Extracellular cystatin SN and cathepsin B prevent cellular senescence by inhibiting abnormal glycogen accumulation

Cystatin SN (CST1), a known inhibitor of cathepsin B (CatB), has important roles in tumor development. Paradoxically, CatB is a member of the cysteine cathepsin family that acts in cellular processes, such as tumor development and invasion. However, the relationship between CST1 and CatB, and their roles in tumor development are poorly understood. In this study, we observed that the knockdown of CST1 induced the activity of senescence-associated β-galactosidase, a marker of cellular senescence, and expression of senescence-associated secretory phenotype genes, including interleukin-6 and chemokine (C-C motif) ligand 20, in MDA-MB-231 and SW480 cancer cells. Furthermore, CST1 knockdown decreased extracellular CatB activity, and direct CatB inhibition, using specific inhibitors or shCatB, induced cellular senescence. Reconstitution of CST1 restored CatB activity and inhibited cellular senescence in CST1 knockdown cells. CST1 knockdown or CatB inhibition increased glycogen synthase (GS) kinase 3β phosphorylation at serine 9, resulting in the activation of GS and the induction of glycogen accumulation associated with cellular senescence. Importantly, CST1 knockdown suppressed cancer cell proliferation, soft agar colony growth and tumor growth in a xenograft model. These results indicate that CST1-mediated extracellular CatB activity enhances tumor development by preventing cellular senescence. Our findings suggest that antagonists of CST1 or inhibitors of CatB are potential anticancer agents.

Cysteine cathepsins perform various functions, including the processing of proteins during antigen presentation, bone modeling and epidermal homeostasis. 1 Previous reports have shown that the activation of extracellular cathepsins such as cathepsin B (CatB) has an important role in the degradation of extracellular matrix proteins, including collagen, laminin and fibronectin, facilitating tumor metastasis through the remodeling of the extracellular environment. [2][3][4] Moreover, CatB promotes the proliferation, invasion and metastasis of some tumor cells. [3][4][5][6] The proteolytic activities of CatB are negatively regulated by specific inhibitory proteins belonging to the type 2 cystatin family. 7 Cystatin SN (CST1), encoded by CST1, is a member of the type 2 cystatin family, and the induction of CST1 expression is associated with tumorigenesis, increased cancer cell proliferation and invasion, and tumor recurrence. [8][9][10][11] As CatB is a functional protease and CST1 is its inhibitor, it is highly paradoxical that both of them contribute to tumorigenesis. To date, the underlying relationship between CST1 and CatB, and their roles in tumor development, remains poorly understood.
Replicative cellular senescence features a permanent cell cycle arrest, resulting in limited cell proliferation. 12 Repeated DNA replication during normal cell proliferation contributes to the shortening of telomeres, which causes cell cycle arrest and genomic instability. 13,14 Premature senescence, an accelerated senescence phenotype, can be induced by various stresses such as oxidative stress, ionizing radiation 15,16 and anticancer chemotherapy. 17,18 Senescent cells exhibit a diverse range of common characteristics, including the arrest of the cell cycle, activation of tumorsuppressor networks, 19,20 morphologic changes, 21,22 induction of senescence-associated β-galactosidase (SA-β-gal) activity 23 and adoption of a senescence-associated secretory phenotype (SASP). 24 Generally, cancer cells avoid cellular senescence by developing gene during carcinogenesis and malignant cell transformation. 25 Thus, cellular senescence has been considered a tumor-suppressive mechanism that inhibits the uncontrolled proliferation of cancer cells. 26,27 In this study, we observed that CST1 knockdown induced cellular senescence and decreased extracellular CatB activity in MDA-MB-231 and SW480 cancer cell lines. We hypothesized that the relationship between CST1 and CatB contributed to tumorigenesis via the prevention of cellular senescence. Moreover, we further investigated the molecular mechanisms underlying CST1-mediated prevention against cellular senescence in cancer cells.

