Conditional Gene Targeting Reveals Cell Type-Specific Roles of the Lysosomal Protease Cathepsin L in Mammary Tumor Progression

Background: Cathepsin L (Ctsl) is a cysteine protease mainly located within the endosomal/lysosomal cell compartment. High expression of Ctsl indicates poor prognosis in human breast cancer. However, the cell type-specific Ctsl functions responsible for this association remain elusive. Methods: Because constitutive Ctsl−/− mice develop a complex phenotype, we developed a conditional model allowing for cell type-specific inactivation of Ctsl in mammary epithelium or myeloid cells in the transgenic mouse mammary tumor virus (MMTV)-polyoma middle T (PyMT) breast cancer model. Results: Ctsl ablation in mammary epithelial cells resulted in delayed initiation and end-stage of cancers. The latter displayed large dead cell areas. Inducible in vitro deletion of Ctsl in MMTV-PyMT-derived breast cancer cells revealed expansion of the acidic cell compartment, alteration of intracellular amino acid levels, and impaired mTOR signaling. In consequence, Ctsl-deficient cells exhibited slow growth rates and high apoptosis susceptibility. In contrast to Ctsl-deficient mammary epithelium, selective knockout of Ctsl in myeloid cells had no effects on primary tumors, but promoted lung metastasis formation. Conclusions: Our cell type-specific in vivo analysis provides strong evidence for a cancer cell-intrinsic, tumor-promoting role of Ctsl in primary breast cancer, whereas metastasis is negatively regulated by Ctsl expressed by bone marrow-derived cells.


Introduction
Proteolysis is a fundamental event at every single stage of tumorigenesis [1,2]. Cysteine cathepsins modulate physiological as well as pathological processes as important components of the intracellular proteolytic network [3]. This protease family is often overexpressed in tumor cells, resulting in their secretion from the acidic cell compartment into the extracellular space [4]. Along with both tumor-promoting and suppressing roles and different cells of origin, cancer model experiments aided the identification of distinct, non-redundant roles of single cysteine proteases in tumor progression.
Profiling of single cysteine cathepsins demonstrated their increased activity during defined stages of tumor progression [5]. Some of them, such as cathepsin Z (Ctsz), were found to be upregulated in tumor-associated macrophages (TAMs) following cathepsin B (Ctsb) knockout [6,7]. Cysteine cathepsins released in the tumor microenvironment (TME) promote tumorigenesis in several ways, e.g., by processing different growth factors, cytokines, and chemokines; by cleaving cell-cell junction proteins; or by remodeling the extracellular matrix (ECM) [8,9]. They also play a role in many other tumor-suppressing processes, such as cell death and autophagy, which can contribute to worsening by the development of drug resistance [10].
Although the function of cathepsin L (Ctsl) in the complex process of tumorigenesis is not yet fully understood, the upregulation of its mRNA and protein levels especially in breast cancer correlates with a higher risk of relapse, poor therapy outcome, and worse overall survival [11][12][13]. Ctsl has been shown to have many unique cell type-specific functions crucial for the maintenance of tissue homeostasis, which cannot be compensated by other cathepsins or other cysteine proteases. By means of Ctsl null mice, the roles of this protease in epidermal homeostasis, hair follicle morphogenesis and cycling, cardiac function, and MHC-II-mediated antigen presentation of cortical thymic epithelial cells were described previously [14][15][16]. Many other substrates of Ctsl lysosomal activity are arising, together with crucial functions in the development and homeostasis of diverse tissues, e.g., as part of vesicles, Ctsl takes part in the proteolytic processing of neurotransmitters and hormones [17][18][19].
Due to its tissue-specific functions, the role of Ctsl in several cancer types is versatile. Tumor promoting effects were reported for the RIP1-Tag 2 pancreatic islet cell carcinogenesis model [20]. This report established a reduction in tumor growth in Ctsl −/− animals, resulting from the combination of impaired proliferation and enhanced cell death. A further Ctsl knockout study using the MycER TAM -Bcl xL pancreatic neuroendocrine cancer revealed an impairment in tumor progression toward the latest stages, an increase in tumor cell death, and elevated expression of autophagy markers, together with defective fusion of autophagosomes and lysosomes [21]. In contrast, several other studies revealed protective functions of Ctsl expression towards carcinogenesis. In a report of intestinal tumorigenesis using the Apc Min model, Ctsl deficiency resulted in an increased tumor incidence as a result of the interplay between Ctsl and the tight-junction protein claudin 1 [22]. Consistently, Ctsl knockout in two squamous cell carcinoma models showed an earlier onset of tumors accompanied by an increase in tumor burden and invasiveness, which was explained by hyper-responsiveness to growth factor signals and hyper-activation of the MAPK/AKT pathways [23,24]. A previous study using the mouse mammary tumor virus (MMTV)-polyoma middle T (PyMT) breast cancer model revealed a massively enhanced metastatic burden in the lungs following transgenic overexpression of human Ctsl [25].
Multiple approaches have been employed to surpass the complex phenotype caused by the lack of Ctsl in mice and, at the same time, to enable the study of the cell type-specific contribution of Ctsl to carcinogenesis. In order to analyze to what extent Ctsl supplied by TAMs contributes to tumorigenesis, bone marrow from Ctsl −/− donor mice was transplanted to RIP1-Tag 2 recipient mice. It could be established that the tumor-promoting functions of Ctsl must be derived from either cancer cells or cells other than from the bone marrow [26]. Additional studies highlighted that restoring the Ctsl catalytic Cancers 2020, 12,2004 3 of 25 activity in epidermal keratinocytes in a tissue-specific manner can counteract the enhanced malignant phenotype observed in Ctsl −/− skin cancers [23].
Contrary to the aforementioned efforts, the present study reports a direct approach for exploring cell type-specific Ctsl functions in primary cancers by targeting the protease using a Cre/loxP strategy. Conditional Ctsl deletion in mammary epithelium and the cancer cells derived therefrom or, alternatively, in myeloid cells capable of infiltrating breast tumors displayed distinct, tissue-specific functions of Ctsl in the maintenance of cell homeostasis, survival, and proliferation in breast cancer. We further provide evidence for an important intracellular function of Ctsl related to lysosomal homeostasis and lysosome-dependent mTOR signaling.

