Tumor Cell–Derived and Macrophage-Derived Cathepsin B Promotes Progression and Lung Metastasis of Mammary Cancer

Invasion and metastasis of cancer result from several interdependent processes in which proteolytic enzymes have been implicated (1). Among other tumor-related features, release of tumor-derived proteases is thought to facilitate the breakdown of basement membranes and extracellular matrix (ECM), thereby promoting cancer cell invasion into surrounding normal tissues. The proteolytic enzymes implicated in tumor progression and metastasis belong to four major catalytic classes of proteases: metalloproteinases [soluble and integral membrane matrix metalloproteinases (MMP), adamalysin-related disintegrin and metalloproteinases (ADAMS), or bone morphogenetic protein-1-type proteases], serine proteases, and cysteine- and aspartyl-type lysosomal proteases (2–4). Lysosomal cysteine proteases belong to the family of papain-like proteolytic enzymes (clan CA, family C1) principally localized subcellularly in the endosomal/lysosomal compartment. Seven of these proteases, cathepsins B (CTSB), C (dipeptidyl peptidase I), F, H, L, O, and X (alternative names: cathepsin Z; cathepsin P), exhibit ubiquitous but, nevertheless, differential expression in mammalian tissues, whereas other papain-like cysteine proteases (i.e., cathepsins J, K, S, V, and W) are only expressed by specific cell types (5, 6). Traditionally, lysosomal cysteine proteases are considered to execute nonspecific bulk proteolysis within the lysosomes (7). Yet, there is growing evidence for specific functions of these enzymes (8–11). Lysosomal proteases are secreted in significant quantities to function extracellularly in certain pathologic conditions and cancer (12). Most notably, early investigations identified CTSB as highly expressed in human breast carcinomas at the invasive fronts of tumors (13).

Furthermore, increased expression and proteolytic activity of lysosomal proteases (e.g., the aspartyl proteases cathepsin D and the cysteine proteases CTSB and cathepsin L) have often been positively correlated with poor prognosis for patients with a variety of malignancies, including mammary adenocarcinomas (14–17). The majority of experimental evidence for the involvement of lysosomal cysteine proteases in tumorigenesis has been obtained using ex vivo culture models (18–20). Recently, a study using a broad-spectrum inhibitor for cysteine cathepsins in the Rip1-Tag2 mouse model identified roles for cysteine cathepsins in angiogenesis and tumor growth and invasiveness within pancreatic islet tumors (21). Although these studies clearly confirm the importance of this class of proteases, they have not provided specific roles for CTSB in the progression and metastasis of cancer in vivo.

To address the role of CTSB in cancer growth and metastasis, we established a mouse mammary cancer model in the absence of CTSB by crossing CTSB-deficient mice (22) with a mammary cancer-susceptible strain of mice [polyoma middle T antigen (PyMT) mice], in which the expression of the PyMT is directed to the mammary epithelium by the mouse mammary tumor virus (MMTV)-long terminal repeat (23). Here, we show a major effect of CTSB on the progression of PyMT-induced mammary carcinomas and the formation of lung metastases. Furthermore, the findings indicate a substantial contribution of CTSB from cancer cells as well as from cells of the tumor microenvironment to the growth of experimental lung metastases.

Animals. To reduce the genetic background influence as a variable in the experiment, CTSB-deficient mice (22) were backcrossed for seven generations to the transgenic mouse strain FVB/N-TgN(MMTVPyVT)634Mul (PyMT; ref. 23), which was obtained from The Jackson Laboratory (Bar Harbor, ME). After the intercross, tumor mice being wild-type, heterozygous, and deficient for CTSB have been named PyMT;ctsb^+/+, PyMT;ctsb^+/−, and PyMT;ctsb^−/−, respectively.

Tumor progression study. Beginning at age 30 days, female mice were examined thrice a week for the development of mammary tumors by finger palpation in a genotype-blinded fashion. After tumors were first detected, diameters of tumors were measured with calipers every 2nd day until age 98 days. For the male tumor progression study, mice were palpated twice a week.

