B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ loca te /b ra in res

Research Report

Murine central and peripheral nervous system transcriptomes:Comparative gene expression

Mark S. LeDoux⁎, Lijing Xu, Jianfeng Xiao, Brett Ferrell,Daniel L. Menkes, Ramin HomayouniUniversity of Tennessee Health Science Center, Departments of Neurology and Anatomy and Neurobiology and Center of Genomics andBioinformatics, 855 Monroe Avenue, Link Building-Suite 415, Memphis, TN 38163, USA

A R T I C L E I N F O

⁎ Corresponding author. Fax: +1 901 448 7440E-mail address: mledoux@utmem.edu (M

0006-8993/$ ­ see front matter © 2006 Elsevidoi:10.1016/j.brainres.2006.05.101

A B S T R A C T

Article history:Accepted 28 May 2006Available online 7 July 2006

The central and peripheral nervous systems exhibit significant embryological,morphological, and functional differences. Moreover, the pathology of most acquired andhereditary neurological diseases preferentially targets specific components of the nervoussystem. In order to test the hypothesis that central and peripheral neural transcriptomesshow fundamental quantitative differences, Affymetrix GeneChip® expression arrays wereused to compare murine lumbar spinal cord (SC) and dorsal root ganglion (DRG) geneexpression. As the crucial component of a novel technique to preserve RNA integrity, micewere perfusion-fixed with RNAlater™ before the SC and DRG were harvested. As perAffymetrix terminology, a total of 111 transcripts were present (P) on all DRG arrays, absent(A) on all SC arrays, and demonstrated at least 10-fold greater expression in DRG than in SC.Conversely, a total of 112 transcripts were present on all SC arrays, absent on all DRG arrays,and showed at least 10-fold greater expression in SC than in DRG. For a subset of transcripts,quantitative real-time RT-PCR was used to corroborate and validate microarray results.Among those genes enriched in DRG, many belonged to a few distinct functional classes: G-protein coupled receptor–protein signaling pathways, potassium transport, sodiumtransport, sensory perception, and cell-surface receptor-linked signal transduction. Incontrast, genes associated with synaptic transmission, organic acid transport,neurotransmitter transport, and circulation were enriched in SC. Notably, the majority ofgenes causally associated with hereditary neuropathies were highly enriched in DRG. Thesedifferential neural gene expression profiles provide a robust framework for futuremolecularand genetic studies of neuropathy and SC diseases.

© 2006 Elsevier B.V. All rights reserved.

Keywords:Dorsal root ganglionSpinal cordMicroarrayRNAG-proteinSodium channelNeuropathy

1. Introduction

Neuroanatomical localization of neural dysfunction is thecornerstone of clinical neurology and is governed by thetopological and temporal specificity of gene expression. Acommon tenet of molecular medicine is that the propensity of

.

.S. LeDoux).

er B.V. All rights reserved

particular organs and tissues to exhibit disease manifesta-tions is closely related to the cell-type specificity of proteinexpression. Along these lines, the cell bodies, axons, andmyelin sheaths of peripheral nerves are either preferentiallyor selectively involved in a large number of acquired andhereditary diseases of the human nervous system.

.

25B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

The central nervous system (CNS) consists of the brain andspinal cord whereas the peripheral nervous system (PNS)includes 31 pairs of spinal nerves and 10 pairs of cranial nervesalong with their associated ganglia, (N.B., the olfactory bulband optic nerves are CNS projections). Spinal nerves containgeneral afferent and general efferent fibers. General afferentfibers originate from dorsal root ganglia (DRG) cells and aredivided into two subtypes: general somatic afferent (GSA) andgeneral visceral afferent (GVA). GSA fibers transduce andtransmit exteroceptive, proprioceptive, and kinesthetic infor-mation. GVA fibers transduce and transmit information frominteroceptors (visceroceptors) which are related to functionsof the autonomic nervous system such as excretion, circula-tion, digestion, and respiration. In contrast to general afferentfibers, general efferent fibers arise from neurons within theCNS. Specifically, general somatic efferent (GSE) and visceralefferent (GVE) fibers originate from motor neurons in laminaIX of the anterior horn whereas preganglionic sympathetic orparasympathetic neurons are located in the intermediolateralregion of the spinal cord (SC). Therefore, to study geneexpression in the afferent division of the PNS, one mustfocus on the DRG. Moreover, based on the functional specia-

Fig. 1 – Electropherograms of total RNA generated by the Agilenincluding a control lane (L). The spike controls appear as a green bDRG3, respectively. Lanes 4, 5, and 6 correspond to pooled SC1, SCeach sample. The ordinate for each graph is fluorescence where

lizations of general afferent fibers, the DRG can be predicted touniquely express genes forming the molecular machinery formechanoreception, thermoreception, and nociception.

Herein, we describe a novel method to facilitate RNAacquisition from mouse DRG and the analysis of this geneticmaterial with high density oligonucleotide arrays to studydifferences in gene expression between the CNS and afferentdivision of the PNS. The resultant dataset provides funda-mental support for a range of physiological, molecular, andclinical studies related to the PNS, particularly its sensorycomponent. Importantly, this dataset can also be used toisolate candidate genes for mutation screening within hered-itary neuropathy loci.

2. Results

2.1. RNA integrity and data reliability

An average of 13.6 μg (10.8 μg, 13.2 μg, and 16.8 μg) of total RNAwas obtained for the 3 independent pooled (i.e., 48 ganglia)DRG samples and an average of 35.2 μg (36.0 μg, 37.2 μg, and

t Bioanalyzer 2100. (A) Composite of all electropherogramsand. Lanes 1, 2, and 3 correspond to pooled DRG1, DRG2, and2, and SC3, respectively. (B) Individual electropherograms foras the abscissa for each graph is time (s).

26 B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

32.4 μg) of total RNA was obtained for the 3 independentpooled (i.e., 12 lumbar cord segments) SC samples. As seen inFigs. 1A and B, 28S and 18S bands and peaks, respectively,were crisp for all DRG and SC RNA specimens. Furthermore,low baseline fluorescence was obtained for all samples. Smallpeaks migrating at approximately 25 s represent 5S rRNA,possibly in combination with tRNA. In Fig. 1B, note that theheight of the spike control relative to the heights of the 5S, 18S,and 28S peaks shows moderate variability which is due todifferences in RNA concentrations among the 6 samples.

Total RNAwas used to generate labeled cRNA probeswhichwere hybridized to Mouse 430 2.0 GeneChip® arrays asdescribed in Experimental procedures. Affymetrix QualityControl data (see Supplementary Material Online) were verysimilar for the DRG and SC arrays. The expression values foreach GeneChip® were pre-processed and normalized. Toevaluate whether there were any intensity dependent biasesin the dataset, we generated MA plots, where the averagedintensity ratio (M) for each probe set in the DRG and SCsamples was plotted against the product of their expressionintensities (A). Using this method, we did not observe anyintensity dependent effects in the normalized data (Fig. 2).Scatter plots of the normalized expression values between

Fig. 2 – MA plots before (A) and after (B) normalization of themicroarray data.

replicate experiments demonstrated that the surgical andgene expression array procedures used in this study were veryreproducible. The lowest R2 was 0.9786 and very few outlierswere seen on the six plots. Analysis of Figs. 3D and F indicatesonly minor discordance of highly expressed genes in the SC2sample in comparison with the SC1 and SC3 samples.Therefore, we conclude that overall reproducibility wassimilar among the DRG and SC arrays.

For statisticalmeasures of data reliability, we examined thedistribution of P and Q values calculated from pairwisecomparisons (two-sided t-tests) of the averaged expressionvalues for each probe set in the DRG (N = 3) and SC (N = 3)samples. The Q value is derived from the distribution of Pvalues for the experiment and is a measure of the expectednumber of false positives at a specific P value cutoff. In otherwords, if the difference in gene expression between twoconditions is random, then the distribution of P values isexpected to be flat (i.e., a null distribution; Fig. 4A). In ourstudy, both P and Q value histograms were unimodal withexponential distributions (Figs. 4A and B). The markedconcentration of genes that are differentially expressed withlow P and Q values indicates that the expression differencesbetween the DRG and SC presented here are reliable.We found12,171 probe sets showing differential expression withP < 0.05. Since the estimated Q value for a P of 0.05 is 0.0248,approximately 302 of these 12,171 probe sets may be falsepositives (Fig. 4C). As illustrated with a volcano plot, similarnumbers of genes were either enriched in DRG or SC across abroad range of P values (Fig. 5). Those genes showing strikingenrichment in either DRG or SC are presented in Tables 1 and2. QRT-PCR results for a subset of these highly enriched genesare provided in Table 3.

2.2. Genes enriched in the DRG

Among the filtered collection of 22,946 probe sets, 3452probe sets were expressed greater than 2-fold in DRGcompared to SC. Among these, 3329 probe sets showed a Pvalue <0.05 and 1497 probe sets showed a P value <0.001(Excel Workbook Online). Notably, 595 probe sets showed aDRG/SC fold difference ≥5 along with a P value <0.05. Tofocus the functional analysis on a smaller subset of genes,probe sets were selected if they were present on all 3 DRGarrays (PPP) and absent on all 3 SC arrays (AAA) along witha fold difference ≥10 and P and Q values <0.05. Thisprocedure resulted in identification of 111 genes that wererobustly expressed in DRG compared with SC (Tables 1 and3). Among these, the transmembrane 4 superfamily mem-ber 3 (Tm4sf3) and FXYD domain-containing ion transportregulator 2 (Fxyd2) showed the largest expression differ-ences (>200).

Enrichment in DRG was confirmed for 10 transcriptsanalyzed with QRT-PCR (Table 3). For 7 of these 10 transcripts,mean values for differential expression were higher with QRT-PCR thanwithmicroarray. Scn10a showed the largest QRT-PCRfold difference.

Among the heavily-filtered group of 111 DRG enriched genespresented in Table 1, GOTM analysis identified five importantparental biological process categories: cell communication, trans-port, organ morphogenesis, calcium homeostasis, and sensory

Fig. 3 – Scatter plots of filtered and normalized expression values for the three DRG (DRG1, DRG2, and DRG3) and three SC (SC1,SC2, and SC3) arrays.

27B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

perception (Fig. 6). Within the cell communication parentalcategory, a large number of genes were parceled into the G-protein-coupled receptor–protein signaling sub-category: specif-ic examples include endothelial differentiation sphingolipid G-protein-coupled receptors 3 and 7(Edg3 andEdg7); G-protein-coupled receptors 35, 64, and 73 (Gpr35, Gpr64, and Gpr73); andserotonin receptor 3B (Htr3b). Two genes functionally associatedwith the sensory perception of pain, amiloride-sensitive cationchannel 3 (Accn3) and calsenilin (Csen), were also categorized as

cation transporters. Similarly, docking protein 4 (Dok4) andneurotrophic tyrosine kinase receptor type 1 (Ntrk1) wereclassified into both the cell communication and organmorphogenesis parental categories.

