Experimental Neurology 263 (2015) 339–349

Contents lists available at ScienceDirect

Experimental Neurology

j ourna l homepage: www.e lsev ie r .com/ locate /yexnr

Regular Article

Gene expression profiling studies in regenerating nerves in a mousemodel for CMT1X: Uninjured Cx32-knockout peripheral nerves displayexpression profile of injured wild type nerves

Mona Freidin a,⁎,1, Samantha Asche-Godin b,1, Charles K. Abrams c

a Department of Neuroscience, Albert Einstein College of Medicine, Bronx, NY 10461, USAb Department of Neurology, State University of New York, Downstate Medical Center, Brooklyn, NY 11203, USAc Department of Neurology, State University of New York, Downstate Medical Center, Brooklyn, NY 11203, USA

⁎ Corresponding author at: Albert Einstein College of1300 Morris Park Avenue, Bronx, NY 10461, USA.

E-mail address: mona.freidin@einstein.yu.edu (M. Fre1 M. Freidin and S. Asche-Godin contributed equally to

http://dx.doi.org/10.1016/j.expneurol.2014.10.0140014-4886/© 2014 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 23 July 2014Revised 13 October 2014Accepted 18 October 2014Available online 23 October 2014

Keywords:CMT1XRegenerationPeripheral nerve injurySchwann cellsNon-myelinatingDifferential gene expression

X-linked Charcot–Marie–Tooth disease (CMT1X) is an inherited peripheral neuropathy caused by mutations inGJB1, the human gene for Connexin32 (Cx32). This present study uses Ilumina Ref8-v2 BeadArray to examinethe expression profiles of injured and uninjured sciatic nerves at 5, 7, and 14 days post-crush injury (dpi) fromWild Type (WT) and Cx32-knockout (Cx32KO) mice to identify the genes and signaling pathways that aredysregulated in the absence of Schwann cell Cx32. Given the assumption that loss of Schwann cell Cx32 disruptsthe regeneration andmaintenance ofmyelinated nerve leading to a demyelinating neuropathy in CMT1X,we ini-tially hypothesized that nerve crush injury would result in significant increases in differential gene expression inCx32KO mice relative to WT nerves. However, microarray analysis revealed a striking collapse in the number ofdifferentially expressed genes at 5 and 7 dpi in Cx32KO nerves relative to WT, while uninjured and 14 dpi timepoints showed large numbers of differentially regulated genes. Further comparisons within each genotypeshowed limited changes in Cx32KO gene expression following crush injury when compared to uninjuredCx32KO nerves. By contrast, WT nerves exhibited robust changes in gene expression at 5 and 7 dpi with no sig-nificant differences in gene expression by 14 dpi relative to uninjured WT nerve samples. Taken together, thesedata suggest that the gene expression profile in uninjured Cx32KO sciatic nerve strongly resembles that of a WTnerve following injury and that loss of Schwann cell Cx32 leads to a basal state of gene expression similar to thatof an injuredWT nerve. These findings support a role for Cx32 in non-myelinating and regenerating populationsof Schwann cells in normal axonal maintenance in re-myelination, and regeneration of peripheral nerve follow-ing injury. Disruption of Schwann cell–axonal communication in CMT1X may cause dysregulation of signalingpathways that are essential for the maintenance of intact myelinated peripheral nerves and to establish the nec-essary conditions for successful regeneration and remyelination following nerve injury.

© 2014 Elsevier Inc. All rights reserved.

Introduction

Connexins are a large family of homologous integralmembrane pro-teins that form cell–cell channels or gap junctions as well as single-cellmembrane hemichannels (Abrams and Rash, 2009; Kar et al., 2012) andprovide a low resistance pathway for the diffusion of small moleculesand ions between coupled cells (Kumar and Gilula, 1996). Recent dataalso suggests that connexins help regulate cell growth and apoptoticor necrotic cell death, independent of the formation of functional gapjunction channels (Lin et al., 2003; Omori et al., 2007; Vinken et al.,

Medicine, Dept. Neuroscience,

idin).this manuscript.

2011). X-linked Charcot–Marie–Tooth disease (CMT1X) is an inheritedperipheral neuropathy associated with mutations in the human genefor Connexin32 (Cx32), GJB1. Cx32 gap junction proteins are found inthe non-compact myelin of the paranodes and Schmidt–Lantermanincisures in myelinated peripheral nerve providing intracellular chan-nels between adjacent myelin loops shortening the radial diffusionof small molecules between the Schwann Cell nucleus and adaxonalmembrane at least 300 times (Abrams and Bennett, 2000; Balice-Gordon et al., 1998; Scherer et al., 1995). Over 300 CMT1X mutationshave been identified (http://www.molgen.ua.ac.be/cmtmutations/Mutations/Mutations.cfm). Disruption of the myelin communicationpathway by mutations in Cx32 has been proposed as a mechanismin CMT1X. Typical features of CMT1X include a delayed onset until thesecond decade, initial involvement of distal nerves, and axonal lossand moderate reductions in peripheral nerve conduction velocities(Ionasescu, 1995; Yiu et al., 2011).

340 M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

Cx32 in non-myelinating Schwann cells and peripheral nerve regeneration

Cx32 protein and mRNA are co-regulated with the expression ofmyelin-specific genes during development and during myelination aswell as during regeneration and repair of injured peripheral nerves(Bondurand et al., 2001; Nagarajan et al., 2002). Sohl et al. (1996) re-ported that following sciatic nerve injury (a period of increased SC pro-liferation), SC expression of Cx32 followed the same time course asseveralmyelin genes; indicating that expression of Cx32 is related to re-generation associated re-myelination. Thus,mutations in or loss of Cx32would be predicted to disrupt the repair and regeneration of injured pe-ripheral nerve.

Xenograph transplant experiments in nude mice demonstratedthat Schwann cells with a loss-of-function CMT1X mutation in Cx32(Val181Ala) were severely impaired in their ability to support the earli-est stages of regeneration of myelinated fibers (Abrams et al., 2003),while Schwann cells expressing a less severe mutation (Glu102Gly)supported normal early regeneration (Sahenk and Chen, 1998). Theabnormalities associated with the Val181Ala mutation in these ex-periments occur at a time of Schwann cell proliferation and de-differentiation, when no myelin is present, indicating an additionalfunctional role for Cx32 in non-myelinating Schwann cells. Furtherevidence from our lab demonstrates that loss of Cx32 decreases prolif-eration in cultured Schwann cells from Cx32-knockout (Cx32KO) miceas compared to Wild Type (WT) (Freidin et al., 2009). These findingsand others (Freidin et al., 2009; Mambetisaeva et al., 1999; Sohl et al.,1996) suggest that mutations in Cx32 might influence both non-myelinating and myelinating SC functions.

Progress of peripheral neuropathy in Cx32KO mice

Cx32KO mice develop histopathologic signs of demyelinating neu-ropathy beginning at approximately three months that progresseswith increasing age (Anzini et al., 1997; Scherer et al., 1998).Myelinatedmotor fibers are more affected than myelinated sensory fibers at allages. The Cx32 gene is located on the X-chromosome and, like otherX-linked genes, is randomly inactivated. Heterozygous Cx32KO+/−female mice have a milder phenotype with fewer demyelinated andre-myelinated axons than age-matched homozygous Cx32KO−/−female and Cx32KO−/Y male mice reflecting the clinical patterns forhuman CMT1X. These findings indicate that loss of function of Cx32Schwann cells is sufficient to cause an inherited demyelinating neurop-athy and supports the hypothesis that Cx32 has an essential role inSchwann cells both in mice and in humans.

Gene profiling studies

Gene expression profiling facilitates the measurement of the rel-ative regulation of thousands of genes in a single RNA sample. Usingmicroarray analysis, Iacobas and colleagues compared Cx32KO andCx43KO brains and found strong overlaps in sets of regulated genes,leading the authors to suggest that the brain transcriptome contains“connexin-dependent regulomes” (Iacobas et al., 2007). Similar studiesexamining peripheral nerves of WT or connexin-null mice have notbeen published. Earlier microarray studies validated the use of mousemodels of peripheral nerve injury and have identified several Schwanncell-specific and cell cycle genes involved in proliferation and differen-tiation during sciatic nerve regeneration (Kury et al., 2002; TenAsbroek et al., 2006).

The current study uses IIlumina MouseRef8-v2 BeadArray to com-pare the expression profiles in uninjured and injured sciatic nervesfrom age-matched Cx32KO and WT mice. The present studies used5-week old mice to limit the potential effects of axonal pathologiesthat develop in Cx32KOmice beginning between two and threemonthsof age (Sargiannidou et al., 2009; Scherer et al., 1998). Given theapparent limited developmental effects in Cx32KO mice, we predicted

significant changes in gene expression following peripheral nervecrush injury and during regeneration in Cx32KO mice as comparedto WT. Microarray analysis, however, showed a biphasic response,with a dramatic decline in the number of differentially regulatedgenes (DEG) at 5 and 7 days post injury (5 and 7 dpi) and significantdifferences in DEG from uninjured Cx32KO mice as compared to WT.Further comparisons of uninjured and injured nerves within each geno-type demonstrated that Cx32KO nerves underwent relatively fewchanges in gene expression in following crush injury as compared touninjured Cx32KO nerves. As predicted, WT samples exhibited robustchanges in gene expression. Taken together, these data suggest thatthe gene expression profile in uninjured Cx32KO sciatic nerve stronglyresembles that of a WT nerve following a crush injury and that loss ofSchwann cell Cx32 leads to a basal state of gene expression similar tothat in injured WT nerve.

