Journal of Dermatological Science 55 (2009) 27–33

Gene expression in canine atopic dermatitis and correlation with clinicalseverity scores

Shona H. Wood a,b,*, Dylan N. Clements d, William E. Ollier b, Tim Nuttall c,Neil A. McEwan c, Stuart D. Carter a

a Department of Veterinary Pathology, Faculty of Veterinary Science, University of Liverpool, Liverpool L69 3ZJ, UKb Centre for Integrated Genomic Medical Research, The Stopford Building, University of Manchester, Manchester, UKc Department of Veterinary Clinical Science, The University of Liverpool, Leahurst, Cheshire, UKd Royal (Dick) School of Veterinary Studies, University of Edinburgh, Easter Bush, The Roslin Institute, Midlothian, Scotland, EH25 9RG, UK

A R T I C L E I N F O

Article history:

Received 3 November 2008

Received in revised form 27 February 2009

Accepted 20 March 2009

Keywords:

Atopic dermatitis

Canine

Severity scores

qPCR

A B S T R A C T

Background: Canine atopic dermatitis (cAD) is a common condition in dogs that may be a naturally

occurring model for human atopic dermatitis (hAD). Despite this, comparative research is limited,

particularly into the genetic background of cAD.

Objectives:

1. Measure candidate gene expression in cAD skin using quantitative real time PCR (qPCR)

2. Correlate gene expression to clinical cAD scores (Canine Atopic Dermatitis Extent and Severity Index

[CADESI]-03 and intradermal allergen test [IDT]).

Methods: mRNA was extracted from biopsies of non-lesional and lesional skin from atopic dogs, and

healthy skin from non-atopic dogs. Gene expression was quantified using qPCR, and compared between

non-lesional atopic, lesional atopic and healthy skin. Gene expression in atopic skin was correlated with

clinical severity (CADESI-03) and the number of positive reactions on an IDT.

Results: Of the 20 quantified genes, 11 demonstrated statistically significant altered mRNA expression

between atopic and healthy skin; dipeptidyl-peptidase-4 (DPP4), phosphatidylinositol-3,4,5-trispho-

sphate-5-phosphatase-2 (INPPL1), serine protease inhibitor kazal type-5 (SPINK5), sphingosine-1-

phosphate lyase-1 (SGPL1), peroxisome proliferator-activated receptor gamma (PPARg), S100 calcium-

binding protein A8 (S100A8), Plakophilin-2 (PKP2), Periostin (POSTN), Cullin4A, TNF-a and

metalloproteinase inhibitor-1 (TIMP-1). Three genes correlated with CADESI-03: serum amyloid A 1

(SAA-1), S100A8, and PKP2; and four with IDT results: mast cell protease I (CMA1), SAA-1, S100A8 and

SPINK5.

Conclusion: Genes with altered expression included those relevant to skin barrier formation and

immune function, suggesting both are relevant in the pathogenesis of AD. Many of these genes reflect the

proposed pathogenesis in hAD, supporting the use of dogs as a model for hAD. Furthermore, these genes

may be considered suitable targets for future genetic and protein function studies in human and canine

AD.� 2009 Japanese Society for Investigative Dermatology. Published by Elsevier Ireland Ltd. All rights reserved.

Contents lists available at ScienceDirect

Journal of Dermatological Science

journa l homepage: www. int l .e lsev ierhea l th .com/ journa ls / jods

1. Introduction

Canine atopic dermatitis (cAD) is a common condition that canbe considered a naturally occurring, spontaneous model of humanatopic dermatitis (hAD). There are many animal models whichdisplay clinical signs consistent with hAD but dogs are of particular

* Corresponding author at: Department of Veterinary Pathology, Faculty of

Veterinary Science, University of Liverpool, Liverpool L69 3ZJ, UK.

E-mail address: shona.wood@liv.ac.uk (S.H. Wood).

0923-1811/$36.00 � 2009 Japanese Society for Investigative Dermatology. Published b

doi:10.1016/j.jdermsci.2009.03.005

interest as they share the same environment as humans, unlike thecommonly used rodent models. The pathogenesis of the disease inboth humans and dogs is strongly associated with immunologicalhyper-reactivity, although skin barrier function, microbial coloni-sation and infection may also have a role [1,2]. The prevalence ofAD over the last three decades has increased in both dogs [3] andhumans [4], suggesting common environmental and/or geneticcomponents.

Unlike most murine models cAD is not caused by a single genedefect, and, like hAD, is a polygenic disorder with complexinheritance and interactions with environmental influences. It is

y Elsevier Ireland Ltd. All rights reserved.

S.H. Wood et al. / Journal of Dermatological Science 55 (2009) 27–3328

therefore a potential complex model for the human condition.Using dogs as a model to study of the genetic basis of AD isadvantageous because dog breeds form genetically isolatedpopulations in strong linkage disequilibrium [5]. Fewer geneticmarkers are required to identify an association with a givenphenotype and smaller sample sizes can be used to find theseassociations compared to human genetic studies [6]. However,research into cAD has been limited to reporting cytokine andchemokine profile changes [7] until recently, when a message RNA(mRNA) expression microarray study was performed [8]. Simila-rities between dysregulated genes in hAD and cAD were identified,although the number and relevance of the genes tested werelimited. In hAD, microarray and qPCR analysis of gene expression,linkage studies and candidate gene analyses have been used toidentify causative or susceptibility alleles. These studies haveimplicated a large number of genes and regions [9]. Twelve geneswith relevant epidermal or immune functions (Cystatin A [10,11],CARD4 [12], P-selectin [13,14], PKP2 [15], PPARg [16,17], SGPL1[18], TNF-a [19,9], Cadherin-13 [20], CMA1 [21,22], DPP4 [23],SPINK5 [24,25], SAA-1 [26]), were selected as potential candidategenes for AD. Eight further genes (ARTS-1, Cullin-4A, INPPL1,S100A8, POSTN, SCCA2, STAT2, TIMP-1) were selected using datafrom a canine mRNA expression microarray [8], showing that theywere dysregulated in cAD.