2.3. Genes enriched in the SC

Among the filtered collection of 22,946 probe sets, 1708 wereexpressed greater than 2-fold in SC compared with DRG.

28 B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

Among these, 1515 probe sets showed a P value <0.05 and 536probe sets showed a P value <0.001 (ExcelWorkbook Online). Atotal of 406 probe sets showed a SC/DRG fold difference ≥5

Fig. 4 – Summary statistics for the DRG and SC datasets.Density histograms of P (A) and Q (B) values for differentialgene expression between the DRG and SC. (C) The expectednumber of false positives as determined with the Q statisticas a function of the number of significant tests (P < 0.05).

Fig. 5 – A volcano plot displaying mean fold differences foreach probe set in the DRG and SC samples as a function ofP value. Genes showing greater than 2-fold differentialexpression with P values <0.05 are depicted with black dots.Other genes are represented with gray dots. The vertical linerepresents a 10-fold difference in gene expression whereasthe horizontal line corresponds to a P value of 0.001.

along with a P value <0.05. Using the strictest criteria, a total of112 genes were selected that were present on all SC arrays(PPP) and absent on all DRG arrays (AAA) along with a folddifference ≥10 and P and Q values <0.05 (Table 2). Prominentamong those genes showing the largest SC/DRG fold differ-ence were glial fibrillary acidic protein (Gfap) and two genesconnected with oligodendroglia, myelin-associated oligoden-drocytic basic protein (Mobp) and oligodendrocyte transcrip-tion factor 1 (Olig1).

Enrichment in SC was confirmed for 10 transcriptsanalyzed with QRT-PCR (Table 3). For 6 of these 10 transcripts,mean values for differential expression were higher with QRT-PCR than with microarray. Slc18a3 showed the largest QRT-PCR fold difference.

GOTM identified three parental biological process catego-ries enriched in SC: cell communication, transport, andcirculation (Fig. 6). Although the cell communication andtransport parental categories were shared with DRG, theindividual genes within these categories differed in importantways. For instance, the SC cell communication parentalcategory contained a synaptic transmission sub-categoryincluding genes for glutamic acid decarboxylase 1 (Gad1) andglycine receptor, alpha 1 (Glra1) and alpha 2 (Glra2) subunits.Likewise, the SC transport parental category contained aneurotransmitter transport sub-category with genes forgamma-aminobutyric acid (GABAA) transporter 4 (Gabt4) andsyntaxin 1B2 (Stx1b2).

2.4. Expression of genes linked to hereditary neuropathiesin DRG

To test the hypothesis that genes associated with PNSdiseases are enriched in the DRG, we examined the expressionlevelsof genes thathavebeen linkedtohereditaryneuropathies.

Table 1 – Genes that are enriched in DRG compared to SC

Symbol Gene name Fold difference P value Q value

Tm4sf3 Transmembrane 4 superfamily member 3 232.7 1.20E-03 2.28E-03Fxyd2 FXYD domain-containing ion

transport regulator 2202.3 4.23E-03 4.65E-03

Accn3 Amiloride-sensitive cation channel 3 193.5 1.89E-04 9.49E-04Scn10a Sodium channel, voltage-gated,

type X, alpha polypeptide170.3 2.29E-03 3.24E-03

Runx1 Runt related transcription factor 1 150.7 9.64E-04 2.02E-03Acpp Acid phosphatase, prostate 147.9 7.82E-03 6.75E-03Avil Advillin 128.0 1.68E-03 2.74E-03Tmem45b Transmembrane protein 45b 111.7 1.29E-03 2.37E-03MGI:1919137 Protein phosphatase 2C zeta 108.4 2.41E-02 1.46E-02Ngfr Nerve growth factor receptor

(TNFR superfamily, member 16)108.0 1.07E-02 8.29E-03

Scn7a Sodium channel, voltage-gated,type VI, alpha polypeptide

94.5 7.58E-04 1.79E-03

Edg7 Endothelial differentiation,lysophosphatidic acidG-protein-coupled receptor 7

92.5 2.07E-02 1.31E-02

Scn9a Sodium channel, voltage-gated,type IX, alpha polypeptide

88.4 1.38E-03 2.44E-03

P2rx3 Purinergic receptor P2X,ligand-gated ion channel, 3

88.0 1.55E-03 2.61E-03

Tusc5 Tumor suppressor candidate 5 76.9 1.63E-03 2.70E-03C130034I18Rik RIKEN cDNA C130034I18 gene 61.8 4.29E-03 4.69E-03Gcnt2 Glucosaminyl (N-acetyl) transferase 2,

I-branching enzyme58.5 6.07E-04 1.59E-03

Htr3b 5-hydroxytryptamine(serotonin) receptor 3B

53.8 1.72E-04 9.12E-04

2310042N02Rik RIKEN cDNA 2310042N02 gene 50.1 4.87E-03 5.05E-03Isl1 ISL1 transcription factor,

LIM/homeodomain (islet 1)45.0 8.01E-04 1.84E-03

Gpr64 G protein-coupled receptor 64 44.1 3.59E-04 1.25E-03Aqp1 Aquaporin 1 42.6 1.93E-03 2.94E-03Timeless Timeless homolog (Drosophila) 41.9 3.63E-02 1.95E-02Slc43a3 Solute carrier family 43, member 3 41.3 2.11E-02 1.32E-02C1qtnf1 C1q and tumor necrosis

factor related protein 139.3 2.20E-03 3.16E-03

Galnt5 N-acetylgalactosaminyltransferase 5 37.9 2.17E-02 1.35E-02S100a4 S100 calcium binding protein A4 37.5 3.31E-02 1.83E-02Myo1b Myosin IB 36.5 7.61E-03 6.62E-03Cd44 CD44 antigen 35.8 4.17E-03 4.61E-03Ttn Titin 35.7 7.19E-03 6.39E-03Nmb Neuromedin B 35.2 1.15E-05 2.90E-04Osbpl3 Oxysterol binding protein-like 3 35.1 1.78E-03 2.81E-03Serpina3g Serine (or cysteine) proteinase

inhibitor, clade A, member 3G34.8 1.21E-03 2.28E-03

Il31ra Interleukin 31 receptor A 33.8 7.65E-03 6.64E-03Daf1 Decay accelerating factor 1 32.3 4.77E-04 1.41E-03Rab3d RAB3D, member RAS oncogene family 31.8 1.32E-02 9.60E-03Nptx2 Neuronal pentraxin 2 30.5 8.72E-04 1.91E-03Blvrb Biliverdin reductase B

(flavin reductase (NADPH))30.4 5.67E-04 1.53E-03

Prkg1 Protein kinase, cGMP-dependent, type I 29.7 1.76E-04 9.19E-04Fmo2 Flavin containing monooxygenase 2 28.7 5.17E-04 1.47E-03Edg3 Endothelial differentiation,

sphingolipid G-protein-coupled receptor, 325.7 8.89E-03 7.33E-03

Ceacam10 CEA-related cell adhesion molecule 10 25.2 1.14E-02 8.65E-03Gm337 Gene model 337, (NCBI) 24.6 2.25E-04 1.00E-03Lect1 Leukocyte cell derived chemotaxin 1 23.6 2.30E-02 1.41E-02Steap Six transmembrane epithelial

antigen of the prostate23.5 2.62E-02 1.55E-02

Eva1 Epithelial V-like antigen 1 23.4 3.09E-02 1.74E-02Gpr73 G protein-coupled receptor 73 23.2 3.78E-03 4.34E-03Csen Calsenilin, presenilin binding protein,

EF hand transcription factor22.6 1.18E-03 2.26E-03

(continued on next page)

29B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

Table 1 (continued)

Symbol Gene name Fold difference P value Q value

2310047C17Rik RIKEN cDNA 2310047C17 gene 22.1 2.05E-02 1.30E-02E130310N06 Hypothetical protein E130310N06 21.3 1.56E-02 1.07E-02Nical NEDD9 interacting protein with

calponin homology and LIM domains20.7 1.05E-02 8.17E-03

Trpc6 Transient receptor potential cationchannel, subfamily C, member 6

20.2 1.29E-02 9.45E-03

MGC54896 Similar to CTCL tumor antigen se57-1 20.0 7.85E-06 2.63E-04Kcnv1 Potassium channel, subfamily V, member 1 19.8 1.23E-02 9.14E-03BC025206 cDNA sequence BC025206 18.9 1.92E-02 1.24E-02Dok4 Docking protein 4 18.7 1.69E-03 2.74E-03Ntrk1 Neurotrophic tyrosine

kinase, receptor, type 118.7 8.52E-04 1.90E-03

Fads3 Fatty acid desaturase 3 18.4 4.58E-05 5.15E-04Bnc2 Basonuclin 2 18.3 6.52E-03 6.01E-03Pcdh21 Protocadherin 21 18.1 1.12E-02 8.54E-03Cldn1 Claudin 1 17.8 4.90E-03 5.06E-03Slc6a19 Solute carrier family 6

(neurotransmitter transporter), member 1917.8 3.08E-03 3.84E-03

Tde2l Tumor differentially expressed 2-like 17.6 1.63E-02 1.10E-02Slc12a7 Solute carrier family 12, member 7 17.2 3.78E-07 1.18E-04Fut11 Fucosyltransferase 11 16.6 3.59E-02 1.94E-02Smoc2 SPARC related modular

calcium binding 216.4 6.19E-03 5.83E-03

Rarres1 Retinoic acid receptorresponder (tazarotene induced) 1

16.4 8.99E-05 6.77E-04

1300006M19Rik RIKEN cDNA 1300006M19 gene 16.1 1.92E-03 2.94E-03Trpa1 Transient receptor potential cation

channel, subfamily A, member 115.9 2.31E-02 1.41E-02

Osbpl10 Oxysterol binding protein-like 10 15.6 1.81E-03 2.84E-03Kctd8 Potassium channel

tetramerisation domain containing 815.6 3.38E-03 4.08E-03

Ccl11 Small chemokine (C-C motif) ligand 11 14.7 9.39E-03 7.59E-031300013J15Rik RIKEN cDNA 1300013J15 gene 14.4 3.85E-02 2.04E-02Onecut1 One cut domain, family member 1 14.2 2.88E-02 1.65E-02Uaca Uveal autoantigen with coiled-coil

domains and ankyrin repeats14.2 1.67E-02 1.12E-02

Klhl5 Kelch-like 5 (Drosophila) 14.1 1.72E-02 1.15E-02Antxr2 Anthrax toxin receptor 2 14.0 4.49E-03 4.83E-03Hoxd1 Homeo box D1 14.0 6.34E-06 2.50E-04Arpc1b Actin related protein 2/3

