Cathepsin B processes thyroglobulin under physiological conditions in the extracellular follicle lumen as well as in endo-lysosomal compartments which is followed by the release of thyroid hormones from the thyroid gland [15–17]. Hence, transport of cathepsin B to the apical plasma membrane domain of normal thyroid epithelial cells is a prerequisite for its TSH-stimulated secretion into the follicle lumen in order to maintain thyroid homeostasis . In pathological conditions, however, such as papillary thyroid carcinoma, cathepsin B has been localized to the basement membrane . Here we provide evidence that such re-routing of cathepsin B transport from apical-to-basolateral poles is a hallmark also of neoplastic cells in FTC (see Figure 2). Therefore, we propose that cathepsin B transport towards basal poles is characteristic for cells in both, papillary and follicular thyroid carcinoma, whereas an apical-directed transport that is characteristic for thyrocytes of normal thyroid tissue, is also still displayed in cells of PTC and FTC-derived tissue areas with intact follicle structures. This notion motivated us to analyze the pathways resulting in altered cathepsin B trafficking and leading to its secretion into the extrafollicular space, which most probably enhances the invasive potential of thyroid carcinoma cells due to cathepsin B's ability to degrade ECM components [1, 10–12, 35, 36].
The data achieved by 3-dimensional immunolocalization of endogenous cathepsin B and experiments employing activity based probes indicated that the thyroid carcinoma cell lines investigated in this study were characterized by cathepsin B trafficking that is destined to endo-lysosomes and, in addition, that cathepsin B is secreted into the extracellular space in a proteolytically active form (see Figures 4 and 5). Moreover, secretion of cathepsin B and related cysteine peptidases from KTC-1 and HTh74 cells was non-directed. We conclude that active cysteine peptidases are likely to reach extrafollicular locations in thyroid carcinoma tissue.
From the trafficking studies with GFP-tagged chimeras, it can be deduced that the active site mutant of cathepsin B, which is transport competent and reaches endo-lysosomes of FRT and HTh74 cells, is retained in the endoplasmic reticulum of KTC-1 and HTh7 cells (see Figures 6 and 7). Furthermore, eGFP-tagged wild type cathepsin B was retained in the Golgi of HTh7 cells. Hence, trafficking of cathepsin B is largely independent of signals intrinsic in the primary structure of the protease, rather transport pathways differ in the thyroid cell lines tested with trafficking defects being more prominent in the thyroid carcinoma cell lines KTC-1 and HTh7, while HTh74 cells remained transport competent and sorted cathepsin B into endo-lysosomes. This is likely to be the prerequisite for the massive secretion of cysteine peptidases like cathepsin B into the extracellular space of HTh74 cell cultures (see Figure 5) and it is likely to explain, why this cell line in particular has lost its contact inhibition and acquired an invasive phenotype.
Protease transport in mammalian cells
The molecular mechanisms underlying protease trafficking to their points-of-action have been studied in a variety of tissues and cell types [6, 27, 37–39]. However, the transport of proteases in mammalian cells is still not fully understood up until today . Cysteine cathepsins, i.e. endo-lysosomal proteases that may also act extracellularly, are synthesized as inactive pre-pro-enzymes at the rough ER (rER). There, the signal peptide is cleaved-off co-translationally. In the oxidizing milieu of the ER lumen, disulfide bridges are formed with the help of protein disulfide isomerase (PDI), an ER-resident enzyme, assisting in the correct folding of the proteins. Further, N-glycosylation of the synthesized proteases may be performed upon recognition of an Asn-X-Ser/Thr-Y motif (amino-acids given in three-letter code, with ‘X’ indicating any amino-acid, and ‘Y’ indicating any amino-acid except proline) by means of oligosaccharyl transferase. The pro-forms of cysteine cathepsins are further transported to the Golgi apparatus, where the zymogenes are modified in terms of N-linked oligosaccharide processing resulting in the addition of mannose-6-phosphate residues by a phosphotransferase and a phosphodiesterase. The mannose-6-phosphate tags are recognized in the trans-Golgi network (TGN) by highly specific mannose-6-phosphate receptors (M6P-R), which sort the pro-forms of cysteine proteases directly to the endo-lysosomal compartments . However, some M6P-tagged proteins like for instance thyroglobulin or the aspartic lysosomal protease cathepsin D, escape endo-lysosomal targeting in thyroid epithelial cells and become secreted instead [41, 42].
These results are similar to observations made in lysosomal storage diseases, such as I-cell disease, where it was shown that transport pathways of lysosomal enzymes may differ tremendously with respect to the cell type. For instance, I-cell disease patients lack an enzyme responsible for the addition of the M6P-tag, i.e. phosphotransferase, thus the lysosomal enzymes are not transported to the endo-lysosomal compartments, but become secreted . The mis-routing of lysosomal enzymes was also examined in fibroblasts isolated from mice deficient in M6P-receptors and displaying an I-cell disease-like phenotype . Interestingly, isolated hepatocytes from the same mice exhibited the complete set of enzymes within their endo-lysosomal compartments  highlighting that alternative pathways of endo-lysosomal targeting exist. Furthermore, it has been shown that cathepsin B can reach peripherally located vesicles in cancer cells by a pathway that is independent of M6P and most probably driven by sorting signals located within the pro-peptide region of the enzyme .