The aim of this study was to quantify the expression of thesecandidate genes in lesional atopic, non-lesional atopic and healthycanine skin. Gene expression in atopic skin was also correlatedwith two clinical measures: the Canine Atopic Dermatitis andSeverity Index (CADESI-03), which is similar to the human SCORADindex, and the number of positive reactions on intradermal testing(similar to prick tests in human patients) with individual allergens(IDT). Determining whether cAD and hAD share a similar geneticbackground will provide evidence that cAD is a suitable complexmodel for the human condition, and help identify novel targetgenes for further study of pathogenesis and intervention.

2. Materials and methods

2.1. Animals

Lesional and non-lesional skin was obtained from 14 dogs withAD (seven males and seven females), with a mean age of 3.2 years(age range 11 months to 8 years 7 months). The breeds were: Boxer(5), Staffordshire bull terrier, Labrador, crossbreed (2), Springerspaniel, Rhodesian ridgeback, Bulldog, West Highland white terrierand Neapolitan mastiff. Lesional samples were also taken fromthree more atopic dogs (all female) with a mean age of 5.3 years(age range 1 year 9 months to 9 years 2 months). The breeds were:Staffordshire bull terrier, Scottish terrier and Labrador. Three morenon-lesional samples were also taken from different atopic dogs(two females and one male) with a mean age of 1.25 years (agerange 1 year to 2 years). The breeds were: crossbreed, Staffordshirebull terrier and Labrador.

The control dogs had a mean age of 4.8 years (range 10 months to10 years) and comprised 10 females and 7 males. The breeds were:crossbreed (14), boxer, German shepherd dog and Siberian husky.

2.2. cAD diagnosis and sample collection

The diagnosis of cAD was based on compatible history andclinical signs, and exclusion of other causes of pruritus [27]. Coatbrushings, skin scrapings, Sarcoptes specific IgG serology and trialtherapy were used to eliminate the possibility of an ectoparasiteinfestation. All atopic dogs underwent a 6–8 week diet trial witheither a home-cooked or commercial single, novel protein diet, or acommercial hydrolysed protein diet and water only to eliminate the

possibility of an adverse food reaction. No anti-inflammatorymedication was given for at least 3 weeks prior to examination.All dogs with a clinical diagnosis of AD had an IDT with 54environmental allergens (Greer Laboratories Inc., Lenoir, NC, USA)performed and interpreted according to accepted criteria [27].Clinical lesions were scored using the CADESI-03. This is a validatedassessment of clinical lesions (erythema, excoriation, lichenificationand self-induced alopecia) at 62 anatomical sites, each measuredfrom 0 (normal) to 5 (most severe) yielding a score of 0–1240 [28].

6 mm diameter punch biopsies were taken from atopic skinwhilst the dogs were sedated with 0.5–1.0 ml of 2% lignocaine(without adrenalin) infiltrated into the subcutis. Non-lesionalsamples were taken from clinically unaffected skin on the flank ofeach atopic dog. Lesional samples were taken from areas oferythema and macular–papular dermatitis from the ventral body.Excoriation, staphylococcal pyoderma and Malassezia dermatitiswere avoided. Control samples were taken from healthy dogs thathad been euthanatized for reasons unconnected with the study,with no history and clinical signs of pruritus or conditions likely toalter immune function. There was a delay of approximately 30 minbetween euthanasia and sample collection. All biopsies wereimmediately snap-frozen in liquid nitrogen or placed intoRNAlaterTM (Ambion Inc., Austin, TX, USA) before storage at�80 8C. The tissues used in this study were in excess of clinicalrequirements, stored and used with informed consent from theowners of each dog. The study followed ethical guidelines as laiddown by The University of Liverpool, The University of Manchesterand the Biology and Biotechnology Research Council.

2.3. RNA extraction and cDNA synthesis

The frozen biopsies were disrupted using a dismembranator(Mikro-Dismembrator, Sartorius Stedim Biotech, France) usingliquid nitrogen to maintain a low temperature. The resultingpowder was placed in Trizol1 (Sigma, Poole, UK) and RNA extractedusing a chloroform alcohol technique [29] and a RNA extraction kit(Qiagen, Crawley, UK). The integrity and quality of the RNA waschecked using an Agilent 2100 Bioanalyzer (Agilent BiotechnologiesInc., Santa Clara, CA, USA). cDNA synthesis was performed viareverse transcription of 200 mg (10 ml) of total RNA using Super-script II reverse transcriptase (Invitrogen, Paisley, UK) according tothe manufacturer’s instructions (http://www.invitrogen.com).