Generation and Characterization of Conditional Ctsl Knockout Mice
We made use of the Cre/loxP technology to address cell-specific functions of Ctsl in murine breast cancer. Ctsl was targeted by flanking exons 3-6 with loxP sites (Figure S1A III). Cre-mediated recombination was predicted to result in the deletion of those exons and in a frameshift-mutation terminating Ctsl translation (Figure S1A IV).
As a proof of concept, Ctsl fl/fl mice were crossed with Sox2-Cre mice, thereby giving rise to litters bearing a ubiquitous deletion of Ctsl ( Figure S1B). Accordingly, Ctsl protein was absent in the kidney and liver of those animals. Furthermore, Ctsl mRNA levels were also undetectable by primers located between the exons 1 and 4, confirming the accuracy of the Cre/loxP strategy for our purposes. We also found the well-described skin phenotype of constitutive Ctsl null mice in the Sox2-Cre/Ctsl −/− animals ( Figure S1C). This phenotype is characterized by disturbed hair follicle cycling as well as epidermal thickening due to keratinocyte hyperproliferation and hyperkeratosis, thereby contrasting with normal skin in the Ctsl fl/fl mice [15].
In the next step, we deleted Ctsl specifically in mammary epithelium and consequently in the cancer cells originating from them, or, alternatively, in myeloid cells. This was achieved by tissue-specific expression of the Cre recombinase directed either by the MMTV-long terminal repeat or the myeloid cell-specific lysozyme M (LysM) promoter, respectively [27,28]. The use of these cell-specific Ctsl deletions enabled us to overcome the defective positive selection of T helper cells at the thymic epithelium of Ctsl null mice [14]. This was proven by flow cytometric analysis of CD4 + /CD8 + T cell ratios in blood, thymus and spleen of 10 week old MMTV-Cre/Ctsl −/− and LysM-Cre/Ctsl −/− female mice ( Figure S1D,E). The CD4 + /CD8 + cell ratios of those mice were unchanged when compared to the Ctsl fl/fl control mice, whereas mice with constitutive Ctsl deficiency showed the expected decline in CD4 + T cells. This result allowed the further analysis of Ctsl in breast cancer without the interference of a T cell imbalance.

Ctsl Deletion in Mammary Epithelial Cells Delays Tumor Onset
For the analysis of Ctsl in breast cancer, MMTV-Cre/Ctsl −/− and LysM-Cre/Ctsl −/− mice were further crossed with the transgenic MMTV-PyMT mouse model of metastasizing breast cancer in the C57BL/6 genetic background [29,30]. To determine the effects of Ctsl deletion on the development of pre-malignant adenomas, whole mammary fat pads of eight-week-old female mice bearing the PyMT oncogene were examined ( Figure 1A). Quantification of the pre-malignant, carmine red-stained areas showed significantly less adenomas in MMTV-Cre/Ctsl −/− breasts than in Ctsl fl/fl or LysM-Cre/Ctsl −/− glands. All MMTV-PyMT mice developed multifocal tumors, as previously described [29]. Consistent with fewer precursor lesions, first palpable tumors in MMTV-Cre/Ctsl −/− mice occurred at an median age of 17.5 weeks, which represents a six week delay when compared to Ctsl fl/fl and LysM-Cre/Ctsl −/− mice ( Figure 1B).
Cancers 2020, 12, x 4 of 28 age of 17.5 weeks, which represents a six week delay when compared to Ctsl fl/fl and LysM-Cre/Ctsl −/− mice ( Figure 1B). Tumors of Ctsl fl/fl and LysM-Cre/Ctsl −/− mice reached end-stage size at week 21.5, while MMTV-Cre/Ctsl −/− cancers reached end-stage size after week 27.5 ( Figure 1C). MMTV-Cre/Ctsl −/− mice showed a reduced tumor incidence per mouse, supporting our findings of impaired tumorigenesis upon specific deletion of Ctsl in mammary epithelium ( Figure 1D). As expected, tumor lysates show the presence of Ctsl in Ctsl fl/fl and LysM-Cre/Ctsl −/− samples, and a marked reduction of the protease in MMTV-Cre/Ctsl −/− tumor lysates ( Figure 1E). The faint Ctsl bands in MMTV-Cre/Ctsl −/− samples are derived from non-epithelial cells in the mammary tumors, e.g., fibroblasts and immune cells. The strong reduction of Ctsl in those samples also indicates an efficient deletion of Ctsl in the cancer cells excluding a mosaic gene deletion. Results were not statistically significant among groups. In the box plots, the boundary of the box closest to zero indicates the 25th percentile, the line within the box marks the median, and the boundary of the box farthest from zero indicates the 75th percentile. Whiskers indicate the 10th and 90th percentiles. (E) Immunoblot analysis of Ctsl expression levels in tumor lysates. Expression levels of the three characteristic bands expected for Ctsl (proform at 37kDa; sc, single chain at 25 kDa; hc, heavy chain at 20 kDa) vary among mice and tumor (Ctsl fl/fl n = 3; MMTV-Cre/Ctsl −/− n = 3; LysM-Cre/Ctsl −/− n = 3). Minute amounts of any of the three forms of Ctsl can be detected in the MMTV-Cre/Ctsl −/− tumors due to the specificity of the Cre-mediated recombination, which enables deletion only in breast epithelial cancer cells, conserving the Ctsl expression in the tumor microenvironment. Tumors of Ctsl fl/fl and LysM-Cre/Ctsl −/− mice reached end-stage size at week 21.5, while MMTV-Cre/Ctsl −/− cancers reached end-stage size after week 27.5 ( Figure 1C). MMTV-Cre/Ctsl −/− mice showed a reduced tumor incidence per mouse, supporting our findings of impaired tumorigenesis upon specific deletion of Ctsl in mammary epithelium ( Figure 1D). As expected, tumor lysates show the presence of Ctsl in Ctsl fl/fl and LysM-Cre/Ctsl −/− samples, and a marked reduction of the protease in MMTV-Cre/Ctsl −/− tumor lysates ( Figure 1E). The faint Ctsl bands in MMTV-Cre/Ctsl −/− samples are derived from non-epithelial cells in the mammary tumors, e.g., fibroblasts and immune cells. The strong reduction of Ctsl in those samples also indicates an efficient deletion of Ctsl in the cancer cells excluding a mosaic gene deletion.

