Interplay between Podoplanin, CD44s and CD44v in Squamous Carcinoma Cells

Abstract

Podoplanin and CD44 are transmembrane glycoproteins involved in inflammation and cancer. In this paper, we report that podoplanin is coordinately expressed with the CD44 standard (CD44s) and variant (CD44v) isoforms in vivo-in hyperplastic skin after a pro-inflammatory stimulus with 12-O-tetradecanoylphorbol-13-acetate (TPA)-and in vitro-in cell lines representative of different stages of mouse-skin chemical carcinogenesis, as well as in human squamous carcinoma cell (SCC) lines. Moreover, we identify CD44v10 in the mouse-skin carcinogenesis model as the only CD44 variant isoform expressed in highly aggressive spindle carcinoma cell lines together with CD44s and podoplanin. We also characterized CD44v3-10, CD44v6-10 and CD44v8-10 as the major variant isoforms co-expressed with CD44s and podoplanin in human SCC cell lines. Immunofluorescence confocal microscopy experiments show that these CD44v isoforms colocalize with podoplanin at plasma membrane protrusions and cell-cell contacts of SCC cells, as previously reported for CD44s. Furthermore, CD44v isoforms colocalize with podoplanin in chemically induced mouse-skin SCCs in vivo. Co-immunoprecipitation experiments indicate that podoplanin physically binds to CD44v3-10, CD44v6-10 and CD44v8-10 isoforms, as well as to CD44s. Podoplanin-CD44 interaction is mediated by the transmembrane and cytosolic regions and is negatively modulated by glycosylation of the extracellular domain. These results point to a functional interplay of podoplanin with both CD44v and CD44s isoforms in SCCs and give insight into the regulation of the podoplanin-CD44 association.

Keywords: CD44s; CD44v; podoplanin; protein–protein interaction; squamous carcinoma cells.

Figure 1

12-O-tetradecanoylphorbol-13-acetate (TPA) induces coordinately podoplanin, CD44 standard (CD44s) and CD44 variant (CD44v) expression in mouse-skin. (A) Hematoxylin–eosin staining of skin sections at the indicated times after treatment with TPA; e—epidermis; d—dermis; hf—hair follicle; gr—granulation tissue; (B) Western-blot analysis of podoplanin (PDPN), CD44v and CD44s in whole skin at the indicated times after treatment with TPA. Treatment with vehicle (acetone) was also included as a control. Note that vehicle slightly induced podoplanin expression, likely as a result of a cold shock. GAPDH used as a control of protein loading. Results represent three independent experiments.

Figure 2

Colocalization of podoplanin with CD44v isoforms in chemically induced mouse-skin squamous carcinoma cell (SCC). Tumor skin sections were double stained with a rat mAb specific for podoplanin (PDPN) and rabbit polyclonal Abs recognizing sequences encoded by the (A) CD44 v3 or (B) v6 exon and appropriate secondary Abs. Nuclei stained with 4′,6-diamino-2-phenilindole (DAPI). Merge images show a colocalization of podoplanin and CD44 isoforms containing v3 or v6 exons in the tumors. Representative images from three different tumors in each case are shown.

Figure 3

Podoplanin is co-expressed with CD44v and CD44s in transformed mouse epidermal cell lines. CD44v10 is specifically expressed in spindle carcinoma cells. (A) RT–PCR analysis of podoplanin and CD44 mRNA expression in the cell lines. The asterisk indicates a transcript of ~380 bp specifically expressed in differentiated spindle cell carcinomas (SpCCs) cells that after isolation (right panel) and sequencing was identified as CD44v10. β-actin was used as a control of RNA loading; (B) Western-blot analysis of podoplanin and CD44 protein expression in the cell lines. Asterisk indicates a protein band putatively corresponding to CD44v10. β-actin and α-tubulin (blot on the right) were used as controls of protein loading. Letters below the blots indicate the phenotypes of the cell lines: E—epithelial; F—fibroblastic. Results represent three independent experiments.

Figure 4

Podoplanin is co-expressed with CD44v and CD44s isoforms in human SCC cell lines. (A) Schematic representation of the human CD44 gene indicating the positions of the primers used in the RT–PCR analysis. Primer positions are indicated by arrowheads. Numbers above constant (C) and variant (v) exons indicate their size in base pairs. Exon v1 is not represented since this sequence is not expressed in human cells [47]. EC—extracellular domain; TM—transmembrane domain; CT—cytosolic domain; (B) RT–PCR analysis of podoplanin and CD44 expression in the indicated cell lines. The pair of primers hs5-hs3 was used for CD44. GAPDH was used as a control of RNA loading; (C) RT–PCR analysis of CD44 expression in the indicated cell lines using primers hs5-hs3. Three main bands of ~1500, ~1013 and ~800 bp, denoted as 1, 2 and 3, respectively were observed; (D) RT–PCR analysis of CD44 expression in HN5 cells using primers C5–C9; (E) exon-specific RT–PCR analysis of CD44 expression in the indicated cell lines. v2–v10, 5′ primers were used in combination with 3′ primer hs3. Exon v3 is covered by two primers (v3a and v3b) because of the existence of an alternative splice acceptor site in this exon [48]; (F) Western-blot analysis of CD44 protein expression. α-tubulin was used as a control of protein loading; (G) Western-blot analysis of podoplanin protein expression. GAPDH was used as a control of protein loading. Data represent three independent experiments.