2.4. Primer design

Transcript sequences obtained from Ensembl were cross-checked with NCBI canine genome transcript sequences. Theassays were designed in areas that showed 100% homologybetween the Ensembl and NCBI sequences. To design the primersthe Roche universal probe library designer was used (https://www.roche-applied-science.com/servlet/). Where possible, theprimers were designed to cross an exon–exon boundary. If thiswas not possible, the primers were designed to be intron spanningto ensure that the primer hybridised on different exons. In thesecases the intron was greater than 1000 bp to maintain specificity.This approach to primer design increases the chance that theprimers are transcript specific and minimises genomic DNAcontamination. Finally, the specificity of the primers was checkedusing BLAST (http://www.ncbi.nlm.nih.gov/BLAST/). Table 1 showsthe primers, amplicons and probes used.

2.5. q-PCR assay design

The q-PCR assays were all performed in triplicate using aTaqManTM ABI PRISM 7900 SDS (Applied Biosystems, Foster City,CA, USA) in 384-well plate format. A 10-ml reaction volume was

Table 1Primers and probes.

Gene Forward primer Reverse primer Probe number

(Roche)

Amplicon

SCCA2 agaattctggctggacaagg tggtatttctaggatcttggcttg 65 agaattctggctggacaaggacacaagcaaacctgtgcagatgatgagacaatccaacg

tttttaatttcacctcactggaggacttgcaagccaagatcctagaaatacca

Cystatin A ccaagaatttgaagccgtaga ctattatcacctacccgcacct 83 ccaagaatttgaagccgtagagtataaaactcaagtggtggctggaataaattactacattaa

ggtgcgggtaggtgataatag

Card 4 tcgtcctgcaccacttcc tgatctggtttacgctgagtct 85 tcgtcctgcaccacttccgcaagcggcttgccctcgacctggacaacaacaatctcaacgactacg

gcgtgagggagctgcagccctgcttcagccgcctcaccgtcctcagactcagcgtaaaccagatca

P-selectin ctgcaccaatctgcaaagc atgagggctggacactgaac 17 ctgcaccaatctgcaaagcaggcatagtgtcagctcctacttcaaaggttcagtgtccagccctcat

PKP2 aagcatctttgggagctctg ggccattttccttctggac 4 aagcatctttgggagctctgcagaatctcactgcaggaagtggaccaatgccgacatcag

tagctcagacagttgtccagaaggaaaatggcc

PPARg caggaaagacaacagacaaatca ggggtgatgtgtttgaacttg 7 caggaaagacaacagacaaatcaccatttgttatctatgacatgaattccttaatgatggg

agaagataaaatcaagttcaaacacatcacccc

SGPL1 cgggttccactgaacaaaat ggagatggctctcctcattg 10 cgggttccactgaacaaaatgatggaggtggatgttcgggcaatgaggagagccatctcc

TNF-a atggcctccaactaatcagc cttggggttcgagaagatga 61 atggcctccaactaatcagccctcttgcccagacagtcaaatcatcttctcgaaccccaag

Cadherin 13 gccctcttcctggcaatc tccagagttatcagcaaagttcc 59 gccctcttcctggcaatcgacagtggcaaccctcccgctaccggtaccggaactttgctg

ataactctgga

CMA1 aggcggaacttcgtactgac cccagggtgaccattatga 4 aggcggaacttcgtactgacagctgctcactgtgcaggaaggttcataatggtcaccctggg

DPP4 agacgcaaagtactatcaactgagat gctgctcctatgcagggtat 81 agacgcaaagtactatcaactgagatgttcaggccctggtctgcccctctataccctgcat

aggagcagc

POSTN gggaagaacgaatcattacagg ttgcaacaatttcttcagagtttc 77 gggaagaacgaatcattacaggtcctgaaataaaatatactaggatttctactggtggtg

gagaaacagaagaaactctgaagaaattgttgcaa

SPINK5 gaaagaggaggacaacttgagaa gaattcgtggcactgatcct 2 gaaagaggaggacaacttgagaaacacaggagaaaagagtaatgaaaaccaggatca

gtgccacgaattc

SAA-1 ttgtgctccctggtcctg gagtaggctctccacatgtctct 66 ttgtgctccctggtcctgggtgtcagcagccagagatggttgacattcctcaaggaagcggg

tcaagggactagagacatgtggagagcctactc

ARTS-1 cctcatctgtccacgtctga tgaagtggaaaatcagttcaagg 66 tgaagtggaaaatcagttcaaggcctttctcatcaggctgctgagggacctcattgataa

tcagacgtggacagatgagg

INPPL1 tctcgaagctcttcttgtactcc cgcaccaagttcttcattgag 39 cgcaccaagttcttcattgagttctactccacctgcctggaggagtacaagaagagcttc gaga

Cullin4A ccttggagagttccatgtcc tctatgttgtcaaaactaaagcatga 11 tctatgttgtcaaaactaaagcatgaatgcggcgctgctttcaccagcaagctggaaggc

atgttcaaggacatggaactctccaagg

TIMP-1 gtggggcacaggtacagg cccagagagactcaccagaga 3 cccagagagactcaccagagaacccaccatggcaccctttgcgcccctggcctcctgcat

cctgctgttgctgtggctgaccgcccccagcagggcctgtacctgtgccccac

S100A8 caatgagggagtttatggcact aaacctggtggggcagat 154 aaacctggtggggcagatccttgggcaccatgctgacggaactggagagtgccataaact

ccctcattg

STAT2 tctccagctccaaggactct aggctcattgtggtctctaacag 36 aggctcattgtggtctctaacagacaggtggatgagctgcaacaaccgctggagcttaag

ccggagccagaagcagagtccttggagctggaga

S.H. Wood et al. / Journal of Dermatological Science 55 (2009) 27–33 29

used per well; this consisted of 5 ml Taqman 2X PCR master mix(Universal PCR Mastermix; Applied Biosystems), 0.1 ml each of20 mM forward and reverse primers, 0.1 ml of 10 mM probe(Exiqon; Roche Diagnostics Ltd.), 0.1 ml distilled water and 4.6 mlof sample cDNA or water for the negative controls.

