Gene 481 (2011) 83–92

Contents lists available at ScienceDirect

Gene

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

Claudin-18 gene structure, regulation, and expression is evolutionaryconserved in mammals

Özlem Türeci a,⁎, Michael Koslowski b, Gerd Helftenbein a, John Castle b, Christoph Rohde a, Karl Dhaene c,Gerhard Seitz d, Ugur Sahin e,b

a Ganymed Pharmaceuticals AG, Freiligrathstr. 12, 55131 Mainz, Germanyb TRON—Translational Oncology and Immunology GmbH, Langenbeckstr. 1, 55131 Mainz, Germanyc Algemeen Stedelijk Ziekenhuis Aalst, Department of Pathology, Merestraat 80, B-9300 Aalst, Belgiumd Hospital Bamberg, Department of Pathology, Buger Str. 80, 96049 Bamberg, Germanye III. Department of Internal Medicine, Division of Translational and Experimental Oncology, Johannes Gutenberg University, Obere Zahlbacherstr. 63, 55131 Mainz, Germany

Abbreviations: CLDN18.2, Claudin-18 isoform 2; CLDCREB, cyclic AMP response element-binding protein;polymerase chain reaction; TJ, tight junction; BLAST, basnr, non-redundant protein database; RefSeq, NCBI RT/EBP/NKX2.1, homeodomain transcription factor; EInstitute; IgG, immunoglobulin G; HRP, horseradishisothiocyanate; FFPE, formalin fixed paraffin-embedded;buffered saline; DAB, 3,3′-diaminobenzidine; kb, kiCalifornia Santa Cruz; aa, amino acids; bp, base pairs;ECL, extracellular loop; PDZ, PDZ protein domain; MUPP1; ICH, The International Conference on HarmonisationRegistration of Pharmaceuticals for Human Use; CPMMedicinal Products; ID, identifier; IHC, immunohistocontrol.⁎ Corresponding author at: Ganymed Pharmaceutica

Mainz, Germany. Tel.: +49 6131 101; fax: +49 6131 1E-mail address: o.tureci@ganymed.ag (Ö. Türeci).

0378-1119/$ – see front matter © 2011 Elsevier B.V. Aldoi:10.1016/j.gene.2011.04.007

a b s t r a c t

a r t i c l e i n f o

Article history:Accepted 15 April 2011Available online 4 May 2011

Received by A.J. van Wijnen

Keywords:Tight junctionClaudinsOrthologConservation

Claudin-18 isoform 2 (CLDN18.2) is one of the few members of the human claudin family of tight junctionmolecules with strict restriction to one cell lineage. The objective of the current study was to comparemolecular structure and tissue distribution of this gastrocyte specific molecule in mammals. We show herethat the CLDN18.2 protein sequence is highly conserved, in particular with regard to functionally relevantdomains in mouse, rat, rabbit, dog, monkey and human and also in lizards. Moreover, promoter regions oforthologs are highly homologous, including the binding site of the transcription factor cyclic AMP–responsiveelement binding protein (CREB), which is known to regulate activation of human CLDN18.2. Employing RT-PCR and immunohistochemistry, we found that, analogous to the human gene, all orthologous CLDN18.2transcripts and proteins are exclusively expressed in differentiated gastric cells. Gene structure, promoterelements and RNA expression pattern of the lung-tissue specific Claudin-18 isoform 1 (CLDN18.1) as well, arehomologous across species. These findings exemplify phylogenetic conservation of lineage-specific membersof a multigene family. Given that CLDN18.2 is a novel drug target candidate, our data is also relevant for drugdevelopment as it reveals all six investigated mammalian species as suitable models for testing safety ofCLDN18.2 targeting regimen.

N18.1, Claudin-18 isoform 1;RT-PCR, reverse transcriptionic local alignment search tool;eference Sequence database;BI, European Bioinformaticsperoxidase; FITC, fluoresceinmin, minutes; PBS, phosphatelobases; UCSC, University ofTM, transmembrane domain;-1, multi-PDZ domain proteinof Technical Requirements forP, Committee for Proprietarychemistry; NTC, no template

ls AG, Freiligrathstr. 12, 5513114.

l rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

In multicellular organisms, maintenance of compartments thatdiffer in fluid and solute composition is ensured by epithelial andendothelial cell layers, which control the passage of water and solutes

(Smetana, 1947). The paracellular aspect of this control is mainlyattributed to cell–cell contact sites known as tight junction (TJ)strands (Farquhar and Palade, 1963). TJs have additional roles, such asmaintaining cell polarity by forming a barrier that prevents lateraldiffusion of membrane proteins and lipids and recruitment ofsignaling molecules for regulation of proliferation, differentiation,motility and other cellular functions (Stevenson et al., 1988).

Tight junctions require claudin molecules for their formation(Furuse et al., 1998). The claudin family comprises more than twodozens of four-pass transmembrane proteins with similar sequenceand structure, but divergent tissue distribution. Whereas mostclaudins are active in multiple tissues or even expressed ubiquitously,a few claudins are restricted to single tissues (reviewed in (Krauseet al., 2008)). Some cell types express unique claudin species (Fujitaet al., 2006). For correct tissue-specificity of transcriptional activation,individual claudin genes are controlled by unique regulatorymechanisms.

