Atic version of HT168; WM983B, cultured from a lymph node metastasis from the patient whose primary tumour gave rise to WM983A. Since CD44, as a cell surface glycoprotein, plays an important role in cell-matrix interaction, it was important to examine whether different matrix components change the alternative splicing pattern, or whether the ASP is stable and possibly inherent to melanoma-specific behavior. Therefore as a first step we determined the CD44 Tetracosactide web fingerprint of HT168M1 human melanoma cell line Thiazole Orange web growing in vitro on different matrices, namely fibronectin, laminin, collagen and matrigel. As shown in Fig. 5 after 48 hours incubation time the CD44 fingerprint was found to be unchanged in the case of every matrix type (Fig. 5). This fingerprint was found to be consistent through all examined cell lines growing on different matrices (only HT168M1 shown). It is interesting, that the fingerprint is retained 
in the cell lines derived from the primary tumours and their metastases alike (HT168 versus HT168M1 and WM983A versus WM983B).Modeling the Effects of the Microenvironment in vitroTo decide whether the in vitro melanoma CD44 fingerprint is maintained in vivo despite the influence of the microenvironment, we compared the CD44 splicing pattern of several, genetically different human melanoma cell lines (A2058, HT199, WM35, WM983A, M35) growing on plastic or different matrices. We also investigated HT168, a cell line cultured from the in vivoThe CD44 Melanoma Fingerprint in vivo in Our Animal ModelAs the in vivo microenvironment is far more complex than the influences of the extracellular matrix, we used an animal model to evaluate the CD44 melanoma fingerprint in vivo. This model has been developed by our group, following the observation that semiorthotopically (subcutaneously) implanted human melanomasCD44 Alternative Splicing Pattern of MelanomaFigure 2. Cloned PCR products from the 59 (exon 4, italic) and 39 (exon 16, bold) primer (squared) combination of CD44 in A2058 human melanoma cell line. Direct sequencing shows a CD44 isoform with no v1 or any other variable exons (A) as well as one with truncated v1 (underlined). doi:10.1371/journal.pone.0053883.galways formed metastases in newborn scid mice (permissive host), yet never did so in adult ones (nonpermissive host). This model made it possible to examine the melanoma `fingerprint’ during the metastatic processes. In vivo expression patterns were evaluated on two human melanoma cell lines HT199 and 15755315 HT168M1. We performed our PCR reaction series on theprimary subcutaneous tumour, circulating tumour cells obtained from blood and lung metastases from transplanted newborn scid mice, as well as the primary subcutaneous tumours from transplanted adult mice. In addition lung tumours were generated in adult animals by intravenous injection (Fig. S4). For HT199 we found that the CD44 fingerprint demonstrated in vitro was unchanged throughout the sampled sites (Fig. 6B). These findings do not explain published observations, that the expression of certain CD44 exons correlate with metastatic potential. Our results suggest that the CD44 ASP behind the `fingerprint’ is the same in all these cases, meaning that the same isoforms are present. The cited quantitative expression changes of single variable exons should therefore be explained differently.We made a further quantitative PCR analysis with our variable exon specific primers on the same samples. We examined the quantitative changes of the i.Atic version of HT168; WM983B, cultured from a lymph node metastasis from the patient whose primary tumour gave rise to WM983A. Since CD44, as a cell surface glycoprotein, plays an important role in cell-matrix interaction, it was important to examine whether different matrix components change the alternative splicing pattern, or whether the ASP is stable and possibly inherent to melanoma-specific behavior. Therefore as a first step we determined the CD44 fingerprint of HT168M1 human melanoma cell line growing in vitro on different matrices, namely fibronectin, laminin, collagen and matrigel. As shown in Fig. 5 after 48 hours incubation time the CD44 fingerprint was found to be unchanged in the case of every matrix type (Fig. 5). This fingerprint was found to be consistent through all examined cell lines growing on different matrices (only HT168M1 shown). It is interesting, that the fingerprint is retained in the cell lines derived from the primary tumours and their metastases alike (HT168 versus HT168M1 and WM983A versus WM983B).Modeling the Effects of the Microenvironment in vitroTo decide whether the in vitro melanoma CD44 fingerprint is maintained in vivo despite the influence of the microenvironment, we compared the CD44 splicing pattern of several, genetically different human melanoma cell lines (A2058, HT199, WM35, WM983A, M35) growing on plastic or different matrices. We also investigated HT168, a cell line cultured from the in vivoThe CD44 Melanoma Fingerprint in vivo in Our Animal ModelAs the in vivo microenvironment is far more complex than the influences of the extracellular matrix, we used an animal model to evaluate the CD44 melanoma fingerprint in vivo. This model has been developed by our group, following the observation that semiorthotopically (subcutaneously) implanted human melanomasCD44 Alternative Splicing Pattern of MelanomaFigure 2. Cloned PCR products from the 59 (exon 4, italic) and 39 (exon 16, bold) primer (squared) combination of CD44 in A2058 human melanoma cell line. Direct sequencing shows a CD44 isoform with no v1 or any other variable exons (A) as well as one with truncated v1 (underlined). doi:10.1371/journal.pone.0053883.galways formed metastases in newborn scid mice (permissive host), yet never did so in adult ones (nonpermissive host). This model made it possible to examine the melanoma `fingerprint’ during the metastatic processes. In vivo expression patterns were evaluated on two human melanoma cell lines HT199 and 15755315 HT168M1. We performed our PCR reaction series on theprimary subcutaneous tumour, circulating tumour cells obtained from blood and lung metastases from transplanted newborn scid mice, as well as the primary subcutaneous tumours from transplanted adult mice. In addition lung tumours were generated in adult animals by intravenous injection (Fig. S4). For HT199 we found that the CD44 fingerprint demonstrated in vitro was unchanged throughout the sampled sites (Fig. 6B). These findings do not explain published observations, that the expression of certain CD44 exons correlate with metastatic potential. Our results suggest that the CD44 ASP behind the `fingerprint’ is the same in all these cases, meaning that the same isoforms are present. The cited quantitative expression changes of single variable exons should therefore be explained differently.We made a further quantitative PCR analysis with our variable exon specific primers on the same samples. We examined the quantitative changes of the i.