Methods

Animals and surgery

A colony of C57Bl/6 WT and Cx32KO mice was bred in-house. TheCx32KO mice were bred from founders that had been extensivelybackcrossed into C57Bl/6 (generously providedby S. Bennett, U. Ottawa,Canada). All animals were provided food and water ad libitum andhoused according to the Guide for the Care and Use of LaboratoryAnimals.

For nerve crush studies, age-matched, 35–37 day-old WT andCx32KO male mice were induced and anesthetized using isoflorane.Withminimal perturbation of the overlyingmuscle and fascia, the sciaticnerve was exposed and crushed 5 mm distal to the sciatic notch for 30 swith ultrafine curved hemostats until translucent. The crush site wasthen marked with India ink and the skin closed with surgical staples.The mice were allowed to recover in their home cages. To minimizepain, allmice received ibuprofen (4.7ml children's ibuprofen suspensionin 500ml water) for 12 h prior to surgery and for 36 h following surgery,as per IPUPAC guidelines. Cage mates that did not undergo surgeryserved as uninjured age-matched controls.

The mice were sacrificed at 5, 7, and 14 dpi for tissue collection. A5 mm segment was collected distal to the crush site and stored inRNALater (Ambion) at −80 °C, as per manufacturer's suggestions. Foruninjured nerve samples, a 5 mm segment was collected distal to thesciatic notch and processed as for crush samples. Nerve samples fromfour mice were pooled for each time point in an experimental series.Each experimental time series was repeated six times, for a total of sixbiological replicates.

RNA purification and labeling

RNA was extracted and purified from pooled nerve samples usingRNAeasy Lipid Mini kit (Qiagen, Country), as per manufacturer's in-structions. Purified RNA (500 ng) was amplified and labeled using theIllumina TotalPrep RNA Amplification Kit (Ambion/Life Technologies).Biotin-labeled cRNA was stored at−80 °C until needed.

Microarray hybridization, quality control, and normalization

Labeled cRNAwas outsourced to the KeckMicroarray Section of YaleCenter for GenomeAnalysis (Yale University, NewHaven, CT) formicro-array processing, hybridization, and background normalization usingIIlumina MouseRef8-v2 BeadArray BeadChip technology (Illumina, Inc.San Diego, CA). cRNA quality was confirmed prior to hybridization bythe Keck microarray facility. Following hybridization and labeling, thearrays were scanned and the output normalized in BeadStudio(Illumina) for quality control, background, and data export.

Each BeadArray chip contained eight cRNA samples permittinghybridization of single biological replicate for the full time series:

Fig. 1. Gene expression profile of Cx32KO nerve samples crush injury relative to WT. Thenumber of differentially regulated genes in Cx32KO nerves relative to WT is dramaticallyreduced at 5 and 7 dpi. Ratio of all corrected Log2 transformed raw gene expression valuesfor Cx32KO versus WT nerves at 5, 7, and 14 days post injury (dpi).

341M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

uninjured nerve and nerve at 5, 7 and 14 dpi for Cx32KO and WT sam-ples. Six Illumina MouseRef-8 v2.0 BeadChips were used for the geneexpression analysis, corresponding to six replicate time series and sixbiological replicates for each time point.

Experiments were designed, performed, and analyzed based onMAIME (Minimum Information About aMicroarray Experiment) guide-lines (Brazma, 2009).

Microarray analysis and statistical tests

Log2-transformed data from all BeadChip experiments wereimported fromBeadStudio (Illumina, Inc. San Diego, CA) into Partek Ge-nomic Suite (Partek Inc., St Louis, MO) for statistical analysis. Followingremoval of batch effects, mixed-model ANOVAwas used to identify dif-ferentially expressed genes. Comparisons between Cx32KO and WT ateach time point were used to generate lists of DEG based on relative ex-pression levels (fold-change) greater than 1.5 and false discovery rate(FDR) corrected p-values (B-H p-Value) derived from the Benjamini–Hochberg step-up procedure (Benjamini and Hochberg, 1995) to ac-count for multiple testing. FDR-corrected gene lists were also deter-mined for each genotype with respect to time based on comparisonsof uninjured versus 5, 7, or 14 dpi.

Functional and pathway analyses of gene lists at each timewere con-ducted using Ingenuity Pathway Analysis (IPA, QIAGEN Redwood City,CA, www.qiagen.com/ingenuity), GOMiner (Zeeberg et al., 2003,2005), Mouse Genome Database (Blake et al., 2014), and Mouse Ge-nome Informatics (MGI) Batch Query (URL: http://www.informatics.jax.org) online analysis tools. Gene ontology (GO) annotations andGOSlim datasets were obtained or investigator-defined using EMBLQuickGO (URL: http://www.ebi.ac.uk/QuickGO/) and MGI Batch Querytools.

Real time PCR validation

RNA was extracted and purified from pooled nerve samples asdescribed. RNA samples were subjected to DNAse treatment (TurboDNA-free Kit, Ambion) to remove chromosomal DNA contaminationand quantitated by absorbance at OD260. cDNA from total RNA (5 μg)was synthesized using the Superscript VILO cDNA synthesis kit (LifeTechnologies), according to manufacturer's suggestions. QualitativeReal TimePCRwas performed using 15 μl reactions containing 75ng sci-atic nerve cDNA and SsoFast Sybr Green Supermix (Bio-rad) on a Chro-mo4 DNA Engine system with Opticon3 Monitor software (Bio-Rad).Validated primer pairs for TATA binding protein, Actin-b, GAPDH,S100b, and, c-fos were obtained from SABiosciences (Qiagen). Se-quences for validated primers for CANX, CYC1, and p75NGFR were ob-tained through PrimerBank (Spandidos et al., 2010). The sequences forthese primers were as follows: CANX-For ATGGAAGGGAAGTGGTTACTGT; CANX-Rev GCTTTGTAGGTGACCTTTGGAG; CYC1-For CAGCTTCCATTGCGGACAC; CYC1-Rev GGCACTCACGGCAGAATGAA; p75NGFR-ForCAATGTCCACCAGGAAAACGA; p75NGFR-Rev GGAAGTTAGGGGCAAGTCG. Data were preprocessed and analyzed for relative expression levelsusing GenEx5.4 (MultiD Analyses, Germany). Samples with CTN33 wereconsidered above the limit of detection. Stable housekeeping referencegenes were determined for sciatic nerve samples using geNorm andNormFinder programs embedded in GenEx5.4 prior to performing geneexpression studies (Vandesompele et al., 2002; Mestdagh et al., 2009).Real time PCR experiments were conducted using MIQE guidelines andsuggested checklists (Bustin et al., 2009, 2010).

Results

Differential gene regulation between WT and Cx32KO at each time point

To address the role of Cx32 in the maintenance and regeneration ofperipheral nerve, this study compared gene expression profiles of sciatic

nerves from uninjuredWT and Cx32KOmice and injured nerve samplesat 5, 7, and 14 after crush. Pooled sciatic nerve samples from age-matched uninjured and injured WT and Cx32KO mice were takenfrom sites distal to the crush and processed for microarray usingIllumina MouseRef8-v2 BeadArray.

ANOVA was conducted on the complete data set using PartekGenomics Suite (Partek, Inc, St. Louis MO). Post-test comparisons withBenjamini–Hochberg (Benjamini and Hochberg, 1995; Hochberg andBenjamini, 1990) multiple test correction and false discovery rate (FDR)(Noble, 2009) created lists of significantly regulated genes in Cx32KO ascompared to WT for each experimental group. In the uninjured nervesamples, 640 differentially expressed genes (DEG) were identified in theuninjured Cx32KO relative to WT, with down-regulation of 316 andup-regulation of 324 genes (Fig. 1, Table 1). Relative expression levelsof several differentially regulated Schwann cell genes (c-Fos, p75NGFR,and S100b) were validated by real time PCR (Supplementary Fig. 1)using RNA collected from uninjured WT and Cx32KO sciatic nerves.

The number of DEG decreased dramatically at early times afterinjury with differential expression in Cx32KO nerves of only 24 and 20genes at 5 dpi and 7 dpi, respectively. At 14 dpi, the differences in thenumber of altered genes re-emerged, rising to 891 DEG in Cx32KO nerverelative to WT. Thus, initial responses to nerve injury are similar in WTand Cx32KO, but significant differences in gene expression appear in unin-jured and regenerating (14dpi) Cx32KO nerves relative to WT.

Uninjured and 14dpi Cx32KO nerves shared 311 DEG relative toWT,representing 48.6% and 34.9% of regulated genes for each time point, re-spectively (Fig. 2). Further comparison at the other time points showedfive genes in common at 5 dpi and 7 dpi, comprising 20.8% and 23.8% ofDEG at each time point. These findings indicate that the greatest differ-ential gene regulation occurs in uninjured and regenerating (14 dpi)Cx32KO nerves.