Histomorphometry. For volumetric measurement of total lung metastases or disseminated tumor colonies in lungs, the cryoembedded lungs were cut transversally to the trachea into systematic random 2-mm-thick slabs. The 5 to 7 slabs obtained from each lung were embedded with Tissue-Tek into a single block as specified previously (24). From each block, 5 μm cryostat sections were obtained and stained with H&E. Stereologic determination of the metastasis volumes was done by computer-assisted point counting as described previously (24).

Immunohistochemistry and immunofluorescence. Goat anti-rat CTSB antibody (provided by E. Weber, Martin-Luther-University, Halle, Germany; ref. 22) has been used for the detection of CTSB. For CTSB-F4/80 double-label immunofluorescence microscopy, the goat anti-rat CTSB antibody was visualized by Alexa Fluor 488 donkey anti-goat antibody (Invitrogen Molecular Probes, Karlsruhe, Germany). After blocking with goat IgG, the F4/80 macrophage antigen was detected by the rat anti-mouse F4/80 antibody (Serotec, Dusseldorf, Germany) followed by Alexa Fluor 546–labeled goat anti-rat antibody (Invitrogen Molecular Probes). Nuclear staining was done with Hoechst 33342.

Isolation of primary tumor cells from PyMT-induced mammary carcinomas. Primary PyMT tumor cells were obtained by mechanical and enzymatic dissociation of solid PyMT-induced carcinomas. The obtained suspension was sequentially transferred through 100 and 40 μm cell strainers, washed twice with PBS, and cultured for 12 hours. Immunodetection with a pan-cytokeratin antibody (Sigma, Hamburg, Germany) revealed 98% of cytokeratin-positive tumor cells after 12 hours in culture. Vital and adherent cells were harvested by trypsinization, PBS washed, and immediately frozen in medium containing 10% DMSO for further use.

Lung colonization assay following i.v. injection of LacZ-tagged PyMT tumor cells. Isolated primary tumor cells were genetically tagged by retroviral transduction of the bacterial LacZ gene according to the manufacturer's instructions (BioCarta, Ann Arbor, MI; ref. 25). LacZ tagging was verified by X-gal staining, and only pools with 100% transduction efficacy were used for the assay. Two days after transduction, 1 × 10⁵ LacZ-tagged primary tumor cells of the genotypes PyMT;ctsb^+/+, PyMT;ctsb^+/−, or PyMT;ctsb^−/− (four donor mice per genotype) were inoculated i.v. into the tail veins of ctsb^+/+, ctsb^+/−, and ctsb^−/− congenic recipient mice (five recipients per donor). Twenty-one days after tumor cell inoculation, lungs were isolated and stained with X-gal, and the number of surface colonies on the lungs were quantified.

In vitro cell invasion assay. Cell culture inserts sealed by an 8-μm pore-size microporous membrane filled with Matrigel (BD Biosciences, Heidelberg, Germany) were used for invasion studies. The Matrigel was dried in a laminar hood and reconstituted with 500 μL DMEM for 1 hour at 37°C. The inner compartments of the cell culture inserts were filled with 5 × 10⁴ PyMT cells in DMEM without FCS. The general cysteine cathepsin inhibitor E64 was added immediately after cell plating (final concentration, 10 μmol/L), and neutralizing cathepsin X antibody (R&D Systems, Minneapolis, MN) at final concentration of 1 μg/mL was preincubated with cells for 15 minutes at 37°C before the assay. The lower compartments were filled with 1 mL DMEM containing 10% FCS as chemoattractant and, where applicable, 10 μmol/L E64 or 1 μg/mL neutralizing cathepsin X antibody. After 20 hours at 37°C and 5% CO₂, the inserts were removed, and cells attached to the outside of the porous membrane were obtained by trypsinization and gentle scraping. Vital cells were counted with the CASY cell counter (Schärfe Systems, Reutlingen, Germany).