complex, subunit 1B13.3 1.73E-02 1.15E-02

Panx1 Pannexin 1 13.0 1.63E-02 1.10E-02Pou4f2 POU domain, class 4,

transcription factor 212.9 1.61E-03 2.67E-03

Sdc1 Syndecan 1 12.6 1.17E-02 8.82E-03Slco5a1 Solute carrier organic anion

transporter family, member 5A112.6 3.30E-02 1.82E-02

C730009D12 Hypothetical protein C730009D12 12.6 1.70E-03 2.75E-03Plcb3 Phospholipase C, beta 3 11.9 4.26E-02 2.20E-024930422I22Rik RIKEN cDNA 4930422I22 gene 11.6 2.63E-02 1.55E-02Ptgir Prostaglandin I receptor (IP) 11.6 9.77E-04 2.03E-03Chrna6 Cholinergic receptor,

nicotinic, alpha polypeptide 611.6 1.36E-02 9.80E-03

Cpne3 Copine III 11.6 6.17E-03 5.82E-03Syt9 Synaptotagmin 9 11.5 1.64E-03 2.70E-03BQ952480 Expressed sequence BQ952480 11.5 1.08E-04 7.43E-04Tbx2 T-box 2 11.4 1.07E-02 8.29E-03Sostdc1 Sclerostin domain containing 1 11.4 1.78E-02 1.18E-02Pawr PRKC, apoptosis, WT1, regulator 11.2 2.85E-02 1.64E-020610010I15Rik RIKEN cDNA 0610010I15 gene 11.2 2.51E-03 3.42E-03Plxnd1 Plexin D1 11.2 1.62E-04 8.84E-04Nol3 Nucleolar protein 3

(apoptosis repressor with CARD domain)11.1 9.05E-04 1.96E-03

5730596K20Rik RIKEN cDNA 5730596K20 gene 11.1 2.23E-04 1.00E-03Stat6 Signal transducer and activator of transcription 6 11.0 1.40E-03 2.46E-03

30 B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

Table 1 (continued)

Symbol Gene name Fold difference P value Q value

Th Tyrosine hydroxylase 11.0 5.63E-05 5.52E-04BB146404 Expressed sequence BB146404 10.8 3.60E-02 1.94E-021810020C19Rik RIKEN cDNA 1810020C19 gene 10.7 4.69E-05 5.18E-04Pqlc1 PQ loop repeat containing 1 10.6 3.34E-03 4.05E-03Slc7a7 Solute carrier family 7

(cationic amino acid transporter,y+ system), member 7

10.5 4.54E-05 5.14E-04

Gna14 Guanine nucleotidebinding protein, alpha 14

10.5 9.39E-04 2.00E-03

E030024M05Rik RIKEN cDNA E030024M05 gene 10.4 1.40E-05 3.04E-04Kcns1 K+ voltage-gated channel, subfamily S, 1 10.4 4.20E-02 2.17E-02Oasl2 2′-5′ oligoadenylate synthetase-like 2 10.4 5.45E-05 5.43E-04Mrgpra2 MAS-related GPR, member A2 10.3 2.36E-04 1.02E-03Hmgcs2 3-hydroxy-3-methylglutaryl-Coenzyme A synthase 2 10.3 1.38E-02 9.89E-03Gpr35 G protein-coupled receptor 35 10.2 6.94E-03 6.26E-03

31B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

Conspicuously, the majority of genes causally associated withhereditary neuropathies are preferentially expressed in DRG(Table 4). Among the hereditary sensory and autonomicneuropathies, the two genes associated with HSN-IV, nervegrowth factor receptor (Ngfr) and neurotrophic tyrosine kinasereceptor type 1 (Ntrk1), were markedly (108- and 18.7-fold,respectively) enriched in DRG. The sodium channel gene(Scn9a) linked to a distinctive disorder of the PNS, erythrome-lalgia, showed an 88-fold difference in DRG in comparison withSC.

Most transcripts linked to hereditary motor sensoryneuropathies (i.e., Charcot–Marie–Tooth disease or CMT)were moderately enriched in DRG relative to SC. The heatshock 27 kDa protein 8 which is linked to CMT 2L showed thelargest fold difference (5.8-fold) among the various CMTs. Incontrast, transcripts linked to disorders that exhibit both CNSand PNS demyelination such as Krabbe's disease, Refsum'sdisease, and metachromatic leukodystrophy demonstratedeither minimal or no enrichment in the DRG. In analogousfashion, preponderant transcript expression in DRG was notseen in multisystem disorders with neuropathy such asMachado–Joseph disease. Interestingly, transcripts associatedwith many of the hereditary motor neuron (HMN) syndromesshowed somewhat greater expression in DRG than in SC.Another heat shock protein gene, Hspb1, which is associatedwith distal HMN, exhibited prominent differential expressionin DRG.

3. Discussion

The value of the work presented herein is bipartite.Foremost, we generated a potent dataset applicable tostudies of the general afferent component of the PNS anddisorders thereof. Additionally, we demonstrated the feasi-bility and utility of transcardiac perfusion of mice with apopular solution used for stabilizing and protecting RNA,RNAlater™. This approach should be widely applicable toRNA extraction from tissues that may require prolongedsurgical dissection for removal such as sympathetic andparasympathetic ganglia, the pituitary gland, cranial ganglia,and urogenital structures.

The gene expression profiles presented in Tables 1 and 2and the Supplementary Material Online should be viewedas trustworthy for several key reasons. First, our data arelargely compatible with what is known about the cellularand molecular biology of the CNS and PNS. For instance, asanticipated, transcripts for myelin-associated oligodendro-cytic basic protein, oligodendrocyte transcription factor 1,glial fibrillary acidic protein, glutamic acid decarboxlase,glycine receptors, and metabotropic glutamate receptorswere markedly enriched in SC in comparison with DRG(Table 2). Second, a growing body of literature has shownhigh concordance rates between data obtained with Affy-metrix oligonucleotide cDNA arrays and QRT-PCR (Mutch etal., 2002; Petersen et al., 2005; Rogojina et al., 2003).Moreover, directional concordance rates are particularlyhigh when fold differences are large as they were for verymany genes in our study (Mutch et al., 2002). The QRT-PCRresults presented herein provide additional support for thisprincipal. Third, the P value distribution and estimation ofthe false positive rates based on Q value calculations forour dataset suggests that the microarray experiment wasstatistically robust. Fourth, quite stringent criteria wereused to generate the culled datasets presented in Tables 1and 2. Finally, our findings are decidedly consistent withprevious studies of gene expression in the DRG and CNS.For instance, Wada and colleagues (1990) used semi-quantitative in situ hybridization to show that neuromedinB transcript is concentrated in the DRG and olfactory bulbwith only very weak expression in SC. As another example,Van der Zwaag et al. (2002) found that the distribution ofplexin D1 transcript exhibits strong topological specificitywith concentrations in the DRG, trigeminal ganglia, granu-lar layer of the cerebellum, and striatum. In our work,neuromedin B and plexin D1 showed DRG/SC fold differ-ences of 35.2 and 11.2, respectively.

3.1. Comparative histology and embryology of theDRG and SC

It should be emphasized that our experiment was not afocused comparison of CNS and PNS neurons. The datapresented hereinmust be interpreted in the context of normal

Table 2 – Genes that are enriched in SC compared to DRG

Symbol Gene name Fold difference P value Q value

Mobp Myelin-associated oligodendrocytic basic protein 220.1 2.57E-03 3.46E-03Ppp1r14a Protein phosphatase 1, regulatory (inhibitor) subunit 14A 197.6 2.17E-03 3.14E-03G3bp Ras-GTPase-activating protein SH3-domain binding protein 187.4 4.88E-03 5.05E-03Slc12a5 Solute carrier family 12, member 5 153.5 1.53E-03 2.59E-03Gfap Glial fibrillary acidic protein 153.1 1.11E-02 8.50E-034933402E03Rik RIKEN cDNA 4933402E03 gene 144.3 2.52E-03 3.42E-036330527O06Rik RIKEN cDNA 6330527O06 gene 137.7 2.14E-03 3.12E-03Itih3 Inter-alpha trypsin inhibitor, heavy chain 3 118.6 1.22E-02 9.09E-03Olig1 Oligodendrocyte transcription factor 1 117.0 2.22E-02 1.37E-029630019K15Rik RIKEN cDNA 9630019K15 gene 105.7 2.09E-02 1.32E-02Grm5 Glutamate receptor, metabotropic 5 104.3 2.41E-03 3.34E-03Adcyap1r1 Adenylate cyclase activating polypeptide 1 receptor 1 102.4 8.81E-03 7.29E-03Slc6a5 Solute carrier family 6 (neurotransmitter transporter, glycine), member 5 91.7 7.31E-03 6.46E-03BC030477 cDNA sequence BC030477 88.8 6.51E-05 5.91E-04Cck Cholecystokinin 88.1 9.10E-03 7.44E-03Slc39a12 RIKEN cDNA C030041M11 gene 84.2 1.44E-02 1.02E-02Edg8 Endothelial differentiation, sphingolipid G-protein-coupled receptor, 8 80.8 9.00E-03 7.38E-03Prss18 Protease, serine, 18 73.5 5.51E-07 1.18E-04Fscn1 Fascin homolog 1, actin bundling protein (Strongylocentrotus) purpuratus) 69.1 7.69E-04 1.80E-03Agt Angiotensinogen 68.0 3.73E-04 1.27E-03Rfx4 Regulatory factor X, 4 (influences HLA class II expression) 66.0 3.95E-03 4.46E-032900019G14Rik RIKEN cDNA 2900019G14 gene 65.3 3.39E-04 1.22E-03A330104H05Rik RIKEN cDNA A330104H05 gene 56.9 1.35E-03 2.42E-03Cxcl14 Chemokine (C-X-C motif) ligand 14 50.5 3.26E-02 1.81E-02Fgf14 Fibroblast growth factor 14 50.4 8.57E-03 7.15E-03Slc1a1 Solute carrier family 1 (neuronal/epithelial high affinity glutamate transporter,

