The claudin gene family: expression in normal and neoplastic tissues

Abstract

Background: The claudin (CLDN) genes encode a family of proteins important in tight junction formation and function. Recently, it has become apparent that CLDN gene expression is frequently altered in several human cancers. However, the exact patterns of CLDN expression in various cancers is unknown, as only a limited number of CLDN genes have been investigated in a few tumors.

Methods: We identified all the human CLDN genes from Genbank and we used the large public SAGE database to ascertain the gene expression of all 21 CLDN in 266 normal and neoplastic tissues. Using real-time RT-PCR, we also surveyed a subset of 13 CLDN genes in 24 normal and 24 neoplastic tissues.

Results: We show that claudins represent a family of highly related proteins, with claudin-16, and -23 being the most different from the others. From in silico analysis and RT-PCR data, we find that most claudin genes appear decreased in cancer, while CLDN3, CLDN4, and CLDN7 are elevated in several malignancies such as those originating from the pancreas, bladder, thyroid, fallopian tubes, ovary, stomach, colon, breast, uterus, and the prostate. Interestingly, CLDN5 is highly expressed in vascular endothelial cells, providing a possible target for antiangiogenic therapy. CLDN18 might represent a biomarker for gastric cancer.

Conclusion: Our study confirms previously known CLDN gene expression patterns and identifies new ones, which may have applications in the detection, prognosis and therapy of several human cancers. In particular we identify several malignancies that express CLDN3 and CLDN4. These cancers may represent ideal candidates for a novel therapy being developed based on CPE, a toxin that specifically binds claudin-3 and claudin-4.

Figure 1

Multiple alignment and family tree of claudin proteins. A. ClustalW was used to generate alignment of all the human claudin protein sequences and the residues were then colored using Jalview [21] according to amino acid conservation among family members. Red bars underneath alignment indicate predicted transmembrane domains. The level of conservation and the consensus are also shown below the alignment. B. Phylogenetic tree of claudin family members. The tree was generated using ClustalW and visualize in Jalview. The numbers on the branches are Blosum62 scores, indicating distances among family members.

Figure 2

In silico analysis of claudin gene expression in various normal and neoplastic tissues. 266 SAGE libraries were examined for the expression of all 21 CLDN genes using SAGE Genie [23]. SAGE data was compiled from both normal and cancerous tissue, and analyzed using dChip software program, which assigned darker shades of red to higher number of tags. GAPDH and ACTN levels were also analyzed as controls. While some CLDN genes are ubiquitously expressed, others exhibit highly tissue specific patterns. A detailed spreadsheet of the data is available as supplementary information.

Figure 3

Real-time RT-PCR survey of CLDN gene expression in normal and neoplastic tissus. Expression data of 13 selected CLDN family members from 24 normal and 24 tumor tissue cDNAs. The dChip software was used to analyze and cluster the data. Clustering of the CLDN genes and the tissues were performed with distances based on 1-rank correlation and the centroid linkage method. Darker shades of red indicate higher CLDN expression.