Tight junction strands generally contain multiple claudin speciesthat may interact in homo- and heterophilic ways (Gonzalez-Mariscalet al., 2003; Morita et al., 1999). It is hypothesized that the distinctcomposition of claudins plays a major role in variable physiological

84 Ö. Türeci et al. / Gene 481 (2011) 83–92

properties of TJs in different tissues (Anderson, 2001; Mitic et al.,2000; Rahner et al., 2001). Furthermore, due to their crucial functionin compartment separation, TJs are an evolutionary well conservedcomponent of vertebrate cytoarchitecture (Kollmar et al., 2001).However, phylogenetic analysis of claudins has been largely donein the context of evolutionary aspects concerning tight junctionconservation (Krause et al., 2008) and little is known about theevolutionary conservation of single claudins.

Recently, a member of the claudin multigene family, isoform 2of claudin-18 (CLDN18.2), appeared among the hits in a combinedin silico data mining and wet bench strategy to identify gastrocytelineage–specific cell surface molecules for use as therapeutic antibody

Fig. 1. Phylogenetic conservation of gene structure, transcript sequence and cis-acting regugene locus on chromosome 3q22, which was used as reference. CLDN18.1 and CLDN18.2 resushows an alignment of the human CLDN18 gene locus against genomic sequences of primatetranscripts encoding isoform 1 or isoform 2 (Kent et al., 2002). (Middle) Summarizedconservation (Siepel et al., 2005). (Lower) Cross-species alignments for selected species, withand Claudin18.2 promoter sequences. Cis-acting regulatory elements described for the humapromoter was not available.

targets (Sahin et al., 2008). We showed that expression of CLDN18.2 innormal tissues is strictly confined to differentiated epithelial cells of thegastricmucosa and is absent from thegastric stemcell zone. CLDN18.2 isretained upon malignant transformation and is expressed in asignificantproportionof primary gastric cancers andmetastases thereof.Moreover, we found frequent ectopic activation of CLDN18.2 inpancreatic, esophageal, ovarian, and lung tumors. The closely relatedsplice variant isoform1 (CLDN18.1), in contrast, is exquisitely restrictedto cells of lung tissue (Niimi et al., 2001).

The strict lineage specificity of both CLDN18 variants in normalhuman tissues prompted us to consider this gene as a model to in-vestigate how gene structure, regulation and expression of highly

latory promoter elements of CLDN18. (A) Genomic organization of the human CLDN18lt from alternative splicing of the first exon. (B) Screenshot of the UCSC genome browser, mammals, and vertebrates. (Top) RefSeq transcripts feature alternative first exons forconservation across 46 vertebrates by PhastCons, with taller bars indicating higherbars indicating conservation (Blanchette et al., 2004). (C) Alignment of the Claudin18.1n CLDN18 isoforms are boxed. The genomic sequence of O. cuniculus for the Claudin18.1

Fig.

1(con

tinu

ed).

85Ö. Türeci et al. / Gene 481 (2011) 83–92

86 Ö. Türeci et al. / Gene 481 (2011) 83–92

selective members of multigene families with housekeeping functionevolve across species.

2. Materials and methods

2.1. Sequence retrieval and analysis

Orthologs of human CLDN18 isoforms were detected by BLASTsearches either in the non-redundant protein database (nr) at NCBI(http://blast.ncbi.nlm.nih.gov/Blast.cgi) or in all available species data-bases at the Ensembl server site (http://www.ensembl.org/Multi/blastview). In each case the human CLDN18.2 protein sequence RefSeqID: NP_001002026 served as query. Failing RefSeq NM transcripts andNP proteins, we selected Ensembl identifier. A BLAST search of thehuman CLDN18.1 and CLDN18.2 proximal core promoter sequences upto−280 bp encompassing the T/EBP/NKX2.1 binding sites (CLDN18.1)and the CREB binding site (CLDN18.2) against the genomic sequencedatabases of all selected species led to the orthologous promotersequences. All alignments were conducted by use of the ClustalW2programvia theweb interface at the EBI webserver (http://www.ebi.ac.uk/Tools/clustalw2/index.html).

2.2. RNA extraction and cDNA synthesis

Fresh frozen human tissue specimens were purchased fromdifferent commercial providers. Monkey (Macaca fascicularis) tissuesections were purchased from LPT (Hamburg, Germany), dog (Canisfamiliaris, Beagle) tissues from Aurigon (Tutzing, Germany) and rabbit(Oryctolagus cuniculus), rat and mouse tissue sections were preparedin house. RNA preparation and first strand cDNA synthesis wereperformed as described elsewhere (Grunwald et al., 2006; Koslowskiet al., 2006).