Distinct Effects of Cell Specific Ctsl-Deficiency on Histological Appearance and Metastasis of MMTV-PyMT Breast Cancers
Histopathological grading of breast cancer was applied using the Elston/Ellis scoring system. Elston/Ellis unifies nuclear pleiomorphy, mitotic rate, and tubule formation [31]. Grading hematoxylin and eosin (HE)-stained murine tumor slides with this system revealed that MMTV-Cre/Ctsl −/− cancers appear highly pleomorphic, bear less tubuli, and display more often a solid structure, with similar mitotic rates to Ctsl fl/fl tumors (Figure 2A Figure S2. Notably, the mitotic index of cancer cells was similar in all three mouse cohorts ( Figure 2C).
In order to quantify visible dead cell areas in the center of MMTV-Cre/Ctsl −/− tumors, histological sections were stained with the TdT-mediated dUTP-biotin nick end labeling (TUNEL) method and cleaved caspase 3 as markers for DNA fragmentation and apoptosis, respectively ( Figure 2A). Significantly larger necrotic areas comprising dead cells, together with an increased number of apoptotic cells within the tumor mass were detected in MMTV-Cre/Ctsl −/− cancers ( Figure 2D). These results were confirmed by quantification of TUNEL stained areas, which also yielded an increase of dead cells in the MMTV-Cre/Ctsl −/− group ( Figure 2E).
To address insufficient blood supply as a possible cause of cell death in MMTV-Cre/Ctsl −/− breast tumors we tested their vascularization by detecting the endothelial marker PECAM1/CD31 by immunohistochemistry (IHC). Vascularization of these tumors was not affected by the Ctsl knockout, because flow cytometry analysis yielded similar percentages of CD31 + cells in end-stage tumors of all genotypes ( Figure 2F). In addition, microvessel density was evaluated by means of CD31 + IHC staining of tumor slides by an experienced pathologist in a blinded manner ( Figure 2G). Solid clusters, single cells, and endothelial cell clusters with lumina were quantified in the tumor mass within 10 high power fields/400-fold magnification by light microscopy, as reported by Weidner et al. [32]. The vessels were mainly located in the stroma, and there were no differences in microvessel density (MVD) across tumors. Indeed, IHC detection of CD31 + cells in Ctsl fl/fl and MMTV-Cre/Ctsl −/− tumor slides revealed normal blood vessels even near dead cell areas, confirming no impairment in vessel formation upon Ctsl deletion ( Figure 2H).
We further addressed whether metastasis formation could be affected by Ctsl deficiency. Representative pictures of lungs of all three Ctsl genotypes stained for the proliferation marker Ki67 pointed to considerable differences in metastasis ( Figure 3A). Strikingly, the metastatic burden was significantly increased in LysM-Cre/Ctsl −/− mice ( Figure 3B). We identified an increased number of metastases as the cause of the increment of the metastatic burden ( Figure 3C), whereas the size of the metastases per lung and their Ki67 Proliferation index was not affected ( Figure 3D,E). In spite of the undifferentiated high grade cancers of MMTV-Cre/Ctsl −/− mice, the lung metastatic burden of those animals was identical to the control group. The relevance of pro-inflammatory immune cell infiltration for the migratory and invasive behavior of cancer cells has been widely proven for the PyMT model [33][34][35]. To address the enhanced metastatic burden found in LysM-Cre/Ctsl −/− mice, we analyzed the numbers of neutrophils, dendritic cells, macrophages, monocytes, T helper cells, cytotoxic T cells, and B cells of end-stage tumors and the corresponding lungs ( Figure 4). No significant changes could be found in the percentages of immune cell infiltration either in tumors or in lungs across the three genotypes, pointing towards cell-intrinsic effects of Ctsl in tumor growth and for the development of metastasis.     In summary, Ctsl deficiency in mammary epithelium appears to have opposing phenotypic consequences for the progression of mammary tumors. On the one hand, Ctsl deletion causes pro-malignant loss of differentiation; on the other hand, Ctsl loss induces massive death of tumor cells, which would restrain tumor progression. To address this, we next developed a cell culture model for Ctsl-deletion in MMTV-PyMT breast cancer cells.

Ctsl −/− Mammary Epithelial Cancer Cells De-Differentiate In Vitro
To elucidate cell-specific functions that underlie the previous findings, a Ctsl-deficient cell line was generated by transducing Ctsl fl/fl breast tumor cells with a doxycycline-inducible Cre-recombinase construct, causing the subsequent recombination of the targeted gene locus. The efficient and homogenous genetic deletion was corroborated by droplet digital PCR ( Figure 4A). After three days of culture, Ctsl −/− breast epithelial cells showed an elongated mesenchymal phenotype, contrasting with the epithelial clusters formed by the parental control cells ( Figure 4B). Thus, we analyzed the mRNA expression levels of the epithelial marker E-cadherin, as well as fibronectin, N-cadherin, and vimentin as mesenchymal markers ( Figure 4C).
We found that E-cadherin was transcriptionally downregulated, whereas mRNA of N-cadherin and vimentin was increased in Ctsl −/− cells, with no changes in fibronectin, supporting the previous observation of de-differentiation in MMTV-Cre/Ctsl −/− cancers. However, in terms of motility, Ctsl fl/fl and Ctsl −/− cells showed comparable speed and also analogous track length per time period ( Figure 4D). Furthermore, the expression of key transcriptional drivers of epithelial-to-mesenchymal transition (EMT), such as Snail1, Zeb1, and Lef1 were not altered in Ctsl −/− cells, whereas a decrease in the transcription of Twist1 could be found ( Figure 4E). Therefore, the morphological changes and the increase in mesenchymal markers at the transcriptional level in cultured Ctsl −/− cells is consistent with the de-differentiation observed in vivo for MMTV-Cre/Ctsl −/− tumors. However, the lack of enhanced motility and the absence of induced EMT-transcription factors largely exclude a canonical EMT process as the cause of the morphological changes in Ctsl −/− breast tumor cells. In this context, we addressed the transcription levels of p63 ( Figure 4F), known for its role in sustaining proliferative potential and stemness of breast epithelial cells [36]. The level of p63 was significantly reduced in four independent batches of Ctsl −/− cells.