Surface labeling of cysteine proteases by active site probes. Cryopreserved PyMT tumor cells were recultivated for 24 hours, harvested, and counted in a Z1 Series Coulter Cell and Particle Counter (Beckman Coulter, Krefeld, Germany). Cells were plated on six-well plate with 3 × 10⁴/cm² and grown to confluency for another 24 hours. Medium was removed, and cells were incubated for 1 hour at 4°C with ice-cold cell medium containing 50 mmol/L HEPES supplemented with 10 μmol/L DCG-04 non-cell-permeable active site probe (26), specific for labeling of cysteine proteases and containing a biotin tag for recognition by streptavidin. Cells were thoroughly washed and harvested by scraping and lysed on ice in 250 mmol/L Tris-HCl (pH 6.8) and 0.1% Triton X-100 containing 10 mmol/L EDTA for 30 minutes. After clearing by centrifugation at 10,000 × g for 10 minutes at 4°C, the supernatants were used as cell lysates. Protein content was determined with DC Protein Assay (Bio-Rad Laboratories, Inc., Hercules, CA), and lysates were normalized to equal amounts of protein. Proteins were separated by SDS-PAGE, blotted onto a polyvinylidene difluoride (PVDF) membrane, and detected with streptavidin peroxidase. As molecular mass markers, prestained protein standards were used (MBI Fermentas, St. Leon-Rot, Germany).

Cathepsin X immunolabeling assay. For immunodetection of cell surface cathepsin X, cysteine proteases were labeled by DCG-04 as described above and purified by use of magnetic beads covalently coupled with streptavidin (Dynabeads M-280 Streptavidin, Dynal, Oslo, Norway). Precipitated surface cysteine proteases were resolved by 15% SDS-PAGE, transferred onto PVDF membranes, and probed with goat anti-cathepsin X antibodies (R&D Systems). Total protein expression of cathepsin X was determined by Western blot of total cell lysate with cathepsin X antibodies and revealed using enhanced chemiluminescence reaction (Pierce, Rockford, IL). Results were normalized to β-actin levels.

Quantitative real-time PCR. Total RNA was isolated from PyMT mammary tumors with a RNeasy kit (Qiagen, Valencia, CA). First strand cDNA templates were reversely transcribed from 5 μg total RNA (Invitrogen, Karlsruhe, Germany). Formation of PCR products was measured continuously by SYBR Green incorporation with the iCycler Real-time PCR Detection System (Bio-Rad Laboratories), and the relative amounts of CTSB, cathepsin X, and PyMT mRNA were normalized to the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) transcript. Gene specific primers were as follows: CTSB, 5′-TGCGTTCGGTGAGGACATAG-3′ (forward) and 5′-CGGGCAGTTGGACCATTG-3′ (reverse); cathepsin X, 5′-TATGCCAGCGTCACCAGGAAC-3′ (forward) and 5′-CCTCTTGATGTTGATTCGGTCTGC-3′(reverse); GAPDH, 5′-TGCACCACCAACTGCTTAG-3′ (forward) and 5′-GATGCAGGGATGATGTTC-3′ (reverse); and PyMT, 5′-CTCCAACAGATACACCCGCACATACT-3′ (forward) and 5′-GCTGGTCTTGGTCGCTTTCTGGATAC-3′ (reverse). For quantification of the PyMT transgene in lung tumor colonies, genomic DNA was isolated from lungs of PyMT cell-recipient mice. Real-time PCR for PyMT and GAPDH quantification was as described above.

Detection of CTSB enzyme activity. CTSB proteolytic activity was determined in lysosomal fractions (10 μg protein) in the presence of the fluorogenic substrate Z-Phe-Arg-4-methyl-coumarin-7-amide (20 μmol/L; Bachem, Bubendorf, Switzerland) and in the presence or absence of the CTSB-specific inhibitor CA074 (20 nmol/L; Bachem). The release of 7-amino-4-methyl-coumarin was monitored by spectrofluorometry at excitation and emission wavelengths of 360 and 460 nm, respectively.

