Gene Expression in the Skin of Dogs Sensitized to the House Dust Mite Dermatophagoides farinae

Atopic dermatitis is a multifactorial allergic skin disease in humans and dogs. Genetic predisposition, immunologic hyperreactivity, a defective skin barrier, and environmental factors play a role in its pathogenesis. The aim of this study was to analyze gene expression in the skin of dogs sensitized to house dust mite antigens. Skin biopsy samples were collected from six sensitized and six nonsensitized Beagle dogs before and 6 hr and 24 hr after challenge using skin patches with allergen or saline as a negative control. Transcriptome analysis was performed by the use of DNA microarrays and expression of selected genes was validated by quantitative real-time RT-PCR. Expression data were compared between groups (unpaired design). After 24 hr, 597 differentially expressed genes were detected, 361 with higher and 226 with lower mRNA concentrations in allergen-treated skin of sensitized dogs compared with their saline-treated skin and compared with the control specimens. Functional annotation clustering and pathway- and co-citation analysis showed that the genes with increased expression were involved in inflammation, wound healing, and immune response. In contrast, genes with decreased expression in sensitized dogs were associated with differentiation and barrier function of the skin. Because the sensitized dogs did not show differences in the untreated skin compared with controls, inflammation after allergen patch test probably led to a decrease in the expression of genes important for barrier formation. Our results further confirm the similar pathophysiology of human and canine atopic dermatitis and revealed genes previously not known to be involved in canine atopic dermatitis.

Interleukin 13 receptor subunit alpha 2 (IL13RA2): IL13RA2 mRNA concentration in sensitized dogs was higher in the allergen-than the saline-treated skin. IL13 is a cytokine that plays a pivotal role in activation and maintenance of IgEproduction by interacting with the receptor complex of IL13RA1 and IL4RA (Leung et al. 2004). Previous studies showed that human AD patients show higher IL13 serum concentration (Katagiri et al. 1997) and higher cutaneous IL13 mRNA expression, especially in acute lesions (Hamid et al. 1996). IL13RA2 was also increased in the serum of human AD patients (Hussein et al. 2011). ILR13A2 binds IL13 with high affinity and it is suspected that this binding of IL13 inhibits its P. Schamber et al.

SI
inflammatory effects (Chomarat and Banchereau 1998). The canine receptor is similar to its human counterpart (Tang 2001) and a similar role in dogs is conceivable. In this study, the normal dogs had lower IL13RA2 mRNA concentrations in the allergen treated skin. It is possible that because of an absent inflammatory response there was no positive feedback for an increasing IL13RA2 gene expression. Further studies are needed to elucidate the exact role of this receptor in cAD and its therapeutic potential.
Tumor necrosis factor ligand superfamily member 13B (TNFSF13B): TNFSF13B or "B cell activating factor of the TNF family" (BAFF) is an important survival factor for B cells (Mackay and Schneider 2009). BAFF exists in a membrane-bound and a soluble form (Mackay and Schneider 2009). Increased BAFF serum concentrations were found in people suffering from asthma (Kang et al. 2006) and in children with AD (Jee et al. 2010). In adults with AD increased BAFF concentrations were found only in acute lesions after allergy patch tests (Chen et al. 2011). Thus, BAFF may be more important in early onset, acute lesions of hAD. In our study, gene expression levels of BAFF 24 h after patch test were higher in allergen treated skin of allergic dogs than in control skin, similar to what is seen in humans. In contrast, BAFF levels at the 24 h time point in normal dogs were decreased in allergen treated skin compared to the negative control, indicating BAFF could be involved in a downregulation of inflammation.
TNFSF9: Dendritic cells (DC) are important for antigen (Ag) presentation, T cell stimulation and homing for Ag specific immune response, thus playing a pivotal role in the pathophysiology of AD (Bancherau and Steinmann et al. 1998). DC maturation is associated with an increase in the expression of co-stimulatory molecules. One of these cofactors is TNFSF9, a membrane-bound ligand of the TNF family (Wu et al. 2011). TNFSF9 binds to its receptor TNFSRF9, which is mainly expressed on activated T and B cells and monocytes (Schwarz et al. 1995;Wu et al. 2011). Some studies consider TNFSRF9 as a co-receptor for T cell proliferation (Schwarz et al. 1995;Wu et al. 2011). In contrast, others indicate that activation of this receptor leads to apoptosis of T cells (Langstein et al. 1998). Our study showed that in the allergen treated skin, mRNA concentration of TNFSF9 was decreased in normal dogs but increased in sensitized dogs compared to the negative control.
Higher expression of TNFSF9 in atopic individuals may contribute to the exaggerated immune response. The role of TFNSF9 and its receptor in AD needs to be further investigated.