Ctsl −/− Mammary Epithelial Cancer Cells Are Growth Defective
The delayed development of MMTV-Cre/Ctsl −/− breast tumors, together with the decrease in transcription of p63 in Ctsl −/− breast epithelial cancer cells lead to a detailed characterization of cell proliferation in vitro. In initial co-culture experiments starting with a 1:1 ratio of Ctsl fl/fl and Ctsl −/− cells, quantification of the non-recombined Ctsl locus by droplet digital PCR showed that Ctsl fl/fl cells clearly outgrew Ctsl −/− cells after six days ( Figure 5A). Subsequently, differences in growth of both cell lines were verified by real-time proliferation monitoring, which revealed that Ctsl −/− cells multiplied significantly less than Ctsl fl/fl , regardless of serum concentration, i.e., 10% or 1% fetal calf serum (FCS) ( Figure 5B). In these experiments, the negative growth slopes in 1% FCS conditions suggest detachment of eventually dying cells from the plates. Therefore the marked impairment in growth of Ctsl −/− breast cancer cells appeared to be in large part caused by an increase in cell death, as we observed for tumors in vivo ( Figure 2). Indeed, the quantification of annexin V binding to apoptotic cells revealed an at least a two-fold increase in apoptosis in Ctsl −/− cells after three days of culture, which was further propagated upon FCS deprivation ( Figure 5C).  In spite of growth impairments of Ctsl −/− breast cancer cells, they were able to establish metastasis in vivo. Thus, we investigated their migratory capacity in vitro, by means of competitive lung colonization experiments ( Figure 5D). We injected 40% or 80% Ctsl −/− tumor cells into the tail-vein of four WT mice per experiment in two independent experiments. The lungs of the animals were harvested on day 28 post-injection. We quantified the occurrence of the PyMT oncogene and the recombination status of the Ctsl locus of PyMT +/T cells by droplet digital PCR on genomic DNA. In both settings the vast majority of PyMT +/T cells in the lung proved to contain an intact homozygous Ctsl gene, thereby further substantiating the growth advantage of Ctsl fl/fl cells over the Ctsl −/− PyMT +/T cells.
Next, IHC for detection of Ctsl was performed on lung slides of these mice ( Figure 5D). Large macrometastases were positive for Ctsl, whereas micrometastases were devoid of Ctsl staining. We conclude that Ctsl −/− as well as Ctsl fl/fl cells are able to disseminate into the lungs and to establish metastatic microlesions. Importantly, only Ctsl fl/fl cancer cells were able to colonize the lungs during the timespan of the experiment. For interpretation of these results it has to be noted that Ctsl fl/fl and MMTV-Cre/Ctsl −/− mice in the primary MMTV-PyMT model had comparable metastatic burdens ( Figure 3B). This must be interpreted in the context of Figure 3, presenting results for mice with end-stage tumors, which occur in Ctsl −/− mice at a higher age ( Figure 1C). Hence, Ctsl −/− breast cancer cells were shown to be able to colonize the lungs only with a long timespan ahead. Comparisons in matched time settings (as in Figure 5D,E), however, underline the compromised in vivo growth of Ctsl −/− mammary cancer cells.

Expansion of the Lysosomal Compartment of Ctsl −/− Mammary Epithelial Cancer Cells
Enlargement and accumulation of lysosomes accompanied by defects in the termination of autophagy have been previously observed for Ctsl −/− cancerous and noncancerous cells [21,23]. In line with this, labeling of Ctsl fl/fl and Ctsl −/− breast cancer cells by the acidophilic dye Lysotracker TM revealed a marked accumulation of acidic organelles in Ctsl-deficient cells ( Figure 6A). The microscopic aspect was validated by flow cytometry showing a 1000-fold increase of Lysotracker TM mean fluorescence intensity of Ctsl −/− cells as compared to Ctsl fl/fl ( Figure 6B). Interestingly, Ctsl fl/fl cancer cells could expand their acidic compartment upon FCS starvation, whereas Ctsl −/− cancer cells had only a minor increase in staining in this condition. In line with this finding, Ctsl −/− cells showed an accumulation of lysosomal acidic β-galactosidase activity ( Figure 6C), as well as increased occurrence of vesicles positive for the lysosomal membrane protein Lamp1 ( Figure 6D). The increase in lysosomal Lamp1 in Ctsl −/− was further supported by Western blots as well as by mRNA quantification ( Figure 6E). In terms of transcription, we also found a trend for increased expression of other lysosomal proteins, such as the proteases cathepsin B and cathepsin D ( Figure 6F). Lysosomal biogenesis is mostly triggered by the transcription factors of the MiT/TFE family, especially by the transcription factor EB (TFEB), of which mRNA expression is also augmented in Ctsl −/− breast cancer cells ( Figure 6F). The upregulation of protein transcripts involved in lysosomal biogenesis supports the enlargement of the lysosomal compartment, enabling the cell to maximize its hydrolytic capacity to overcome the accumulation of undegraded cargo and the consequent stress.

Impaired mTORC1 Activity in Ctsl −/− Cancer cells
The mammalian target of rapamycin (mTOR) is a key signaling hub in homeostasis, also involved in the coordination of cellular stress responses [37]. The mTOR complex 1 (mTORC1) requires physical association with adaptor proteins at the cytosolic site of the lysosomal membrane for full activity. mTORC1 inactivation limits cell growth while triggering multiple cellular stress responses, such as lysosomal biogenesis and macroautophagy. Because we had evidence for increased lysosome formation in Ctsl −/− breast cancer cells (Figure 6), other aspects of mTORC1 signaling were addressed. First, we investigated the phosphorylation state of the p70 S6 Kinase (S6K) isoform as a main mTORC1 downstream target ( Figure 7A). We quantified phosphorylation of p70 S6K at threonine 389, as well as the p85 S6K isoform upon phosphorylation at threonine 412. Indeed, quantification revealed a significant decrease in phospho-p70 S6K, indicating reduced mTORC1 activity in Ctsl −/− cells. Alanine, glutamic acid, and glutamine were significantly increased. Significantly reduced levels of isoleucine, leucine, tryptophan, phenylalanine, glycine, lysine, asparagine, and aspartic acid were recorded (n = 3, 3 independently generated batches); mean ± S.E.M. (E) Representative mTOR and phospho-mTOR Western blot images and its quantification(n = 4, p < 0.05, one sample t-test); mean ± S.E.M. Phospho-mTOR was detected at its serine 2448 residue. Alphatubulin was used for normalization.