2.3. RT-PCR profiling

For end-point RT-PCR analysis, degenerated sense primers specificfor CLDN18.1 (5′-GTG TTC CAr TAy GAr GGG CTs TGG-3′) or CLDN18.2(5′-CTG ATy GGG wTT GCv GGC ATy ATT GC-3′) orthologs frommonkey, dog, rabbit, rat and mouse were used in combination with adegenerated antisense primer (5′-CCA GAA GTT rGT nAC sAG CAT GTTGG-3′) at an annealing temperature of 64 °C in a 35 cycle PCR reaction.

Real-time quantitative analysis of transcript expression was per-formed using the ABI PRISM 7900 Sequence Detection Systeminstrument and software (Applied Biosystems). Expression of 18srRNA (sense: 5′-CGA TGC TCT TAG CTG AGT GTC-3′; antisense: 5′-TAACCA GAC AAA TCG CTC CAC-3′; 65 °C) was assessed in hexamer-primedcDNAs using QuantiTect SYBR Green PCR Kit (Qiagen). CLDN18transcripts were amplified in a 40 cycle PCR reaction using the sameprimers used for conventional RT-PCR. The relative expression levels ofCLDN18 transcripts was computed usingΔΔCT calculation with respectto the internal 18 s rRNA standard to normalize for variances in thequality of RNA and the amount of input cDNA.

Table 1CLDN18 splice variants in six mammalian species are listed by either RefSeq IDs or Ensnomenclature.

Species Chromosome Isoform 1

Transcript Protein

Human 3 NM_016369 NP_05745Dog 23 ENSCAFT00000011986 ENSCAFP0Macaque 3 ENSMMUT00000013455 ENSMMUPMouse 9 NM_019815 NP_06278Rabbit 14 ENSOCUT00000027967 ENSOCUP0Rat 8 NM_001014096 NP_00101

2.4. First and second antibodies

For immunohistochemistry, four different antibodies with IgGconstant regions from three different species were available. Thisfacilitated detection of CLDN18 across different species. Generallyspeaking, we chose secondary antibodies to be used from differentspecies as the first antibody they were combinedwith. Rabbit antiseraaCLDN18.2/n-term (generated at Eurogentec) and anti-CLDN18/mid(purchased from Zymed) were used on fixed paraffin embeddedtissue in combination with the second antibody Powervision Poly HRPgoat-anti-rabbit (ImmunoLogic). anti-CLDN18/mid was the only oneamong the four antibodies which was cross-reactive with CLDN18.1.Mouse IgG2a muMAB362 and its chimerized human IgG1 variantiMAB362, which recognize a complex conformational epitope wereused for staining of cryo sections. Here, Powervision Poly HRP goat-anti-mouse (ImmunoLogic) and Rabbit-anti-human IgG (JacksonImmunoResearch Laboratories) followed by incubation with Power-vision Poly HRP goat-anti-rabbit (ImmunoLogic), respectively wereused as second antibodies. None of these first/second antibody setswas used for staining of tissues from species from which one of thetwo antibodies were derived.

The anti-CD20 antibody Rituximab was used as negative control.For some experiments, antibodies were conjugated with FITC (bySquarix) and used in combination with Rabbit-anti-FITC-HRP (AbDSerotec) as second antibody, again to prevent unspecific staining bydirect binding of the secondary antibodies to tissue sections.

2.5. Immunohistochemistry

Tissue microarrays were either prepared by standard techniquesfrom frozen (cryo) or formalin fixed paraffin-embedded (FFPE) tissuesections or purchased from the companies Biomax (Rockville, USA)and Biochain (Hayward, USA). Commercial tissue microarrays weretreated according to the suppliers' instructions prior to antibodytreatment.

Sections of FFPE tissues were deparaffinized and treated for 30 minat 95 °C with DakoCytomation Target Retrieval solution (DakoCyto-mation, Glostrup, Denmark) for epitope retrieval. Cryo sections werefixed for 10 min in cold acetone at −20 °C, air-dried and eitherdirectly processed or stored at −80 °C. Sections were rehydrated inPBS before staining. Staining of both FFPE and cryo sections wasperformed according to the same protocol. Endogenous peroxidaseswere quenched by incubation in 0.3% H2O2 in PBS. After washingwith PBS, unspecific antibody binding sites were blocked withblocking buffer (10% goat serum in PBS) followed by incubationwith the primary antibody diluted in blocking buffer. Samples werethen washed 3 times with PBS and incubated with the respectivesecondary antibody. Visualization was performed using the peroxi-dase substrates DAB (DakoCytomation) or Nova Red substrate kit(Vector Laboratories, Burlingame, CA, USA). After counterstainingwith Hematoxylin, dehydration andmounting sections were analyzedmicroscopically.

embl IDs. Isoforms 1 and 2 nomenclature are standardized according to the human

Isoform 2

Transcript Protein

3 NM_001002026 NP_0010020260000011114 ENSCAFT00000011985 ENSCAFP0000001111300000012607 ENSMMUT00000013453 ENSMMUP000000126059 NM_001194921 NP_0011818500000018995 ENSOCUT00000014966 ENSOCUP000000128624118 ENSRNOT00000044524 ENSRNOP00000041474

Fig. 2. Conservation of protein sequence and composition of CLDN18 isoforms across mammals. Alignment of the Claudin18.1 and Claudin18.2 protein sequences of primate and mammalian species. Transmembrane domains (TMs) 1–4 areshown in gray. The consensus claudin family signature located in the first extracellular loop (ECL1) is boxed. Negative residues in ECL1 are in red, positive charged residues in green. Conserved cysteines are depicted in yellow.