The amplification was performed according to a standard 7900protocol: 10 min at 50 8C followed by 40 cycles of 95 8C for 1 minand 60 8C for 15 s, as recommended by the manufacturer (AppliedBiosystems). The real time data was analysed by SequenceDetection Systems software, version 2.2.1 (Applied Biosystems).The detection threshold was set manually at 0.05 for all assays.Standard curves were generated for each assay, to confirm theefficiency of the assay was between 93% and 107% and the R2 valuewas >0.98. All primer sets were tested and run on a 2% agarose gelwith 10 bp molecular weight markers to confirm the amplificationof a single amplicon under the cycling conditions used.

Prior to this study, an assessment of potentially suitablereference genes for skin was performed. This demonstrated thatRPL13A and CG14980 were the most stably expressed [30] incanine skin and therefore the most suitable reference genes tonormalise the rest of the samples [31].

For each gene, synthesised oligos of the amplified region(Eurogentec) were used to calibrate the amount of observedfluorescence to a known concentration. This approach enabledquantification of normalised transcript number in all samples,which compensated for the differences in mRNA concentrationsbetween the samples [32]. The normalised transcript numberswere calculated as follows:

calibrator template number ¼ volume of synthesized oligo used

�molar concentration

� Avogadro’s constant ðaltered to the volume usedÞ

Ct synthesized oligo� Ct experimental sample ¼ Ct difference

E power ¼ power ðE value; Ct differenceÞ

E power� calibrator template number ¼ transcript number

To normalise the data, the geometric means of the referencegene transcript numbers are used for each experimental sample inturn:ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

RPL13A� CG14980p

¼ geometric mean of reference genes

reference geometric mean

template number of experimental sample

¼ normalised template number

2.6. Statistical analysis

Samples were excluded if their RNA integrity number (RIN), asmeasured by the Bioanalyser 2100, were lower than 6 and/or ifthey fell outside the limits of detection established by the standardcurves. Normalised transcript numbers obtained from calibrationwith synthesised oligos were used for the statistical analysis of theqPCR data.

Comparisons between microarray and qPCR data used theactual or true fold change as calculated in the previousmicroarray study [8]. This was used because normalisedtranscript numbers cannot be quantified from microarrayresults. The data were not normally distributed and thereforenon-parametric analyses were undertaken. Wilcoxon rankedsign tests (Systat 121, SSI, USA) were used to compare matchedlesional and non-lesional samples from the same individual.Mann–Whitney U tests were used to compare data from the

S.H. Wood et al. / Journal of Dermatological Science 55 (2009) 27–3330

unmatched lesional/non-lesional samples and for the compar-isons between atopic and control samples. The correlationanalyses were carried out using the Pearson R test (Excel,Microsoft, USA). The statistical significance level was set atp < 0.05.

3. Results

3.1. Gene expression in cAD lesional, cAD non-lesional and healthy

dog skin

Table 2 shows the fold change in gene-specific mRNAexpression among lesional, non-lesional and control skin. Statis-tically significant changes in fold change of gene expression wereseen between lesional atopic and control skin for SPINK5, INPPL1,DPP4, SGPL1, PPARg, S100, PKP2, POSTN and Cullin-4A. Non-lesional skin also showed statistically significant changes inexpression compared to control skin for DPP4, INPPL1, PPARgand S100, though to a lesser extent than in lesional skin. OnlyPPARg and S100 showed a statistically significant difference inexpression in lesional and non-lesional skin. TNFa was down-regulated in non-lesional skin compared to controls and lesional,although this was only statistically significant when compared tocontrols. There were large changes in expression of SAA-1 andSCCA2 in lesional and non-lesional skin respectively whencompared to control skin, but these were not statisticallysignificant due to the large variation between individual cases.

3.2. Correlation of gene expression with IDT

CMA1 and SAA-1 expression correlated with the number ofpositive reactions on IDT (Fig. 1a and c), although their expressionwas not increased in atopic skin per se (Table 2). S100A8, incontrast, had a strong correlation with IDT (Fig. 1b) and itsexpression was significantly increased in cAD (Table 2). Finally,SPINK5 expression showed a positive correlation with IDT (Fig. 1d),although this was weaker than the correlations for CMA1, SAA-1and S100A8.