Statistical analyses. Comparisons of arithmetic means between several groups were analyzed by t test (two sided) or ANOVA for experiments with more than two subgroups. Post hoc range tests and pair-wise multiple comparisons were done with t test (two sided). Proportions were compared by χ² test, and statistical analysis of Kaplan-Meier plots was done by log-rank test. P ≤ 0.05 was considered as statistically significant. Statistical analyses of lung metastasis were done by analysis of covariance (ANCOVA; Supplementary Data).

Progression of PyMT-induced primary carcinomas in female and male mice. To examine the effect of CTSB on formation of PyMT-induced mammary tumors, ctsb^−/− mice were bred with transgenic mice predisposed to mammary adenocarcinoma due to expression of the PyMT oncogene in the mammary gland epithelium (23). Female mice, hemizygous for the PyMT transgene, were grouped according to their respective CTSB genotype: (a) PyMT;ctsb^+/+, (b) PyMT;ctsb^+/−, and (c) PyMT;ctsb^−/−. Beginning at age 5 weeks, mice were palpated thrice a week for the development of mammary tumors in a genotype-blinded fashion. No tumors were found in nontransgenic control mice, indicating that the development of the tumor is due to expression of the PyMT transgene. In the female PyMT;ctsb^−/− group, palpable mammary tumors were detected in all mice at age 72 days, which is 16 days later compared with the control PyMT;ctsb^+/+ mice (Fig. 1A). On the average, PyMT;ctsb^−/− mice possessed a palpable tumor in 1 of their 10 mammary glands by age 57 days, which is ∼1 week later than in the PyMT;ctsb^+/+ mice (P < 0.01, log-rank test). After appearance of the first palpable tumor, tumor diameters were measured every 2nd day. We determined that carcinomas developed more slowly in PyMT;ctsb^−/− mice, which showed tumors in all 10 mammary glands 12 days later than PyMT;ctsb^+/+ (data not shown). However, no statistical significance was observed for the difference of total number of tumors developed by the three groups of PyMT transgenic mice with different genotypes of CTSB (data not shown). To further determine whether the lack of CTSB in PyMT transgenic mice inhibited tumor growth, we divided the tumors according to their size into three groups: <0.5, 0.5 to 1, and >1 cm in diameter (Fig. 1B). PyMT;ctsb^−/− mice developed significantly more small tumors as well as less medium and large size tumors after age 72 days compared with PyMT;ctsb^+/+ and PyMT;ctsb^+/− mice (Fig. 1B). In contrast, PyMT;ctsb^+/+ and PyMT;ctsb^+/− mice display faster tumor growth with distinctive strong decrease in the number of small tumors after 72 days (Fig. 1B,, right) and a corresponding increase of medium-sized tumors (Fig. 1B,, middle and left). Figure 1C shows the size of tumors according to their diameters 30 days after their first detection. Within these 30 days, significantly less tumors from PyMT;ctsb^−/− mice developed in medium and large size mammary cancers compared with PyMT;ctsb^+/+, indicating a growth permissive function of CTSB.

In addition, we analyzed the mammary tumor development in PyMT;ctsb^+/+ (n = 15), PyMT;ctsb^+/− (n = 16), and PyMT;ctsb^−/− (n = 22) male mice (Supplementary Fig. S1A). The mammary tumor development in male CTSB-deficient tumor mice was even more significantly inhibited compared with PyMT;ctsb^+/+ male mice. Although 50% of PyMT;ctsb^+/+ and PyMT;ctsb^+/− male mice developed palpable tumors at ages 15 and 18 weeks, respectively, it took 26 weeks to develop mammary tumors in 50% of PyMT;ctsb^−/− mice (P < 0.001, log-rank test). In contrast to female mice, in which multifocal adenocarcinomas developed in all of the 10 mammary glands, PyMT transgenic male mice on the average developed two mammary tumors without significant difference between the three experimental groups with different CTSB genotypes. Again, the mammary cancers were divided into three groups according to their size at 30 days after the first detection of the tumor (Supplementary Fig. S1B). As the females, male PyMT mice showed significant, genotype-dependent, gradually decreased mammary tumor growth rates in ctsb^+/+, ctsb^−/−, and ctsb^−/− mice. In contrast to the other genotypes, PyMT;ctsb^−/− mice did not develop tumors larger than 1 cm in diameter within the 1-month observation period. In summary, these results show increased tumor latency and decreased tumor growth on deletion of CTSB in female and male PyMT mice.