Results
Knockdown of CST1 induces cellular senescence. To investigate the role of CST1 in tumorigenesis, we screened human colon and breast cancer cell lines and discovered that CST1 was highly expressed in and was secreted into the culture medium by SW480 and MDA-MB-231 cells (Supplementary Figure 1). In SW480 colon cancer and MDA-MB-231 breast cancer cells, CST1 knockdown using the pLenti expression system containing a short hairpin RNA (shRNA) for CST1 (shCST1; Figure 1a) induced striking morphologic changes, including an enlarged nucleus and flattened cytoplasm, and increased SA-β-gal activity compared with pLenti-mock control cells (Figure 1b (upper). Senescence under the indicated condition was quantified as the percentage of SA-β-gal-positive cells (lower). Scale bars, 50 μm. (c) The expression of SASP genes induced by CST1 knockdown. RNAs were extracted from cells, and real-time PCR analysis was performed to quantify the transcripts of CST1, IL-6 and chemokine ligand 20 (CCL20). (d) Analysis of anchorage-dependent cell proliferation. Cells were seeded on to six-well plates (5 × 10 4 cells per well) and cultured for the indicated number of days. The number of cells was counted using a hemocytometer. One-way ANOVA was used for statistical analysis (*Po0.05). (e) Analysis of anchorage-independent cell proliferation. Cells were seeded on to complete DMEM containing agarose (1 × 10 5 cells per well) and cultured for 12 days. Formed colonies were stained with 0.005% crystal violet and quantified under a light microscope. Scale bar, 10 μm. (f) Analysis of tumor growth in the xenograft model. MDA-MB-231 cells transduced with a control shRNA or shCST1 (1 × 10 6 cells) were subcutaneously implanted into the left or right flanks of athymic nude mice (n = 8). Tumor volume was measured on the indicated days. One-way ANOVA was used for statistical analysis (*Po0.05). (g) Protein expression of p-Rb, total Rb, p21, cyclin D1 or CST1 was evaluated by western blotting. *Po0.05; **Po0.01; ***Po0.001 by Student's t-test exhibited significantly suppressed in vivo tumor growth ( Figure 1f). The inhibition of tumor cell growth may result from G 0 /G 1 -phase cell cycle arrest ( Supplementary Figures 3a  and b) induced by a decrease in the expression of cyclin D1 and cyclin-dependent kinase 2 (CDK2) expression ( Supplementary Figures 3c and d). To confirm that the G 0 /G 1 -phase cell cycle arrest was caused by CST1 knockdown, we conducted western blotting and found that CST1 knockdown suppressed cyclin D1 and phosphoretinoblastoma (p-Rb) and induced p21 (Figure 1g). These results suggest that CST1 knockdown reduces cancer cell proliferation through the induction of cellular senescence.  Figure 1). We found that SA-β-gal activity in shCST1-MDA-MB-231 cells was inhibited by supplementation with conditioned medium obtained from MCF-7 cells stably expressing CST1, but not by supplementation with wild-type MCF-7 conditioned medium ( Figure 2b). To determine whether the increased SA-β-gal activity was solely induced by knockdown of CST1 in MDA-MB-231, MDA-MB-231 cells were seeded in a sixwell plates coated with immunoglobulin G (IgG) or purified recombinant CST1-Ig (rCys-SN) (Supplementary Figure 4b). rCys-SN effectively suppressed SA-β-gal activity induced by CST1 knockdown compared with those treated with phosphate-buffered saline (PBS) or IgG alone (Figure 2c). These results show that extracellular CST1 is critical for the inhibition of cellular senescence.
Downregulation of extracellular CatB activity by CST1 knockdown induces cellular senescence. As CST1 is a known representative inhibitor of CatB that is located in the extracellular compartment and lysosomes, we examined the role of CatB on senescence induced by CST1 knockdown. Unexpectedly, we found that CST1 knockdown suppressed  Figure 5). CatB protein levels were unchanged under CST1 knockdown conditions. To examine the CST1-mediated positive regulation on CatB activity, we analyzed CatB activity in the culture supernatant obtained from CST1 knockdown MDA-MB-231 cells replenished with rCys-SN and observed that the reduced CatB activity caused by CST1 knockdown was restored with the addition of rCys-SN ( Figure 3b). To confirm whether this reduced CatB activity was directly associated with CST1 knockdown-mediated cellular senescence, we inhibited CatB activity using a shRNA for CatB (Figure 3c), or chemical inhibitors of CatB (CA-074 or inhibitor III; Figure 3e). Notably, the SA-β-gal-positive cell population was significantly increased by CatB knockdown (Figure 3d) and CatB inhibitor treatment ( Figure 3f).