3.3. Correlation of gene expression with CADESI-03

SAA-1 (Fig. 2a), S100A8 (Fig. 2b) and PKP2 (Fig. 2c) expressionshowed a correlation with CADESI-03, although the r value was

Table 2Fold change in mRNA expression in cAD lesional, cAD non-lesional and control dog sk

Gene Lesional vs. control

skin-true fold change

p-value Non-lesional

skin-true fold

DPP4 �1.11 0.002 �0.53

INPPL1 2.48 0.002 0.78

SPINK5 4.68 0.004 0.22

SGPL1 0.74 0.004 0.94

PPARg �0.79 0.01 �0.17

S100A8 34.47 0.02 0.85

PKP2 0.38 0.03 0.52

POSTN �1.09 0.03 �0.87

Cullin4A 0.80 0.03 0.40

Cadherin-13 �0.51 0.06 �0.25

TNF-a �0.71 0.12 �0.92

SAA-1 2.39 0.17 0.13

P-selectin �0.22 0.18 �0.32

TIMP-1 0.20 0.23 �0.18

SCCA2 1.12 0.27 3.56

STAT2 �0.22 0.32 �0.30

Cystatin-A 0.07 0.33 0.37

CMA1 0.15 0.38 0.56

CARD4 �0.13 0.39 �0.04

ARTS-1 �0.08 0.49 �0.37

Shaded cells show the statistically significant fold change (p < 0.05).

generally lower than the correlation with the IDT results and onlyPKP2 expression was increased in cAD (Table 2).

4. Discussion

In this study, we used data from previous microarray studiesand from the published literature to select candidate genesimplicated in the pathogenesis of AD. The data showed that 11 outof the 20 quantified genes demonstrated statistically significantaltered mRNA expression between atopic and healthy skin. Thefunctions of these genes vary, but include both immunological andskin barrier functions. Seven out of the 11 genes have shownprevious associations or altered expression in hAD. The remainingfour genes were shown to be dysregulated in cAD on an earliercanine RNA microarray study [8].

S100A8 levels correlated with CADESI-03, suggesting thatS100A8 expression in dogs correlates with disease severity.Moreover, expression was markedly increased in lesional atopicskin compared to both non-lesional atopic (18.15-fold change) andhealthy skin (34.5-fold change). Expression in non-lesional atopicskin correlated with the number of positive reactions on an IDT,although the clinical relevance of this is unclear. S100A8 is a pro-inflammatory molecule which has been shown to ‘track’ theseverity of inflammatory diseases in humans [33]. These resultstherefore suggest that S100A8 could be involved in the develop-ment of TH2 mediated acute inflammation and TH1 mediatedchronic inflammation in cAD. Elevated expression in non-lesionalskin may also be a measure of cutaneous hyper-reactivity. INPPL1mRNA expression was also up-regulated in both lesional (2.48-foldchange) and non-lesional (0.78-fold change) atopic compared tohealthy skin, suggesting that it may be a good candidate for furtherresearch. In humans, INPPL1 has roles in calcium transport, B cellactivation, FceRIb mediated mast cell and Langerhans cellactivation, regulation of IgE mediated IL-6 production [34] and Tcell development [35]. Dysregulation of INPPL1 could thereforeaffect allergen specific IgE responses, allergen presentation anduptake, T-cell activation and inflammatory responses in AD.

The DPP4 gene, which produces a protein identical to the T cellactivation antigen CD26, was down-regulated in cAD. This issimilar to findings in human microarray [20] and flow cytometerystudies [23]. Reduced DPP4 expression could lead to less effectiveregulation of inflammation [36], delayed wound healing [37] andultimately lesion formation in AD.

in.

vs. control

change

p-value Lesional vs. non-lesional

skin-true fold change

p-value

0.03 �0.38 0.12

0.01 0.95 0.21

0.07 3.66 0.10

0.001 �0.12 0.11

0.39 �0.53 0.03

0.05 18.15 0.02

0.001 �0.10 0.95

0.08 �0.12 0.45

0.28 0.28 0.45

0.12 �0.21 0.33

0.04 0.12 0.16

0.11 2.00 0.15

0.08 0.08 0.29

0.83 0.42 0.05

0.44 �1.14 0.50

0.19 0.06 0.21

0.43 �0.28 0.16

0.17 �0.36 0.29

0.27 �0.09 0.54

0.11 0.27 0.21

Fig. 1. Correlation between the total number of positive reactions on an IDT with 54 environmental allergens and normalised transcript number in non-lesional atopic skin.

S.H. Wood et al. / Journal of Dermatological Science 55 (2009) 27–33 31

PKP2 was slightly up-regulated in lesional and non-lesionalskin, and expression correlated with CADESI-03. This is in contrastto the reduced expression reported in a human microarray,although this studied cultured fibroblasts and not whole skin [20].Increased PKP2 expression up-regulates beta catenin/T cell factorsignalling activity [38], enhancing T-cell survival [39]. Enhanced Tcell survival and activity is thought to be a key factor in human andcanine AD [7,40].

PPARg expression was down-regulated in atopic lesional dogskin but not in non-lesional skin. Similar findings were seen in ahuman microarray study of atopic skin [20]. Other members of thePPAR family (PPARb/d and PPARa) may represent potentialpathogenic candidates as they too have been associated withinflammation and skin barrier function, and have been implicatedin the pathogenesis of hAD [41]. PPARg may be important in thedevelopment of chronic inflammation as it is an anti-inflammatorymolecule. However, reduced expression of PPARg is also associatedwith reduced lamellar body formation and lipid processing, andincreased epidermal permeability. The epidermal barrier is ofincreasing significance in AD research; particularly following theassociation of two independent loss of function filaggrin variants(R501X and 2282del4) with human AD [42]. Filaggrin was notincluded in this study because of a lack of reliable primers toquantify its expression in canine skin.