Pulmonary metastases of PyMT carcinomas and experimental lung colonization by PyMT cancer cells. A characteristic feature of the MMTV-PyMT adenocarcinoma mouse model is the development of numerous lung metastases that can be observed both macroscopically and microscopically by age 3 to 4 months (23). We estimated the total pulmonary metastases volumes by computer-assisted stereology on systematically sampled histologic lung sections from PyMT female mice of all three experimental groups at age 14 weeks (24). ANCOVA showed that besides genotype (ctsb^+/+, ctsb^+/−, and ctsb^−/−), primary tumor weight, and backcross, generation also had a significant effect on metastasis volume. Furthermore, compared with the PyMT;ctsb^+/+ group, the volume of pulmonary metastases was reduced to 45% in the PyMT;ctsb^+/− group (P = 0.01) and to 65% in the PyMT;ctsb^−/− group (Fig. 2), although the latter difference in volume of metastatic tumor nodules did not reach statistical significance (P = 0.13).

To gain more insight into the role of tumor CTSB in the process of tumor metastasis, we did lung colonization experiments with LacZ-tagged primary tumor cells from 98-day-old PyMT;ctsb^+/+, PyMT;ctsb^+/−, and PyMT;ctsb^−/− female mice. Immunocytochemistry revealed 98% cytokeratin-positive tumor cells in primary cell cultures (data not shown). Adherent LacZ-positive tumor cells were injected into the tail vein of congenic female FVB/N recipients. After 21 days, the number of developed lung metastases was determined by counting of LacZ-positive colonies (Supplementary Fig. S2). The number of colonies represents a measure of efficiency of tumor cell seeding and survival in the host tissue. Interestingly, the injection of PyMT;ctsb^+/− cells resulted in significantly reduced number of lung colonies compared with PyMT;ctsb^+/+, an effect that was not further enhanced for the PyMT;ctsb^−/− cells (Fig. 3A). As a further measure of efficiency of metastasis, the amount of the PyMT transgene in genomic DNA from lungs of animals after PyMT cell injection was quantified and divided by the number of lung tumor nodules present in the individual mouse (Fig. 3B). For the PyMT;ctsb^+/− and PyMT;ctsb^−/− cells, the assay revealed a 90% reduction of lung colony size compared with the PyMT;ctsb^+/+ group (Fig. 3B). Thus cancer cell–derived CTSB promotes both seeding and growth of experimental metastases.

Cell surface cathepsin X compensates for CTSB in PyMT;ctsb^−/− tumor cells. To further investigate the effects of CTSB in primary and experimental lung metastasis, we studied primary PyMT tumor cells in cell culture. First, we examined the gene expression and activity of CTSB in tumor cells by quantitative reverse transcription-PCR and degradation of a fluorogenic CTSB peptide substrate, respectively (Fig. 4A). Tumor cells isolated from PyMT;ctsb^+/− mice showed 30% reduced CTSB expression and CTSB activity compared with PyMT;ctsb^+/+ cells, and neither CTSB mRNA nor activity was detectable in PyMT;ctsb^−/− tumor cells. Viability, growth, and proliferation rate of tumor cells were not affected by the CTSB genotypes (Fig. 4B). Consistent with the in vivo findings in metastasis of primary PyMT carcinomas and experimental lung colonization, significantly reduced of invasion across Matrigel-occluded porous membranes was seen in tumor cells heterozygous for CTSB, whereas deletion of the second CTSB allele did not result in additional reduction of invasion (Fig. 4C). Notably, after treating the cells with the membrane-impermeable, broad-spectrum inhibitor of cysteine proteases E64, the Matrigel invasion of PyMT tumor cells was decreased to baseline levels in all three tumor cell types with different CTSB genotypes (Fig. 4C). This result suggests a compensation for loss of CTSB by other cysteine protease/s that are sensitive to E64.