87Ö.Türeci

etal./

Gene

481(2011)

83–92

88 Ö. Türeci et al. / Gene 481 (2011) 83–92

3. Results

3.1. Gene structure and transcript sequence of CLDN18 orthologs

The human CLDN18 gene locus on chromosome 3q22 coversapproximately 35 kb and is organized in 6 exons and 5 introns(Fig. 1A). Alternative usage of exons 1a and 1b gives rise to the twoisoforms CLDN18.1 (RefSeq ID: NM_016369, NP_057453) andCLDN18.2 (RefSeq ID: NM_001002026, NP_001002026), which codefor closely related proteins (Fig. 1A and B top). To gain insight into theevolutionary conservation of Claudin-18, we used the human CLDN18protein to search for orthologous proteins in macaque, mouse, rat,dog, and rabbit utilizing sequence similarity search and genomealignments. Starting with the human CLDN18 protein sequences to

Fig. 3. Conserved cell lineage specificity of CLDN18 isoforms. Tissue distribution of CLDN18PCR. NTC; no template control. (C) Immunohistochemistry with FITC-labeled splice(D) Immunohistochemical analysis of CLDN18 protein expression with anti-claudin-18/miregion of the stomach were stained with anti-claudin-18/n-term antiserum.

predict and annotate the gene structures of orthologs, wewere able tofind both isoforms in all species (Table 1). Interspecies sequenceconservation can be computed using a probabilistic model thatdescribes the process of DNA substitution at each site (Siepel et al.,2005). This summarization of conservation can be viewed using the“Conservation by PhastCons” genome track found in the UCSCGenome Browser (Kent et al., 2002) and allowed us to examineinterspecies conservation (Fig. 1B middle). Additionally, we analyzedthe alignments of individual species genomes to the human genome,generated through the “multiz” algorithm that performs a progressivegenome-genome alignment to generate the best-in-genome pairwisealignment (Blanchette et al., 2004). We exploited this tool to obtaincross-species “multiz” alignments for selected representatives ofmammals (Pan troglodytes, Macaca mulatta, Mus musculus, Rattus

isoforms in different species was analyzed by (A) end point and by (B) quantitative RT--variant 2 specific antibody IMAB362. Rituximab was used as negative control.d antiserum, which is reactive with both splice variants. (E) Epithelia from the antral

Fig. 3 (continued).

89Ö. Türeci et al. / Gene 481 (2011) 83–92

norvegicus, Oryctolagus cuniculus, Canis familiaris) and vertebrates(chicken, lizard, Xenopus tropicalis) (Fig. 1B, lower, “Multiz Align-ments of 46 Vertebrates”).

Across the entire locus these analyses revealed higher conserva-tion across primates, some conservation in mammals, and lessin chicken, lizard and frog. The intronic and intergenic regions ofchicken, lizard and frog have no sequence conservation to human. Allexons and thus the entire coding region, in contrast, show high levelsof conservation, which extends to chicken, lizard and frog. Forexample, between the human and rabbit coding regions of CLDN18.2,243 of 261 aa residues (94%) are identical, 254 aa residues are similarwithout any gaps. The protein coding region of CLDN18.2 exon 1contains 220 nucleotides. Between human and rabbit, 71 of 73 aaresidues are identical (97%). At the nucleotide level, there are noinsertions or deletions in this alternative exon across human,macaque, mouse, rat, dog and rabbit, and 81% of the nucleotides areidentical. The CLDN18.1 first exon shows a similarly high level ofconservation, with only four non-identical residues and no gapsbetween human and rabbit, for example. In summary, the CLDN18isoforms are phylogenetically highly conserved with regard to genestructure and nucleotide sequence.

3.2. Promoter regions of CLDN18 orthologs

We have previously reported that the activation of humanCLDN18.2 depends on the binding of the transcription factor cyclicAMP–responsive element binding protein (CREB) to its unmethylatedconsensus site within a CpG island (Sahin et al., 2008). To determinewhether this mechanism plays a role for regulation of other CLDN18.2orthologs, we fragmented the CLDN18.2 promoter along lines ofevolutionary conservation. A well-conserved 180 bp hypotheticalcore promoter exists immediately upstream of the transcription startsite and the overall sequence conservation shows a sharp decline

further upstream (Fig. 1B). We analyzed this region and found highconservation even to lizards and chicken. Moreover, sequencealignment shows that among the six primates and mammals, theCREB binding site TGACGTG is perfectly conserved, as is an unusuallyhigh number of nucleotides (37 of 55) between the start of the CREBbinding site and the transcription start site.