Suppressor of cytokine signaling proteins (SOCS) 3:
The SOCS3 gene is a member of the SOCS family, a group of transcription factors, induced by cytokines (Elliott and Johnston 2004). SOCS proteins are important for the balance of Th1 and Th2 immune responses (Arakawa et al. 2004). SOCS3 is mainly expressed by Th2 cells. In people suffering from AD, SOCS3 expression correlates with the severity of clinical signs and further supports Th2 cell differentiation (Seki et al. 2003).
In cAD a higher expression of SOCS3 was found in lesional and nonlesional skin of allergic dogs compared to non-allergic controls (Schlotter et al. 2011). In our study the non-sensitized dogs showed a decrease in SOCS3 expression 24 h after allergen PT, compared to the other groups. The expression in the control skin was similar to the expression seen in the skin 4 SI P. Schamber et al.
of sensitized dogs. This result could indicate that non-allergic individuals actively downregulate pro inflammatory pathways which may fail in allergic individuals. If this observation can be confirmed, SOCS3 may a potential target for AD therapy.
B-cell lymphoma 3-encoded protein (BCL3): BCL3 was first detected as a proto-oncogene in B cell leukemia (McKeithan et al. 1990). BCL3 is expressed by different cell types such as lymphocytes (Brasier et al. 2001) and keratinocytes (Massoumi et al. 2006). In keratinocytes BCL3 expression is stimulated by Th2 cytokines (IL4, IL13), the molecule acts as a transcriptional factor downregulating the expression of genes important for the innate immune response, mainly antimicrobial peptides and upregulating the TNF alpha dependent expression of IL6 and IL8 (Buchau et al. 2009). An increase in the BCL3 concentration in lesional skin could be suppressed by vitamin D3 in vitro and in vivo (Buchau et al. 2009). In our study the sensitized dogs showed an increased expression of BCL3 after allergen PT. The non-sensitized dogs showed a prominent decrease of cutaneous BCL3 expression at the site of the allergen patch compared to the saline patch and compared to the expression in sensitized skin at both PT sites. In a similar fashion to the expression of nearly all inflammatory genes evaluated in this study the decrease in the skin of non sensitized dogs at the site of allergen application was prominent, compared to its negative control and to the expression in sensitized skin (at both allergen and saline PT sites). These results suggest that BCL3 may influence secondary skin infections in atopic individuals. Vitamin D3 may reduce the occurrence of such infections by suppressing BCL3 and by increasing cutaneous adenosin monophosphate expression. We can hypothesize that nonallergic individuals actively downregulate exaggerated inflammatory responses.
Adenosine A2B receptor (ADORA) 2B: The endogenous signaling purine molecule adenosine is an important mediator for many different biochemical processes (Berne et al. 1983) and has been implicated in playing an increasingly important role in the pathogenesis of asthma (Driver et al. 1993). Depending on the binding receptor, adenosine has anti-or proinflammatory properties (Rorke and Holgate 2002). In patients suffering from asthma an increased adenosine level was found in the liquid of bronchioalveolar lavage (Driver et al. 1993). Adenosine monophosphate causes bronchoconstriction in asthmatic, but not healthy individuals (by interacting with the ADORA2B-adenosine-receptor and mast cell degranulation) (Marquartd and Walker 1990). ADORA2B may have anti-or pro-inflammatory effects or both. Twenty-four hours after allergen PT, sensitized dogs showed an increased ADORA2B mRNA concentration in contrast to normal dogs. To the authors' knowledge a role of adenosine in canine atopic dermatitis has not yet been reported. It is possible that ADORA2B is involved in mast cell degranulation in atopic skin, similar to what has been reported for human asthma. Further studies are needed to investigate if antagonists of ADORA2B could be targets for new therapies in AD.
Fc gamma R: IgE and IgG both may play a role in the pathophysiology of cAD (Willemse et al. 1985;Halliwell and DeBoer 2001). IgG may have a protective role and higher IgG serum concentrations have been reported in normal compared to atopic dogs (Lian and Halliwell 1998). In humans patients with extrinsic AD show elevated allergen specific serum IgE and IgG concentrations (Sicherer and Leung 2006). Most investigations on FC-receptors focused on FCepsilon receptors because P. Schamber et al.