Based on the assumption that most of the protease activity required for tumor cell invasion is supplied by extracellularly exposed proteases, we determined the activity profiles of cysteine cathepsins at the cell surface of PyMT cells (Fig. 5A). Cell surface labeling of active cysteine cathepsins with the membrane-impermeable active site probe DCG-04 (26) revealed the presence of cell surface-associated CTSB in PyMT;ctsb^+/+ and PyMT;ctsb^+/− cells, although CTSB was not detectable on PyMT;ctsb^−/− cells (Fig. 5A). Interestingly, overexpression of distinct cysteine protease bands was detected at the surface of CTSB-deficient tumor cells (Fig. 5A). The major up-regulated band detected at the surface of PyMT;ctsb^−/− cells comigrated in the SDS-PAGE with the cysteine protease cathepsin X (alternative names: cathepsin Z; cathepsin P) as detected by Western blotting (Fig. 5A). Precipitation of DCG-04-labeled proteases via its biotin tag and subsequent immunoblotting further confirmed the identification of the up-regulated cell surface cysteine protease cathepsin X (Fig. 5B). To assess the contribution of cathepsin X for invasion of PyMT tumor cells, a neutralizing anti-cathepsin X antibody, specifically inhibiting cathepsin X, has been used for Matrigel assays (Fig. 5C). In an initial experiment, several antibody concentrations in the range of 0.01 to 10 μg/mL were tested in Matrigel assays with PyMT;ctsb^−/− cells, and the antibody concentration of 1 μg/mL was found to result in 40% inhibition of invasion, an effect that was not further enhanced by higher amounts of antibody (data not shown). Remarkably, neutralizing anti-cathepsin X antibody treatment significantly reduced Matrigel invasion of PyMT;ctsb^−/− cells but did not affect invasion of PyMT;ctsb^+/+ or PyMT;ctsb^+/− cells (Fig. 5C). To further determine whether cathepsin X is up-regulated at the transcriptional or post-translational level, the expression of cathepsin X was measured by quantitative real-time PCR (data not shown), and the protein amount was determined by Western blot of total cell extract (Fig. 5D). No significant difference in cathepsin X mRNA or protein expression was found in the tumor cells with different CTSB genotypes, suggesting that enhanced surface activity of cathepsin X results from the redistribution of the enzyme to the cell surface. Thus, these results indicate that the absence of CTSB in PyMT tumor cells induces an increase of extracellular cathepsin X activity via its recruitment to the cell surface, compensating at least partially for the CTSB deficiency.

Host effects on lung colonization by PyMT tumor cells. CTSB in a tumor can be either derived from the tumor cells or expressed by cells of the tumor stroma. Recent studies have suggested a role for non-tumor-derived proteases in promotion of tumor growth and metastasis (21, 27). To explore the possible functional role of CTSB derived from the tumor microenvironment, we i.v. injected four different LacZ-tagged PyMT;ctsb^+/+ primary tumor cell lines, each isolated from individual primary PyMT tumors, into congenic ctsb^+/+, ctsb^+/−, and ctsb^−/− female mice that did not carry the PyMT transgene. These studies revealed that the complete loss of CTSB expression in recipient mice resulted in 50% reduction of pulmonary tumor colony number relative to ctsb^+/+ and ctsb^+/− recipients (Fig. 6A). In addition, the average volume of nodules in the lungs of ctsb^−/− recipients was reduced by 90% compared with the ctsb^+/+ and ctsb^+/− recipients (Fig. 6B). Subsequently, CTSB expression in PyMT lung colonies was investigated by immunohistochemistry to more precisely determine the source of CTSB within developing tumors (Supplementary Fig. S3). As expected, tumor colonies derived from PyMT;ctsb^+/+ tumor cells injected to ctsb^+/+ showed strong CTSB expression throughout the tumor mass. Similarly, PyMT;ctsb^−/− tumor cells injected into ctsb^−/− recipients showed no staining for CTSB. The majority of CTSB expression in lung colonies was derived from the PyMT;ctsb^+/+ tumor cells injected into ctsb^−/− recipients. Interestingly, when PyMT;ctsb^−/− cells were injected into ctsb^+/+ recipients, a strong CTSB staining was found in several distinct cells, mainly located near the edge of the lung colony. Pervious work has shown that lung metastasis formation in the MMTV-PyMT breast cancer model positively correlates with the extent of macrophage recruitment to the tumor (28). Therefore, double immunoflourescence was done for macrophages (antibody F4/80) and CTSB (Fig. 6C). Tumor colonies produced by PyMT;ctsb^−/− cells injected into ctsb^+/+ recipient mice showed an infiltration by macrophages with high expression of CTSB (Fig. 6C,, yellow). In contrast, macrophages distant from the tumor colony did not show any considerable expression of CTSB (Fig. 6C , red). These data suggest that macrophages up-regulate CTSB expression on being recruited to the tumor node and that macrophage-derived CTSB promotes growth of PyMT colonies in the lung.