Interestingly, the lung-specific isoform 1 of mouse CLDN18 hasbeen shown to be activated by T/EBP/NKX2.1 (Niimi et al., 2001), ahomeodomain-containing transcription factor that is restricted toalveolar type II and Clara cells of the lung and to cells in brain andthyroid. We found that the two T/EBP/NKX2.1 binding sites reportedas being responsible for activation of this gene in lung cells are alsopresent in promoter regions of all investigated orthologs (Fig. 1C).Moreover, the entire 280 bp sequence region proximal to thetranscription start is nearly identical for all orthologous genes.

3.3. Protein sequences of CLDN18 orthologs

Next, we compared protein sequences of orthologs from mouse,rat, rabbit, dog and primates for the splice variants of human CLDN18(Fig. 2) and found an overall sequence homology higher than 84%across orthologs of each isoform.

Overall sequence identity between human CLDN18.1 andCLDN18.2 is 91%, with the differences restricted to the N-terminusthat is encoded by exons 1a and 1b. The 73 amino acid sequenceencoded by CLDN18.2 exon 1a is about 95% identical across the sixspecies, as is the N-terminal sequence region encoded by CLDN18.1exon 1b.

Both splice variants of human CLDN18 harbor four hydrophobictransmembrane (TM) domains, a short intracellular N-terminal se-quence, two extracellular loops (ECL) of 73 and 23 aa and a cytoplasmicC-terminal sequence (71 aa). This overall structure and the framing ofaforementioned elements, which dictate the tetraspanin membrane

Fig. 3 (continued).

90 Ö. Türeci et al. / Gene 481 (2011) 83–92

topology of CLDN18, are conserved in all analyzed orthologs (Fig. 2).Only in themouse sequence is a four amino acid insertion, located in thesecond extracellular domain.

Most interestingly, other essential sequence elements were alsorepresented in orthologs. These include the consensus claudin familysignature [GN]-L-W-x(2)-C-x(7,9)-[STDENQH]-C located in the firstECL, which harbors two cysteines (aa 52 and 63) that potentially forman intramolecular disulfide bond to stabilize protein conformation(Angelow and Yu, 2009). Moreover, the two conserved cysteines atpositions 103 and 193, which serve as palmitoylation sites, as well asthe C-terminal PDZ binding motif, which mediates direct interactionwith tight junction-associated cytoplasmic proteins such as ZO-1, -2,and -3 or multi-PDZ domain protein (MUPP-1), are also conserved.

The larger first ECL of claudin family members affects theparacellular tightness and the selective ion permeability (Krause et al.,2009) whereas the shorter second ECL is thought to be responsible for

narrowing of the paracellular cleft and has a holding function betweenthe opposing cell membranes. When we compared human CLDN18splice variants, we found a considerable difference in the total numberand distribution of negatively and positively charged residues in ECL1 ofCLDN18.1 and CLDN18.2, suggesting differences in selective ionpermeability (Fig. 2). Interestingly, in orthologs of each splice variantthe pattern of charged amino acids was found to be conserved.

Altogether, this data implies conservation of the cellular functionof CLDN18 variants in mammals and primates.

3.4. Tissue distribution of CLDN18 orthologs

To determine experimentally, whether cell type specificity ofCLDN18.2 expression has evolutionary evolved, we assayed a compre-hensive set of more than 30 different tissue and organ types in sixmammals.

91Ö. Türeci et al. / Gene 481 (2011) 83–92

CLDN18.2 transcript-specific expression was analyzed both by endpoint (Fig. 3A) aswell as quantitativeRT-PCR (Fig. 3B).Multiple samplesin up to 34different tissue types per specieswere investigated (Table 2).We classified CLDN18.2 as present when the signal was reproducibleand greater than 1/10 of the maximum signal obtained in gastricmucosa. Meta-analysis of these data showed expression to be restrictedto gastric mucosa in the primates and mammals we studied. Splicevariant 1 of CLDN18, in contrast, was transcriptionally activated only inlung tissue of all species.

Exclusive expression of CLDN18.2 in stomach was confirmed onprotein level by screening up to 16 different tissues per species byimmunohistochemistry. For human, monkey and mouse tissues threedifferent antibodies were used to confirm data with independent

Table 2Expression of CLDN18.2 transcript and protein orthologs as assayed by RT-PCR and immunohsummarized in this table (and others as independent confirmation for data not shown hereterm (FFPE), muMab362 and IMAB362-FITC (cryo) and data were combined in the table.stained with IMAB362 (cryo).