system Xag), member 150.0 3.57E-03 4.21E-03

Copg2as2 Coatomer protein complex, subunit gamma 2, antisense 2 47.9 1.12E-04 7.56E-04Viaat Vesicular inhibitory amino acid transporter 46.3 7.49E-03 6.56E-03Prox1 Prospero-related homeobox 1 43.4 2.33E-02 1.42E-02Gad1 Glutamic acid decarboxylase 1 41.2 1.16E-02 8.76E-03Cacnb2 Calcium channel, voltage-dependent, beta 2 subunit 39.4 1.52E-04 8.51E-04AW490526 Expressed sequence AW490526 39.1 4.87E-02 2.43E-02Glra1 Glycine receptor, alpha 1 subunit 38.5 2.08E-05 3.71E-04Lfng Lunatic fringe gene homolog (Drosophila) 38.0 1.13E-04 7.58E-04Scrg1 Scrapie responsive gene 1 37.7 7.06E-03 6.31E-03Tac2 Tachykinin 2 36.7 1.18E-02 8.88E-036330404F12Rik RIKEN cDNA 6330404F12 gene 36.2 5.34E-04 1.49E-03Gjb1 Gap junction membrane channel protein beta 1 34.8 2.94E-02 1.67E-02Lrrtm4 Leucine rich repeat transmembrane neuronal 4 34.4 5.19E-03 5.25E-03Nkx2_2 NK2 transcription factor related, locus 2 (Drosophila) 34.1 6.16E-03 5.82E-03Slc14a1 Solute carrier family 14 (urea transporter), member 1 33.5 8.32E-03 7.01E-03Sox11 SRY-box containing gene 1 32.8 8.46E-03 7.09E-03Ptprd Protein tyrosine phosphatase, receptor type, D 31.7 2.71E-02 1.58E-02Cacna1g Calcium channel, voltage-dependent, T type, alpha 1G subunit 31.2 2.18E-02 1.36E-021500009C09Rik RIKEN cDNA 1500009C09 gene 30.9 1.12E-02 8.54E-03AI841794 Expressed sequence AI841794 30.5 6.64E-03 6.09E-03Gabt4 Gamma-aminobutyric acid (GABA-A) transporter 4 29.8 2.00E-06 1.79E-04Grm3 Glutamate receptor, metabotropic 3 29.6 3.33E-02 1.83E-02Adssl1 Adenylosuccinate synthetase like 1 28.8 1.25E-05 2.90E-04Kcnd2 Potassium voltage-gated channel, Shal-related family, member 2 28.1 1.64E-03 2.70E-03Cabp7 Calcium binding protein 7 27.3 3.41E-02 1.87E-02Hapln2 Hyaluronan and proteoglycan link protein 2 26.8 2.24E-03 3.20E-03AU067636 Expressed sequence AU067636 26.7 9.31E-03 7.55E-03E330037M01Rik RIKEN cDNA E330037M01 gene 25.9 3.18E-02 1.77E-02Gja12 Gap junction membrane channel protein alpha 12 25.7 9.13E-03 7.45E-03St18 Suppression of tumorigenicity 18 25.0 1.41E-02 1.00E-02Fgfr3 Fibroblast growth factor receptor 3 24.9 2.07E-03 3.06E-03Spag5 Sperm associated antigen 5 24.9 9.77E-04 2.03E-03Gnao1 Guanine nucleotide binding protein, alpha o 24.7 1.56E-03 2.62E-03Slc1a2 Solute carrier family 1 (glial high affinity glutamate transporter), member 2 24.7 3.36E-02 1.85E-02Lrrc4b Leucine rich repeat containing 4B 24.6 2.05E-02 1.30E-02Pou3f3 POU domain, class 3, transcription factor 3 24.5 3.68E-02 1.97E-02Slc8a1 Solute carrier family 8 (sodium/calcium exchanger), member 1 24.4 2.69E-04 1.09E-03Grp Gastrin releasing peptide 23.8 9.90E-04 2.05E-03

32 B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

Table 2 (continued)

Symbol Gene name Fold difference P value Q value

A930009L07Rik RIKEN cDNA A930009L07 gene 22.4 2.77E-03 3.61E-03Dmp1 Dentin matrix protein 1 22.0 1.71E-02 1.14E-02Trhr Thyrotropin releasing hormone receptor 21.9 1.31E-03 2.38E-03Arhgap23 Rho GTPase activating protein 23 21.0 1.46E-04 8.40E-04Slco1c1 Solute carrier organic anion transporter family, member 1c1 20.6 1.87E-02 1.22E-02Aqp4 Aquaporin 4 19.9 5.80E-03 5.61E-03Cldn10 Claudin 10 19.9 3.60E-03 4.23E-03Trf Transferrin 19.0 3.33E-03 4.04E-03Trhr Thyrotropin releasing hormone receptor 18.9 4.39E-02 2.25E-02Uts2 Urotensin 2 18.8 9.84E-04 2.04E-03B3gat1 Beta-1,3-glucuronyltransferase 1 (glucuronosyltransferase P) 18.6 2.51E-03 3.42E-03Stx1b2 Syntaxin 1B2 18.3 1.08E-02 8.35E-03E130304D01 Hypothetical protein E130304D01 17.9 6.29E-06 2.49E-04Slc7a10 Solute carrier family 7 (cationic amino acid transporter, y+ system), member 10 17.1 9.28E-03 7.54E-03AI746471 Expressed sequence AI746471 17.0 7.65E-03 6.64E-03Kcnab3 Potassium voltage-gated channel, shaker-related subfamily, beta member 3 16.8 2.94E-02 1.67E-02Psd2 Pleckstrin and Sec7 domain containing 2 16.5 7.12E-04 1.73E-03Sall1 Sal-like 1 (Drosophila) 16.1 5.73E-03 5.57E-03LOC208158 Similar to hypothetical protein FLJ12748 15.8 3.07E-02 1.73E-02B830045N13Rik RIKEN cDNA B830045N13 gene 15.6 1.03E-02 8.07E-03Hipk2 Homeodomain interacting protein kinase 2 15.5 8.65E-05 6.66E-04Mtap2 Microtubule-associated protein 2 15.5 7.36E-05 6.13E-04Nkx6_2 NK6 transcription factor related, locus 2 (Drosophila) 15.5 2.34E-05 3.95E-04Cx3cl1 Chemokine (C-X3-C motif) ligand 1 14.9 4.76E-05 5.19E-04BC017634 cDNA sequence BC017634 14.9 2.81E-02 1.62E-02Inpp4b Inositol polyphosphate-4-phosphatase, type II 14.8 2.29E-02 1.41E-022900001G08Rik RIKEN cDNA 2900001G08 gene 14.8 5.50E-03 5.44E-03Hpcal4 Hippocalcin-like 4 14.5 3.35E-02 1.84E-02Olfml1 Olfactomedin-like 1 14.3 3.15E-02 1.76E-02Gm98 Gene model 98, (NCBI) 14.1 5.42E-03 5.39E-03Glra2 Glycine receptor, alpha 2 subunit 13.9 1.20E-02 8.99E-03Slc1a3 Solute carrier family 1 (glial high affinity glutamate transporter), member 3 13.9 1.68E-04 9.00E-046430547I21Rik RIKEN cDNA 6430547I21 gene 13.7 9.07E-04 1.96E-03Tacr3 Tachykinin receptor 3 13.6 2.61E-02 1.54E-02Calb2 Calbindin 2 13.3 1.02E-03 2.07E-03Slc18a3 Solute carrier family 18 (vesicular monoamine), member 3 13.0 5.48E-03 5.42E-03AW060763 Expressed sequence AW060763 12.7 1.40E-02 9.99E-03B3gat1 Beta-1,3-glucuronyltransferase 1 (glucuronosyltransferase P) 12.5 1.04E-05 2.84E-04BB086117 Expressed sequence BB086117 11.9 5.89E-03 5.65E-033110035E14Rik RIKEN cDNA 3110035E14 gene 11.6 3.82E-03 4.36E-03BC037006 cDNA sequence BC037006 11.5 4.79E-02 2.40E-02Phyhip Phytanoyl-CoA hydroxylase interacting protein 11.3 3.43E-03 4.11E-03Pygm Muscle glycogen phosphorylase 10.9 1.30E-03 2.37E-03Bbox1 Butyrobetaine (gamma), 2-oxoglutarate dioxygenase 1

(gamma-butyrobetaine hydroxylase)10.7 2.48E-02 1.49E-02

Ptgs1 Prostaglandin-endoperoxide synthase 1 10.5 3.67E-02 1.97E-024933426K21Rik RIKEN cDNA 4933426K21 gene 10.2 6.13E-03 5.80E-03Matk Megakaryocyte-associated tyrosine kinase 10.1 2.06E-02 1.30E-029630044O09Rik RIKEN cDNA 9630044O09 gene 10.0 2.53E-05 4.07E-04

33B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

DRG and SC histology. In particular, the DRG contains unipolarneurons (i.e., ganglion cells) and harbors no synaptic connec-tions. The DRG is surrounded by connective tissue containingsmall blood vessels. Ganglion cells, the major constituent ofthe DRG, are aligned in rows separated by the fasciculi ofmyelinated and unmyelinated nerves. In the DRG, myelinatedfibers are surrounded by Schwann cells. Each ganglion cell issurrounded by a layer of small cuboidal supporting cellsknown as satellite cells. In addition, the complex of eachganglion cell and its satellite cells is enveloped by a thinfibrous layer and its associated fibroblasts. The cellular andcorresponding molecular complexity of ganglion cells has

been well described. For illustration, medium to large but notsmall ganglion cells express neurofilament (Hall et al., 1997)whereas serotonin 5HT2A receptors and neuropeptides arerelatively concentrated in small ganglion cells (Hall et al., 1997;Nicholson et al., 2003).

The embryological differences between the DRG and SCalso bear on the gene expression profiles presented in thisstudy. Neural elements from the central and peripheralnervous systems originate from the neural plate, a thickenedarea of embryonic ectoderm overlying the notochord. Subse-quently, the neural tube gives rise to a neural tube and neuralcrest. The neural tube differentiates into the brain and SC

Table 3 – QRT_PCR analysis of differentially expressed genes in mice DRG and SC

Gene symbol Gene title Forward/reverse primers Microarrayfold

difference

QRT-PCRmedian folddifference

Mean ±standarderror

DRG enriched genesAccn3 Amiloride-sensitive cation channel 3 CAGCCCTGTGGACCTGAGAA/ 193.5 300.8 321.6 ± 65.9

GCCAGCACCGCCCTTAGScn10a Sodium channel, voltage-gated, type X,

alpha polypeptideTCCCTTGTTGTGTGGCAATG/ 170.3 4467.8 6300.0 ± 1746.5CCCACGCAAAGGAATCAAAG

Edg7 Endothelial differentiation,lysophosphatidic acid G-protein-coupledreceptor 7

ACACACCAGTGGCTCCATCA/ 92.5 99.5 107.6 ± 13.1GCCGTCCAGCAGCAGAAC

Htr3b 5-hydroxytryptamine (serotonin)receptor 3B

GACCTTTGCGTCCATGCTGTA/ 53.8 410.1 460.7 ± 128.5CAGGACAAAAACTCATCGTTCCA

Gpr64 G protein-coupled receptor 64 CAGTCCTCCATTCCCTCTCCTA/ 44.1 39.3 32.3 ± 6.1TACTTGGGCTTCCAGATCAGAAA