During progression to high malignancy, tumor cells acquire multiple mutations that affect proliferation, programmed death, and genetic stability (29). Furthermore, a favorable microenvironment is essential for growth, invasion, and metastasis of a tumor (1). Proteases capable of degrading ECM proteins have recently been identified as critical regulators of the tumor microenvironment. Here, we show that the lysosomal cysteine protease CTSB contributes to growth, invasion, and metastatic process of PyMT-induced mammary carcinoma.

CTSB is overexpressed and also secreted by malignant cells and, in some studies, was found to be secreted from stromal cells closely associated with the tumor (12, 13). Mature CTSB, as well as the proteolytically inactive proenzyme, was detected in soluble form within the extracellular space and also found to be associated with the plasma membrane (30–32). These observations are consistent with our findings for cell surface expression of CTSB in PyMT mouse mammary cancer cells. Labeling with the broad-spectrum cathepsin probe DCG-04 revealed that CTSB is the most active cysteine cathepsin on the cell surface of wild-type PyMT cells. Cell surface association of CTSB can be achieved through an interaction with Annexin II heterotetramers that direct CTSB to caveolae (33). Furthermore, CTSB exerts proteolytic activity at the cell surface at neutral or slightly acidic pH (34) and can initiate proteolytic cascades through activation of MMP (e.g., MMP-3) and/or urokinase-type plasminogen activator and is capable of degrading the ECM proteins laminin, type IV collagen, and fibronectin (35). Accordingly, CTSB inhibition by synthetic inhibitors, antibodies, or antisense RNA clearly reduces tumor cell migration and invasion into ECM-like matrices in vitro (18, 19), whereas CTSB overexpression facilitates these processes (20, 36). These potential roles for CTSB in tumor cells are supported by our analysis of the PyMT mammary cancer model crossed with CTSB-deficient mice. Female and male mice exhibit a CTSB genotype-correlated delay in development of palpable tumors and reduced tumor growth rates, but absolute tumor numbers are not altered by the CTSB genotype. This clearly indicates that CTSB is not critical for tumorigenic transformation of selected cell clones but rather contributes to overcoming tumor cell dormancy by promoting invasive growth, a process that occurs most likely by degradation of the ECM.