SI
of the dominant role of IgE in hAD (Kinet 1999). However, IgG can also be increased (DeBoer 1998). Most cells of the immune system have receptors for IgG (FcgammaR). Three different FCgamma receptors, FCgammaI (Cluster of differentiation (CD) 64) FCgammaRII (CD32) and FCgammaRIII (CD16), are known (Ravetch and Kinet 1991). In humans an enhanced expression of CD64 and CD16 was found in acute and chronic atopic skin lesions (Kiekens et al. 2000). In an atopic mouse model it was shown that FcepsilonRI (an IgE receptor) and CD16 (an IgG receptor) have overlapping roles (Abboud et al. 2009). The normal dogs in our study had a reduced expression of CD32 and CD16 at the allergen PT.
Sensitized dogs showed higher mRNA expression of both receptors in the allergen PT skin. Our results suggest that IgGreceptors may be involved in canine atopic skin reactions.

Chitin receptors and chitinases:
Chitin is the second most abundant polysaccharide in the environment, after cellulose and it is a component part of different species such as bacteria, mushrooms and insects, but it is not synthesized by mammalians. Recently, it was shown that chitin activates macrophages by interacting with different surface receptors, such as macrophage mannose receptor 1 (MRC1), toll-like receptor 2 (TLR2) (Da , dectin 1 (CLEC7A) (Lee 2009), and leukotriene B4 receptor (BLT1) (Reese et al. 2007). In our study normal dogs showed a diminished mRNA concentration of MRC1, TLR2 and CLEC7A. In contrast the concentration of these three receptors was increased in the allergen PT skin of sensitized dogs. As chitin is a component of the exoskeleton of house dust mites, these results could indicate possible macrophage activation via chitin-receptor-interaction, contributing to the Th1 activation seen in chronic atopic lesions.
Chitinases are hydrolytic enzymes that are able to degrade chitin. Functional mammalian chitinase genes and chitinase-like proteins (CLP), which are able to bind chitin, but which have lost their enzymatic ability (Renkema et al. 1998;Chang et al. 2001) have been identified. Recent studies in mammals suggest that both chitinases and CLP are potent regulators of the innate immune response through interaction with chitin molecules (Shibata et al. 1997;Lee et al. 2008). Some authors suggest an association between CLP and the development and progression of allergic diseases and tissue remodeling Ober and Chupp 2009). One example for such a CLP is chitinase3-like1 (CHI3L1) coding for its protein YKL40. Its expression is stimulated by IL13 ). Increased YKL40 concentrations were found in the serum and lungs of asthma patients and were correlated with the severity of clinical signs. Polymorphisms found in the CHI3L1-gene were correlated with YKL40 concentration and asthma (Chupp et al. 2007;Ober and Chupp 2009). Another polymorphism in CHI3L1 has been proposed to be related to atopy in Korean children (Sohn et al. 2009). In the present study non sensitized dogs showed a decreased and sensitized dogs an increased CHI3L mRNA concentration in the allergen PT skin. CHI3L1 may play a role in the maintenance of the Th2 inflammation. In previous studies it has been shown that YKL40 prevents the apoptosis of T-cells and macrophages by the inhibition of Fas expression (Lee 2009). Further studies are needed to elucidate the possible role of CHI3L1 and its protein in canine and human atopic dermatitis. 6 SI P. Schamber et al.