ERK 1/2 Signaling Is Crucial for Viability and Survival of Ctsl −/− Mammary Cancer Cells
In an attempt to identify intracellular signaling pathways that might help Ctsl −/− breast cancer cells to cope with the cellular stress resulting from lysosomal dysfunction, we found a trend showing an increase of the phosphorylated isoform of the extra cellular-signal regulated kinases (ERK) 1 and Alanine, glutamic acid, and glutamine were significantly increased. Significantly reduced levels of isoleucine, leucine, tryptophan, phenylalanine, glycine, lysine, asparagine, and aspartic acid were recorded (n = 3, 3 independently generated batches); mean ± S.E.M. (E) Representative mTOR and phospho-mTOR Western blot images and its quantification(n = 4, p < 0.05, one sample t-test); mean ± S.E.M. Phospho-mTOR was detected at its serine 2448 residue. Alpha-tubulin was used for normalization. We next focused on the process of macroautophagy. One of its fundamental steps is the conversion of the microtubule-associated protein 1A/1B-light chain 3 I (LC3I) to LC3II. However, the LC3 I/II ratio was similar in Ctsl fl/fl and Ctsl −/− cells ( Figure 7B). In addition, mRNA expression of LC3 and p62 was unchanged in Ctsl −/− cells, indicating no differences in the initiation of autophagy ( Figure 7C). Our results rule out an increased induction of macroautophagy in Ctsl −/− breast cancer cells. Moreover, defective terminal degradation is very likely to contribute to the formation of large acidic vesicles in the cells as reported in the previous section.
Therefore, we hypothesized that deficient protein degradation caused by the loss of Ctsl could result in alterations of amino acid levels, which in turn are sensed by mTORC1, modulating its activity [38]. Interestingly, amino acid profiling showed a significant decrease in the ketogenic amino acids, as well as of asparagine and aspartic acid in Ctsl −/− cells ( Figure 7D). In contrast, the remaining glucogenic amino acids were increased or unaffected by the Ctsl-deficiency. This result indicates a considerable metabolic alteration in the Ctsl −/− breast cancer cells. With regard to the key amino acids critical for mTORC1 activation, leucine concentration was significantly reduced in Ctsl −/− cells, whereas arginine is also represented at lower levels with a greater variability between experiments. This suggests sensing of those reduced amino acid levels by mTORC1 and hence a lower activity of this complex. To address this, we assessed the status of mTOR phosphorylation at serine 2448 ( Figure 7E). Indeed, a significant decrease in mTOR phosphorylation could be established for Ctsl −/− breast cancer cells.
These results suggest an altered lysosomal turnover as the primary cause of the described cell phenotypes in absence of the lysosomal endoproteinase Ctsl. This affects amino acid availability in cells and mediates a stress response that is, at least in part, triggered by mTORC1 inactivation.

ERK 1/2 Signaling Is Crucial for Viability and Survival of Ctsl −/− Mammary Cancer Cells
In an attempt to identify intracellular signaling pathways that might help Ctsl −/− breast cancer cells to cope with the cellular stress resulting from lysosomal dysfunction, we found a trend showing an increase of the phosphorylated isoform of the extra cellular-signal regulated kinases (ERK) 1 and 2 in Ctsl −/− cells ( Figure 8A). In order to study the dependence of our Ctsl −/− cells on the mitogen-activated protein kinase (MAPK) pathway, cells were treated with Trametinib, a clinically approved inhibitor of the kinase upstream of ERK, i.e., the mitogen-activated protein kinase kinase 1 (MEK) [39]. Cell viability of Ctsl −/− cells was reduced to 50% after six-day Trametinib treatment ( Figure 8B). Apoptotic cell death of Ctsl −/− cells was massively increased upon treatment, confirming the reliance of the cells on MAPK signaling for survival ( Figure 8C). These results point to MAPK signaling as an adaptive cell response to lysosomal stress and suggest a therapeutic benefit for dual targeting of MAPK and lysosomal pathways. A substantial reduction of the phosphorylation of both ERK isoforms, p44 and p42, was corroborated by Western blot upon Trametinib treatment ( Figure 8D).

Discussion
Making use of a newly established mouse line that enables the conditional genetic ablation of Ctsl in the transgenic MMTV-PyMT breast cancer model, we report that Ctsl-deficiency in mammary epithelial tumor cells resulted in a considerable delay in the emergence of tumors. Additionally, end-stage tumors were delayed, morphologically altered and displayed large dead cell areas. In contrast, the selective knockout of Ctsl in myeloid cells had no effects on primary tumors, affecting exclusively the metastatic seeding of PyMT tumor cells to lungs. Hence, our data support a cancer-cell intrinsic role of Ctsl in primary tumor growth while the development of metastasis is influenced by the tumor microenvironment.
inhibitor of the kinase upstream of ERK, i.e., the mitogen-activated protein kinase kinase 1 (MEK) [39]. Cell viability of Ctsl −/− cells was reduced to 50% after six-day Trametinib treatment ( Figure 8B). Apoptotic cell death of Ctsl −/− cells was massively increased upon treatment, confirming the reliance of the cells on MAPK signaling for survival ( Figure 8C). These results point to MAPK signaling as an adaptive cell response to lysosomal stress and suggest a therapeutic benefit for dual targeting of MAPK and lysosomal pathways. A substantial reduction of the phosphorylation of both ERK isoforms, p44 and p42, was corroborated by Western blot upon Trametinib treatment ( Figure 8D).