Differential gene regulation over time for each genotype

To evaluate differences in gene regulation between uninjured andinjured nerves for each genotype, DEG for WT and Cx32KO wereassessedwith respect to time following crush injury. Comparisonswith-in each genotype group were made to determine differential geneexpression in uninjured mice against that of mice at 5, 7, and 14 dpi.

Table 1Differentially expressed genes in uninjured and injured Cx32KO nerves at 5, 7, and 14 dpirelative to WT. Following removal of batch effects, mixed-model Analysis of Variation(ANOVA) was used to identify differentially expressed genes (DEG) from Log2 trans-formed raw data. Gene lists were also determined for each genotype with respect to timebased on comparisons of uninjured versus 5, 7, or 14 dpi. (Uninj: FDR p b 0.05; 5 dpi FDRp b 0.1; 7 dpi FDR p b 0.2; 14 dpi p b 0.05).

Days post injury (dpi) Total DEG 32KO vs WT (#↑, #↓)

Uninjured 640 (316↑, 324↓)5 dpi 24 (11↑, 13↓)7 dpi 20 (7↑, 13↓)14 dpi 891 (383↑, 508↓)

Table 2Differentially expressed genes for WT and Cx32KO nerves at 5, 7, and 14 dpi relative touninjured nerves for each genotype. Corrected gene lists for each genotype with respectto timewere determined comparinguninjured versus 5, 7, or 14 dpi for relative expressionlevels (fold-change) greater than 1.5 and FDR p b 0.05 for all comparisons.

WT total DEG (#↑, #↓) 32KO total DEG (#↑, #↓)

Uninj vs 5 dpi 1399 (669↑,730↓) 221 (102↑,119↓)Uninj vs 7 dpi 1392 (667↑,725↓) 21 (12↑,9↓)Uninj vs 14 dpi 0 159 (123↑,36↓)

342 M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

As summarized in Table 2, dramatic differences in gene regulation wereobserved following crush injury in WT nerves relative to uninjuredWTnerves. 1399 and 1392WT genes changed at 5 dpi and 7 dpi, respec-tively, compared to uninjured WT. Strikingly, by 14 dpi, no significantdifferences were detected in WT nerves compared to uninjured. Crushinjury followed a different pattern in Cx32KO nerves with comparative-ly fewDEG in Cx32KOnerves at 5 dpi and 7 dpi (221 and 21genes) com-pared to uninjured Cx32KO. UnlikeWT, 14 dpi Cx32KO samples showeda significant regulation of 159 genes nerves versus uninjured.

Comparison of differentially regulated genes in uninjured Cx32KO nervesand genes regulated in WT nerves following injury

To test the observation that uninjured Cx32KO gene expression fol-lows a pattern similar to injured WT, the number of regulated genes inuninjured Cx32KO relative to uninjured WT (see Table 1) were com-pared with WT DEG (relative to uninjured WT; see Table 2) at 5 and7 dpi (Fig. 3). 520 and 1162 genes, or approximately 83% of DEG, areshared between injured WT nerves at 5 and 7 dpi (versus uninjuredWT) and uninjured Cx32KO (versus uninjured WT) (Fig. 3A). Similarcomparisons of DEG within each genotype reveal a high degree of

Fig. 2. Venn diagram comparing differentially expressed genes in Cx32KO nerves relative to Wgenes in Cx32KO nerve samples relative to WT at each time point (A). Venn diagrams of up-re

overlap in the WT gene lists at 5 and 7 dpi (relative to uninjured WT)(Fig. 2B), with 1160 shared genes or 88.9% and 83.5% of 5 and 7 dpiWT DEG in common. Fig. 2C compares gene lists in injured Cx32KOnerves relative to uninjured with 21 shared genes shared between at5 and 7 dpi; representing only 10% (21/221 DEG) of genes in theCx32KO 5 dpi group and the entire 7 dpi gene list. These observationsare particularly striking given the large number of regulated WT genesrelative to uninjuredWT at these early time points (1307 and 1392, re-spectively; Table 2) and the relatively small number of DEG in Cx32KOmice (compared to uninjured Cx32KO) at 5 and 7 dpi (221 and 21, re-spectively; Table 2). Thus, the gene expression profile in uninjuredCx32KO peripheral nerve shows limited changes following injury andclosely shares the expression profile of injured WT nerves.

Gene ontology (GO) functional analysis of gene expression profiles inuninjured and injured Cx32KO mice compared to WT

Gene ontology (GO) enrichment analysis assigns biological andfunctional groupings to lists of differentially expressed genes based oncommercially and publically assembled GO databases. Numerous soft-ware tools are available for GO analysis and provide detailed statisticalresults to illustrate the biological significance of regulated genes.

Lists of differentially regulated genes from each Cx32KO versus WTtime point were probed for GO term enrichment using the GOminer

T following crush injury. Venn diagram illustrating the overlap of differentially expressedgulated (B) and down-regulated (C) genes in Cx32KO nerves relative to WT.

Fig. 3. Comparison of shared genes in uninjured Cx32KO nerves and injured WT. DEG for uninjured Cx32KO (relative to uninjured WT) were compared to the gene lists in WT nerves(relative to uninjured WT) at 5 and 7 dpi demonstrate a high degree of overlap between injured WT nerves and uninjured Cx32KO nerves (A); with greater than 75% of DEG sharedbetween uninjured Cx32KO and injured WT at 7 dpi versus uninjured WT. Comparison of DEG fromWT at 5 and 7 dpi relative to uninjured WT (B) and Cx32KO at 5 and 7 dpi relativeto uninjured Cx32KO (C) shows a high degree of overlap in WT nerves.

343M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

(Zeeberg et al., 2005) software tool assigning biological and functionalGO terms for each gene list. This analysis yielded significant enrichmentscores for over 900 GO functional groups which, in turn, encompassed alarge number of overlapping genes, GO terms, and GO identifiers, mak-ing meaningful interpretation difficult without further refinement.

Fig. 4. Schwann/myelin GOSlim functional analysis of Cx32KO vsWT for uninjured and injuredanalyzed for enrichment of Schwann/Myelin GOSlim functional categories using the GOMinertest, with corrections. By convention, a functional category is considered over-expressed or enexpression (p b 0.05).

Given the inherent redundancy of GO databases, the creation ofconsolidated lists of GO terms (GOSlims) that are based on existingrelationships between terms in the ontologies provide a correctedsummary of the larger GO annotation sets. GOSlims, whether curatedor user-defined, facilitate a more circumscribed classification and

groups at 5, 7, and 14 days post-injury. Gene lists for each experimental time group wereonline tool. Enrichment scores for each GOSlim term were calculated using Fisher's exactriched with enrichment scores N1. Enrichment scores N3 correspond to significant over-

Fig. 5. Immune/stress response GOSlim functional analysis of Cx32KO vsWT for uninjuredand injured groups at 5, 7, and 14 days post-injury. Gene lists for each experimental groupwere analyzed for enrichment of Immune/Stress Response GOSlim functional categoriesusing the GOMiner online too. Enrichment for Immune/Stress Response terms wereobserved in uninjured and 14 dpi injured Cx32KO nerves relative to WT.

Fig. 6. Differential gene expression of genes associated with Schwann/myelin GOSlimterms in uninjured and injured Cx32KO nerves vs WT. After compensating for overlapsacross Schwann/Myelin GO gene annotations, a list of 157 Schwann/Myelin GOSlimgenes was generated. This list was compared to DEG from uninjured and injuredCx32KO nerves relative to WT. 25 genes were significantly regulated in Cx32KO nervesrelative toWT for at least one time point. A:Down-regulated genes. B:Up-regulated genes.

344 M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

visualization of regulated functional groups for a particular data set.CustomMyelin/Schwann GOSlim and Immune/Stress Response GOSlimterm lists were generated based on GO annotations associated withmyelination, regeneration, peripheral nerve injury, and Schwann cellbehavior for the Myelin/Schwann GOSlim and enriched immune re-sponse GO annotations for the Immune/Stress Response GOSlim usingthe QuickGO term tool (European Bioinformatics Institute). EmployingGOMiner, these custom GOSlims were subsequently used to reprobeeach list of significantly regulated genes in uninjured, 5, 7, and 14 dpiCx32KO samples relative to WT. The analysis generated enrichmentscores for the GOSlim terms. By convention, a GO term is considered“enriched” with a score greater than one and “significantly enriched”with a score greater than three.