SPINK5 was upregulated in lesional atopic compared tohealthy skin (fold change 4.68), and its expression correlatedwith the number of positive IDT reactions. SPINK5, which hasbeen associated with hAD [24,25], regulates hair and skinmorphogenesis [43], particularly proteolysis during epithelial

formation and keratinocyte differentiation. This suggests thataltered SPINK5 expression will affect skin barrier function [44].SPINK5 is also expressed in the thymus, where alteredexpression has been hypothesised to cause abnormal matura-tion of T-lymphocytes and TH2 polarisation [45]. SPINK5protein, in contrast, is down-regulated when associated withatopic symptoms in Netherton Syndrome in humans. Thisapparent contradiction in expression levels between humansand dogs may reflect differences in pathogenesis, SPINK5activity in different disease states, and/or poor correlationbetween RNA and protein expression [46].

Expression of SGPL1 was increased in both lesional and non-lesional skin, as observed in a human real time PCR study [18].Sphingomyelin metabolism enzymes are highly conserved acrossspecies [47], and are likely to have similar functions in dogs andhumans. SGPL1 is involved in the skin barrier through regulation ofsphingosine 1 phosphate and enhanced catalytic cleavage [48],influencing antimicrobial activity as well as keratinocyte prolif-eration and differentiation [49].

Genes correlated with CADESI-03 scores were pro-inflamma-tory except PKP2, although PKP2 has been shown to have T-cellrelated functions. The correlations with CADESI suggest anassociation between specific gene activity and disease severity.The strength of the correlations could have been affected by the useof different clinicians in CADESI assessments, although this hasbeen shown to have high inter- and intra-observer reliability [28].Genes correlated with the IDT results were also mainly inflam-matory related, with the exception of SPINK5, although this alsohas T-cell related activity.

Fig. 2. Correlation between CADESI-03 and normalised transcript number in non-lesional atopic skin.

S.H. Wood et al. / Journal of Dermatological Science 55 (2009) 27–3332

The number of positive reactions to IDT was used as a measureof allergen specific IgE because, unlike humans, total IgE levels donot correlate with atopic status or clinical signs, and the specificityand sensitivity of IgE specific serology is variable in dogs [50,51].There was no correlation between CADESI-03 and number ofpositive reactions on IDT (data not shown; r = 0.29 p = 0.26),suggesting that IDTs are a measure of individual susceptibility andexposure to allergens but not AD severity. The clinical significanceof the associations with the number of positive IDT reactions isunclear and further investigation of the clinical relevance of IDT isrecommended.

A weakness of this study was the inclusion of multiple dogbreeds and the variable time scale of the lesions at presentation.This was dictated by the nature of obtaining samples fromclinical cases. It is possible, however, that different breeds mayhave different genetic traits that result in clinical AD, andtherefore it would be valuable to study expression andpolymorphisms of the genes identified in this study with largersingle breed cohorts. It would also be interesting to repeat thiswork on samples taken at defined time points following allergenexposure in experimental canine models of AD. It is also possiblethat the results were influenced by the differing homeenvironments for each dog. The effect of the environment,including diet, on AD is recognised in both canine and human AD[52,53], and this was not assessed or controlled for in this study(apart from exclusion of food allergy by a 6 week food trialperformed as part of the diagnosis).

In conclusion, this study identified 11 genes significantlydysregulated in cAD. They are either associated or dysregulated inhAD, or regulate immune reactions and skin barrier function. It istherefore likely that dogs with naturally occurring AD are models

for the human condition. Moreover, this gene expression studysupports the selection of the identified genes for a large scale genepolymorphism study which will help to elucidate their roles in thepathogenesis of AD.

Acknowledgements

This study was funded by the Biotechnology and BiologicalSciences Research Council and Pfizer Animal Health

References

[1] Olivry T, Hill PB. The ACVD task force on canine atopic dermatitis (VIII): is theepidermal lipid barrier defective? Vet Immunol Immunopathol 2001;81(Sep-tember):215–8.

[2] VickeryF B.P.. Skin barrier function in atopic dermatitis. Curr Opin Pediatr2007;19:89–93.

[3] Hiller A, Griffin CE. The ACVD task force on canine atopic dermatitis (I):incidence and prevalence. Vet Immunol Immunopathol 2001;81(Septem-ber):147–51.

[4] Williams HC. Is the prevalence of atopic dermatitis increasing? Clin ExpDermatol 1992;17:385–91.

[5] Sutter NB, Eberle MA, Parker HG, Puller BJ, Kirkness EF, Kruglyak L, OstranderEA. Extensive and breed-specific linkage disequilibrium in Canis familiaris.Genome Res 2004;14(December):2388–96.

[6] Starkey MP, Scase TJ, Mellersh CS, Murphy S. Dogs really are man’s bestfriend—canine genomics has applications in veterinary and human medicine!Brief Funct Genomic Proteomic 2005;4:112–28.

[7] Nuttall TJ, Knight PA, McAleese SM. Expression of T-helper 1, T-helper 2 andimmunosuppressive cytokines in canine atopic dermatitis. Clin Exp Allergy2002;32:789–95.

[8] Merryman-Simpson AE, Wood SH, Fretwell N, Jones PG, McLaren WM, McE-wan NA, et al. Gene (mRNA) expression in canine atopic dermatitis: microarrayanalysis. Vet Dermatol 2008;19(April):59–66.

[9] Morar N, Willis-Owen S AG, Moffatt MF, Cookson William OCM. The genetics ofatopic dermatitis. J Allergy Clin Immunol 2006;118:24–34.