GATA3
: GATA transcription factors are a family of transcription factors characterized by their ability to bind to the DNA sequence "GATA". GATA binding protein 3 (GATA3) is a transcription factor inhibiting the Th1-and promoting the Th2response (Zheng and Flavell 1997;Nawijn et al. 2001). In transgenic mice, over-expression of the human GATA3 gene led to an augmented Th2 immune response, which was reversible by inhibiting GATA3 expression (Bae et al. 2011).
Polymorphisms in GATA3 where shown to be associated with AD in British children (Arshad et al. 2008) but not German children (Suttner et al. 2009). In contrast to the results in humans the sensitized dogs in this study showed a decreased concentration of GATA3 mRNA. This could be due to the fact that these dogs did not suffer from naturally occurring cAD, but were sensitized. The findings may also be related to this gene pool of dog, if one interprets that the variation between German and British children may be genetically determined but alternatively, GATA3 may not be involved in cAD. been detected in hAD. However, a recent study showed a decreased expression of FLG2, desmoglein 1 (DSG1), desmocollin and transglutaminase 3 (TGM3) in human AD using comparative proteomic profiling (Broccardo et al. 2011). Future studies will help to elucidate the role of FLG2 in atopic dermatitis. Our results suggest a putative role of FLG2 in cAD.
Skin-specific aspartic peptidase retroviral-like 1 (ASPRV1): ASPRV1 seems to be important for posttranslational processing of profilagrin to filaggrin, a key event during epidermal differentiation (Matsui et al. 2011). It was shown that ASPRV1 deficient hairless mice developed dry skin and a thicker and less hydrated stratum corneum. Missense mutations in hAD patients and normal individuals were shown to have a negative effect on the ability of ASPRV1 to cleave the profilaggrin linker peptide (Matsui et al. 2011). Another study revealed impaired skin regeneration and remodeling in mice with impaired ASPRV1 expression (Hildenbrand et al. 2010). In contrast, a study evaluating atopic Europeans failed to find an association between ASPRV1 gene mutations and atopic eczema (Sandilands et al. 2012). The mRNA concentration of ASPRV1 in sensitized dogs after allergen treatment was strongly decreased, in contrast to control dogs. Possibly there is an increased filaggrin demand in allergen-exposed skin that cannot be met by atopic dogs due to a decreased ASPRV1 P. Schamber et al.

SI
expression in cAD. This could explain the previous findings in atopic dogs where a loss of function mutation or modified expression of FLG was not present (Chervet et al. 2010), but immunohistochemical changes pointing to a filaggrin deficiency were found (Marsella et al. 2009).

