Epithelial tight junction proteins as potential antibody targets for pancarcinoma therapy

Abstract

Recombinant monoclonal antibodies are beginning to revolutionize cancer therapy. In combination with standard chemotherapy, high response rates have been reported with antibodies of the human IgG1 isotype for treatment of non-Hodgkin's lymphoma and breast cancer. It is becoming apparent that targets for antibody-based therapies do not necessarily need to be absent from normal tissues but can be present there either in low copy numbers or with binding epitopes shielded from the therapeutic antibody. Here, we studied whether claudin proteins that form tight junctions in normal epithelia are still expressed on carcinoma cells and whether their extracellular domains can be recognized by antibodies. We show that mRNAs of claudins 1, 3, 4, and 7 are all expressed in different human carcinoma cell lines, while claudin 8 was selectively expressed in breast and pancreas cancer lines. Chicken polyclonal antibodies were raised against peptides contained within predicted extracellular domains of claudins 1, 3, and 4. Affinity-purified IgG fractions for claudins 3 and 4 were monospecific and bound to human breast and colon carcinoma lines, but not to a line of monocytic origin. Claudin 3 antibodies also homogeneously stained human renal cell carcinoma tissue and micrometastatic tumor cells as identified by cytokeratin staining in bone marrow biopsies of breast cancer patients. Fluorescence-activated cell sorting and immunocytochemistry indicated that claudin antibodies bound to the surface of tumor cells. By analogy to other tumor-associated antigens that are differentially accessible to antibodies on tumor vs normal tissue, we propose that certain claudin proteins have potential as targets for novel antibody-based therapies of carcinomas.

Fig. 1

Expression of claudin 1, 3, 4, 7, and 8 mRNAs in various cancer cell lines. Products from RT-PCR reactions were separated on agrose gels, and DNA stained by ethidium bromide. Ep-CAM expression was included as epithelial marker, and β-actin expression to document comparability and quality of samples. The arrow heads mark the 500-bp position within the DNA size markers added on left and right most lanes. A reagent control was done in the penultimate lane.

Fig. 2

a Schematic model for the membrane arrangement and orientation of claudin proteins according to Tsukita [60]. b Primary sequence alignment of claudins 1–18. Italics putative transmembrane domains; bold putative extracellular domains. The numbers in brackets give the amino acid positions of the first amino acid. Underlined amino acids mark highly conserved residues. c Alignment of protein sequences of human and chicken (gallus) claudins 3 and 4. Sequences for human and chicken claudins were obtained from the Protein Database of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov). Sequence of chicken claudin 4 is only known in part (amino acid 1–56) which was aligned to human claudin 4. Differing amino acids in human and chicken sequences are boxed, changes to similar amino acids are underlined. The amino acid identities between claudins 3 and 4 of human and chicken origin were 73.6% and 78.6%, respectively. Sequences for chicken claudins 1, 7, and 8 were not available in any database.

Fig. 3

Cell surface expression claudins 3 and 4 on cancer cell lines. FACS histograms using the colon cancer line SW480 (a), breast cancer line CAMA-1 (b), and monocytic line U937 (c) for surface staining with chicken anti-claudin peptide IgG CLDN1.1, 3.2, 3.3, and 4.3 are shown (see also Table 1). An anti-human Ep-CAM mouse monoclonal antibody 3B10 was used as positive control (upper panel). Faint line IgG of a nonimmunized chicken; bold line chicken anti-claudin IgG or anti-Ep-CAM mab.

Fig. 4

Binding of chicken anti-claudin IgGs to vital SW480 and CAMA-1 cells. Bright field microscopic images are depicted in the first and third vertical panels. Immunofluorescence images are shown in the second and fourth vertical panels. Staining was performed with chicken anti-claudin IgG against peptides CLDN3.2, 3.3, and 4.3, as shown on the left. Bound chicken IgG was visualized with Cy3-conjugated rabbit anti-chicken IgG. IgG from a nonimmunized chicken egg served as negative control. Cell staining was analyzed by immunofluorescence using a phase contrast microscope.

Fig. 5

Immunohistochemical staining of frozen tissues with chicken anti-human claudin IgG. Normal tissue from human kidney (right panel) was stained with mouse anti-Ep-CAM (3B10) antibody and anti-claudin 3 chicken IgG (CLDN3.3). Human renal carcinoma tissue (left panel) was stained with anti-claudin 3 chicken IgG (CLDN3.2). IgG from a nonimmunized chicken egg served as negative controls. Cell staining was analyzed by phase contrast microscopy.

Fig. 6

Double staining of cells isolated from bone marrow aspirates of four mamma carcinoma patients for claudin 3 and cytokeratin. The upper panels show in red claudin- and in green cytokeratin-specific staining of bone marrow cells of breast carcinoma patients BM1117, BM412, BM520, BM573 with corresponding bright field images. Air-fixed slides were prepared from mononucleated cells and stained with chicken anti-CLDN3.3 peptide IgG, and the mouse anti-cytokeratin antibody A45/B/B3 detected by Cy3-labeled anti-chicken and FITC-labeled anti-mouse antibodies, respectively. The lower left panels show claudin 3- and cytokeratin-specific staining of CAMA-1 breast cancer cells as positive control and the corresponding bright fields. Chicken control IgG, and mouse MOPC21 antibody were used as negative controls. Exposure times with Cy3 filter were 100 ms, with FITC filter 500 ms, and bright field 1 ms.