Functional analysis of myelin and Schwann cell GO termsHighly significant Myelin/Schwann GOSlim enrichment scores were

obtained for compact myelin, myelin sheath, myelin assembly, andmyelination GOSlim terms at 5 dpi and, to a lesser value, at 14 dpi(Fig. 4A). The 7 dpi gene list returned a significant GO enrichmentscore of 150 for the compact myelin term, as well. Schwann cell differ-entiation and Schwann cell development GOSlim terms generatedenrichment scores greater than 3 for the 14 dpi gene list (Fig. 4B). Fur-ther classification showed that only the uninjured Cx32KO (relative touninjured WT) gene list achieved enrichment scores greater than one;specifically for genes associated with the myelin maintenance, negativeregulation of myelination, and myelination GOSlim terms. Myelin/Schwann GOSlim terms for NGF binding, myelination of peripheralnerve, peripheral nerve development, and constituent myelin sheathwere enriched (score N 1) at 14 dpi. Taken together, the Myelin/Schwann GOSlim functional analysis suggests that loss of Cx32 resultsin a dysregulation of genes associated with the temporal response toperipheral nerve injury aswell as genes involved in functional pathwaysinvolved with regeneration.

Functional analysis of immune response GO termsTo examine the regulation of GO functional categories with respect

to immune response, the gene lists for uninjured and injured Cx32KOnerve samples (relative toWT) were reprobed with the Immune/StressResponse GOSlim. The majority of enrichment scores greater than onewere obtained for uninjured and 14 dpi genes lists, including regulationof response to biotic stimulus, response to stress, defense response, andinflammatory response (Fig. 5). Analysis of the 5 dpi gene list did showGO enrichment for the regulation of immune system process and im-mune response GOSlim terms; reflecting a further shift in functionalgene regulation of Cx32KO nerves toward biological response termsassociated with injured nerve relative to WT.

Expression of genes associated with functional pathways

GOannotations are the gene products associatedwith aGO termandare generated by a combination of electronic/bioinformatic strategiesand manual curation. Using QuickGo and MGI batch inquiry tools, GOgene annotations associated with the Schwann/Myelin GOSlim andImmune Response GOSlims were created and compared to the lists ofdifferentially regulated genes for the Cx32KO versus WT uninjuredand 5, 7, and 14 dpi groups.

Gene expression profiles within the Schwann/Myelin GOSlimThe Schwann/Myelin GOSlim terms comprised 157 genes, after

compensating for overlaps across GO annotations. Of these commonSchwann/Myelin GOSlim genes, 25 genes were significantly regulatedin Cx32KO nerves relative to WT for at least one time point (Fig. 6).Four genes, p75NGFR, Pou3f1, Cxcr4, andEdnrb,were significantly upreg-ulated in both uninjured and regenerating 14 dpi Cx32KO nerves rela-tive to WT (Fig. 6B). The majority of Schwann/Myelin GOSlim DEG(15/25 genes) showed significant regulation in only 14 dpi Cx32KO

nerves relative to WT; with decreased expression in twelve genes andincreases in three genes (Figs. 6A and B). The myelin genes Mag andP0 were significantly reduced in Cx32KO relative to WT after crushinjury, with down-regulation of P0 at 7dpi and Mag at both 5 and7 dpi. Only Myo5a, a motor protein gene associated with RNA transportin Schwann cells (Canclini et al., 2014), was increased in uninjuredCx32KO compared to WT nerves.

To begin to examine the potential interconnectivity of regulatedSchwann/Myelin GOSlim genes, an interaction network of Schwann/Myelin GOSlim DEG was generated using the Ingenuity PathwayAnalysis (IPA) tool (Supplementary Fig. 2). This user-curated networkillustrates the potential relationships between differentially regulated

345M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

genes associated with Schwann cell differentiation, myelination, andregeneration and the impact and the effects due to loss of Schwanncell Cx32.

Gene expression profiles within the Immune GOSlimThe Immune GOSlim is associatedwith 5413 gene annotations. After

compensating for overlaps, the Immune GOSlim gene list was reducedto 191 unique genes. Differences in genes associated with immune re-sponse did not significantly differ between Cx32KO and WT followingcrush injury at 5 and 7 dpi. However, 75.4% (144/191 genes) of theImmune GOSlim gene list were differentially regulated in Cx32KO ascompared to WT in uninjured and 14 dpi nerves. Thus, early responsesto injury do not differ betweenWTandCx32KO but the gene expressionpathways associated with immune responses are altered in uninjuredand regenerating (14 dpi) Cx32KO nerves relative to WT nerves.

Given the complexity and interrelationships of the genes associatedwith immune and stress response signaling pathways, Ingenuity Path-way Analysis (IPA) was conducted comparing Cx32KO and WT genelists for each time point. Consistent with Immune GOSlim gene annota-tion analysis, IPA identified significant regulation of the chemokine/cy-tokine immune response signaling pathway in uninjured and 14 dpiCx32KO nerve samples (relative toWT). Chemokines belong to a familyof inflammatory cytokines that regulate a host of inflammatory andimmune biological processes involved with cellular activation, differen-tiation, and survival (Comerford and McColl, 2011; Pineau and Lacroix,2009). Responses to peripheral nerve injury, including Wallerian de-generation and nerve regeneration, involve coordinated activation ofchemokine and cytokine pathways (Dubovy et al., 2013; Patodia andRaivich, 2012).

To begin the analysis, a consensus chemokine/cytokine gene list wasgenerated by cross-referencing genes associated with chemokine andcytokine signaling as identified by IPA, KEGG (Kanehisa et al., 2012)and InnateDB tools (Breuer et al., 2013; Lynn et al., 2008). This final con-sensus list of 343 chemokine/cytokine genes was then compared to thelists of DEG identified from the Immune GOSlim gene list. The resultingcross comparison found 23 regulated genes in common (Fig. 7). Nineof these cytokine/chemokine genes were regulated in both uninjuredand 14 dpi Cx32KO samples relative to WT (Fig. 7A.), while nine were

Fig. 7. DEG associated with cytokine/chemokine signaling pathways in uninjured and14dpi Cx32KOnerves vsWT. A consensus gene list for the chemokine and cytokine signal-ing pathwaywas generated and cross-referenced to the Immune/Stress Response GOSlimgene creating a consensus cytokine/chemokine gene list for analysis. This gene listwas used to probe DEG from uninjured and injured Cx32KO nerves relative to WT. A:Down-regulated genes. B: Up-regulated genes.

regulated in only uninjured Cx32KO and three in 14 dpi Cx32KO nervesas compared to WT (Figs. 7A and B).

An interaction network composed of differentially expressed cyto-kine and chemokine genes from the uninjured and 14dpi data sets wasgenerated using IPA (Supplementary Fig. 3), illustrating the interconnec-tions associated with genes involved in inflammatory and immuneresponse pathways in Cx32KO peripheral nerves relative to WT.

Regulation of p75NGFR and Jun signaling pathways

p75NGFR, a member of the TNFR-superfamily of cytokine receptors,is expressed in non-myelinating and immature Schwann cells and ishighly regulated during development, myelination, and in regeneratingperipheral nerve (Tomita et al., 2007; Zhou and Li, 2007). p75NGFR

serves as a marker for non-myelinating Schwann cells in mature nerveand de-differentiated Schwann cells following nerve injury and is sig-nificantly upregulated in uninjured and regenerating 14 dpi Cx32KOnerves compared to WT.

The transcription factor c-Jun is a master regulator of the Schwanncell injury response (Arthur-Farraj et al., 2012; Fontana et al., 2012),providing the switches necessary for the de-differentiation and dif-ferentiation of the Schwann cell program following nerve crush oraxotomy (Arthur-Farraj et al., 2012; Fontana et al., 2012). Examinationof c-Jun expression inWTnerves shows increased levels at 5 dpi relativeto uninjured WT, in agreement with earlier reports showing inductionof c-Jun in Schwann cells following crush injury (Shy et al., 1996;Soares et al., 2001; Stewart, 1995). By contrast, differential regulationof c-Junwas not observed in Cx32KOnerves at any time following injuryrelative to uninjured Cx32KO nerves. Direct comparison of WT andCx32KO at each time point following injury revealed upregulation ofc-Jun in regenerating Cx32KO nerves at 14 dpi relative to WT.

To explore the intersection of c-Jun signaling pathwayswith the reg-ulation of p75NGFR, a c-Jun GOSlimwas generated using QuickGo, KEGG,and MGI batch inquiry tools. A consensus gene list was subsequentlycreated using the c-Jun GOSlim gene annotations. The c-Jun GOSlimcontained 11 distinct GO categories and 3844 gene annotations. Follow-ing removal of duplicates, a consensus c-Jun GOSlim gene list was creat-ed. Of the 413 unique c-Jun GOSlim genes, 23 and 18 genes overlappedwith the Schwann/Myelin and Immune GOSlim gene annotations, re-spectively. Comparison of the three gene lists found only p75NGFR incommon.

Analysis of the c-Jun GOSlim gene list identified 36 significantlyregulated genes in uninjured and 14 dpi Cx32KO samples relative toWT (Fig. 8). Cross comparison of regulated c-Jun GOSlim genes withSchwann/Myelin GOSlim genes revealed that only p75NGFR and TGFβ1were shared between the two gene lists; with differential regulationof p75NGFR in both uninjured and 14 dpi and TGFβ1 in 14 dpi Cx32KOnerves relative to WT. Similar evaluation of regulated c-Jun GOSlimgenes with Immune GOSlim gene lists revealed 11 common genes;again sharing p75NGFR as a differentially regulated gene in uninjuredand 14 dpi Cx32KO nerves. Finally, a network diagram of differentiallyexpressed c-Jun GOSlim genes was generated using IPA (SupplementaryFig. 4); illustrating the connections between regulated genes in the c-Junpathway and DEG associated with immune response, Schwann cell, andmyelin genes in Cx32KO nerves relative to WT.