Figure 5

Colocalization of podoplanin with CD44v and CD44sC9 isoforms in SCC cells. HN5 cells were co-transfected with PDPN-mCherry and CD44v3-10-eGFP, CD44v8-10-eGFP, CD44v6-10-eGFP or CD44sC9-eGFP constructs and the subcellular distribution of the proteins studied by immunofluorescence and confocal microscopy. Grayscale images show the specific localization of each single protein while merged images are the result of GFP (green) and mCherry (red) combined images. Data represent 10–12 cells analyzed per condition from three independent experiments. Scalebar, 10 μm.

Figure 6

Podoplanin binds both CD44s and CD44v isoforms. (A) Western-blot analysis of constructs encoding the indicated CD44v and CD44s isoforms tagged with Ha after expression in HEK293T cells; (B) co-immunoprecipitation analysis of podoplanin and CD44 isoforms. HEK293T cells were co-transfected with FLAG-tagged human podoplanin and Ha-tagged human CD44v and CD44s isoforms. Lysates were immunoprecipitated with an ant-FLAG Ab and immunoprecipitates were run by SDS-PAGE and immunoblotted with an anti-Ha Ab (right panel). Lane 8L shows a longer exposure of CD44sC9 co-immunoprecipitate. Left panel shows the expression levels of the different CD44 constructs after transfection (input). Presence of podoplanin in the cell lysates and precipitates determined with an anti-podoplanin Ab (NZ1). Note CD44 proteins co-precipitated with podoplanin had lower sizes than fully glycosylated mature forms. Results represent three independent experiments.

Figure 7

Co-immunoprecipitation of podoplanin mutants with CD44s. (A) Schematic representation of podoplanin wild-type and mutant constructs used for co-immunoprecipitation assays. SP—signal peptide; EC—extracellular domain; TM—transmembrane domain; CT—cytosolic domain; ΔEC—the podoplanin EC domain was substituted by GFP; QN.N—positive charged residues (RK.R) in podoplanin juxtamembrane CT domain were substituted by uncharged polar amino acids (QN.N) to impede binding to ERM proteins; S/Tm—potential O-glycosylation Ser and Thr EC residues were stochastically mutated to Ala; PLAG3m—several residues, including Thr52, whose glycosylation was found to be crucial for podoplanin interaction with CLEC-2 were mutated to Ala; ΔPLAG3—the whole PLAG3 motif was deleted; TMCD45—the podoplanin TM region was substituted by that of CD45; G137L—the GXXXG motif (GIIVG) within the TM region involved in podoplanin clustering and association with lipid rafts was mutated to GXXXL; ΔCT—the whole podoplanin CT domain was deleted; (B) co-immunoprecipitation analysis of podoplanin wild-type and mutant forms with CD44s. HEK293T cells were co-transfected with FLAG-tagged human podoplanin constructs and Ha-tagged human CD44s. Lysates were immunoprecipitated with an ant-FLAG Ab, and immunoprecipitates were run by SDS-PAGE and immunoblotted with an anti-Ha Ab (right panel above). Left panel shows the expression levels of the different podoplanin constructs and CD44 after transfection (input). Presence of podoplanin in the cell lysates and precipitates was determined with an anti-podoplanin mAb (NZ1). Note that this mAb does not recognize the podoplanin constructs in which the PLAG3 domain is mutated (PLAG3m or ΔPLAG3) as previously reported [51]. Accordingly, the ΔEC mutants are neither detected by the NZ1 mAb but can be recognized by an anti-GFP Ab instead. Results represent three to four independent experiments.

Figure 8

Co-immunoprecipitation of podoplanin glycosylation and transmembrane mutants with CD44s. (A) Schematic representation of podoplanin extracellular (EC) domain showing the amino acid sequences of four regions found to be glycosylated. Ser and Thr residues within these regions were mutated to Ala to generate the indicated podoplanin mutant constructs; (B) schematic representation of podoplanin wild-type and TM mutants. The podoplanin TM region was substituted by those of synaptobrevin (SYN), glycophorin A (GPA) and ERBB2 to generate the indicated chimeric mutants; (C,D) co-immunoprecipitation analysis of CD44s and (C) podoplanin glycosylation and (D) chimeric TM mutant constructs. HEK293T cells co-transfected with FLAG-tagged human podoplanin constructs and Ha-tagged human CD44s. Lysates were immunoprecipitated with an ant-FLAG Ab and immunoprecipitates were run by SDS-PAGE and immunoblotted with an anti-Ha Ab (right panels above). Left panels show the expression levels of the different podoplanin constructs and CD44s after transfection (input). The presence of podoplanin in the cell lysates and precipitates was determined with an anti-podoplanin mAb (NZ1). Results represent three to four independent experiments.