Discussion
Making use of a newly established mouse line that enables the conditional genetic ablation of Ctsl in the transgenic MMTV-PyMT breast cancer model, we report that Ctsl-deficiency in mammary epithelial tumor cells resulted in a considerable delay in the emergence of tumors. Additionally, endstage tumors were delayed, morphologically altered and displayed large dead cell areas. In contrast, the selective knockout of Ctsl in myeloid cells had no effects on primary tumors, affecting exclusively the metastatic seeding of PyMT tumor cells to lungs. Hence, our data support a cancer-cell intrinsic role of Ctsl in primary tumor growth while the development of metastasis is influenced by the tumor microenvironment.
A hyperactive PI3K/AKT/mTOR pathway is a characteristic of up to 70% of breast cancers of all molecular breast cancer subtypes, i.e., Luminal A, Luminal B, Her2 (ErbB2)-enriched, as well as basal- A hyperactive PI3K/AKT/mTOR pathway is a characteristic of up to 70% of breast cancers of all molecular breast cancer subtypes, i.e., Luminal A, Luminal B, Her2 (ErbB2)-enriched, as well as basal-like, triple-negative cancers [40]. In MMTV-PyMT mice the expression of the polyoma-virus middle T antigen (PyMT) in mammary epithelium starts in an estrogen-dependent manner in adolescent female mice. PyMT is anchored to the inner leaflet of the plasma membrane, where it becomes swiftly phosphorylated by the SRC proto-oncogene product. The phospho-moieties serve as activating docking sites for intracellular proteins including Shc and PI3K/p85 [41]. Therefore, the MMTV-PyMT mice develop metastasizing breast cancer mainly due to hyperactive mitogen-activated-protein (MAP)-kinase and PI3K signaling pathways, thereby modelling the oncogenic signaling present in a majority of human breast cancers. In addition, breast cancer progression and spontaneous metastasis to the lung of MMTV-PyMT mice strongly depend on pro-tumor polarization of myeloid cells [42]. In summary, the MMTV-PyMT mice are an excellent and widely used model for studying PI3K-related oncogenic signaling and myeloid cell function in breast cancer. Based on transcription profiles, MMTV-PyMT cancers have been related to Luminal-type human breast cancers [43].
The contribution of cysteine cathepsins' proteolytic activity in the tumor microenvironment has been previously documented [1,44]. Studies transferring PyMT cells in Ctsb +/+ or Ctsb −/− mice revealed fewer lung colonies, together with an enhanced expression of Ctsb in the lung-infiltrating macrophages, demonstrating a substantial contribution of this protease in promoting lung colonization [6]. In addition, cathepsin S (Ctss) has been shown to be a relevant factor in the promotion of breast-to-brain metastasis, where experimental brain metastasis was only reduced by the combined depletion in macrophages and tumor cells [45]. The deletion of cysteine cathepsins such as Ctss and Ctsz in tumor-associated macrophages (TAMs) contributed not only to reduced metastases but also to limit the invasiveness of lesions in pancreatic cancer [46,47]. However, in contrast to the previous work on Ctsb, Ctss, and Ctsz, in the present study the conditional genetic ablation of Ctsl in myeloid cells increased the metastatic seeding of breast cancer cells to the lungs of the mice. Notably, neither in vitro differentiation of LysM-Cre/Ctsl −/− bone marrow-derived macrophages ( Figure S2), nor the number of immune cells in primary tumors and lungs were significantly altered ( Figure S4). Therefore, we hypothesize that Ctsl-deficiency alters polarization and/or function of bone marrow-derived TAMs.
The deletion of the Ctsl in mammary epithelial cells has been shown to have considerable effects on diverse breast cancer traits, exhibiting reduced proliferation, enhanced cell death, and morphological changes. Noteworthy, non-tumor Ctsl −/− mice raise their litters normally, implying that the lack of Ctsl does not impair the normal structure nor the function of mammary glands [15]. In contrast, we observed a worse histopathological tumor grading in breast cancers lacking Ctsl, with cellular dedifferentiation and loss of normal structural features such as tubuli. We planned to address the mentioned alteration in cell-based approaches, but initially we failed several attempts to establish a Ctsl −/− cell line by isolating the cells directly from MMTV-PyMT primary cancers and immortalize them in vitro. As presented in Figure 4, we overcame this hurdle by isolating a cell line from a Ctsl-expressing PyMT/Ctsl fl/fl tumor and subsequently inducing the deletion of the protease by transduction of a Cre recombinase expression system. This allowed to establish a Ctsl-deficient cell model, with following experiments confirming that Ctsl −/− MMTV-PyMT +/T breast cancer cells display an increased susceptibility for cell death and diminished cell growth. In the search for mechanisms that could explain the observed effects, we first found an increased accumulation of enlarged acidic vesicles. An increase in the amount and volume of vesicles has been already described for many of the spontaneous phenotypes of constitutive Ctsl −/− mice, in diverse cell types such as keratinocytes [23,24,48], as well as cardiomyocytes [16,49].
The nutrient sensing capability of lysosomes depends on the mTORC1 protein complex [50]. The activation and function of its main effector, the serine/threonine protein kinase mTOR depends on the correct assembly of mTORC1 on the cytosolic side of the lysosomal membrane, where it integrates growth signals and nutrient status such as amino acid levels [37,51]. Importantly, the PyMT oncogene that drives the cancer model used in our study is a strong inducer of the PI3K-AKT-mTOR signaling axis [41].This is relevant because the majority of human breast cancers show a hyper-activation of this pathway due to activating point mutations, gene amplifications, or inactivation of phosphatases such as PTEN [52,53]. In terms of the metabolic state of Ctsl −/− epithelial breast cancer cells, our data show reduced levels of ketogenic amino acids, suggesting considerable metabolic changes in those cells. Furthermore, the levels of major activators of mTORC1, such as arginine and leucine, are decreased. In line with this, we found evidence of decreased mTOR phosphorylation, as well as of decreased phosphorylation of one of its major targets, the p70 S6 kinase. We propose that deficiency of the lysosomal endoproteinase Ctsl in mammary cancers results in a lysosomal storage phenotype, altered amino acid levels, and interference with mTOR signaling, causing the six-week delay observed for MMTV-Cre/Ctsl −/− breast tumors. mTORC1 positively regulates cell growth and survival, while negatively regulating lysosomal biogenesis and autophagy [35]. Our in vitro data provided evidence for increased lysosomal biogenesis, rather than macroautophagy induction. Although autophagy can be started by both Ctsl fl/fl and Ctsl-deficient breast cancer cells, its termination might be impaired in the last cell line. In addition, our findings provide evidence for a growth disadvantage of Ctsl −/− PyMT +/T cells. This prompted us to search for mechanisms that enable Ctsl −/− cancer cells to survive and slowly proliferate further. It has been shown for Ctsl −/− keratinocytes that Ctsl deficiency enhanced recycling of growth factors, such as epidermal growth factor (EGF) [23,54]. PyMT also activates the Raf/ERK cascade to induce tumorigenesis [55]. Thus, we explored the mitogen-activated protein (MAP)-kinase signaling in the Ctsl −/− PyMT +/T cells. We could find a trend towards increased ERK phosphorylation, which resulted, however, in a significant reduction of cell viability and a major increase in cell death upon MEK inhibition.