S.H. Wood et al. / Journal of Dermatological Science 55 (2009) 27–33 33

[10] Kato T, Takai T, Mitsuishi K, Okumura K, Ogawa H. Cystatin A inhibits IL-8production by keratinocytes stimulated with Der p 1 and Der f 1: Biochemicalskin barrier against mite cysteine proteases. J Allergy Clin Immunol2005;116:169–76.

[11] Vasilopoulos Y, Cork MJ, Teare D, Marinou I, Ward SJ, Duff GW, et al. Anonsynonymous substitution of cystatin A, a cysteine protease inhibitor ofhouse dust mite protease, leads to decreased mRNA stability and shows asignificant association with atopic dermatitis. Allergy 2007;62:514–9.

[12] Weidinger S, Klopp N, Rummler L, Wagenpfeil S, Novak N, Baurecht HJ, et al.Association of NOD1 polymorphisms with atopic eczema and related pheno-types. J Allergy Clin Immunol 2005;116:177–84.

[13] de Mora F, de la Fuente C, Jasmin P, Gatto H, Marco A, Ferrer L, et al. Evaluationof the expression of P-selectin, ICAM-1, and TNF-alpha in bacteria-free lesionalskin of atopic dogs with low-to-mild inflammation. Vet Immunol Immuno-pathol 2007;115(February):223–9.

[14] Wang HB, Wang JT, Zhang L, Geng ZH, Xu WL, Xu TH, et al. P-selectin primesleukocyte integrin activation during inflammation. Nat Immunol 2007;8:882–92.

[15] Green KJ, Jones JC. Desmosomes and hemidesmosomes: structure and functionof molecular components. FASEB J 1996;10:871–81.

[16] Dahten A, Mergemeier S, Worm M. PPARgamma expression profile and itscytokine driven regulation in atopic dermatitis. Allergy 2007;62:926–33.

[17] Schmuth M, Jiang YJ, Dubrac S, Elias PM, Feingold KR. Peroxisome proliferator-activated receptors (PPAR) and liver X receptors (LXR) in epidermal biology. JLipid Res 2008;49(3):499–509 (Epub January).

[18] Seo E-Y, Park GY, Lee K-M, Kim J-A, Lee J-H, Yang J-M. Identification of targetgenes of atopic dermatitis by real time PCR. J Invest Dermatol 2006;126:1187–9.

[19] Bonness S, Bieber T. Molecular basis of atopic dermatitis. Curr Opin AllergyClin Immunol 2007;7:382–6.

[20] Park YD, Lyou YJ, Lee KJ, Lee DY, Yang JM. Towards profiling the geneexpression of fibroblasts from atopic dermatitis patients: human 8K comple-mentary DNA microarray. Clin Exp Allergy 2006;36:649–57.

[21] Max XQ, Shirakawa T, Yoshikawa T, Yoshikawa K, Kawai M, Sasaki S, et al.Association between genetic variants of mast-cell chymase and eczema.Lancet 1996;348(August):581–3.

[22] Weidinger S, Rummler L, Klopp N, Wagenpfeil S, Baurecht HJ, Fischer G, et al.Association study of mast cell chymase polymorphisms with atopy. Allergy2005;60:1256–61.

[23] Bock O. Expression of dipeptidyl-peptidase IV (CD26) on CD8+ T cells issignificantly decreased in patients with psoriasis vulgaris and atopic derma-titis. Exp Dermatol 2001;10:414–9.

[24] Nishio Y, Noguchi E, Shibasaki M, Kamioka M, Ichikawa E, Ichikawa K, et al.Association between polymorphisms in the SPINK5 gene and atopic dermatitisin the Japanese. Genes Immun 2003;4:515–7.

[25] Kato A, Fukai K, Oiso N, Hosomi N, Murakami T, Ishii M. Association of SPINK5gene polymorphisms with atopic dermatitis in the Japanese population. Br JDermatol 2003;148:665–9.

[26] Thorn ZYL. Regulation of the human acute phase serum amyloid A genes bytumour necrosis factor-a, interleukin-6 and glucocorticoids in hepatic andepithelial cell lines. Scand J Immunol 2004;59:152–8.

[27] DeBoer DJ, Hillier A. The ACVD task force on canine atopic dermatitis (XV):fundamental concepts in clinical diagnosis. Vet Immunol Immunopathol2001;81(September):271–6.

[28] Olivry T, Marsella R, Iwasaki T, Mueller R. Validation of CADESI-03, a severityscale for clinical trials enrolling dogs with atopic dermatitis. Vet Dermatol2007;18(April):78–86.

[29] Clements DN, Vaughan-Thomas A, Peansukmanee S, Carter SD, Innes JF, OllierWE, et al. Assessment of the use of RNA quality metrics for the screening ofarticular cartilage specimens from clinically normal dogs and dogs withosteoarthritis. Am J Vet Res 2006;67:1438–44.

[30] Wood SH, Clements DN, McEwan NA, Nuttall T, Carter SD. Reference genes forcanine skin when using quantitative real time PCR. Vet Immunol Immuno-pathol 2008;126((3–4) August):392–5.

[31] Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, et al.Accurate normalization of real-time quantitative RT-PCR data by geometricaveraging of multiple internal control genes. Genome Biol 2002;3(7).research0034.1–research0034.11.

[32] Mohammadi M, Day FJR. Oligonucleotides used as template calibrators forgeneral application in quantitative polymerase chain reaction. Anal Biochem2004;335(December):299–304.