Hence, even though compelling evidence for alternative trafficking mechanisms has been published, the underlying sorting signals or alternative transport routes were not fully elucidated until today (for review see ). An excellent model system for the study of transport differences are cells which are characterized by distinct plasma membrane domains thus polarized into a basolateral and an apical plasma membrane domain. In order to elucidate the mechanisms that trigger apically or basolaterally-directed transport, Madin-Darby canine kidney cells (MDCK) or the thyroid epithelial cell line FRT have been intensively studied. Interestingly, FRT cells transport plasma membrane proteins to opposite cell poles as MDCK cells, even though both cell lines are polarized and display apparently morphological features of differentiated epithelial cells [27, 34, 47, 48]. Interestingly, the precise mechanisms that explain why e.g. transmembrane proteins are inserted into either the basolateral or the apical plasma membrane domain of MDCK or FRT cells, respectively, remain elusive. However, because thyrocytes are able to perform vesicular protein transport to opposite cell poles, they qualify as excellent models in order to study protein trafficking in epithelial cells.
GFP-tagging and activity based probes as tools to study protease trafficking in thyroid epithelial and carcinoma cells
Previously, we have constructed a mammalian expression vector encoding cathepsin B-eGFP that proved suitable for trafficking studies of cathepsin B in the fully differentiated and polarized rat thyroid epithelial cell line FRT as well as in TSH-responsive FRTL-5 cells . This vector can also be used to analyze cathepsin B transport in Chinese Hamster Ovary cells and in a number of other cell types indicating that eGFP tagging of cathepsin B does not grossly alter its trafficking in mammalian cells. More recently, our original pCathB-eGFP vector has been modified in the eGFP portion in order to improve signal-to-noise ratios  and it was sub-cloned into a modified plasmid for tissue-specific expression under the control of the A33-antigen promoter . In these cases, the cathepsin B-encoding sequence of the original vector was not altered.
In contrast, here we describe the construction of a vector coding for an inactive mutant counter-part of cathepsin B, in which the active site cysteine was substituted for an alanine. It was taken care to exchange cysteine with alanine instead of the more likely exchange of cysteine with serine (sulfhydryl side chain would then be exchanged by hydroxyl group), because we wanted to exclude the possibility of creating a serine protease-like protein by site directed mutagenesis of the cDNA coding for the cysteine peptidase cathepsin B. A serine exchange could have meant to create a catalytic dyad consisting of serine and histidine. Hence, our site-directed mutagenesis and cloning strategy aimed at the generation of an inactive enzyme with subtle changes in the active site cleft. The goal was to modify the primary structure of cathepsin B in such a way that the protein would still fold properly and thus, would not induce an unfolded protein response due to mis-folding and retention in the ER. In fact, these aims were achieved as is obvious from the observation that cathepsin B-C29A-eGFP chimeras proved fully transport-competent in the normal thyroid epithelial cell line FRT (see Figure 6), where it reached endo-lysosomes. In addition, the active site mutant counterpart of cathepsin B was sorted into endocytic compartments of the thyroid carcinoma cell line HTh74 (see Figure 7).
The cathepsin B-eGFP and cathepsin B-C29A-eGFP chimeric proteins were not only expressed in normal thyrocytes and in thyroid carcinoma cells, rather cathepsin trafficking was also investigated in combination with the activity based probe GB117 [13, 25] in order to specify its sorting into transport vesicles. Hence, several aspects of protease transport were addressed in this study. (i) We analyzed whether active cysteine proteases are directed to vesicles different from those that are reached by inactive proteases. Thus, mature enzymes would display specific sorting signals to direct them into distinct sub-cellular compartments. (ii) As an alternative explanation of re-routing of cathepsin B transport in thyroid carcinoma cells, it was tested whether cathepsin B can be transported differently when expressed in normal epithelial cells versus tumor-transformed cells. Thus, assuming that sorting of proteases is governed by the features of the different cell-types themselves.
We provide evidence for the notion that HTh74 cells, although representing anaplastic thyroid carcinoma cells, maintain transport competence and directed cathepsin B-eGFP in both versions, active and inactive, to endo-lysosomes. In this respect, HTh74 cells clearly resembled normal, non-transformed FRT cells that transported both chimeric proteins to identical destinations. However, the non-TSH receptor bearing anaplastic thyroid carcinoma cell line HTh7 and the papillary thyroid carcinoma cell line KTC-1 exhibited trafficking defects. Here inactive cathepsin B was retained within the ER and only the active cathepsin B-eGFP was transported further, i.e. up to the Golgi apparatus and to the endo-lysosomes, respectively.