V.; Gerstein, M. The importance of bottlenecks in protein networks: Correlation with gene essentiality and expression dynamics. PLoS Comput. Biol. 2007, 3, e59. [CrossRef] 18. Cami, J.; Farre, M. Drug addiction. N. Engl. J. Med. 2003, 349, 975–986. [CrossRef] 19. Goldstein, R.Z.; Volkow, N.D. Drug addiction and its underlying neurobiological basis: Neuroimaging evidence for the involvement of the frontal cortex. Am. J. Psychiatry 2002, 159, 1642–1652. [CrossRef] 20. Olkin, I.; Sampson, A.R. Multivariate Analysis: Overview. In International Encyclopedia of the Social & Behavioral Sciences, 1st ed.; Smelser, N.J., Baltes, P.B., Eds.; Pergamon: Oxford, UK, 2001; pp. 10240–10247. [CrossRef] 21. Marcello Manfredi, E.R. Biomarkers Discovery through Multivariate Statistical Methods: A Review of Recently Developed Methods and Applications in Proteomics. J. Proteom. Bioinform. 2013, s3.[CrossRef] 22. Vella, D.; Marini, S.; Vitali, F.; Di Silvestre, D.; Mauri, G.; Bellazzi, R. MTGO: PPI Network Analysis Via Topological and Functional Module Identification. Sci. Rep. 2018, 8, 5499. [CrossRef][PubMed] 23. Gupta, A.; Mohanty, P.; Bhatnagar, S. Integrative analysis of ocular complications in atherosclerosis unveils pathway convergence and crosstalk. J. Recept. Signal. Transduct. Res. 2015, 35, 149–164. [CrossRef][PubMed] 24. Cadet, J.L.; Jayanthi, S.; McCoy, M.T.; Vawter, M.; Ladenheim, B. Temporal profiling of methamphetamine-induced changes in gene expression in the mouse brain: Evidence from cDNA array. Synapse 2001, 41, 40–48. [CrossRef][PubMed] Int. J. Mol. Sci. 2020, 21, 8856 19 of 20