Ntrk1 Neurotrophic tyrosine kinase, receptor,type 1

TGAGAAGCCTAACCATCGTGAA/ 18.7 524.9 478.1 ± 69.2TTCCAGGAGAGGGACTCCAA

Slc6a19 Solute carrier family 6 (neurotransmittertransporter), member 19

GCTCACCTGTGTGGGCTTTT/ 17.8 19.3 13.2 ± 4.5TCCCTCGAACACCAGAAGGA

Chrna6 Cholinergic receptor, nicotinic, alphapolypeptide 6

CACTCGTCCGATGTTGAAGATG/ 11.6 89.6 123.7 ± 30.3TGTCCACCACCATAGCCATGT

Syt9 Synaptotagmin 9 CCGAGAGCATCGACCAGATC/ 11.5 8.2 10.2 ± 2.1CGCTCAGCCTCGTTTCCTACT

Th Tyrosine hydroxylase TGCTGTTCTCAACCTGCTCTTC/ 11.0 22.6 26.1 ± 5.5AGTGGTGGATTTTGGCTTCAA

SC enriched genesSlc6a5 Solute carrier family 6 (neurotransmitter

transporter, glycine), member 5GGCTGCCTCTCTCTCCATTCT/ 91.7 376.0 528.4 ± 215.7TTGGGAAACTCATCCGAGATG

Slc1a1 Solute carrier family 1 (neuronal/epithelial high affinity glutamatetransporter, system Xag), member 1

TGACATCAACAGGACGGGTAAA/ 50.0 27.5 26.2 ± 2.0TACTGCTGAAAACAGGCTTGGA

Gad1 Glutamic acid decarboxylase 1 TGGTGAATGGCTGACATCGA/ 41.2 93.1 102.3 ± 31.1ACCATCCAACGATCTCTCTCATC

Cacnb2 Calcium channel, voltage-dependent,beta 2 subunit

AGAGGACCGGGAGGCAGTAC/ 39.4 11.5 13.6 ± 3.3CTGCGCTGTATCTGACATTGG

Lrrtm4 Leucine rich repeat transmembraneneuronal 4

TTTATCTGGGAATGGCAAAAGC/ 34.4 14.4 12.7 ± 2.1CAGCAGCAGGGTGGGAAATA

Cacna1g Calcium channel, voltage-dependent,T type, alpha 1G subunit

CGATGCTTCCTCCCTGAGAA/ 31.2 40.8 39.2 ± 9.7CCCGAGGCTGAGAGCAGAT

Gabt4 Gamma-aminobutyric acid (GABA-A)transporter 4

GGAGTTCGTGTTGAGCGTAGCT/ 29.8 209.1 210.2 ± 40.5TGAAAAACACCACGTAAGGAATCA

Grm3 Glutamate receptor, metabotropic 3 CAACACCACCAAGCTCTGTGA/ 29.6 30.6 28.3 ± 3.2GCTCCTTTATTTGGGTTGAATGG

Calb2 Calbindin 2 GCATCCCAGTTCCTGGAAATC/ 13.3 28.9 33.2 ± 6.0TCATGCCAGAACCCTTCCTT

Slc18a3 Solute carrier family 18 (vesicularmonoamine), member 3

GCGTTGCACTGTTACTGGACAA/ 13.0 392.9 557.8 ± 188.4GGTTCCCACACCTCAGAGATCA

34 B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

whereas the PNS arises from the neural crest. More specifi-cally, neural crest cells differentiate into cranial, dorsal root,and autonomic ganglion cells, satellite cells, and Schwanncells. The adhesion molecule CD44, which we found to behighly enriched in DRG (Table 1 and Fig. 6), provides a potentexample of embryologically-specific gene expression (Ikeda etal., 1996). Most migrating neural crest cells strongly expressCD44 whereas expression is absent in the neural tube. Inparticular, CD44 transcripts are localized to neural crest, DRGganglion, and Schwann cells.

Within the SC, all neurons, astrocytes, oligodendrocytes,and ependymal cells are derived from the neural plate. Incontrast,microglia, which are found throughout thewhite andgray matter of the SC, originate from mesodermal mesenchy-

mal cells. Similarly, the connective tissues of the DRG andblood vessels which permeate the SC are derived frommesodermal mesenchymal cells.

3.2. Comparison with other DRG gene expression arraystudies

The term “peripheral neuropathy” encompasses a multitudeof acquired and hereditary disorders of the PNS. Althoughfrequently caused by metabolic disorders such diabetesmellitus, acquired neuropathies can arise from trauma (e.g.,laceration), infection (e.g., herpes zoster), chemotherapeuticagents (e.g., cisplatinum), poisonings (e.g., arsenic), and a hostof autoimmune, neoplastic, and inflammatory conditions

Fig. 6 – GOTM gene ontology of biological processes for genes that were highly expressed in either the DRG (A) or SC (B). Thehierarchical organization of the categories is shown by arrows from the most proximal (orange) to distal (green) branches.Genes within each category are shown on the right.

35B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

(Bromberg, 2005; Pratt andWeimer, 2005). Rather than placinga narrow focus on etiological validity, animal models ofneuropathy and associated gene expression studies havebroadly examined the molecular biology of regeneration andperipheral neuropathic pain (Wang and Wang, 2003). Inparticular, the scientific community has exhibited a growinginterest in mechanisms of peripheral neural regeneration andsurvival in hopes of 1 day being able to significantly stimulatethe growth of previously damaged nerves (Sahenk et al., 2005).

Several relatively recent studies have employed cDNAoligonucleotide microarray technology for gene expressionanalysis in the DRG after a variety of surgical and pharmaco-logical perturbations (Linnarsson et al., 2001; Xiao et al., 2002;Costigan et al., 2002; Tanabe et al., 2003; Valder et al., 2003).Despite inadequate coverage of the genome under consider-ation, either rat or mouse, in previous work, microarrays haveproven to be powerful tools for identifying relevant molecularpathways in the PNS. For example, Linnarsson et al. (2001)explored the differential effects of glial cell line-derivedneurotrophic factor (GDNF) on embryonic and adult mouseDRG using Affymetrix Mu19K and Mu11K GeneChips andfound important interactions between treatment and devel-opmental stage for several hundred genes, particularly thoserelated to translation. Then, in confirmatory experiments, itwas shown that GDNF activates a transcriptional programrepressing neurite outgrowth.

Other cDNA microarray analyses of the DRG have utilizedthe sciatic nerve transection model system to define geneexpression profiles related to regeneration in the PNS. Xiaoand colleagues (2002) used custommicroarrays to quantify ratlumbar DRG transcripts at multiple time points after sciaticnerve transection. In their study, 2-fold changes from controlrats were considered significant. Among the less than 200genes showing significant fold-changes were many encodingneuropeptides, ion channels, synaptic vesicle proteins, andcell-surface receptors. Correspondingly, we found that tran-scripts for several of these same genes, including calcitoningene related peptide (DRG/SC = 15.85), galanin (DRG/SC = 4.50),and synaptotagmin (DRG/SC = 2.31), were enriched in DRGrelative to SC. Two additional studies of lumbar DRG geneexpression after sciatic nerve transection employed Affyme-trix oligonucleotide arrays (Costigan et al., 2002; Tanabe et al.,2003). In rats, 3 days after sciatic nerve transection, a 2-fold cutoff identified 178 upregulated genes. Many of these upregu-lated genes such as galanin and neuropeptide Ywere commonto those recognized in the report by Xiao et al. (2002).Furthermore, several genes shown to be upregulated in DRGby Costigan et al. (2002) were also enriched in DRG relative toSC. Examples include heat shock protein 27 kDa (DRG/SC = 5.82), syndecan (DRG/SC = 14.62), and small proline-richprotein 1A (DRG/SC = 3.97). A study in mice using MouseGenome U74 A, B, and C arrays (Affymetrix) examined DRG

Table 4 – Expression differences between DRG and SC for genes linked to hereditary neuropathies

Human disorder Mousegene ⁎

Gene name Folddifference

P value Q value

Hereditary sensory and autonomic neuropathiesHSN-I Sptlc1 Serine palmitoyltransferase, long chain base

subunit 11.4 1.74E-01 2.55E-02

HSN-IIHSN-II Ikbkap Inhibitor of kappa light polypeptide enhancer in

B-cells, kinase complex-associated protein1.9 3.14E-03 3.88E-03

HSN-IV Ntrk1 Neurotrophic tyrosine kinase, receptor, type 1 18.7 8.52E-04 1.90E-03HSN-IV Ngfr Nerve growth factor receptor (TNFR superfamily,

member 16)108.0 1.07E-02 8.29E-03

HSN-Absent Pain Ngfb Nerve growth factor, beta NA NA NA

Hereditary painful neuropathiesα-Galactosidase Deficiency Gla Galactosidase, alpha 2.0 1.94E-02 1.25E-02Erythromelalgia Scn9a Sodium channel, voltage-gated, type IX, alpha

polypeptide88.4 1.38E-03 2.44E-03

Reflex Sympathetic Dystrophy H2-Aa Histocompatibility 2, class II antigen A, alpha 8.7 2.87E-06 2.03E-04

Hereditary motor sensory neuropathiesCharcot-Marie-Tooth (CMT) 1A Pmp22 Peripheral myelin protein 22 3.4 1.63E-04 8.87E-04CMT 1B Mpz Myelin protein zero 2.3 4.77E-03 5.00E-03CMT 1C Litaf LPS-induced TN factor −2.7 1.46E-02 1.03E-02CMT 1D Egr2 Early growth response 2 2.9 1.01E-02 7.96E-03CMT 1E Mpz Myelin protein zero 2.3 4.77E-03 5.00E-03CMT 1F Nefl Neurofilament, light polypeptide 1.5 1.29E-02 9.45E-03CMT 2A Kif1b Kinesin family member 1B −1.9 8.02E-03 6.86E-03CMT 2A Mfn2 Mitofusin 2 1.8 4.15E-05 5.00E-04CMT 2B Rab7 RAB7, member RAS oncogene family 1.7 8.74E-05 6.66E-04CMT 2D Gars Glycyl-tRNA synthetase 2.3 1.42E-04 8.36E-04CMT 2I Mpz Myelin protein zero 2.3 4.77E-03 5.00E-03CMT 2J Mpz Myelin protein zero 2.3 4.77E-03 5.00E-03CMT 2L Hspb8 Heat shock 27 kDa protein 8 5.8 6.95E-04 1.70E-03AR-CMT2A Lmna Lamin A 5.0 1.09E-03 2.16E-03AR-CMT + hoarseness Gdap1 Ganglioside-induced differentiation-associated-

protein 11.8 1.68E-04 9.51E-03

CMT 4A Gdap1 Ganglioside-induced differentiation-associated-protein 1

1.8 1.68E-04 9.51E-03

CMT 4B Mtmr2 Myotubularin related protein 2 1.2 1.18E-01 4.82E-02CMT 4B2 Mtmr13 Myotubularin related protein 13 1.2 6.64E-04 1.01E-01CMT 4C Sh3tc2 SH3 domain and tetratricopeptide repeats 2 NA NA NACMT 4D (Lom) Ndrg1 N-myc downstream regulated gene 1 1.3 2.45E-02 1.48E-02CMT 4E Egr2 Early growth response 2 2.9 1.01E-02 7.96E-03CMT 4F Prx Periaxin 2.5 6.29E-03 5.89E-03CMT DIB Dnm2 Dynamin 2 1.0 8.34E-01 1.11E-02CMT DI3 Mpz Myelin protein zero 2.3 4.77E-03 5.00E-03HNPP Pmp22 Peripheral myelin protein 22 3.4 1.63E-04 8.87E-04HMSN 3 Pmp22 Peripheral myelin protein 22 3.4 1.63E-04 8.87E-04PNS and CNS hypomyelination Sox10 SRY-box containing gene 10 −2.3 7.89E-04 1.82E-03Sensory PN + hearing loss Gjb3 NA NA NACongenital hypomyelinating neuropathy Arhgef10 Rho guanine nucleotide exchange factor (GEF) 10 −2.1 1.00E-03 2.38E-03Congenital cataracts, facial dysmorphism,and neuropathy syndrome