Discussion

This study examined the expression profiles of injured and unin-jured sciatic nerves from WT and Cx32KO mice to identify the genesand functional or signaling pathways that are dysregulated in theabsence of Schwann cell Cx32. Based on the assumption that loss ofSchwann cell Cx32 disrupts the regeneration and maintenance ofmyelinated nerve leading to a demyelinating neuropathy in CMT1X,we initially hypothesized that nerve crush injury would result in signif-icant increases in differential gene expression in Cx32KO mice relative

Fig. 8. c-JunGOSlimDEG in uninjured and injured 14 dpi Cx32KOnerves vsWT. A consensusc-Jun GOSlim gene list was created and used to probe DEG from uninjured and injuredCx32KO relative to WT. 36 significantly regulated genes were identified in uninjured and14 dpi Cx32KO samples relative to WT. A: Down-regulated genes. B: Up-regulated genes.

346 M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

to WT. However, microarray analysis revealed that the uninjured andregenerating 14dpi groups showed the most DEG with 640 and 891genes, respectively for uninjured and 14 dpi Cx32KO versus WT and astriking collapse in the number of DEG at 5 and 7 dpi in Cx32KO nervesrelative toWT. A comparison of uninjured nerves to injured nerves at 5,7, and 14 dpi within each genotype showed limited changes in Cx32KOgene expression at all three timepoints;while expression profiles inWTexhibited robust changes at 5 and 7 dpi and no significant differences at14 dpi. These findings suggest that the gene expression profiles of unin-jured and regenerating 14 dpi nerves lacking functional Schwann cellCx32 most closely resemble that of injured WT nerves at 5 and 7 dpi;providing a revised perspective for understanding the cellular mecha-nisms in CMT1X.

Disruption of peripheral nerve myelin does not occur in Cx32KOmice until three months of age (Scherer et al., 1998; Vavlitou et al.,2010). At two months, prior to the appearance of active demyelination,axonal degeneration becomes evident in uninjured peripheral nerves ofCx32KO mice with increases in β-amyloid precursor protein, increasesin neurofilament packing, and reduced diameters of large myelinat-ed fibers (Vavlitou et al., 2010). The present study specifically used5-week old mice to limit the potential effects of axonal pathologiesthat appear in Cx32KOmice at twomonths of age. The significant differ-ences in the gene expression profiles of uninjured Cx32KOmice relativeto WT further suggest that loss of Cx32 disturbs normal Schwann cellsignaling pathways at developmental time points preceding signs ofaxonopathy.

Peripheral nerve injury results in axonal degeneration and loss of ax-onal contact, a process during which Schwann cells of the distal stumpde-differentiate to an immature phenotype (Mirsky et al., 2008; Yanget al., 2012); with down-regulation of pro-myelinating genes andincreases in immature or non-myelinating Schwann cell genes (Allodiet al., 2012). Peak up-regulation of the immature Schwann cell markers,such as p75NGFR, typically occurs between five and eight days followinginjury in C57Bl/6 mice, concurrent with down-regulation of Cx32,P0 (MPZ), MBP, and other major myelin genes (Barrette et al., 2010;Bolin and Shooter, 1993). Genes associated with immature or de-differentiated Schwann cells decline to uninjured levels between

days 14 and 28 post-injury (Hall et al., 1997) as Schwann cells re-differentiate to mature myelinating and non-myelinating phenotypesupon contact with regenerating axons (Hall et al., 1997; Jessen andMirsky, 2008). As predicted, WT nerves responded with a significantregulation of large numbers of genes following crush injury, with1399 and 1392 genes at 5 and 7 dpi, respectively and no significant dif-ferences in gene expression at 14 dpi compared to uninjuredWT nerve;consistent with a return to uninjured expression levels in successfullyregenerating nerves. By contrast, Cx32KO nerves showed limitedchanges in gene expression following injury as compared to uninjuredCx32KO nerves, with significant regulation of only 221 and 21 genesat 5 and 7 dpi and significant regulation of 159 genes at 14 dpi. Theinitial and reparative responses to nerve injury require controlledmod-ulation of the timing, degree, and specificity of gene expression forsuccessful nerve regeneration to occur. Loss of functional Schwanncell Cx32 appears to disrupt these processes, contributing to a dysregu-lation of regenerative signals in Cx32KO nerves.

Isoformsof axonal- and Schwann cell-derivedNrg1 have been impli-cated in the regulation of peripheral nerve demyelination and regener-ation following injury (Chang et al., 2013; Fricker et al., 2011; Stassartet al., 2013. Significant changes in Nrg1 expression were not detectedin Cx32KO nerves relative to WT at any time points examined. Whileregulation of Nrg1 receptors ErbB2 and ErbB3 is essential to peripheralnerve responses to injury (Mukhatyar et al., 2014; Nicolino et al.,2009), the probes for ErbB2 and ErbB3 were not present on IlluminaMouse Ref8-v2 chips, precluding the ability to address regulation ofErbB2 and ErbB3 genes in these studies.

c-Jun is a master regulator of Schwann cell response to peripheralnerve injury (Arthur-Farraj et al., 2012; Pham et al., 2009). It is rapidlyupregulated following nerve injury in mice and rats in concurrencewith Wallerian degeneration (Camara-Lemarroy et al., 2010; Parkinsonet al., 2008). Activation of c-Jun governs Schwann cell response to injuryincluding myelin clearance, trophic factor expression, and regenerativepotential in injured peripheral nerves (Jessen and Mirsky, 2008).Accordingly, mice with a Schwann cell specific deletion of c-Jun displayresults in neuronal death, disruption of the Schwann repair cell, delayedmyelin degeneration, impaired myelin clearing, and impaired regen-eration following injury (Arthur-Farraj et al., 2012) Klein et al. exam-ined c-Jun in three mouse models of CMT1 and found increasedimmunolabeling for c-Jun in myelinating Cx32KO Schwann cells inadultmice in a spatial associationwith demyelinating axons, suggestingthat c-Junmight also be amarker for pre-demyelinating axonal process-es in disease models of peripheral nerves (Klein et al., 2014). We foundan increased expression of c-Jun in Cx32KO nerves relative to WT at14 dpi, a period of axonal regeneration and re-myelination, but didnot detect significant changes in Cx32KO nerves relative to WT atother time points. Failure to detect differences in c-Jun expressionmay reflect the relatively young ages used these studies,whichprecedesthe age at which c-Jun might be activated due to demyelination andother axonal changes that appear in older Cx32KO peripheral nerves(Sargiannidou et al., 2009; Scherer et al., 1998).

Nerve injury is associated with the de-differentiation and re-programming of Schwann cells to a repair cell phenotype; with corre-sponding increases in c-Jun and declines in Egr2, a pro-myelinatingtranscription regulator (Parkinson et al., 2008; Pham et al., 2009).Subsequent transient upregulation of Pou3f1 (Oct-6/SCIP) triggers asustained re-induction of Egr2 phenotype which, in turn, triggersSchwann cell re-differentiation. Induction of myelin-specific genes en-sues with successful re-differentiation toward a mature myelinatingphenotype (Mandemakers et al., 2000; Svaren and Meijer, 2008). In-creases in c-Jun and Pou3f1 along with decreased Egr2 were observedin 14dpi Cx32KO nerves relative toWT; further supporting a dysregula-tion of the Schwann cell program in the absence of functional Cx32.However, no significant changes in c-Jun were observed within geno-type in Cx32KO nerves relative to uninjured at any time point, suggest-ing either a baseline shift toward less mature or injured Cx32KO

347M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

Schwann cell phenotypes or an inability to properly respond to the inju-ry. Examination of WT responses to injury followed predicted patternswith, significant increases in c-Jun at 5dpi relative to uninjured WT.

p75NGFR is amarker for immature or non-myelinating Schwann cellsand is co-localized with c-Jun following nerve injury (Fontana et al.,2012). We found an increased expression of p75NGFR in uninjured and14dpi Cx32KO nerves relative toWT, further supporting the conclusionthat Cx32KO Schwann cells display a basal immature or injuredphenotype. These findings are consistent with recent reports showingincreased immunolabeling for p75NGFR in non-myelinating and super-numerary Schwann cells in three CMT1 mouse models (Guenard et al.,1996; Klein et al., 2014). The effects of p75NGFR on Schwann cells afterinjury are complex as p75NGFR has roles in both Schwann cell apoptosis(Syroid et al., 2000) andmyelination, depending on ligand presentationand intracellular binding partners (Cosgaya et al., 2002; Provenzanoet al., 2008; Teng et al., 2005; Tep et al., 2012). Signaling throughp75NGFR can lead to activation of c-Jun-dependent downstream signal-ing pathways to regulate Schwann cell survival and differentiation(Lindwall Blom et al., 2014; Shin et al., 2013; Yeiser et al., 2004). Inter-estingly, c-Fos, which forms the AP1 transcription factor complex withc-Jun, was significantly upregulated in uninjured and 14dpi Cx32KOnerves, relative to WT. Divergent classes of differentially regulatedgenes in uninjured and injured Cx32KO nerves are associated with thep75NGFR and c-Jun signaling pathways, as illustrated in IPA networksin Supplementary Figs. 2, 3, and 4. That loss of Cx32 influences expres-sion of these major markers of Schwann cell differentiation, suggestsCx32 plays an important role in modulating expression of matureSchwann cell phenotype.