Human Monkey Dog

RT-PCR IHC RT-PCR IHC RT-PCR

Abdominal organsEsophagus 0/6 0/5 0/2 0/2 0/1Stomach 10/11 16/16 2/2 3/3 1/1Intestine – 0/3 0/1 0/5 0/1Colon 0/3 0/13 0/2 0/2 0/1Liver 0/4 0/4 0/2 0/3 0/1Pancreas 0/5 0/4 0/2 0/2 0/1Gall bladder – 0/3 0/2 0/2 –

Urogenital organsBreast 0/5 0/7 0/2 0/2 –

Ovary 0/5 0/7 0/1 0/1 –

Kidney 0/4 0/13 0/2 0/3 0/1Testis 0/4 0/3 0/1 0/1 –

Cervix 0/2 0/2 0/1 – –

Placenta 0/3 0/3 – – –

Uterus 0/4 0/1 0/1 0/1 –

Fallopian tube – – 0/1 0/1 –

Urinary bladder 0/5 0/3 0/2 0/2 –

Ureter – – 0/2 0/2 –

Prostate 0/5 0/6 0/1 0/1 –

Endometrium – 0/3 – – –

Cardiopulmonary systemLung 0/10 0/5 0/2 0/3 0/1Heart 0/5 0/3 0/1 0/2 0/1Endothelia (aorta) – 0/2 0/2 0/2 0/1

Nervous systemTotal brain 0/5 – – – 0/1Cerebellum 0/1 0/3 0/2 0/2 –

Cerebrum – 0/2 0/2 0/2 –

Pituitary gland 0/1 0/3 0/2 0/2 –

Cerebral cortex 0/1 0/4 0/2 0/2 0/1Spinal cord – – 0/1 0/2 –

Retina 0/1 0/3 0/2 0/2 –

Inner ear – – 0/2 0/2 –

Hematopoietic systemBone marrow 0/1 0/5 1/2 0/2 –

Lymph nodes 0/5 0/3 0/4 0/4 –

Thymus 1/2 – 0/2 0/2 –

Tonsil 0/2 – – – –

Spleen 0/3 0/3 0/2 0/3 –

Blood cells 0/5 – – 0/2 –

OthersAdrenal gland – 0/3 0/2 0/2 –

Salivary gland – 0/3 – – –

Sceletal muscle 0/4 0/3 – 0/3 0/1Thyroid 0/3 0/3 0/2 0/2 –

Skin 0/3 0/3 0/1 0/2 0/1

reagents. A selection of IHC data obtained with CLDN18 splice variant2 specific antibodies is summarized in Table 2 and exemplified inFig. 3C. Both clearly confirm restriction to gastric tissue. The isoform-cross-reactive antibody anti-CLDN18/mid, in contrast, additionallystained epithelial cells of lung tissue (Fig. 3D).

We previously reported that in human stomach mucosa CLDN18.2staining is observed in differentiated exocrine and endocrine cells ofgastric glands and not in cells of the neck zone harboring gastric stemcells (Sahin et al., 2008). We found a similar zonated staining patternin monkey and mouse stomach tissue sections (Fig. 3E).

Conservation of tissue expression pattern is in line with ourfinding that promoter gene regions and sites essential for transcrip-tional regulation of CLDN18 are highly homologous in all orthologs.

istochemistry. The following splice variant 2 specific antibodies were used for IHC dataor exemplified in Fig.3): Human and monkey tissues were stained with anti-CLDN18/n-Rabbit tissue was stained with muMAB362, whereas dog, mouse and rat tissues were

Rabbit Rat Mouse

IHC RT-PCR IHC RT-PCR IHC RT-PCR IHC

0/1 1/1 0/1 0/1 0/1 0/2 0/11/1 1/1 1/1 1/1 1/1 2/2 1/10/1 0/1 0/1 0/1 0/1 0/2 0/10/1 0/1 0/1 0/1 0/1 0/2 0/10/1 0/1 0/1 0/1 0/1 0/2 0/10/1 0/1 0/1 0/1 0/1 0/1 0/1– – – – – 0/2 –

– 0/1 0/1 – 0/1 – –

– 0/1 0/1 0/1 0/1 0/1 0/10/1 0/1 0/1 0/1 0/1 0/2 0/1– 0/1 0/1 0/1 0/1 0/1 0/1– – – – – – –

– – – – – – –

– 0/1 0/1 0/1 – 0/1– – – – – –

– 0/1 0/1 0/1 0/1 0/2 0/1– 0/1 0/1 0/1 0/1 – –

– 0/1 0/1 0/1 0/1 – 0/1– – – – – – –

0/1 0/1 0/1 0/1 0/1 0/2 0/10/1 0/1 0/1 0/1 0/1 0/2 0/10/1 0/1 0/1 0/1 0/1 – 0/1

0/1 – 0/1 0/1 – – –

0/1 – – 0/1 0/1 0/2 0/1– – – – – 0/2 –

– – – – – – –

– – – 0/1 – – –

– – – 0/1 0/1 – 0/1– 0/1 0/1 0/1 0/1 – 0/1– – 0/1 – 0/1 0/1

0/1 – – 0/1 0/1 – 0/1– – – 0/1 0/1 0/2 0/1– 0/1 0/1 0/1 0/1 0/2 0/1– – – – – –

– 0/1 0/1 0/1 0/1 0/2 0/10/1 0/1 0/1 0/1 0/1 0/1

0/1 – – 0/1 0/1 – –

– – – – – – –

0/1 0/1 0/1 0/1 0/1 0/2 0/1– 0/1 0/1 0/1 0/1 – 0/10/1 0/1 0/1 0/1 0/1 0/2 0/1

92 Ö. Türeci et al. / Gene 481 (2011) 83–92

4. Discussion

The findings described in this study are of relevance for two areasof investigation.