Ctdp1 CTD phosphatase, subunit 1 1.2 3.49E-02 1.90E-02

Farber's lipogranulomatosis Gba Glucosidase, beta, acid 1.7 1.20E-02 8.99E-03Glycosylation deficient Pmm2 Phosphomannomutase 2 2.0 1.41E-02 1.00E-02Krabbe's disease Galc Galactosylceramidase 1.1 1.52E-01 5.87E-02Metachromatic leukodystrophy Arsa Arylsulfatase A 1.6 5.43E-03 5.39E-03

Refsum's diseaseChildhood Phyh Phytanoyl-CoA hydroxylase −1.2 1.01E-02 7.96E-03Adolescent-adult Pex7 Peroxisome biogenesis factor 7 1.3 6.18E-02 2.92E-02Infant Pex1 Peroxisome biogenesis factor 1 −1.1 5.67E-01 1.69E-01Andermann syndrome Slc12a6 Solute carrier family 12, member 6 −1.1 7.77E-01 1.99E-01

Hereditary amyotrophic lateral sclerosis (ALS)ALS1 Sod1 Superoxide dismutase 1, soluble 1.9 1.16E-02 8.76E-03ALS4 Als4 Amyotrophic lateral sclerosis 4 homolog (human) 1.4 2.39E-02 1.45E-02ALS8 Vapb Vesicle-associated membrane protein, associated

protein B and C1.5 1.54E-02 1.06E-02

36 B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

Table 4 (continued)

Human disorder Mousegene ⁎

Gene name Folddifference

P value Q value

Hereditary amyotrophic lateral sclerosis (ALS)ALS2 Als2 Amyotrophic lateral sclerosis 2 (juvenile) homolog

(human)1.3 1.08E-01 4.49E-02

Bulbar syndromesAchalasia-addisonianism-alacrimiasyndrome

Aaas Achalasia, adrenocortical insufficiency, alacrimia NA NA NA

Kennedy's syndrome Ar Androgen receptor NA NA NAPrimary lateral sclerosis, juvenile Als2 Amyotrophic lateral sclerosis 2 (juvenile) homolog

(human)1.3 1.08E-01 4.49E-02

Multisystem disordersDisinhibition-dementia-Parkinsonism-amyotrophy complex

Mapt Microtubule-associated protein tau 1.9 8.83E-03 7.29E-03

Hexosaminidase A Hexa Hexosaminidase A NA NA NAMachado-Joseph disease Mjd Machado-Joseph disease (spinocerebellar ataxia 3,

olivopontocerebellar ataxia 3, autosomal dominant,ataxin 3) homolog (human)

−1.1 5.43E-01 2.54E-02

Polyglucosan body disease Gbe1 Glucan (1,4-alpha-), branching enzyme 1 2.9 7.24E-03 6.42E-03

Spinal muscular atrophy (SMA)SMA 5q Smn1 Survival motor neuron 1 1.2 2.62E-01 9.05E-02

Distal hereditary motor neuropathy (HMN)HMN2 Hspb8 Heat shock 27 kDa protein 8 5.8 6.95E-04 1.70E-03HMN5 Gars Glycyl-tRNA synthetase 2.3 1.42E-04 8.36E-04HMN5B Bscl2 Bernardinelli-Seip congenital lipodystrophy 2

homolog (human)2.4 4.01E-03 4.50E-03

HMN + vocal cord involvement Dctn1 Dynactin 1 2.0 9.09E-04 1.96E-03HMN + upper motor neuron Als4 Amyotrophic lateral sclerosis 4 homolog (human) 1.4 2.39E-02 1.45E-02Diaphragm + neonatal Ighmbp2 Immunoglobulin mu binding protein 2 1.3 8.10E-02 3.60E-02Distal HMN Hspb1 Heat shock protein 1 6.9 1.58E-04 8.69E-04

NA, data not available.* Gene names in bold are linked to more than one disease.

37B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

transcripts 7 days after sciatic nerve transection. In addition tosubstantial overlap with the earlier work of Costigan et al.(2002) and Xiao et al. (2002), Tanabe and co-workers (2003)found that fibroblast growth factor 14 (Fgf14) was upregulated9.6-fold. In contrast, Fgf14 was enriched in mouse SC relativeto DRG over 50-fold (Table 2). In the periphery, FGF14participates in neural regeneration via a Rac1 GTPase-depen-dent mechanism (Tanabe et al., 2003). Our findings along withrecent mechanistic studies by others suggest that FGF14 mayplay additional roles in the CNS (Lou et al., 2005).

The molecular pathophysiology of peripheral neuropathicpain has been investigated with Affymetrix rat U34A Gene-Chips (Wang et al., 2002; Valder et al., 2003). Both Wang et al.(2002) and Valder et al. (2003) harvested DRGs 2 weeks after L5and L6 spinal nerve ligation (SNL). The SNLmodel causes pain,allodynia, and hyperalgesia. Results from the two groups werevery similar. Basically, immediate early genes, and genesencoding ion channels, neuropeptides, and proteins associat-ed with immune activation and plasticity were regulated inthe SNL model. Many of these same genes (e.g., galanin,neuropeptide Y, leucine zipper protein, muscle LIM protein)were also upregulated in the sciatic nerve transection modelsdescribed above. Importantly, follow-up in situ hybridization

and immunocytochemical experiments showed that SNLstimulated changes in both neuronal and non-neuronal celltypes. One striking examplewas GFAPwhich exhibited limitedbasal expression but marked upregulated in ipsilateral satel-lite cells after SNL. In our own work, Gfap was stronglyenriched in SC relative to DRG (SC/DRG = 153) which suggeststhat basal expression is required in the CNS (Johnson, 2004).

3.3. Biological associations

Thecentral roleof thePNS is the transductionofphysical stimuliinto receptor potentials and transmission of the resultant spiketrain pattern to the CNS. Moreover, transduction ofmechanical,thermal, and chemical stimuli is a function largelyunique to thePNS. The PNS utilizes a large collection of receptors in thesestimulus transduction processeswhich produce local potentialsrestricted to the terminal membranes of nerves in skin, joints,and muscle. The multitude of receptor types (e.g., chemicalnociceptors, mechanical nociceptors, muscle spindles, Golgitendon organs, Merkel's receptors, etc.) and transductionmechanisms (e.g., mechanical, receptor–ligand, thermal, etc.)strongly suggests that a substantial number of unique genefamilies are expressed in DRG ganglion cells.

38 B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

Two members (Trpa1 and Trpc6) of the transient receptorpotential (TRP) superfamily of non-selective cation channels,which are known to play important roles in both the PNS andCNS, were highly enriched (15.9- and 20.2-fold, respectively) inDRG (Moran et al., 2004; Nilius et al., 2005). The six TRP familiesare classical (TRPC), melastatin (TRPM), vanilloid receptor(TRPV), mucolipin (TRPMLs), polycystins (TRPPs) TRPs, andankyrin transmembrane protein 1 (TRPA1). Perhaps the moststudied TRP, TRPV1, is a temperature-sensing channel whichbinds capsaicin. However, TRPV1 was not included in ouranalysis because the probe set signal was not reliable. TRPA1is activated by mustard oil and, as such, also appears tofunction as a nociceptor (Jordt et al., 2004). TRPC6 has beenimmunohistochemically localized to vagal GSA fibers in theheart (Calupca et al., 2002). In contrast, cardiac GVE fibers donot express TRPC6. In humans, TRPC6 is broadly distributed innon-neural tissues and mutations of TRPC6 are associatedwith familial focal segmental glomerulosclerosis (Winn et al.,2005). Other TRPs showed lower levels of DRG enrichmentwhereas no TRPs were enriched in SC: Trpc3 (DRG/SC = 9.86),Trpv2 (DRG/SC = 5.55), and Trpm4 (DRG/SC = 2.22). Of note,TRPV2 is activated by noxious heat (Caterina et al., 1999).

Although GOTM functional assignments are not preciseand often place genes intomultiple subcategories, the clustersof DRG- and SC-“specific” genes shown in Fig. 6 are quiteconsistent with the distinctive biologies of the DRG and SC. Inparticular, a large number of G-protein-coupled receptors(GPCRs) showed preponderant expression in the PNS. Sinceactivation of GPCRs is associated with effects that are bothslower and of longer duration than those produced byionotropic receptors, is it unlikely that GPCRs are directlyinvolved in signal transduction by those large-diametermyelinated sensory fibers responsible for proprioception.Instead, many of these GPCRs may be involved in chemicalnociception or trophic survival of axons. For illustration, themas-related gene (Mrg) family includes a number of G-protein-coupled receptors that are specific or relativelyspecific to sensory neurons of the PNS (Gustafson et al.,2005). Several of these Mrgs are activated by peptides. In ourstudy, four Mrgs (Mrgpra1 DRG/SC = 10.33, Mrgpra3 DRG/SC = 6.54, Mrgpre DRG/SC = 3.60, and Mrgpra1 DRG/SC = 2.61)were significantly enriched in DRG whereas none wereenriched in SC. On the other hand, substantial collections ofgenes encoding metabotropic and ionotropic glutamate,glycine, GABA, and dopamine receptors were significantlyenriched in SC (see Fig. 6 and Excel Workbook Online).