Mice
The embryonic stem (ES) cell clone for conditional targeting of mouse Ctsl by a so-called knock-out first strategy (Ctsl tm1a(EUCOMM) ) was generated by and obtained from the European Conditional Mouse Mutagenesis Program (EUCOMM) (Figure S1A II). The ES cells (JM8.N4; background C57BL/6N) were injected into blastocysts and chimeric mice transmitting the targeted allele were established as founder mice. Subsequently, the neomycine selection cassette and the lacZ reporter were removed by FLPe-mediated recombination of the Frt sites included in the construct ( Figure S1A II and III). The resulting C57BL/6 mice harboring the loxP-flanked (floxed) Ctsl gene (Ctsl fl/fl ) were crossed either with the transgenic mouse strains MMTV-Cre +/T or LysM-Cre +/T , both bearing the MMTV-PyMT (polyomavirus middle T) antigen. Mice were monitored by palpations twice a week starting at 8 weeks of age until tumors reached end-stage. Animal work was performed in accordance with the German law for animal protection (Tierschutzgesetz) as published on 18 May 18, 2006, with last amendment on 20 November, 2019. Ethics approval registration number is G14/18 RP, regional council Freiburg. Female tumor-bearing mice were monitored by palpations twice a week starting at 8 weeks of age, and tumor onset as well as tumor progression were documented until tumors reached end-stage.

Genotyping
Mouse genotyping was accomplished by PCR. Lysates were diluted and combined with FastGene Taq 2× Ready Mix (#LS31, Nippon Genetics, Europe). PCR reactions were run with a pre-established, optimized protocol for genotyping and end-products were analyzed by gel electrophoresis.

Mammary Whole Mounts
Mammary fat pads of 8-week-old mice were removed and extended over a glass microscopy slide. Slides were fixed and stained in carmine red-alum solution (2% carmine red, 5% aluminum potassium sulfate). Slides were imaged in an automated manner at 4× magnification and stitched automatically by means of the microscope software package. Slides were stored in methyl salicylate (#M 6752, Merck, Darmstadt, Germany) after imaging [55]. Stained areas were calculated using the Image J software.

Histological Grading of Tumors
Histopathological grading and estimation of the necrotic surface were performed on HE-stained tumor sections in a blinded manner by an experienced pathologist using the Elston/Ellis scoring system [31].

Quantification of Metastasis
For the quantification of metastasis, lung slides were stained for Ki67 in order to differentiate rapidly proliferating breast tumor cells from the lung parenchyma. For a correct representation of metastasis across the whole lung, three spatially separated planes per lung were quantified. By means of the free software Image J, the total area of all three lung slides was calculated, together with the areas of the metastasis. The metastatic burden was calculated by dividing the sum of the total area of the metastasis present in each slide by the sum of the total area of the lung, with the total area of the lung set to 100%. The mean size was calculated by dividing the area of the metastasis by the number of metastases in each of the three slides per lung.

Flow Cytometry of Tissues and Cells
For T cell phenotyping, 10-week-old mice were anesthetized and blood was withdrawn. The spleen and thymus were harvested and disrupted by passing through a cell strainer. Cells were seeded at a density of 1 × 10 5 cells per well and stained with a mixture of PE Rat Anti-Mouse CD8a Clone 53-6.7 (#553033, BD Biosciences) and FITC Rat Anti-Mouse CD4 Clone RM4-5 (#553046, BD Biosciences).
For immune-cell phenotyping, tumors and lungs were dissected carefully from end-stage tumor-bearing mice and disrupted by enzymatic dissociation as performed for tumor cell isolation. For Annexin V and LysoTracker TM analysis, cells were detached after 3-day culture and washed repeatedly in Annexin V binding buffer. Cells were subsequently stained with 1:500 FITC Annexin V (#556419, BD Biosciences) or with 1:5000 LysoTracker™ Green DND-26 (#L7526, Thermo Fisher Scientific, Waltham, MA, USA).

Microvessel Density (MVD) Measurement
CD31-stained endothelial cells (solid clusters, single cells and endothelial cell clusters with lumina) were quantified within 10 high power fields (HPF)/400-fold magnification by light microscopy by an experienced pathologist in a blinded manner in the tumor mass. The counting was performed according to a modified score reported in Weidner et al. (1991) in tumor hot spots [32].

Generation and Culture of Breast Cancer Primary Cell Lines
End-stage tumors of Ctsl fl/fl mice were carefully harvested and disrupted by enzymatic dissociation with DNAse I (#DN25, Sigma-Aldrich), Hyaluronidase Type I-S (#H3506, Sigma-Aldrich) and Collagenase Type IV (#C5138, Sigma-Aldrich). Cells were kept as monolayer cultures in a cell culture incubator under sterile conditions at 37 • C, 5% CO 2 , and 91.0% rH. Once spontaneously immortalized after 10-12 passages, mammary breast cancer cells were cultured in Dulbecco's modified Eagle's medium (Gibco/Invitrogen, Paisley, UK) supplemented with 1% penicillin/streptomycin, 1% L-glutamine (both from Gibco/Invitrogen), and 10% fetal calf serum (PAN-Biotech, Aidenbach, Germany), in humidified air containing 5% CO 2 . Cells were also kept under low-serum conditions (1%). Cells were transduced with a doxycycline-inducible retroviral Cre-construct (kindly provided by Tilman Brummer). Induction by 2 µg/mL doxycycline took place every 48 h for 6 days. In the indicated experiments, 10 µM of the pan cysteine Cathepsin inhibitor E64d were supplemented to the full medium of Ctsl fl/fl cells every 48h for 4 days prior to lysis.