One is the analysis of gene expression evolution, of which theunderlying patterns are poorly understood. The advancement of high-throughput technologies for oligonucleotide microarrays and se-quencing have enabled characterization of the expression of thou-sands of genes simultaneously, opening the door for studies into thegeneral principles of gene expression evolution (Brazma and Vilo,2000; Holter et al., 2001). Reported findings with regard to disparityof orthologs, however, are ambiguous (Liao and Zhang, 2006; Yanaiet al., 2004). There are in fact genes which, consistent with commonsense, are conserved across species with regard to coding sequence,function as well as tissue distribution (Li et al., 1997). Others incontrast, differ from their orthologous counterpart largely and are f.e.activated in entirely different tissues (Pao et al., 2006). Most likely,sequence and expression profile of different categories of genesevolve under different phylogenetic constraints and there is nogeneral rule. Claudins are particularly interestingmodel genes, as theyhave to fulfill two seemingly contradictory features. Being buildingblocks of tight junctions and thus conferring essential housekeepingfunctions, claudins have to be expressed throughout all epithelialcells regardless of their lineage. On the other hand, their housekeeping function has to manifest itself in a tissue-specific way (f.e.distinct ion and solute permeabilities). These requirements aremet bya combinatorial claudin code, as claudins constitute a family withmultiple members, which occur in various tissue-specific combina-tions. Our data implies with CLDN18 as a model that specialization ofmembers of such gene families occurs early in evolution of mammals.

The second area impacted by our data is drug development.CLDN18.2 is frequently activated in various cancer types, is absentfrom indispensible and vital normal organs, is expressed on the cellsurface and belongs to a gene family, which is involved into tumor-promoting cellular functions. All this makes CLDN18.2 an attractivetarget for antibody based therapies. Two drug development programsare underway. One features iMAB362, a recombinant chimericmonoclonal antibody against CLDN18, which is highly specific forsplice variant 2 (Sahin et al., 2008). This antibody has been shown toexecute potent antitumoral activity in vitro and in animal models(manuscript in preparation) and is currently in clinical development.The other uses virus-like particles decorated with the extra-cellularloop of CLDN18.2 as vaccine to induce immune responses recognizingCLDN18.2 expressing cancer cells (Klamp et al., 2011) and is currentlyevaluated for the pre-clinical proof of efficacy.

According to regulatory guidelines development of such targeteddrugs involves testing of their safety in a relevant animal species.Relevant animal species are those in which the test material ispharmacologically active (ICH S6 CPMP/ICH/302/95). Minimal require-ments are that animals express the target epitope and demonstrate asimilar tissue distribution profile as humans. Our findings show highconservation of the CLDN18.2 sequence, functional domains andexpression,with stomachas the only potential on-target toxicity relevantorgan. Together with the notion that tight junction molecules have thesame functional role across species necessary (though not sufficient)conditions for suitabilityof thesemammals as relevantpreclinicalmodelsfor safety pharmacology are fulfilled.

Acknowledgements

We thank Gunda Brandenburg and Dirk Usener for their assistance.

References

Anderson, J.M., 2001. Molecular structure of tight junctions and their role in epithelialtransport. News Physiol. Sci. 16, 126–130.

Angelow, S., Yu, A.S., 2009. Structure-function studies of claudin extracellular domainsby cysteine-scanning mutagenesis. J. Biol. Chem. 284, 29205–29217.

Blanchette, M., Kent, W.J., Riemer, C., Elnitski, L., Smit, A.F., Roskin, K.M., Baertsch, R.,Rosenbloom, K., Clawson, H., Green, E.D., Haussler, D., Miller, W., 2004. Aligningmultiple genomic sequences with the threaded blockset aligner. Genome Res. 14,708–715.

Brazma, A., Vilo, J., 2000. Gene expression data analysis. FEBS Lett. 480, 17–24.Farquhar, M.G., Palade, G.E., 1963. Junctional complexes in various epithelia. J. Cell Biol.

17, 375–412.Fujita, H., Chiba, H., Yokozaki, H., Sakai, N., Sugimoto, K., Wada, T., Kojima, T., Yamashita,

T., Sawada, N., 2006. Differential expression and subcellular localization of claudin-7, -8, -12, -13, and -15 along the mouse intestine. J. Histochem. Cytochem. 54,933–944.

Furuse, M., Fujita, K., Hiiragi, T., Fujimoto, K., Tsukita, S., 1998. Claudin-1 and -2: novelintegral membrane proteins localizing at tight junctions with no sequencesimilarity to occludin. J. Cell Biol. 141, 1539–1550.

Gonzalez-Mariscal, L., Betanzos, A., Nava, P., Jaramillo, B.E., 2003. Tight junctionproteins. Prog. Biophys. Mol. Biol. 81, 1–44.