25. Jayanthi, S.; McCoy, M.T.; Beauvais, G.; Ladenheim, B.; Gilmore, K.; Wood, W., 3rd; Becker, K.; Cadet, J.L. Methamphetamine induces dopamine D1 receptor-dependent endoplasmic reticulum stress-related molecular events in the rat striatum. PLoS ONE 2009, 4, e6092. [CrossRef] 26. Kodama, M.; Akiyama, K.; Ujike, H.; Shimizu, Y.; Tanaka, Y.; Kuroda, S. A robust increase in expression of arc gene, an effector immediate early gene, in the rat brain after acute and chronic methamphetamine administration. Brain Res. 1998, 796, 273–283. [CrossRef] 27. El-Boustani, S.; Ip, J.P.K.; Breton-Provencher, V.; Knott, G.W.; Okuno, H.; Bito, H.; Sur, M. Locally coordinated synaptic plasticity of visual cortex neurons in vivo. Science 2018, 360, 1349–1354. [CrossRef] 28. Shepherd, J.D.; Bear, M.F. New views of Arc, a master regulator of synaptic plasticity. Nat. Neurosci. 2011, 14, 279–284. [CrossRef] 29. Gao, M.; Sossa, K.; Song, L.; Errington, L.; Cummings, L.; Hwang, H.; Kuhl, D.; Worley, P.; Lee, H.K. A specific requirement of Arc/Arg3.1 for visual experience-induced homeostatic synaptic plasticity in mouse primary visual cortex. J. Neurosci. 2010, 30, 7168–7178. [CrossRef] 30. Shepherd, J.D.; Rumbaugh, G.; Wu, J.; Chowdhury, S.; Plath, N.; Kuhl, D.; Huganir, R.L.; Worley, P.F. Arc/Arg3.1 mediates homeostatic synaptic scaling of AMPA receptors. Neuron 2006, 52, 475–484. [CrossRef] 31. Wu, J.; Petralia, R.S.; Kurushima, H.; Patel, H.; Jung, M.Y.; Volk, L.; Chowdhury, S.; Shepherd, J.D.; Dehoff, M.; Li, Y.; et al. Arc/Arg3.1 regulates an endosomal pathway essential for activity-dependent beta-amyloid generation. Cell 2011, 147, 615–628. [CrossRef] 32. Francois, M.; Leifert, W.; Martins, R.; Thomas, P.; Fenech, M. Biomarkers of Alzheimer’s disease risk in peripheral tissues; focus on buccal cells. Curr. Alzheimer Res. 2014, 11, 519–531. [CrossRef] 33. Falcon, E.; McClung, C.A. A role for the circadian genes in drug addiction. Neuropharmacology 2009, 56 (Suppl. 1), 91–96. [CrossRef][PubMed] 34. Shibley, H.L.; Malcolm, R.J.; Veatch, L.M. Adolescents with insomnia and substance abuse: Consequences and comorbidities. J. Psychiatr. Pract. 2008, 14, 146–153. [CrossRef][PubMed] 35. Nikaido, T.; Akiyama, M.; Moriya, T.; Shibata, S. Sensitized increase of period gene expression in the mouse caudate/putamen caused by repeated injection of methamphetamine. Mol. Pharmacol. 2001, 59, 894–900. [CrossRef][PubMed] 36. Watanabe, M.; Hida, A.; Kitamura, S.; Enomoto, M.; Ohsawa, Y.; Katayose, Y.; Nozaki, K.; Moriguchi, Y.; Aritake, S.; Higuchi, S.; et al. Rhythmic expression of circadian clock genes in human leukocytes and beard hair follicle cells. Biochem. Biophys. Res. Commun. 2012, 425, 902–907. [CrossRef] 37. Yoon, S.S.; Kim, H.; Choi, K.H.; Lee, B.H.; Lee, Y.K.; Lim, S.C.; Choi, S.H.; Hwang, M.; Kim, K.J.; Yang, C.H. Acupuncture suppresses morphine self-administration through the GABA receptors. Brain Res. Bull. 2010, 81, 625–630. [CrossRef] 38. Yoon, S.S.; Yang, E.J.; Lee, B.H.; Jang, E.Y.; Kim, H.Y.; Choi, S.M.; Steffensen, S.C.; Yang, C.H. Effects of acupuncture on stress-induced relapse to cocaine-seeking in rats. Psychopharmacology 2012, 222, 303–311. [CrossRef] 39. Hill, R.A.; Kiss Von Soly, S.; Ratnayake, U.; Klug, M.; Binder, M.D.; Hannan, A.J.; van den Buuse, M. Long-term effects of combined neonatal and adolescent stress on brain-derived neurotrophic factor and dopamine receptor expression in the rat forebrain. Biochim. Biophys. Acta 2014, 1842, 2126–2135. [CrossRef] 40. Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet J. 2011, 17, 3. [CrossRef] 41. Kim, D.; Pertea, G.; Trapnell, C.; Pimentel, H.; Kelley, R.; Salzberg, S.L. TopHat2: Accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol. 2013, 14, R36. [CrossRef] 42. Trapnell, C.; Williams, B.A.; Pertea, G.; Mortazavi, A.; Kwan, G.; van Baren, M.J.; Salzberg, S.L.; Wold, B.J.; Pachter, L. Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nat. Biotechnol. 2010, 28, 511–515. [CrossRef][PubMed] 43. Trapnell, C.; Hendrickson, D.G.; Sauvageau, M.; Goff, L.; Rinn, J.L.; Pachter, L. Differential analysis of gene regulation at transcript resolution with RNA-seq. Nat. Biotechnol. 2013, 31, 46–53. [CrossRef] 44. Xia, J.; Wishart, D.S. Web-based inference of biological patterns, functions and pathways from metabolomic data using MetaboAnalyst. Nat. Protoc. 2011, 6, 743–760. [CrossRef][PubMed] Int. J. Mol. Sci. 2020, 21, 8856 20 of 20

45. Bader, G.D.; Hogue, C.W. An automated method for ﬁnding molecular complexes in large protein interaction networks. BMC Bioinform. 2003, 4, 2. [CrossRef][PubMed] 46. Bindea, G.; Mlecnik, B.; Hackl, H.; Charoentong, P.; Tosolini, M.; Kirilovsky, A.; Fridman, W.H.; Pages, F.; Trajanoski, Z.; Galon, J. ClueGO: A Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics 2009, 25, 1091–1093. [CrossRef][PubMed]

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional aﬃliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).