3.4. Clinical correlates

The pathological and clinical manifestations of hereditaryneurological diseases show reasonable correlations with thetopology of gene expression. For case in point, abnormalities ofCNS myelin are unusual in Charcot–Marie–Tooth (CMT) 1A(Pmp22 DRG/SC = 3.4) but are typically seen in patients withSOX10 (Sox10 DRG/SC = −2.3) mutations. Other associationsderived from review of Table 4 must be interpreted in thecontext of our experimental paradigm in which the SC in totowas compared to DRG. For illustration, three genes that causehereditary motor neuropathies were enriched in the DRGrelative to SC by 2.4- to 6.9-fold (i.e., Bscl2, Dctn1, and Hspb1).

However, this does not allow for the more focused conclusionthat these genes were enriched in ganglion cells relative toventral horn motor neurons. For case in point, Plumier et al.(1997) have shown that the protein encoded by Hspb1, 27-kDaheat shock protein (Hsp27), is robustly expressed in motorneurons of the ventral horn but sparsely expressed in other SCcell populations. This would account for the observed enrich-ment of Hspb1 expression in the DRG relative to SC, since thelevel of Hspb1 transcript in the motor neurons is diluted byother SC transcripts. Thus, the dataset generated in our studywith regard to quantitative differences in gene expressionbetween the CNS and PNS will requiremore focused examina-tions with in situ hybridization, immunocytochemistry, andQRT-PCR of cells acquired via laser capture microdissection.

In summary, the SC and DRG transcriptomes presentedherein provide a powerful dataset for studies of normal andpathological states of the CNS and PNS, respectively. Morespecifically, identification of transcripts enriched in DRG willsupport the development of target-derived pharmacologicaltherapeutics for regeneration, remyelination, and treatmentof neuropathic pain. Candidate genes within disease lociderived from linkage analysis or genome-wide associationstudies can be readily identified by searching the ExcelWorkbook provided online. Finally, mechanistic studies ofgene families enriched in DRG may improve our understand-ing of signal transduction in the PNS.

4. Experimental procedures

4.1. Animals and tissue acquisition

All experiments were performed in accordance with theNational Institutes of Health's Guide for the Care and Use ofLaboratory Animals and the guidelines of the InstitutionalAnimal Care and Use Committee. SC and DRG were harvestedfrom a total of 12 male, 3 month-old C57BL/6J mice (JacksonLaboratories, Bar Harbor, ME). Mice were rapidly and deeplyanesthetized with 100 mg/kg of intraperitoneal pentobarbitalprior to transcardiac perfusion-fixation. After opening thethoracic cavity, the ascending aorta was cannulated with a 25-gauge blunt tip needle. The descending aorta was not cross-clamped. Perfusion–fixation was performed manually usingtwo sterile 30-cc syringes connected to a 3-way stopcock. First,the vascular tree was flushed with approximately 10 cc ofdiethyl pyrocarbonate (DEPC)-treated normal saline. The DEPCtreatment was employed to limit the possibility of exogenousRNAases from entering vascular beds. Next, tissues were“fixed” and RNA was stabilized by perfusion with approxi-mately 25 cc of RNAlater™ (Ambion, Austin, TX). Perfusion–fixation was performed with manual pressure and gauged byefflux of fluid from the right atrium. Progressively increasingmanual force was required during the process of injectingRNAlater™ into the vascular tree.

Mice have 13 thoracic and 6 lumbar vertebrae and spinalcord segments. The spinal cord terminates at the level of the4th lumbar (L4) vertebral body. The 1st lumbar spinal segmentbegins at the bottom of the 11th thoracic (T11) vertebral bodyand the 6th lumbar spinal segment ends in the middle of the1st lumbar (L1) vertebral body. After completion of perfusion–

39B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

fixation, the thoracic, lumbar, and upper sacral portions ofthe vertebral column were sharply dissected from theremainder of the carcass. The vertebral column was thenplaced into a 15-cc conical tube containing RNAlater™. The SCand DRG were exposed for extraction using a surgicalmicroscope and microsurgical instrumentation. RNAlater™was dribbled onto the tissue specimen by the surgicalassistant during the entire microsurgical procedure. Dorsaland ventral roots were sectioned at their entry and exit zonesrespectively. Both left and right DRGs from L1 to L6 weresharply dissected away from their dorsal roots. SC segmentsL1 to L6 and the 12 DRG were placed into separate micro-centrifuge tubes containing Tri-Reagent™ (Ambion) main-tained at 4 °C. In pilot experiments, acquisition of high-quality RNA from mouse DRG required strict adherence to theprocessing steps described above.

4.2. RNA extraction and processing

The SC and DRG were homogenized in Tri-Reagent™ using arotor/stator type tissue homogenizer (Tissue-Tearor™; Bios-pec Products, Bartlesville, OK, USA). Separate homogenizerswere used for SC and DRG. Total RNA was extracted using theTri-Reagent™ protocol. Total RNA from 4 mice was pooled foreach SC and DRG sample and 3 independent samples wereused for the expression array analysis. In this fashion, a totalof 6 Affymetrix Mouse Genome 430 2.0 GeneChip® expressionarrays were used to compare gene expression between thecentral and peripheral nervous systems of adultmale C57BL/6Jmice.

RNA purity and concentration were confirmed fromOD260/280 readings using a dual beam UV spectrophotome-ter. RNA integrity was determined by capillary electropho-resis using the RNA 6000 Nano Lab-on-a-Chip kit and theAgilent Bioanalyzer 2100 prior to synthesis of 1st and 2ndstrand cDNA (Fig. 1).

Synthesis and labeling of cRNA probes, hybridization toGeneChip® expression arrays, and acquisition of fluores-cence intensities was performed by Genome Explorations(Memphis, TN). Biotin labeled cRNA was synthesized usingdouble-stranded cDNA. First and second strand cDNA weresynthesized from 15 μg of total RNA using the SuperScriptDouble-Stranded cDNA Synthesis Kit (Invitrogen, Carlsbad,CA) and oligo-dT24-T7 (5′-GGC CAG TGA ATT GTA ATA CGACTC ACT ATA GGG AGG CGG-3′) primer (PrOligo, Boulder, CO)in accordance with the manufacturer's instructions. ThecRNA was synthesized and labeled with biotinylated UTP andCTP by in vitro transcription using the T7 promoter-coupleddouble stranded cDNA as template and the Bioarray™HighYield™ RNA Transcript Labeling Kit (ENZO DiagnosticsInc., Farmingdale, NY). Double stranded cDNA synthesizedfrom the previous steps were washed twice with 70% ethanoland resuspended in 22 μl RNase-free H2O. The cDNA wasincubated with 4 μl of 10× each Reaction Buffer, Biotin-labeled Ribonucleotides, DTT, RNase Inhibitor Mix and 2 μl20× T7 RNA Polymerase for 5 h at 37 °C. The labeled cRNAwas separated from unincorporated ribonucleotides bypassing through a CHROMA SPIN-100 column (Clontech,Mountain View, CA) and ethanol precipitated at −20 °C for1 h to overnight.

The cRNA pellet was resuspended in 10 μl RNase-free H2Oand 10.0 μg was fragmented by ion-mediated hydrolysis at95 °C for 35 min in 200 mM Tris–acetate (pH 8.1), 500 mMpotassium acetate, and 150 mM magnesium acetate. Thefragmented cRNA was hybridized for 16 h at 45 °C toAffymetrix Mouse 430 2.0 GeneChip® expression arrayswhich contain 45,000 probe sets which analyze the expressionlevels of over 39,000 transcripts from over 34,000 well-characterized mouse genes.

Arrays were washed at 25 °C with 6× SSPE (0.9 M NaCl,60 mMNaH2PO4, 6 mM EDTA+ 0.01% Tween 20) followed by astringent wash at 50 °C with 100 mM MES, 0.1 M [Na+], 0.01%Tween 20. Then, the arrays were stained with phycoerythrein-conjugated streptavidin (Molecular Probes, Eugene, OR) and thefluorescence intensities were determined using the GCS 3000high-resolution confocal laser scanner (Affymetrix). Thescanned images were analyzed using programs resident inGeneChip® Operating System v1.2 (GCOS; Affymetrix). Theexpression data were analyzed as described in a previous report(Lockhart et al., 1996). The signal intensity for each gene wascalculated as the average intensity difference, represented by[Σ(PM − MM)/(number of probe pairs)], where PM denotesperfect-match probes and MM denotes mismatch probes. Foreach probe set, aWilcoxon's rank test is calculated to determinea P value for the difference between the PM value (signal) andMM value (noise). A probe-set is called present (P) if the P valueis less than 0.04, marginal (M) if the P value is between 0.04 and0.06, and absent (A) if the P value is greater than 0.06.

4.3. Relative quantitative real-time reverse transcriptasePCR (QRT-PCR)

QRT-PCR with SYBR Green (Applied Biosystems [ABI], FosterCity, CA, USA) was used to reexamine selected genes showingat least a 10-fold microarray expression difference betweenDRG and SC. Total RNA was extracted with Tri-Reagent™ andgenomic DNAwas removedwith DNA-free™ (Ambion) prior toreverse transcription of RNA (RETROscript™, Ambion). Pri-mers were designed using Primer Express® (ABI) to generateamplicons specific for each gene under study. Agarose gelelectrophoresis (3%) was used to establish that each primerpair generated single amplicons of the correct size. To confirmamplicon identity, bands were cut from gels, purified with theQIAquick Gel Extraction Kit from Qiagen (Valencia, CA), andsequenced with an ABI PRISM® 3100 Genetic Analyzer. QRT-PCR was performed using the ABI 7900 Real Time PCR Systemwith the SYBR Green Master Mix Kit (ABI) based on ABIprotocol. Samples were normalized to 18S rRNA expressionlevels. Differential expression was determined using thecomparative CT method. For each gene, transcript levelswere compared between 4 individual SC and DRG samplesfrom 4 mice which were different from those used for themicroarray experiment.

4.4. Bioinformatics and statistics

The expression values from Affymetrix were pre-processedand analyzed using GeneSpring 7.3 (Agilent Technologies).Probes sets were excluded from further analysis unless theywere called P or M on at least 3 arrays and also produced

40 B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

fluorescent signal intensities above 50. After application ofthis initial filter, a total of 22,946 probe sets remained. In orderto correct for variations in sample loading and staining,expression values for each probe set were normalized to themedian expression value for all probe sets in each chip as wellas themedian expression value for each gene across the set of6 arrays. To evaluate our normalization procedure, MA plotswere generated (Fig. 2), where M = log2 [DRG/SC] and A = 1/2log2 [DRG × SC]. In these two equations, “DRG” and “SC”represent the mean intensity values for each probe set in theDRG and SC samples, respectively.