RT-Cell Proliferation Monitoring
Cells were seeded at a density of 0.5 × 10 5 and cultured for 3 days. Proliferation was monitored by means of an xCELLigence DP device (OLS OMNI Life Sciences, Bremen, Germany) for 72 h, using E-16 plates (#2801032, ACEA Biosciences Inc., San Diego, CA USA) following manufacturer's guidelines.
For Ctsl fl/fl and Ctsl −/− co-culture experiments, cells were seeded at a ratio of 1:1 and cultured for a period of 6 days. Lysates were generated and analyzed by droplet digital PCR, as mentioned before.
For experimental metastasis assay, co-cultures containing a different ratio of Ctsl fl/fl and Ctsl −/− were harvested at a confluence of 80%. One hundred microliter cell suspension of each of the 2 different cell mixtures containing 42% and 75% Ctsl −/− cells were injected via tail vein injection in immunodeficient mice separately. After 28 days, mice were sacrificed and lungs were harvested. Right lungs were fixed and embedded in paraffin for histological analysis. Left lungs were lysed overnight and PyMT and Cre DNA were analyzed by droplet digital PCR, performed following the aforementioned standard protocol. Primers and probe for PyMT are as follows: PyMT Probe

Cell Movement Tracking Assay
Cells were plated at low density, 0.015 × 10 6 cells per well in 24 well plates until attachment was guaranteed (6 h). Plates were imaged using a JuLI TM Stage real time imaging device for a period of 12 h at standard culture conditions. Cell movement was tracked and mean velocity was calculated using the MTrackJ FIJI/ImageJ plugin.

SDS-Page and Western Blotting
Thirty to fifty micrograms of lysates were loaded onto 8%, 10%, or 12% SDS-polyacrylamide gels and electrophoretically separated by SDS-PAGE, and transferred to nitrocellulose membranes (Amersham GE Healthcare, Chalfont St Giles, UK) using a wet-blot system (Bio-Rad). They were blocked with 4% BSA in TBS-0.1% Tween and incubated with the primary antibody at a concentration

RNA Isolation and RT-PCR
An RNeasy Mini Kit ® (#74104, Qiagen) was used to isolate RNA from cell lysates, which was subsequently transcribed to cDNA using the iSCRIPT ® cDNA synthesis system (#1708890, Bio-Rad).
To analyze lysosomal biogenesis, the following primers were used: Cathepsin B forward:

Amino Acid Profiling by Metabolomics
Cells were cultured for 72 h under standard conditions (10% FCS) and washed three times with 0.9% NaCl. Monolayer cultures were harvested on ice with 1 mL Metabolomics lysis buffer (90% Methanol, 2.5 µg/mL Isoguanosine hydrate (#NI07234, Carbosynth Limited, Newbury, UK), 1 µg/mL O-Methyl-L-Tyrosine (#H63096, Alfa Aesar/Thermo Fischer Scientific, Schiltigheim, France). Samples were homogenized and transferred to LC-MS glass vials with the injection volume set to 5 µL. A pool of all samples containing 20 µL of each supernatant was used as quality control. Amino acids were separated by a Waters Acquity UPLC BEH Amide column (150 mm × 2.1 mm, 1.8 µm, Waters Corporation, Milford, MA, USA) at 50 • C using a flow rate of 0.6 mL/min with water + 0.1% formic acid as buffer A and acetonitrile (AE70.2, Roth) + 0.1% formic acid as buffer B. Intensities of all 20 amino acids were measured by liquid chromatography-ESI-MS/MS on a 1290 Infinity UHPLC system coupled to a 6460 triple quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA, USA) via an Agilent Jetstream electrospray ionization source (Agilent Technologies), in a randomized manner. The following gradient was applied: 0.0-0.1 min at 90% B, 0.01-0.02 min to 85% B, 0.02-1 min to 75% B, 1.0-2.0 min to 40% B. Finally, the column was washed for 3 min at 50% B and re-equilibrated for 4 min at 90% B. Total run time was set to 9 min. The applied MS settings were: capillary voltage, 4000 V; nozzle voltage, 500 V; gas temperature, 300 • C; gas flow, 7 L/min; sheath gas temperature, 350 • C; sheath gas flow, 11 L/min; nebulizer pressure, 50 psi. MS/MS spectra were acquired in dynamic multiple reaction monitoring (MRM) mode. Optimized settings and mass spectral transitions were taken from previous reports [56].

MTT Cell Viability Assay
Cells were plated on day 5 of Trametinib treatment at a density of 1 × 10 4 cells per well in a 96 well transparent plate. After 24 h, 150 µL of a 1:10 dilution of the MTT reagent (#Ab146345, Abcam) in Indicator-free full-DMEM were added. Cells were incubated with MTT reagent until DMSO was added for detection. Absorption was measured at 570 nm and 650 nm with an Enspire Perkin Elmer multimode plate reader.

Statistical Analysis
All quantitative data presented are reported as mean ± standard error of the mean (S.E.M.) of at least 5 biological replicates from independent experiments. Single data points were plotted in the cases where n < 5. n value, p value, as well as statistical test used in each analysis, are stated in the figure legends. One or two sample Student´s t-test (two tailed) was used for comparison of the data from experimental groups. Analysis of variance (ANOVA) followed by a post-hoc Tukey test was used for multiple-group comparison. The two-tailed non-parametric Wilcoxon-Mann-Whitney-test was used if data were aberrant from normal distribution. Fischer´s exact test was used to analyze categorical data. Origin 2018 R was used to plot the data as well as for its statistical analysis.

Conclusions
Ctsl deficiency in mammary epithelium strongly impairs lysosomal function, resulting in altered mTOR signaling. To bypass the effects of Ctsl deficiency, Ctsl −/− cancer cells make use of alternative growth pathways such as the MAP-kinase pathway. Taken together, our findings suggest clinically relevant effects of a combined inhibition of Ctsl and the PI3K/MAP-kinase pathways in breast cancer cells. Furthermore, our study indicates a cancer-cell autonomous role of Ctsl in primary tumor growth, whereas metastasis is also dependent on Ctsl in bone marrow-derived cells.