Recently, it has been reported that CST1 neutralizes the inhibitory effect of cystatin C (CST3) on CatB activity. 28 Therefore, we investigated whether cystatin C was associated with the regulatory function of CST1 on extracellular CatB activity. CST3 knockdown rescued extracellular CatB activity CST1 knockdown induces cellular senescence through modulation of GSK3β activity. To explore how CST1 inhibits cellular senescence, a human phospho-mitogenactivated protein kinase (MAPK) array containing p38MAPK, p70S6K and glycogen synthase kinase 3β (GSK3β) was used. Although the active phosphorylation of p70S6K (p-p70S6K) and inhibitory phosphorylation of GSK3β (p-GSK3β) were increased, the active phosphorylation of p38MAPK (p-p38) was decreased (Supplementary Figure 7a). The change in the phosphorylation status of GSK3β or p38MAPK was confirmed in both CST1 knockdown MDA-MB-231 and SW480 cancer cells (Figure 4a). Moreover, we found that the inhibition of GSK3β using GSK3β inhibitors, such as SB415286 and SB216763, induced SA-β-gal activity (Figure 4b). Conversely, a p38MAPK inhibitor, SB203580, did not induce SA-β-gal activity (Supplementary Figure 7b). These results indicate that the GSK3β signaling pathway is involved in CST1 ). To provide more evidence that GSK3β activity is directly involved in CST1 knockdown-mediated cellular senescence, we introduced wild-type or mutant GSK3β (GSK3β-S9A) expression vectors into MDA-MB-231 cells. Although SA-β-gal activity induced by CST1 knockdown was unaltered in mock vector and wild-type GSK3β-expressing cells, the ectopic expression of GSK3β-S9A (active form) significantly suppressed the SA-β-gal activity induced by CST1 knockdown (Figure 4d). CatB knockdown also induced the inhibitory phosphorylation of GSK3β at serine 9 ( Figure 4e). These results show that CST1 knockdown induces cellular senescence through the inhibition of GSK3β activity, which is mediated by extracellular CatB activity.
Inhibition of GSK3β activity by CST1 knockdown induces glycogen accumulation. Reportedly, enhanced glycogenesis has been shown to be involved in cellular senescence via GSK3β/glycogen synthase (GS) modulation. 29 Therefore, we investigated whether the inhibition of GSK3β induced the dephosphorylation of GS, thereby activating it and enhancing glycogenesis, which may induce cellular senescence. First, we confirmed that the dephosphorylation of GS was regulated by extracellular CST1. Knockdown of CST1 induced active GS (Figure 5a), whereas the reconstitution of CST1 knockdown cells with rCys-SN induced the phosphorylation of GS to its inactive form (Figure 5b), a process associated with glycogen accumulation ( Figure 5c). As CST1 downregulated extracellular CatB activity, we confirmed that direct CatB inhibition, using shCatB or CA-074, decreased phosphorylation of GS in MDA-MB-231 cells (Supplementary Figure 8). To examine whether this glycogen accumulation was associated with the CST1-mediated modulation of GSK3β activity, we ectopically expressed wild-type GSK3β or GSK3β-S9A in CST1 knockdown MDA-MB-231 cells. Although glycogen accumulation was strongly observed in mock or wild-type GSK3β-expressing CST1 knockdown cells, the ectopic expression of GSK3β-S9A abrogated glycogen accumulation (Figure 5d). Following the direct inhibition of GSK3β using the chemical inhibitors, SB415286 or SB216763, glycogen accumulation was also observed (Supplementary Figure 9). These results imply that extracellular CST1-mediated CatB activity influences the GSK3β-GS pathway to regulate the glycogen metabolism associated with cellular senescence.
Glycogen accumulation induced by GS is a direct cause of cellular senescence mediated by CST1 knockdown. To investigate the role of glycogen accumulation on cellular senescence in CST1 knockdown cells, we silenced GS in CST1 knockdown MDA-MB-231 cells. We confirmed that the knockdown of GS did not alter the phosphorylation of GSK3β at serine 9, which is upstream of GS (Figure 6a). Furthermore, we confirmed that GS knockdown suppressed glycogen accumulation in CST1 knockdown MDA-MB-231 cells (Figure 6b). Importantly, SA-β-gal-positive cells induced by CST1 knockdown were converted to SA-β-gal-negative cells following GS knockdown (Figure 6c). Associated with the inhibition of cell proliferation during cellular senescence, Rb phosphorylation and cyclin D1 expression, reduced by CST1 knockdown were recovered by GS knockdown in CST1 knockdown MDA-MB-231 cells. In addition, levels of the cell cycle regulator p21, which were increased by CST1 knockdown, were reduced by GS knockdown (Figure 6d). Similarly, the expression of SASP genes, including IL-6 and CCL20, induced by CST1 knockdown was abrogated by GS knockdown (Figure 6e). These results show that extracellular CST1 negatively regulates GS, which induces abnormal glycogen accumulation, resulting in cellular senescence. Together, our results show that extracellular CST1 has an important role in tumorigenesis by preventing the cellular senescence induced by glycogen accumulation.