TGM1 and CE precursor proteins:
The protein envelope component of the cornified envelope (CE) consists of different cross-linked proteins (Credille et al. 2009). It is assumed that the enzyme transglutaminase 1 (TGM1) is responsible for the crosslinking of CE proteins such as involucrine, loricrin, and others (Greenberg et al. 1991;Steinert and Marekov 1995) into a mechanically resistant protein polymer and of attaching lipids (ceramides) to the crosslinked proteins by esterification (Marekov and Steinert 1998;Nemes et al. 1999). Neonatal death is seen in mice that lack the TGM1 gene, which leads to a defective stratum corneum, massive TEWL and subsequent dehydration (Matsuki et al. 1998). After skin transplantation from TGM1 deficient mice to normal ones ichthyosiform changes were observed (Kuramoto et al. 2002). Mutations were found in the TGM1 gene of humans and dogs suffering from ichthyosis (Cao et al. 2009;Credille et al. 2009). In our study, The transmembrane protein DSG1, a Ca2+ binding cadherin binds to the DSG1 molecule of the adjacent keratinocyte in desmosomes (Green and Simpson 2007) and corneodesmosomes (Caubet et al. 2004;Descargues et al. 2006). DSG1 has an important role in the skin barrier and function of the stratum corneum and is decreased in human AD (Broccardo et al. 2011), which is in agreement with our findings, that sensitized dogs (in contrast to healthy dogs) showed a decreased gene expression of the 24 h allergen PT compared to saline PT.
The tight junction (TJ) contributes to intercellular adhesion in epithelial cells and is located at the most apical part of their lateral membranes (Farquhar and Palade 1963). In stratified epithelia TJ are located in the maculae occludentes of the stratum granulosum (Squier 1973). The function of the maculae occludentes in the epidermis remains controversial (Hashimoto 1971;Elias and Friend 1975). In mice, TJ's may be involved in the epidermal barrier integrity (Yamamoto et al. 2008). Occludin (OCLN) is a transmembrane protein of the TJ (Ando-Akatsuka et al. 1996) expressed in the outer layers of the epidermis. Another TJ protein, cingulin (CNG) interacts with other TJ-proteins, such as actin (Bazzoni et al. 2000;D'Atri and Citi 2001) and is involved in the regulation of the gene expression of other proteins as well as cell proliferation (Aijaz et al. 2005;Guillemot and Citi 2006). In the skin of sensitized dogs CNG and OCLN expression was strongly decreased 24 h after allergen PT. This may contribute to an impaired cell proliferation and TJ formation. Whether defective TJ's are involved the pathophysiology of AD should be examined in further studies.

Protease inhibitors:
The balance between cell proliferation, maturation and desquamation is of particular importance for a physiologically sound skin barrier. Endogenous epidermal proteases and exogenous proteases are involved in the process of corneocyte desquamation by corneodesmolysis (Horikoshi et al. 1999). Proteases further have the property to activate or inactivate antimicrobial peptides such as cathelicidines in the skin (Yamasaki et al. 2006). Protease inhibitors are produced by keratinocytes to prevent excessive protease activity and associated uncontrolled desquamation of the stratum corneum resulting in skin barrier defects and inflammation (Hansson et al. 2002;Denecker et al. 2008). The gene "serine peptidase inhibitor, Kazal type 5" (SPINK5) is coding for an important protease inhibitor. Polymorphisms in this gene have been shown to be associated with human AD (Walley et al. 2001;Nishio et al. 2003;Weidinger et al. 2008) and a loss of function mutation with Netherton syndrome (Chavanas et al. 2000). The sensitized dogs in our study showed reduced expression of SPINK5 24 h after allergen PT. On the assumption that SPINK5 in our dog is not mutated a reduced expression would not prevent the skin against excessive protease activitiy and its consequences mentioned before. Another study however, found an increased expression of SPINK5 in dogs with AD (Wood et al. 2009). These contrasting findings could be due to the different biological models used. Wood et al. analyzed skin samples of dogs of different breeds, different age and different disease states. Genes associated with hAD vary between different populations. In dogs it was shown that breed diversity limits the detection of gene association in cAD (Wood et al. 2010). In addition it should be mentioned that SNPs or point mutations can lead to dysfunctional proteins without a reduction in expression or sometimes even with increased P. Schamber et al.

SI
expression due to aberrant feedback loops. A reduced SPINK5 expression due to inflammation like in our study or a lack of SPINK5 function due to a loss of function mutation with normal or increased expression may be involved in cAD in some breeds and protease inhibitors could be a new therapeutic option (Egelrud et al. 2005).