Grunwald, C., Koslowski, M., Arsiray, T., Dhaene, K., Praet, M., Victor, A., Morresi-Hauf,A., Lindner, M., Passlick, B., Lehr, H.A., Schafer, S.C., Seitz, G., Huber, C., Sahin, U.,Tureci, O., 2006. Expression of multiple epigenetically regulated cancer/germlinegenes in nonsmall cell lung cancer. Int. J. Cancer 118, 2522–2528.

Holter, N.S., Maritan, A., Cieplak, M., Fedoroff, N.V., Banavar, J.R., 2001. Dynamicmodeling of gene expression data. Proc. Natl. Acad. Sci. U. S. A. 98, 1693–1698.

Kent, W.J., Sugnet, C.W., Furey, T.S., Roskin, K.M., Pringle, T.H., Zahler, A.M., Haussler, D.,2002. The human genome browser at UCSC. Genome Res. 12, 996–1006.

Klamp, T., Schumacher, J., Huber, G., Kuehne, C., Meissner, U., Selmi, A., Hiller, T., Kreiter,S., Tuereci, O., Sahin, U., 2011. Highly specific auto-antibodies against claudin-18isoform 2 induced by a chimeric HBcAg virus-like particle vaccine kill tumor cellsand inhibit growth of lung metastases. Cancer Res., accepted for publication.

Kollmar, R., Nakamura, S.K., Kappler, J.A., Hudspeth, A.J., 2001. Expression andphylogeny of claudins in vertebrate primordia. Proc. Natl. Acad. Sci. U. S. A. 98,10196–10201.

Koslowski, M., Sahin, U., Huber, C., Tureci, O., 2006. The human X chromosome isenriched for germline genes expressed in premeiotic germ cells of both sexes. Hum.Mol. Genet. 15, 2392–2399.

Krause, G., Winkler, L., Mueller, S.L., Haseloff, R.F., Piontek, J., Blasig, I.E., 2008. Structureand function of claudins. Biochim. Biophys. Acta 1778, 631–645.

Krause, G., Winkler, L., Piehl, C., Blasig, I., Piontek, J., Muller, S.L., 2009. Structure andfunction of extracellular claudin domains. Ann. N.Y. Acad. Sci. 1165, 34–43.

Li, J., Rhodes, J.C., Askew, D.S., 1997. Evolutionary conservation of putative functionaldomains in the human homolog of the murine His-1 gene. Gene 184, 169–176.

Liao, B.Y., Zhang, J., 2006. Evolutionary conservation of expression profiles betweenhuman and mouse orthologous genes. Mol. Biol. Evol. 23, 530–540.

Mitic, L.L., Van Itallie, C.M., Anderson, J.M., 2000. Molecular physiology andpathophysiology of tight junctions I. Tight junction structure and function: lessonsfrom mutant animals and proteins. Am. J. Physiol. Gastrointest. Liver Physiol. 279,G250–G254.

Morita, K., Furuse, M., Fujimoto, K., Tsukita, S., 1999. Claudin multigene family encodingfour-transmembrane domain protein components of tight junction strands. Proc.Natl. Acad. Sci. U. S. A. 96, 511–516.

Niimi, T., Nagashima, K., Ward, J.M., Minoo, P., Zimonjic, D.B., Popescu, N.C., Kimura, S.,2001. Claudin-18, a novel downstream target gene for the T/EBP/NKX2.1 home-odomain transcription factor, encodes lung- and stomach-specific isoformsthrough alternative splicing. Mol. Cell. Biol. 21, 7380–7390.

Pao, S.Y., Lin, W.L., Hwang, M.J., 2006. In silico identification and comparative analysisof differentially expressed genes in human and mouse tissues. BMC Genomics 7,86.

Rahner, C., Mitic, L.L., Anderson, J.M., 2001. Heterogeneity in expression and subcellularlocalization of claudins 2, 3, 4, and 5 in the rat liver, pancreas, and gut.Gastroenterology 120, 411–422.

Sahin, U., Koslowski, M., Dhaene, K., Usener, D., Brandenburg, G., Seitz, G., Huber, C.,Tureci, O., 2008. Claudin-18 splice variant 2 is a pan-cancer target suitable fortherapeutic antibody development. Clin. Cancer Res. 14, 7624–7634.

Siepel, A., Bejerano, G., Pedersen, J.S., Hinrichs, A.S., Hou, M., Rosenbloom, K., Clawson,H., Spieth, J., Hillier, L.W., Richards, S., Weinstock, G.M., Wilson, R.K., Gibbs, R.A.,Kent, W.J., Miller, W., Haussler, D., 2005. Evolutionarily conserved elements invertebrate, insect, worm, and yeast genomes. Genome Res. 15, 1034–1050.

Smetana, H., 1947. The permeability of the renal glomeruli of several mammalianspecies to labelled proteins. Am. J. Pathol. 23, 255–267.

Stevenson, B.R., Anderson, J.M., Bullivant, S., 1988. The epithelial tight junction:structure, function and preliminary biochemical characterization. Mol. Cell.Biochem. 83, 129–145.

Yanai, I., Graur, D., Ophir, R., 2004. Incongruent expression profiles between human andmouse orthologous genes suggest widespread neutral evolution of transcriptioncontrol. OMICS 8, 15–24.