Discussion
CST1 has been known as an endogenous inhibitor protein of cysteine proteases. 7 CatB, a representative cysteine protease, is highly expressed in breast, 30 colorectal 31 and lung cancers. 32 Moreover, CatB-mediated remodeling of the tumor environment directly affects cancer cell behaviors, such as invasion and metastasis. [33][34][35] Paradoxically, CST1 has been proposed to be a novel marker in several tumors, including non-small cell lung cancer, pancreatic cancer, colorectal cancer and gastric cancer. [8][9][10][11] The upregulated expression of CST1 is associated with proliferation, invasion and metastasis of cells, which are the main characteristics of tumor malignancy. 8,10,36 To date, the cooperative effect of CatB and CST1 during tumorigenesis remains poorly defined. In this study, we observed that extracellular CatB activity was positively regulated by CST1 expression and negatively regulated GS activity, inducing abnormal glycogen accumulation, which is a direct cause of cellular senescence. Senescent cells are characterized by growth arrest, resistance to apoptosis, altered gene expression and the expression of senescence markers. 16 Senescent cells usually exhibit G 1 -phase cell cycle arrest, which is caused by the expression of cell cycle inhibitors, such as p21 and p16. 21,27,37 The most useful marker for identifying senescent cells is SA-βgal. 23 Furthermore, senescent cells are known to secrete SASP proteins, such as IL-6, IL-8 and CCL20, into the culture medium. 16,38,39 In this study, senescence phenotypes, including growth arrest accompanying G 1 -phase cell cycle arrest, SA-β-gal activity and increased SASP secretion, were induced by the shRNA-mediated knockdown of CST1. CST1 encodes a typical secretory protein presenting a secretory signal peptide at the N-terminus. 7 When we treated CST1 knockdown cells with CST1-enriched conditioned medium or rCys-SN, CST1 knockdown-mediated cellular senescence was significantly inhibited. These results suggest that extracellular CST1 prevents cellular senescence. Recently, our colleagues reported that CST1 increases CatB activity by neutralizing the function of cystatin C, 28 a known inhibitor of CatB with a broader spectrum of inhibitory activity than CST1. 40 CST1 knockdown significantly decreased extracellular CatB activity without decreasing intracellular CatB activity. Expectedly, CatB knockdown also effectively induced cellular senescence. In addition, two independent inhibitors of CatB, CA-074 and cathepsin inhibitor III, induced cellular senescence. In particular, CA-074 is known to be non-permeable, potent, and specific a CatB inhibitor. 41,42 Our results suggested that extracellular CST1 contributes to the maintenance of extracellular CatB activity, thereby inhibiting cellular senescence. A recent study demonstrated that lysosomal CatB is released into the cytosol where it cleaves sirtuin 1, which mediates stress-induced premature senescence under stress conditions. 43 Therefore, we propose that CatB has two opposing roles on cellular senescence depending on its localization. Although extracellular CatB prevents cellular senescence under normal conditions, intracellular lysosomal CatB induces cellular senescence under stress conditions. We screened and confirmed signaling pathways involved in CST1 knockdown-mediated senescence using a human phospho-MAPK array and western blotting, and discovered that the inhibitory phosphorylation of GSK3β was regulated by extracellular CST1 and CatB. Unfortunately, we did not reveal how GSK3β phosphorylation was regulated under CST1 knockdown conditions. GSK3β is a multifunctional serinethreonine kinase that functions in various signaling pathways. It was first identified as an enzyme that regulates glycogen metabolism through GS inactivation. 44,45 GSK3β phosphorylation at the serine 9 residue creates an inactive form of GSK3β. This inactive form of GSK3β induces an increase in dephosphorylated GS, resulting in glycogen accumulation. 46 Abnormal glycogen accumulation, which inhibits glucose utilization, has been shown to induce cellular senescence in various models. 29,47,48 The abnormal glycogen accumulation induced by CST1 or CatB knockdown may trigger intracellular signal transduction because glycogen directly activates adenosine monophosphate-activated protein kinase, 49 which can lead to cellular senescence if persistently activated. 50 In this study, the inhibitory phosphorylation of GSK3β was increased under CST1 or CatB knockdown, resulting in glycogen accumulation and cellular senescence. Conversely, a constitutively active mutant GSK3β (GSK3β-S9A) inhibited the glycogen accumulation and cellular senescence induced by CST1 knockdown. CST1 or CatB knockdown also induced a decrease in GS phosphorylation and glycogen accumulation, which were effectively inhibited by GSK3β-S9A. GS knockdown significantly diminished cellular senescence phenotypes, including SA-β-gal activity, cell cycle regulator modulation and SASP gene expression, as well as inhibiting glycogen accumulation in CST1 knockdown cells.