Tumor cell metastasis is a complex cascade of sequential steps, in which tumor cells disseminate, extravasate from the vessel, invade new tissue, and proliferate to produce metastatic foci (37). CTSB expression on the gene and protein level has been correlated with metastasis (38, 39); however, little is known about the functional relevance of CTSB for the metastatic process in vivo. In the present study, the role of cancer cell–derived CTSB has been addressed by analysis of spontaneous pulmonary metastases as well as lung metastases, resulting from introduction of primary PyMT cells of different CTSB genotypes and ex vivo migration/invasion of isolated PyMT tumor cells. Taken together, these experiments clearly indicate that CTSB contributes to metastasis in the PyMT mammary cancer model. The reduced migration and invasion of tumor cells with reduced CTSB levels in vitro provides circumstantial evidence for a contribution of CTSB to tumor cell dissemination from primary tumors. More directly, the lung colonization experiments revealed that CTSB supports the sequestration of PyMT cells into the lung tissue because a lower number of tumor nodules were present in lungs after injection of PyMT;ctsb^+/− and PyMT;ctsb^−/− cells. In addition, our results suggest that CTSB is also involved in the support of growth of tumor colonies because the volume of lung nodules derived from PyMT;ctsb^+/− and PyMT;ctsb^−/− cells was highly reduced. A remarkable and unexpected finding in all of these experiments was the strong reduction of metastasis burden, lung colony formation, and Matrigel invasion in ctsb^+/− PyMT tumors or isolated PyMT tumor cells derived from PyMT;ctsb^+/− mice. Remarkably, these effects were not enhanced by total ablation of CTSB expression in the homozygous knockout groups. In an attempt to understand these results, we probed the surface of primary PyMT cells for active cysteine cathepsins using a broad-spectrum activity-based probe. PyMT;ctsb^−/− cells showed a redistribution of active cathepsin X to the cell surface, resulting in at least a partial compensation for loss of CTSB function as shown by reduction of tumor cell invasion on treatment of CTSB-deficient cells with a neutralizing cathepsin X antibody. Cathepsin X is a recently discovered cysteine protease that is up-regulated in human cancers (40, 41). Strikingly, cathepsin X and CTSB are the only carboxypeptidase enzymes in the cysteine cathepsin family, although CTSB exhibits an additional endopeptidase activity (42, 43). This suggests that cathepsin carboxypeptidase activity contributes significantly to invasion.

High levels of CTSB have frequently been detected at the invasive edge of tumors (13, 14, 44). The cellular source of this protease in solid tumors, consisting of both malignant tumor cells and a variety of stromal cells (e.g., fibroblasts, endothelial cells, and bone marrow–derived inflammatory cells), has remained elusive. CTSB has been reported to be expressed in stromal fibroblasts and macrophages in carcinomas of breast, colon, and prostate (14, 45, 46); however, the specific contribution of CTSB from these cell populations to tumor cell growth and invasiveness has not yet been defined. Here, we do lung colonization experiments with PyMT cancer cells i.v. injected into ctsb^+/+, ctsb^+/−, and ctsb^−/− recipient mice. Interestingly, CTSB deficiency of the recipient resulted in reduced seeding efficiency. However, the size reduction of tumor nodules was even more significant in the ctsb^−/− recipients, indicating that host cell–derived CTSB is potentially as important as tumor-derived protease activity for metastatic growth. In the search for host cells that supply CTSB, we identified tumor-infiltrating macrophages, which induce CTSB expression on entry into the tumor mass. This result is in line with a recent report suggesting that macrophages contribute to the development of invasive, metastatic carcinomas in the MMTV-PyMT mammary cancer model (28). This intriguing role of tumor-associated macrophages was recently suggested to be of general significance in the progression of solid tumors (47). In addition to CTSB, macrophages may produce other proteases (e.g., MMP-9) that contributes to carcinogenesis in mice (27). Thus, our present data provide further evidence for functional involvement of proteolytic enzymes of tumor-invading bone marrow–derived cells in cancer progression (48).

In summary, the present results suggest that CTSB derived from both tumor and stromal cells from the microenvironment plays critical roles in multiple stages of tumor growth and metastasis. Thus, pharmacologic inhibition of extracellular CTSB within and around a tumor by plasma membrane-impermeable inhibitors may be a promising route for effective cancer treatment.

Grant support: Deutsche Forschungsgemeinschaft (SFB 364; B12) grant 23-7532.22-33-11/1 of the Ministerium für Wissenschaft und Kunst, Baden-Wuerttemberg, and the European Union Framework Programme 6, Project LSHC-CT-2003-503297, Cancerdegradome (A. Krüger).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Susanne Dollwet-Mack, Anne Schwinde, and Ulrike Stabenow (Institut für Molekulare Medizin und Zellforschung, Freiburg, Germany) and Katja Honert (Institut für Experimentelle Onkologie, Munich, Germany) for excellent technical assistance.