Wallerian degeneration is an innate immune response to peripheralnerve injury in which rapid release of pro-inflammatory cytokines bySchwann cells induces macrophage recruitment and activation to thesite of injury. Macrophage-mediated cytokine and chemokine produc-tion contributes further to the molecular and cellular events in injuredperipheral nerves during the later phases of Wallerian degeneration(Cheepudomwit et al., 2008; Li et al., 2013). Mice with the delayedWallerian degeneration (WldS) mutation have a slower break downof distal axons for weeks after injury (Barrette et al., 2010). Nerve injurytriggers increased differential expression of genes associated with func-tional groups for axonal degeneration, inflammation, immune response,and growth factors/cytokines in WT as compared to WldS mice(Barrette et al., 2010). Interestingly, many of these same immune re-sponse genes and biological processes were upregulated or enrichedin uninjured and 14 dpi Cx32KO nerves (relative to WT), furthersupporting that loss of Cx32 promotes a shift toward a gene expressionprofile resembling an injured or de-differentiated repair ofWT Schwanncell.

Mutations in GJB1 encoding Cx32 cause CMT1X. However, thecellular mechanisms underlying disruption of functional Cx32 and thedemyelinating axonal pathology associated with CMT1X are not under-stood. More than 300 mutations in GJB1 have been identified, affectingall regions of the Cx32 protein. Loss-of-function GJB1/CMT1Xmutationsabolish Cx32 protein expression or limit production of functional Cx32channels (Braathen, 2012; Ionasescu et al., 1996; Scherer et al., 1999)and appear to cause a similar degree of neuropathy. Moreover, studiesexamining GJB1/CMTXmutants that are associatedwith less severe dis-ease pathology support partial loss-of-function with alterations in pro-tein trafficking and biophysical changes in Cx32 channel properties(Scherer and Kleopa, 2012). These findings address the effects of GJB1/CMT1X mutations on Cx32 gap junction activity in mature myelinatingSchwann cells, but fail to fully account for disruption of Cx32 function innon-myelinating, regenerating or immature Schwann cells.

Axonal pathology secondary to demyelination is a feature of CMT1X.As demonstrated in transplant models in which human nerve segmentsare transplanted into sciatic nerves of nudemice, mouse axons regener-ate into the transplanted segments and are re-myelinated by the humanSchwann cells from the graft. Nerve grafts from patients with CMT1X

mutations displayed decreased axonal diameter and delayed regenera-tion as compared to axons that regenerate into grafts of normal humannerve (Abrams et al., 2003; Sahenk and Chen, 1998). Mouse models ofCMT1X provide additional evidence of axonal pathology in CMT1X.Recent studies re-examining Cx32KO mice identified axonal pathologyprior to demyelination, including a reduction in the diameter of myelin-ated axons and other cytoskeletal changes associated with slowed axo-nal transport and axonopathy (Vavlitou et al., 2010). Disturbance ofSchwann cell–axonal signaling and support of axonal function wouldprovide a means for myelination-independent axonal pathology ob-served in CMT1X.

Loss of functional Schwann cell Cx32 in uninjured and 14 dpi pe-ripheral nerves leads to a gene expression profile similar to that of aninjuredWT nerve, with upregulation of genes associated with imma-ture and injured Schwann cell phenotypes. Non-myelinating andregenerating populations of Schwann cells require Cx32 for normalaxonal maintenance and the re-myelination, and regeneration of pe-ripheral nerve following injury. Dysregulation of Schwann cell–axonalcommunication in Cx32KO mice contributes to the disruption of theimmune response and differentiation pathways that are essential forsuccessful nerve regeneration following injury and for themaintenanceof normal intact peripheral nerves.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.expneurol.2014.10.014.

Acknowledgments

This work was supported by grants from NINDS (5R01NS050705-06)and Muscular Dystrophy Association (218214) awarded to CKA.

References

Abrams, C.K., Bennett, M.V.L., 2000. Hereditary Human Diseases caused by connexinmutations. In: Peracchia, C. (Ed.), Gap Junctions — Molecular Basis of Cell Communi-cation in Health and Disease. Academic Press.

Abrams, C.K., Rash, J.E., 2009. Connexins in the nervous system. In: Harris, A.L., Locke, D.(Eds.), Connexins: A Guide. Humana Press, pp. 323–358.

Abrams, C.K., Freidin, M., Bukauskas, F., Dobrenis, K., Bargiello, T.A., Verselis, V.K., Bennett,M.V., Chen, L., Sahenk, Z., 2003. Pathogenesis of X-linked Charcot–Marie–Toothdisease: differential effects of two mutations in connexin 32. J. Neurosci. 23,10548–10558.

Allodi, I., Udina, E., Navarro, X., 2012. Specificity of peripheral nerve regeneration: interac-tions at the axon level. Prog. Neurobiol. 98, 16–37.

Anzini, P., Neuberg, D.H., Schachner, M., Nelles, E., Willecke, K., Zielasek, J., Toyka, K.V.,Suter, U., Martini, R., 1997. Structural abnormalities and deficient maintenance ofperipheral nerve myelin in mice lacking the gap junction protein connexin 32.J. Neurosci. 17, 4545–4551.

Arthur-Farraj, P.J., Latouche, M., Wilton, D.K., Quintes, S., Chabrol, E., Banerjee, A.,Woodhoo, A., Jenkins, B., Rahman, M., Turmaine, M., Wicher, G.K., Mitter, R.,Greensmith, L., Behrens, A., Raivich, G., Mirsky, R., Jessen, K.R., 2012. c-Jun reprogramsSchwann cells of injured nerves to generate a repair cell essential for regeneration.Neuron 75, 633–647.

Balice-Gordon, R.J., Bone, L.J., Scherer, S.S., 1998. Functional gap junctions in the Schwanncell myelin sheath. J. Cell Biol. 142, 1095–1104.

Barrette, B., Calvo, E., Vallieres, N., Lacroix, S., 2010. Transcriptional profiling of the injuredsciatic nerve of mice carrying the Wld(S) mutant gene: identification of genesinvolved in neuroprotection, neuroinflammation, and nerve regeneration. BrainBehav. Immun. 24, 1254–1267.

Benjamini, Y., Hochberg, Y., 1995. Controlling the false discovery rate: a practicaland powerful approach to multiple testing. J. R. Stat. Soc. Ser. B Methodol. 57,289–300.

Blake, J.A., Bult, C.J., Eppig, J.T., Kadin, J.A., Richardson, J.E., Mouse Genome Database, G.,2014. The Mouse Genome Database: integration of and access to knowledge aboutthe laboratory mouse. Nucleic Acids Res. 42, D810–D817.

Bolin, L.M., Shooter, E.M., 1993. Neurons regulate Schwann cell genes by diffusible mole-cules. J. Cell Biol. 123, 237–243.

Bondurand, N., Girard, M., Pingault, V., Lemort, N., Dubourg, O., Goossens, M., 2001.Human Connexin 32, a gap junction protein altered in the X-linked form ofCharcot–Marie–Tooth disease, is directly regulated by the transcription factor SOX10.Hum. Mol. Genet. 10, 2783–2795.

Braathen, G.J., 2012. Genetic epidemiology of Charcot–Marie–Tooth disease. Acta Neurol.Scand. Suppl. iv-22.

Brazma, A., 2009. Minimum Information About a Microarray Experiment (MIAME) —

successes, failures, challenges. ScientificWorldJournal 9, 420–423.Breuer, K., Foroushani, A.K., Laird, M.R., Chen, C., Sribnaia, A., Lo, R., Winsor, G.L.,

Hancock, R.E., Brinkman, F.S., Lynn, D.J., 2013. InnateDB: systems biology of innate

348 M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

immunity and beyond—recent updates and continuing curation. Nucleic Acids Res.41, D1228–D1233.

Bustin, S.A., Beaulieu, J.F., Huggett, J., Jaggi, R., Kibenge, F.S., Olsvik, P.A., Penning, L.C.,Toegel, S., 2010. MIQE precis: Practical implementation of minimum standard guide-lines for fluorescence-based quantitative real-time PCR experiments. BMC Mol. Biol.11, 74.

Bustin, S.A., Benes, V., Garson, J.A., Hellemans, J., Huggett, J., Kubista, M., Mueller, R., Nolan,T., Pfaffl, M.W., Shipley, G.L., Vandesompele, J., Wittwer, C.T., 2009. The MIQE guide-lines: minimum information for publication of quantitative real-time PCR experi-ments. Clin. Chem. 55, 611–622.

Camara-Lemarroy, C.R., Guzman-de la Garza, F.J., Fernandez-Garza, N.E., 2010. Molecularinflammatory mediators in peripheral nerve degeneration and regeneration.Neuroimmunomodulation 17, 314–324.