To date, studies on extracellular CatB and CST1 have focused on cell proliferation, invasion and metastasis related to extracellular matrix degradation during tumorigenesis. In this study, we revealed a new function of the extracellular CST1-CatB axis in the modulation of glycogen metabolism via the regulation of the GSK3β-GS axis, contributing to the prevention of cellular senescence. Finally, we suggest that extracellular CST1, a positive regulator of extracellular CatB, contributes to tumorigenesis by preventing cellular senescence through the inhibition of abnormal glycogen accumulation. Western blot and phopho-MAPK array. Cell lysates were extracted with CETi protein extraction solution (TransLab, Daejeon, Korea). Thirty to fifty micrograms of protein extract was used for western blotting with antibodies against p-Rbs, p21, cyclin D1, GSK3β, p-GSK3β, CatB, GS, p-GS, p38MAPK, p-p38MAPK, p70S6K, p-p70S6K (Cell Signaling, Denvers, MA, USA), α-tubulin, mouse IgG peroxidase conjugate, rabbit IgG peroxidase conjugate (Sigma-Aldrich Corp.) and Rb (Santa Cruz Biotechnology, Inc., Dallas, TX, USA). The human phospho-MAPK Array Kit (R&D Systems, Inc., Minneapolis, MN, USA) was used according to the manufacturer's instructions. A polyclonal antibody against CST1 was obtained from mice immunized with the purified protein.
SA-β-gal staining. SA-β-gal activity in cells was determined by cytochemical staining. Senescent cells were quantified as the number of SA-β-gal-positive cells per total cell counts in the same microscopic field. The number was determined by counting at least five random fields per sample.
Cell proliferation assay. For the analysis of anchorage-dependent cell proliferation, cells were seeded on to six-well plates at a density of 5 × 10 4 cells per well, and cell numbers were counted daily for 6 days. Anchorage-independent cell proliferation was determined by the colony forming assay. Cells (1 × 10 5 cells per well) were suspended in complete DMEM containing 0.35% agarose and laid over agar consisting of complete DMEM containing 0.5% agarose on a 12-well plate. Cells were cultured for 12 days, and colonies that developed were stained with 0.005% crystal violet and counted using light microscopy. For the analysis of xenograft tumor growth, cells (1 × 10 6 cells) mixed with Matrigel (BD Biosciences, San Jose, CA, USA) were subcutaneously injected into the right flank for shCST1 or left flank for the control shRNA of 6-week-old female athymic nude mice (Orient Bio Co., Seongnam, Republic of Korea). Tumor volume was calculated using the following equation: (maximal length × perpendicular width × height)/2. All procedures were approved by the Animal Ethical Committee of Gyeongsang National University, Jinju, Republic of Korea.
Determination of cathepsin activity. Cells were cultured in complete protein-free medium (Lonza). Cathepsin activity in the cell culture supernatant was quantified using a fluorometric assay kit (BIoVision Inc., Milpitas, CA, USA) according to the manufacturer's recommendations.
PAS staining assay. Glycogen staining was performed using the periodic acid-Schiff (PAS) reagent (Sigma-Aldrich Corp.). Cells were fixed in formalinethanol solution for 1 min at room temperature and stained with 0.5% Schiff's reagent for 15 min. After washing cells with PBS several times, stained cells were observed and photographed. Cells accumulating glycogen were quantified as the number of PAS-positive cells (magenta) per total cell count in the same microscopic field. The number was determined by counting at least five random fields per sample. To determine the origin of the staining, cells were treated with 0.5%