[33] Benoit S, Toksoy A, Ahlmann M, Schmidt M, Sunderkotter C, Foell D, et al.Elevated serum levels of calcium-binding S100 proteins A8 and A9 reflectdisease activity and abnormal differentiation of keratinocytes in psoriasis. Br JDermatol 2006;155:62–6.

[34] Tridandapani S, Wang Y, Yijie M, Clay B, Anderson CL. Src homology 2 domain-containing inositol polyphosphate phosphatase regulates NF-kb-mediatedgene transcription by phagocytic FcgRs in human myeloid cells. J Immunol2002;169(October):4370–8.

[35] Kashiwada M, Cattoretti G, McKeag L, Rouse T, Showalter BM, Al-Alem U, et al.Downstream of tyrosine kinases-1 and Src homology 2-containing inositol 50-phosphatase are required for regulation of CD4+CD25 + T cell development. JImmunol 2006;176(April):3958–65.

[36] Busso N, Wagtmann N, Herling C, Chobaz-Peclat V, Bischof-Delaloye A, So A,et al. Circulating CD26 is negatively associated with inflammation in humanand experimental arthritis. Am J Pathol 2005;166(February):433–42.

[37] Ghersi G, Dong H, Goldstein LA, Yeh Y, Hakkinen L, Larjava HS, et al. Regulationof fibroblast migration on collagenous matrix by a cell surface peptidasecomplex. J Biol Chem 2002;277(August):29231–4.

[38] Chen X, Bonne S, Hatzfeld M, van Roy F, Green KJ. Protein Binding andFunctional Characterization of Plakophilin 2. Evidence for its Diverse rolesin desmosomes and beta -catenin signalling. J Biol Chem 2002;277(March):10512–2.

[39] Ding L, Shen s, Lino AC, Curotto de Lafaille MA, Lafaille JJ. Beta-cateninstabilization extends regulatory T cell survival and induces anergy in non-regulatory T cells. Nat Med 2008;14:162–9.

[40] Avgerinou GG, Andreas V, Stavropoulos PG, Katsambas AD. Elevated serumlevels of calcium-binding S100 proteins A8 and A9 reflect disease activity andabnormal differentiation of keratinocytes in psoriasis. Br J Dermatol2008;47(March):219–24.

[41] Sertznig P, Seifert M, Tilgen W, Reichrath J. Peroxisome Proliferator-activatedreceptors (PPARs) and the human skin: importance of PPARs in skin physiol-ogy and dermatologic diseases. Am J Clin Dermatol 2008;9(1):15–31.

[42] Palmer CN, Irvine AD, Terron-Kwiatkowski A, Zhao Y, Liao H, Lee SP, et al.Common loss-of-function variants of the epidermal barrier protein filaggrinare a major predisposing factor for atopic dermatitis. Nat Genet 2006;38:441–6.

[43] Moffatt MF. SPINK5: A gene for atopic dermatitis and asthma. Clin Exp Allergy2004;34:325–7.

[44] Komatsu N, Takata M, Otsuki N, Ohka R, Amano O, Takehara K, et al. Elevatedstratum corneum hydrolytic Activity in Netherton Syndrome Suggests aninhibitory regulation of desquamation by SPINK5-derived peptide. J InvestDermatol 2002;118:436–43.

[45] Tartaglia-Polcini A, Bonnart C, Micheloni A, Cianfarani F, Andre A, Zambruno G,et al. SPINK5, the defective gene in Netherton Syndrome, encodes multipleLEKTI isoforms derived from alternative pre-mRNA processing. J Invest Der-matol 2005;126(December):315–24.

[46] Greenbaum D, Colangelo C, Williams K, Gerstein M. Comparing proteinabundance and mRNA expression levels on a genomic scale. Genome Biol2003;4:117.

[47] Hla Timothy. Signaling and biological actions of sphingosine 1-phosphate.Pharmacol Res 2003;47:401–7.

[48] Reiss U, Oskouian B, Zhou J, Gupta V, Sooriyakumaran P, Kelly S, Wang E, et al.Sphingosine-phosphate lyase enhances stress-induced ceramide generationand apoptosis. J Biol Chem 2004;279(January):1281–90.

[49] Herzinger T, Kleuser B, Schafer-Korting M, Korting HC. Sphingosine-1-Phos-phate Signaling and the Skin. Am J Clin Dermatol 2007;8:329–36.

[50] Olivry T, Dunston SM, Pluchino K, Porter K, Hammerberg B. Lack of detection ofcirculating skin-specific IgE autoantibodies in dogs with moderate or severeatopic dermatitis. Vet Immunol Immunopathol 2008;122(March):182–7.

[51] Hill PB, Hillier A, Olivry T. The ACVD task force on canine atopic dermatitis (VI):IgE-induced immediate and late-phase reactions, two inflammatorysequences at sites of intradermal allergen injections. Vet Immunol Immuno-pathol 2001;81:199–204.

[52] Nodtvedt A, Bergvall K, Sallander M, Egenvall A, Emanuelson U, Hedhammar A.A case-control study of risk factors for canine atopic dermatitis among boxer,bullterrier and West Highland white terrier dogs in Sweden. Vet Dermatol2007;18:309–15.

[53] Bisgaard H, Simpson A, Palmer CN, Bønnelykke K, Mclean I, Mukhopadhyay S,et al. Gene-environment interaction in the onset of eczema in infancy:filaggrin loss-of-function mutations enhanced by neonatal cat exposure. PLoSMed 2008;5:e131.