Further DEGs
Keratinocyte proline-rich protein (KPRP) is a recently identified marker of epidermal differentiation (Kong et al. 2003). KPRP in humans is expressed in the stratum granulosum and Lee et al. 2005 found an increased expression in patients with psoriasis. In contrast its expression in patients with hAD was decreased (Lee et al. 2005). A decreased expression of KPRP in sensitized compared to control dogs was found in this study and indicates a participation of KPRP in the pathophysiology of cAD. It is not known if KPRP is a structural protein of the cornified envelope or has other function with regard to skin differentiation, thus further studies are needed to elucidate its role in the cutaneous homeostasis.
Calmodulin-like 5 (CALML5) is an epidermal protein related to the calmodulin family of Ca 2+ -binding proteins. CALML5 is highly expressed during keratinocyte differentiation and an increased concentration of CALML5 was found in the skin of psoriasis patients, although it was not clear if this increase was due to an enhanced expression or reduced proteolytic degradation (Mehul et al. 2001). The diminished expression of CALML5 in the skin of sensitized dogs 24 h after allergen PT may indicate disturbed epidermal differentiation.
Peroxisome proliferator-activated receptor alpha (PPARA) is a transcription factor activated by fatty acids that are produced in inflammation (Moraes et al. 2006). In the skin, PPARA is expressed by keratinocytes (Rivier et al. 1998), Langerhans cells (Dubrac et al. 2007), macrophages (Babaev et al. 2007) and T-cells (Cunard et al. 2002). PPARA regulates the proliferation and differentiation of keratinocytes (Komuves et al. 2000) and is involved in wound healing (Michalik et al. 2001). In human atopic skin a diminished expression of PPARA was documented (Plager et al. 2007). The activation of PPARA by topical treatment with PPARA-ligands showed anti-inflammatory effects in humans with AD (Eberlein et al. 2008;Eichenfield et al. 2009) and ultraviolet-B-light-induced skin inflammation (Kippenberger et al. 2001). Törma et al. showed that PPARA expression decreases early in skin inflammation following allergen exposure (Törma and Berne 2009). In the present study, sensitized dogs showed a decrease of PPARA expression 24 h after allergen PT in comparison to the control group. As PPARA is not only involved in keratinocyte differentiation, but also has regulatory activity in skin inflammation (Dubrac and Schmuth 2011). The importance of PPARA-ligands in cAD should be elucidated. Arachidone lipoxygenase 3 (ALOXE3), which codes for the LOX3 protein, which is predominantly expressed in the epidermis (Krieg et al. 2002). The enzymes of ALOXE3 and ALOX12B convert arachidonic acid into epoxyalcohol products, which activate PPARA and therefore seem to play an important role in epidermal differentiation (Yu et al. 2007). Functional impairment of either ALOX12B or ALOXE3 results in ichthyosiform skin disease in humans (Jobard et al. 2002). Expression of ALOXE3 also decreases skin inflammation after 10 SI P. Schamber et al.
allergen exposure (Törmä and Berne 2009). Our results show, for the first time, a similar expression pattern of PPARA and ALOXE3 in canine skin inflammation. We hypothesize that both genes play a role in cAD.

References
Abboud, G., D. Staumont-Salle, A. Kanda, T. Roumier, N. Deruytter et al., 2009 Fc(epsilon)   File S2: Excel spread sheet showing the data after statistical unpaired analysis; the feature number is the sample number in the array, following rows show the different type of gene Ids, thereafter the coefficient of differential expression, the tvalue, the p-value, the adjusted p-value and the F values. The last row indicates if the expression is higher (1) or lower than the mean value (-1), the colored rows show human genes that matched with the canine ones to identify more functions for the non annotated canine genes.
File S3: Excel spread sheet with the results of the DAVID analysis of the gene sets 1 and 2; the Annotation Clusters with the highest enrichment score are shown first, the terms give the functional group in wich the genes are classified; in the row GENES the human orthologous gene IDs are listed.

File S4:
The results of the DAVID analysis of genes that were classified in Clusters 1-4 in the SOTA analysis; the annotation clusters with the highes enrichment score are shown first, the row "Term" shows the functional group where the genes were classified in, in the row F "Genes" gene Ids of the genes matching this category are listed.