Scatter plots andcoefficients of linear determination (R2)weregenerated to evaluate the combined technical and biologicalreproducibility of expression values among arrays (Fig. 3).Differential gene expression was analyzed with a combinationof four parameters: magnitude of differential expression, Qvalues, two-sided t-tests with P values, and detection calls (P, M,or A). The Q value is the positive false discovery rate (pFDR)analog of the P value where pFDR is defined as the expectedproportion of false positives among all rejections of the nullhypothesis (Storey and Tibshirani, 2003). Q values were deter-mined using the QVALUE software (http://faculty.washington.edu/~jstorey/qvalue/) implemented by the free statistical soft-ware program R (http://www.r-project.org/) (Fig. 4). To provide avisual summary of differential expression relative to P values, avolcano plot was generated using GeneSpring 7.3 (Fig. 5).

Functional classification of genes enriched in DRG and SCwere evaluated using Gene Ontology Tree Machine (GOTM,http://genereg.ornl.gov/gotm/; Zhang et al., 2004) based ongreater than 19,000 gene function categories assigned by theGene Ontology Consortium (Ashburner et al., 2000). GOTM is aweb-based tool for the analysis and visualization of gene setsfrom high-throughput platforms like gene expression arrays.Using Fisher's Exact Test, GOTM calculates P values based onthe number of genes in each functional category for a givendataset compared with the number of genes expected byrandom distribution. GOTM categories containing at least twogenes with a P value <0.01 are shown in Fig. 6.

The last step examined the clinical relevance of themurinegene expression array data. Several online resources includ-ing the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/) and Washington University Neuro-muscular Disease Center (http://www.neuro.wustl.edu/neu-romuscular) were used to explore the relationships betweenmurine gene expression in the DRG and orthologous humangenes associated with hereditary human neuropathies.

Acknowledgments

This research effort was supported by the Winston WolfePeripheral Neuropathy Research Laboratory and the Neurop-athy Research Foundation (to DLM) and National Institutes ofHealth (R01 NS048458 and R03 NS050185 to MSL).

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.brainres.2006.05.101.

R E F E R E N C E S

Ashburner, M., Ball, C.A., Blake, J.A., Botstein, D., Butler, H.,Cherry, J.M., Davis, A.P., Dolinski, K., Dwight, S.S., Eppig, J.T.,Harris, M.A., Hill, D.P., Issel-Tarver, L., Kasarskis, A., Lewis, S.,Matese, J.C., Richardson, J.E., Ringwald, M., Rubin, G.M.,Sherlock, G., 2000. Gene ontology: tool for the unification ofbiology. The Gene Ontology Consortium. Nat. Genet. 25, 25–29.

Bromberg, M.B., 2005. An approach to the evaluation of peripheralneuropathies. Semin. Neurol. 25, 153–159.

Calupca, M.A., Locknar, S.A., Parsons, R.L., 2002. TRPC6immunoreactivity is colocalized with neuronal nitric oxidesynthase in extrinsic fibers innervating guinea pig intrinsiccardiac ganglia. J. Comp. Neurol. 450, 283–291.

Caterina, M.D., Rosen, T.A., Tominaga, M., Brake, A.J., Julius, D.,1999. A capsaicin-receptor homologue with a high thresholdfor noxious heat. Nature 398, 436–441.

Costigan, M., Befort, K., Karchewski, L., Griffin, R.S., D'Urso, D.,Allchorne, A., Sitarski, J., Mannion, J.W., Pratt, R.E., Woolf, C.J.,2002. Replicate high-density rat genome oligonucleotidemicroarrays reveal hundreds of regulated genes in the dorsalroot ganglion after peripheral nerve injury. BMCNeurosci. 3, 16.

Gustafson, E.L., Maguire, M., Campanella, M., Tarozzo, G., Jia, Y.,Dong, X.W., Laverty, M., Murgolo, N., Priestley, T., Reggiani, A.,Monsma, F., Beltramo, M., 2005. Regulation of two ratmas-related genes in a model of neuropathic pain. Brain Res.Mol. Brain Res. 142, 58–64.

Hall, A.K., Ai, X., Hickman, G.E., MacPhedran, S.E., Nduaguba, C.O.,Robertson, C.P., 1997. The generation of neuronal heterogeneityin a rat sensory ganglion. J. Neurosci. 17, 2775–2784.

Ikeda, K., Nakao, J., Asou, H., Toya, S., Shinoda, J., Uyemura, K.,1996. Expression of CD44H in the cells of neural crest origin inperipheral nervous system. NeuroReport 7, 1713–1716.

Johnson, A.B., 2004. Alexander disease: a leukodystrophy causedby a mutation in GFAP. Neurochem. Res. 29, 961–964.

Jordt, S.E., Bautista, D.M., Chuang, H.H., McKemy, D.D.,Zygmunt, P.M., Hogestatt, E.D., Meng, I.D., Julius, D., 2004.Mustard oils and cannabinoids excite sensory nerve fibresthrough the TRP channel ANKTM1. Nature 427, 260–265.

Linnarsson, S., Mikaels, A., Baudet, C., Ernfors, P., 2001. Activationby GDNF of a transcriptional program repressing neuritegrowth in dorsal root ganglia. Proc. Natl. Acad. Sci. U. S. A. 98,14681–14686.

Lockhart, D.J., Dong, H., Byrne, M.C., Follettie, M.T., Gallo, M.V.,Chee, M.S., Mittmann, M., Wang, C., Kobayashi, M., Horton, H.,Brown, E.L., 1996. Expression monitoring by hybridization tohigh-density oligonucleotide arrays. Nat. Biotechnol. 14,1675–1680.

Lou, J.Y., Laezza, F., Gerber, B.R., Xiao, M., Yamada, K.A.,Hartmann, H., Craig, A.M., Nerbonne, J.M., Ornitz, D.M., 2005.Fibroblast growth factor 14 is an intracellular modulator ofvoltage-gated sodium channels. J. Physiol. 569, 179–193.

Moran, M.M., Xu, H., Clapham, D.E., 2004. TRP ion channels in thenervous system. Curr. Opin. Neurobiol. 14, 362–369.

Mutch, D.M., Berger, A., Mansourian, R., Rytz, A., Roberts, M.A.,2002. The limit fold change model: a practical approach forselecting differentially expressed genes from microarray data.BMC Bioinform. 3, 17.

Nicholson, R., Small, J., Dixon, A.K., Spanswick, D., Lee, K., 2003.Serotonin receptor mRNA expression in rat dorsal rootganglion neurons. Neurosci. Lett. 337, 119–122.

Nilius, B., Voets, T., Peters, J., 2005. TRP channels in disease. Sci.STKE 295, re8.

Petersen, D., Chandramouli, G.V., Geoghegan, J., Hilburn, J.,Paarlberg, J., Kim, C.H., Munroe, D., Gangi, L., Han, J., Puri, R.,Staudt, L., Weinstein, J., Barrett, J.C., Green, J., Kawasaki, E.S.,2005. Three microarray platforms: an analysis of theirconcordance in profiling gene expression. BMC Genomics 6, 63.

41B R A I N R E S E A R C H 1 1 0 7 ( 2 0 0 6 ) 2 4 – 4 1

Plumier, J.C., Hopkins, D.A., Robertson, H.A., Currie, R.W., 1997.Constitutive expression of the 27-kDa heat shock protein(Hsp27) in sensory and motor neurons of the rat nervoussystem. J. Comp. Neurol. 384, 409–428.

Pratt, R.W., Weimer, L.H., 2005. Medication and toxin-inducedperipheral neuropathy. Semin. Neurol. 25, 204–216.

Rogojina, A.T., Orr, W.E., Song, B.K., Geisert Jr., E.E., 2003.Comparing the use of Affymetrix to spotted oligonucleotidemicroarrays using two retinal pigment epithelium cell lines.Mol. Vision 9, 482–496.

Sahenk, Z., Nagaraja, H.N., McCracken, B.S., King, W.M., Freimer,M.L., Cedarbaum, J.M., Mendell, J.R., 2005. NT-3 promotes nerveregeneration and sensory improvement in CMT1A mousemodels and in patients. Neurology 65, 681–689.

Storey, J.D., Tibshirani, R., 2003. Statistical significance forgenome-wide studies. Proc. Natl. Acad. Sci. U. S. A. 100,9440–9445.

Tanabe, K., Bonilla, I., Winkles, J.A., Strittmatter, S.M., 2003.Fibroblast growth factor-inducible-14 is induced in axotomizedneurons and promotes neurite outgrowth. J. Neurosci. 23,9675–9686.

Valder, C.R., Liu, J.J., Song, Y.H., Luo, Z.D., 2003. Coupling gene chipanalyses and rat genetic variances in identifying potentialtarget genes that may contribute to neuropathic allodyniadevelopment. J. Neurochem. 87, 560–573.

Van der Zwaag, B., Hellemons, A.J., Leenders, W.P., Burbach, J.P.,Brunner, H.G., Padberg, G.W., Van Bokhoven, H., 2002.PLEXIN-D1, a novel plexin family member, is expressed in

vascular endothelium and the central nervous system duringmouse embryogenesis. Dev. Dyn. 225, 336–343.

Wada, E., Way, J., Lebacq-Verheyden, A.M., Battey, J.F., 1990.Neuromedin B and gastrin-releasing peptide mRNAs aredifferentially distributed in the rat nervous system. J. Neurosci.10, 2917–2930.

Wang, L.X., Wang, Z.J., 2003. Animal and cellularmodels of chronicpain. Adv. Drug Delivery Rev. 55, 949–965.

Wang, H., Sun, H., Della Penna, K., Benz, R.J., Xu, J., Gerhold, D.L.,Holder, D.J., Koblan, K.S., 2002. Chronic neuropathic pain isaccompanied by global changes in gene expression and sharespathobiology with neurodegenerative diseases. Neuroscience114, 529–546.

Winn, M.P., Conlon, P.J., Lynn, K.L., Farrington, M.K., Creazzo, T.,Hawkins, A.F., Daskalakis, N., Kwan, S.Y., Ebersviller, S.,Burchette, J.L., Pericak-Vance, M.A., Howell, D.N., Vance, J.M.,Rosenberg, P.B., 2005. A mutation in the TRPC6 cation channelcauses familial focal segmental glomerulosclerosis. Science308, 1801–1804.

Xiao, H.S., Huang, Q.H., Zhang, F.X., Bao, L., Lu, Y.J., Guo, C., Yang,L., Huang, W.J., Fu, G., Xu, S.H., Cheng, X.P., Yan, Q., Zhu, Z.D.,Zhang, X., Chen, Z., Han, Z.G., Zhang, X., 2002. Identification ofgene expression profile of dorsal root ganglion in the ratperipheral axotomy model of neuropathic pain. Proc. Natl.Acad. Sci. U. S. A. 99, 8360–8365.

Zhang, B., Schmoyer, D., Kirov, S., Snoddy, J., 2004. GOTree Machine(GOTM): aweb-based platform for interpreting sets of interestinggenes using Gene Ontology hierarchies. BMC Bioinform. 5, 16.