Canclini, L., Wallrabe, H., Di Paolo, A., Kun, A., Calliari, A., Sotelo-Silveira, J.R., Sotelo, J.R.,2014. Association of Myosin Va and Schwann cells-derived RNA in mammal myelin-ated axons, analyzed by immunocytochemistry and confocal FRET microscopy.Methods San Diego, Calif. 66, 153–161.

Chang, H.M., Shyu, M.K., Tseng, G.F., Liu, C.H., Chang, H.S., Lan, C.T., Hsu, W.M., Liao, W.C.,2013. Neuregulin facilitates nerve regeneration by speeding Schwann cell migrationvia ErbB2/3-dependent FAK pathway. PLoS ONE 8, e53444.

Cheepudomwit, T., Guzelsu, E., Zhou, C., Griffin, J.W., Hoke, A., 2008. Comparison of cyto-kine expression profile during Wallerian degeneration of myelinated and unmyelin-ated peripheral axons. Neurosci. Lett. 430, 230–235.

Comerford, I., McColl, S.R., 2011. Mini-review series: focus on chemokines. Immunol. CellBiol. 89, 183–184.

Cosgaya, J.M., Chan, J.R., Shooter, E.M., 2002. The neurotrophin receptor p75NTR as apositive modulator of myelination. Science 298, 1245–1248.

Dubovy, P., Jancalek, R., Kubek, T., 2013. Role of inflammation and cytokines in peripheralnerve regeneration. Int. Rev. Neurobiol. 108, 173–206.

Fontana, X., Hristova, M., Da Costa, C., Patodia, S., Thei, L., Makwana, M., Spencer-Dene, B.,Latouche, M., Mirsky, R., Jessen, K.R., Klein, R., Raivich, G., Behrens, A., 2012. c-Jun inSchwann cells promotes axonal regeneration and motoneuron survival via paracrinesignaling. J. Cell Biol. 198, 127–141.

Freidin, M., Asche, S., Bargiello, T.A., Bennett, M.V., Abrams, C.K., 2009. Connexin 32increases the proliferative response of Schwann cells to neuregulin-1 (Nrg1). Proc.Natl. Acad. Sci. U. S. A. 106, 3567–3572.

Fricker, F.R., Lago, N., Balarajah, S., Tsantoulas, C., Tanna, S., Zhu, N., Fageiry, S.K., Jenkins,M.,Garratt, A.N., Birchmeier, C., Bennett, D.L., 2011. Axonally derived neuregulin-1 is re-quired for remyelination and regeneration after nerve injury in adulthood. J. Neurosci.31, 3225–3233.

Guenard, V., Montag, D., Schachner, M., Martini, R., 1996. Onion bulb cells in mice defi-cient for myelin genes share molecular properties with immature, differentiatednon-myelinating, and denervated Schwann cells. Glia 18, 27–38.

Hall, S.M., Li, H., Kent, A.P., 1997. Schwann cells responding to primary demyelinationin vivo express p75NTR and c-erbB receptors: a light and electron immunohisto-chemical study. J. Neurocytol. 26, 679–690.

Hochberg, Y., Benjamini, Y., 1990. More powerful procedures for multiple significancetesting. Stat. Med. 9, 811–818.

Iacobas, D.A., Iacobas, S., Spray, D.C., 2007. Connexin-dependent transcellulartranscriptomic networks in mouse brain. Prog. Biophys. Mol. Biol. 94, 169–185.

Ionasescu, V.V., 1995. Charcot–Marie–Tooth neuropathies: from clinical description tomolecular genetics. Muscle Nerve 18, 267–275.

Ionasescu, V., Ionasescu, R., Searby, C., 1996. Correlation between connexin 32 genemutations and clinical phenotype in X-linked dominant Charcot–Marie–Tooth neu-ropathy. Am. J. Med. Genet. 63, 486–491.

Jessen, K.R., Mirsky, R., 2008. Negative regulation of myelination: relevance for develop-ment, injury, and demyelinating disease. Glia 56, 1552–1565.

Kanehisa, M., Goto, S., Sato, Y., Furumichi, M., Tanabe, M., 2012. KEGG for integration andinterpretation of large-scale molecular data sets. Nucleic Acids Res. 40, D109–D114.

Kar, R., Batra, N., Riquelme, M.A., Jiang, J.X., 2012. Biological role of connexin intercellularchannels and hemichannels. Arch. Biochem. Biophys. 524, 2–15.

Klein, D., Groh, J., Wettmarshausen, J., Martini, R., 2014. Nonuniformmolecular features ofmyelinating Schwann cells in models for CMT1: distinct disease patterns are associ-ated with NCAM and c-Jun upregulation. Glia 62, 736–750.

Kumar, N.M., Gilula, N.B., 1996. The gap junction communication channel. Cell 84, 381–388.Kury, P., Greiner-Petter, R., Cornely, C., Jurgens, T., Muller, H.W., 2002. Mammalian

achaete scute homolog 2 is expressed in the adult sciatic nerve and regulates theexpression of Krox24, Mob-1, CXCR4, and p57kip2 in Schwann cells. J. Neurosci. 22,7586–7595.

Li, M., Guo, W., Zhang, P., Li, H., Gu, X., Yao, D., 2013. Signal flow and pathways in re-sponse to early Wallerian degeneration after rat sciatic nerve injury. Neurosci.Lett. 536, 56–63.

Lin, J.H., Yang, J., Liu, S., Takano, T., Wang, X., Gao, Q., Willecke, K., Nedergaard, M., 2003.Connexinmediates gap junction-independent resistance to cellular injury. J. Neurosci.23, 430–441.

Lindwall Blom, C., Martensson, L.B., Dahlin, L.B., 2014. Nerve injury-induced c-Jun activa-tion in Schwann cells is JNK independent. BioMed. Res. Int. 2014, 392971.

Lynn, D.J., Winsor, G.L., Chan, C., Richard, N., Laird, M.R., Barsky, A., Gardy, J.L., Roche, F.M.,Chan, T.H., Shah, N., Lo, R., Naseer, M., Que, J., Yau, M., Acab, M., Tulpan, D., Whiteside,M.D., Chikatamarla, A., Mah, B., Munzner, T., Hokamp, K., Hancock, R.E., Brinkman, F.S.,2008. InnateDB: facilitating systems-level analyses of themammalian innate immuneresponse. Mol. Syst. Biol. 4, 218.

Mambetisaeva, E.T., Gire, V., Evans, W.H., 1999. Multiple connexin expression in peripheralnerve, Schwann cells, and Schwannoma cells. J. Neurosci. Res. 57, 166–175.

Mandemakers, W., Zwart, R., Jaegle, M., Walbeehm, E., Visser, P., Grosveld, F., Meijer, D.,2000. A distal Schwann cell-specific enhancer mediates axonal regulation of the

Oct-6 transcription factor during peripheral nerve development and regeneration.EMBO J. 19, 2992–3003.

Mestdagh, P., Van Vlierberghe, P., De Weer, A., Muth, D., Westermann, F., Speleman, F.,Vandesompele, J., 2009. A novel and universal method for microRNA RT-qPCR datanormalization. Genome Biol. 10, R64.

Mirsky, R.,Woodhoo, A., Parkinson, D.B., Arthur-Farraj, P., Bhaskaran, A., Jessen, K.R., 2008.Novel signals controlling embryonic Schwann cell development, myelination anddedifferentiation. J. Peripher. Nerv. Syst. 13, 122–135.

Mukhatyar, V., Pai, B., Clements, I., Srinivasan, A., Huber, R., Mehta, A., Mukhopadaya, S.,Rudra, S., Patel, G., Karumbaiah, L., Bellamkonda, R., 2014.Molecular sequelae of topo-graphically guided peripheral nerve repair. Ann. Biomed. Eng. 42, 1436–1455.

Nagarajan, R., Le, N., Mahoney, H., Araki, T., Milbrandt, J., 2002. Deciphering peripheralnerve myelination by using Schwann cell expression profiling. Proc. Natl. Acad. Sci.U. S. A. 99, 8998–9003.

Nicolino, S., Panetto, A., Raimondo, S., Gambarotta, G., Guzzini, M., Fornaro, M., Battiston,B., Tos, P., Geuna, S., Perroteau, I., 2009. Denervation and reinnervation of adultskeletal muscle modulate mRNA expression of neuregulin-1 and ErbB receptors.Microsurgery 29, 464–472.

Noble, W.S., 2009. How does multiple testing correction work? Nat. Biotechnol. 27,1135–1137.

Omori, Y., Li, Q., Nishikawa, Y., Yoshioka, T., Yoshida, M., Nishimura, T., Enomoto, K., 2007.Pathological significance of intracytoplasmic connexin proteins: implication in tumorprogression. J. Membr. Biol. 218, 73–77.

Parkinson, D.B., Bhaskaran, A., Arthur-Farraj, P., Noon, L.A., Woodhoo, A., Lloyd, A.C., Feltri,M.L.,Wrabetz, L., Behrens, A.,Mirsky, R., Jessen, K.R., 2008. c-Jun is a negative regulatorof myelination. J. Cell Biol. 181, 625–637.

Patodia, S., Raivich, G., 2012. Downstream effector molecules in successful peripheralnerve regeneration. Cell Tissue Res. 349, 15–26.

Pham, K., Nassiri, N., Gupta, R., 2009. c-Jun, krox-20, and integrin beta4 expressionfollowing chronic nerve compression injury. Neurosci. Lett. 465, 194–198.

Pineau, I., Lacroix, S., 2009. Endogenous signals initiating inflammation in the injurednervous system. Glia 57, 351–361.

Provenzano, M.J., Xu, N., VerMeer,M.R., Clark, J.J., Hansen,M.R., 2008. p75NTR and sortilinincrease after facial nerve injury. Laryngoscope 118, 87–93.

Sahenk, Z., Chen, L., 1998. Abnormalities in the axonal cytoskeleton induced by aconnexin32 mutation in nerve xenografts. J. Neurosci. Res. 51, 174–184.

Sargiannidou, I., Vavlitou, N., Aristodemou, S., Hadjisavvas, A., Kyriacou, K., Scherer, S.S.,Kleopa, K.A., 2009. Connexin32 mutations cause loss of function in Schwann cellsand oligodendrocytes leading to PNS and CNS myelination defects. J. Neurosci. 29,4736–4749.

Scherer, S.S., Kleopa, K.A., 2012. X-linked Charcot–Marie–Tooth disease. J. Peripher. Nerv.Syst. 17 (Suppl. 3), 9–13.

Scherer, S.S., Deschenes, S.M., Xu, Y.T., Grinspan, J.B., Fischbeck, K.H., Paul, D.L.,1995. Connexin32 is a myelin-related protein in the PNS and CNS. J. Neurosci. 15,8281–8294.

Scherer, S.S., Xu, Y.T., Nelles, E., Fischbeck, K., Willecke, K., Bone, L.J., 1998. Connexin32-null mice develop demyelinating peripheral neuropathy. Glia 24, 8–20.

Scherer, S.S., Bone, L.J., Deschenes, S.M., Abel, A., Balice-Gordon, R.J., Fischbeck, K.H.,1999. The role of the gap junction protein connexin32 in the pathogenesis ofX-linked Charcot-Marie-Tooth disease. Novartis Found. Symp. 219, 175–185[discussion 185–177].

Shin, Y.K., Jang, S.Y., Park, J.Y., Park, S.Y., Lee, H.J., Suh, D.J., Park, H.T., 2013. TheNeuregulin-Rac-MKK7 pathway regulates antagonistic c-jun/Krox20 expression inSchwann cell dedifferentiation. Glia 61, 892–904.

Shy, M.E., Shi, Y., Wrabetz, L., Kamholz, J., Scherer, S.S., 1996. Axon-Schwann cell in-teractions regulate the expression of c-jun in Schwann cells. J. Neurosci. Res. 43,511–525.

Soares, H.D., Chen, S.C., Morgan, J.I., 2001. Differential and prolonged expression of Fos-lacZ and Jun-lacZ in neurons, glia, and muscle following sciatic nerve damage. Exp.Neurol. 167, 1–14.

Sohl, G., Gillen, C., Bosse, F., Gleichmann, M., Muller, H.W., Willecke, K., 1996. Asecond alternative transcript of the gap junction gene connexin32 is expressedin murine Schwann cells and modulated in injured sciatic nerve. Eur. J. Cell Biol. 69,267–275.

Spandidos, A., Wang, X., Wang, H., Seed, B., 2010. PrimerBank: a resource of human andmouse PCR primer pairs for gene expression detection and quantification. NucleicAcids Res. 38, D792–D799.

Stassart, R.M., Fledrich, R., Velanac, V., Brinkmann, B.G., Schwab, M.H., Meijer, D., Sereda,M.W., Nave, K.A., 2013. A role for Schwann cell-derived neuregulin-1 in remyelination.Nat. Neurosci. 16, 48–54.

Stewart, H.J., 1995. Expression of c-Jun, Jun B, Jun D and cAMP response element bindingprotein by Schwann cells and their precursors in vivo and in vitro. Eur. J. Neurosci. 7,1366–1375.

Svaren, J., Meijer, D., 2008. The molecular machinery of myelin gene transcription inSchwann cells. Glia 56, 1541–1551.

Syroid, D.E., Maycox, P.J., Soilu-Hanninen, M., Petratos, S., Bucci, T., Burrola, P., Murray, S.,Cheema, S., Lee, K.F., Lemke, G., Kilpatrick, T.J., 2000. Induction of postnatal schwanncell death by the low-affinity neurotrophin receptor in vitro and after axotomy.J. Neurosci. 20, 5741–5747.

Ten Asbroek, A.L., Van Ruissen, F., Ruijter, J.M., Baas, F., 2006. Comparison of Schwann celland sciatic nerve transcriptomes indicates that mouse is a valid model for the humanperipheral nervous system. J. Neurosci. Res. 84, 542–552.

Teng, H.K., Teng, K.K., Lee, R., Wright, S., Tevar, S., Almeida, R.D., Kermani, P., Torkin, R.,Chen, Z.Y., Lee, F.S., Kraemer, R.T., Nykjaer, A., Hempstead, B.L., 2005. ProBDNF in-duces neuronal apoptosis via activation of a receptor complex of p75NTR and sortilin.J. Neurosci. 25, 5455–5463.

349M. Freidin et al. / Experimental Neurology 263 (2015) 339–349

Tep, C., Kim, M.L., Opincariu, L.I., Limpert, A.S., Chan, J.R., Appel, B., Carter, B.D., Yoon, S.O.,2012. Brain-derived neurotrophic factor (BDNF) induces polarized signaling of smallGTPase (Rac1) protein at the onset of Schwann cell myelination through partitioning-defective 3 (Par3) protein. J. Biol. Chem. 287, 1600–1608.

Tomita, K., Kubo, T., Matsuda, K., Fujiwara, T., Yano, K., Winograd, J.M., Tohyama, M.,Hosokawa, K., 2007. The neurotrophin receptor p75NTR in Schwann cells is implicat-ed in remyelination and motor recovery after peripheral nerve injury. Glia 55,1199–1208.

Vandesompele, J., De Preter, K., Pattyn, F., Poppe, B., Van Roy, N., De Paepe, A., Speleman,F., 2002. Accurate normalization of real-time quantitative RT-PCR data by geometricaveraging of multiple internal control genes. Genome Biol. 3, RESEARCH0034.

Vavlitou, N., Sargiannidou, I., Markoullis, K., Kyriacou, K., Scherer, S.S., Kleopa, K.A., 2010. Ax-onal pathology precedes demyelination in a mouse model of X-linked demyelinating/type I Charcot–Marie–Tooth neuropathy. J. Neuropathol. Exp. Neurol. 69, 945–958.

Vinken, M., Decrock, E., De Vuyst, E., Ponsaerts, R., D'Hondt, C., Bultynck, G., Ceelen, L.,Vanhaecke, T., Leybaert, L., Rogiers, V., 2011. Connexins: sensors and regulators ofcell cycling. Biochim. Biophys. Acta 1815, 13–25.

Yang, D.P., Kim, J., Syed, N., Tung, Y.J., Bhaskaran, A., Mindos, T., Mirsky, R., Jessen, K.R.,Maurel, P., Parkinson, D.B., Kim, H.A., 2012. p38MAPK activation promotes dener-vated Schwann cell phenotype and functions as a negative regulator of Schwanncell differentiation and myelination. J. Neurosci. 32, 7158–7168.

Yeiser, E.C., Rutkoski, N.J., Naito, A., Inoue, J., Carter, B.D., 2004. Neurotrophin signalingthrough the p75 receptor is deficient in traf6–/– mice. J. Neurosci. 24, 10521–10529.

Yiu, E.M., Geevasinga, N., Nicholson, G.A., Fagan, E.R., Ryan, M.M., Ouvrier, R.A., 2011. Aretrospective reviewof X-linkedCharcot–Marie–Tooth disease in childhood. Neurology76, 461–466.

Zeeberg, B., Feng, W., Wang, G., Wang, M., Fojo, A., Sunshine, M., Narasimhan, S., Kane, D.,Reinhold,W., Lababidi, S., Bussey, K., Riss, J., Barrett, J.,Weinstein, J., 2003. GoMiner: a re-source for biological interpretation of genomic and proteomic data. GenomeBiol. 4, R28.

Zeeberg, B., Qin, H., Narasimhan, S., Sunshine, M., Cao, H., Kane, D., Reimers, M., Stephens,R., Bryant, D., Burt, S., Elnekave, E., Hari, D., Wynn, T., Cunningham-Rundles, C.,Stewart, D., Nelson, D., Weinstein, J., 2005. High-throughput GoMiner, an ‘industrial-strength’ integrative gene ontology tool for interpretation of multiple-microarrayexperiments, with application to studies of Common Variable Immune Deficiency(CVID). BMC Bioinforma. 6, 168.

Zhou, X.F., Li, H.Y., 2007. Roles of glial p75NTR in axonal regeneration. J. Neurosci. Res. 85,1601–1605.