Soft matrix maintains pancreatic cancer cell dormancy in association with autophagic flux, lysosomal biogenesis, and YAP1 degradation

Gain-of-function mutation of KRas and loss-of-function mutation of p53 are critical for the malignancy of PDAC and the characteristics of human PDAC are recapitulated in transgenic KrasG12D/Trp53R172H mice [18]. We reasoned that even such oncogenically transformed cells might remain dormant if a soft niche is imposed, which may explain the delayed onset of PDAC in light of the nature of pancreatic softness. Tumor cellular dormancy is characterized by slow proliferation and G1 cell cycle arrest, reduced cell size, presence of tight junctions, and acquisition of stem cell markers [19, 20]. To simulate the softness of pancreatic stroma, murine KPC (UN-KPC961, Pdx1 driving KRas [12] and Trp53R172H) and Bxpc-3 cells were embedded in 3D collagen lattices (approximately 0.5 kPa) or seeded on collagen-coated 2D surface to mimic the fibrotic stiffness [21] (Fig. 1A). On the 2D stiff interface, the KPC and BxPC-3 cells swiftly went through growth cycle with a larger cell volume. However, in 3D collagen, the growth of cells was substantially slowed down and the cell cycle was arrested at the G0/1 phase, along with smaller cell size (Fig. S1A-F). Similarly, epithelial-to-mesenchymal transition (EMT) associated factors such as β-catenin, Twist1 and Vimentin were down-regulated, whereas E-cadherin was up-regulated in cancer cells embedded in the soft stroma (Fig. S1G-H). We also found that stem cell markers including Sox2, Sox4, Sox9, Notch1 and Jag1, as well as putative pancreatic cancer stem cell markers (CD24, CD44 and CD133) were all up-regulated in KPC cells embedded in soft 3D compared to those cultured on stiff 2D (Fig. S1I).

Fig. 1

Soft matrix confines pancreatic cancer cells in cellular dormancy in association with autophagic flux, lysosomal biogenesis, and YAP1 degradation. (A) Schematic illustrating cell culture on 2D stifness or in 3D softness. (B) RNA-seq analysis of genes involved in YAP pathway in KPC cells cultured on 2D monolayers or embedded in 3D collagen (N = 2). (C) Western blot analysis of KPC cells cultured on 2D monolayers or embedded in 3D collagen. (D) Western blot analysis of KPC cells embedded in 3D collagen at the indicated time points. (E-G) RNA-seq analysis of genes involved in lysosome biogenesis and autophagy pathways in KPC cells cultured on 2D monolayers or embedded in 3D collagen (N = 2). (H) Lysosome biogenesis was visualized through confocal microscopy analysis of Lysotracker/Hoechst staining in BxPC3 cells cultured on 2D monolayers or embedded in 3D collagen. (I) Confocal microscopy analysis of autophagic flux of murine or human pancreatic cancer cells expressing LC3-tandemRFP-GFP in cells cultured on 2D monolayers or embedded in 3D collagen. The experiments were repeated 2–3 times, and typical results were presented. (J-K) Western blot analysis of KPC and KC cells cultured on 2D monolayers or embedded in 3D collagen. Scale bar = 25 μm

Interestingly, the transcriptome analysis revealed that several target genes of the mechano-sensor YAP1, such as Ctgf, Gli2 and Birc5, were markedly down-regulated in 3D soft stroma (Fig. 1B). Moreover, both the total and phosphorylated levels of YAP1 were found to be markedly down-regulated in both KC and KPC cells embedded in soft 3D collagen (Fig. 1C-D, Fig. S1J), whereas the mRNA levels of Yap1 remained unchanged (Fig. S1K), indicating that YAP1 is regulated at the post-transcriptional level by the cells in 3D soft stroma. The Hippo kinase pathway is well known for its negative regulation of YAP1 though phosphorylation and subsequent ubiquitination and proteasome-mediated degradation [22]. Here, the mammalian counterparts of the Hippo kinases (Lats1 and Lats2), and the proteasome substrate p21 were markedly down-regulated in cells cultured in soft substrates, suggesting that a Hippo-independent mechanism is involved in stromal softness-mediated YAP1 degradation (Fig. 1C-D).

There are two main mechanisms of protein degradation in eukaryotic cells, including ubiquitin-proteasome system (UPS) and autophagy-lysosome pathway (ALP) [23]. Transcriptome expression profiling showed that mRNA levels of many lysosomal hydrolases (such as Ctsh, Ctsl, Ctsb, Ctsf, and Lipa), lysosomal membrane proteins (such as Lamp1, Lamp2, and M6pr) and autophagic genes were up-regulated in cells embedded in the soft 3D collagen lattices compared to those in cells cultured on 2D stiffed interface (Fig. 1E-G). Lysosomal biogenesis was further evaluated by Lysotracker-Red staining in human pancreatic cancer cells (BxPC-3), where increased number and size of lysosomal puncta were observed in cells embedded in the soft 3D collagen lattices compared to those in cells cultured on 2D interface (Fig. 1H). To monitor the impact of stromal rigidity on the autophagic flux of cancer cells, we transfected KPC and BxPC-3 cells with the tandem-fluorescence-tagged LC3 (tfLC3) constructs, followed by fluorescence imaging of GFP and RFP. As shown in Fig. 1I, the formation of acidic autophagic-lysosomal puncta network, indicated by red fluorescence and quenching of green fluorescence, was significantly increased in cells embedded in the soft 3D collagen lattices compared to that in cells cultured on 2D interface. Autophagic flux levels in BxPC-3 cells were further quantified by FACS, where a significant higher red/green fluorescence ratio was observed in cells cultured on soft substrates compared to that in cells cultured on stiff substrates (Fig. S1L). We further performed Western blot analysis to examine the autophagic flux in KC and KPC cells embedded in soft 3D type-I collagen lattices or seeded on plastic coated with collagen. In soft 3D matrix, p62/Sqstm1, a cargo protein that mediates autophagic-lysosomal degradation, underwent prompt turnover (Fig. 1J-K). Phospholipid conjugated LC3 (LC3-II) and Atg7, the markers of phagophore formation and autophagic flux, were increased in cells cultured in soft stroma (Fig. 1J-K). These data suggested that lysosomal biogenesis and autophagic flux levels were up-regulated by the cancer cells in reponse to stromal softness. Taken together, these lines of evidence indicate that stromal softness can promote autophagic-lysosomal flux, attenuate the YAP pathway, and confine tumor cell growth.

Stromal softness induces YAP1 degradation in cancer cells in a lysosomal flux-dependent manner

YAP turnover is attributed to Hippo kinase through YAP phosphorylation, followed by ubiquitination and proteasome digestion, while impediment of YAP degradation is related to tumor growth [22]. Here, we confirmed this notion, showing that both YAP1 and p21 proteins were restored by treatment with MG132, a specific proteasome inhibitor, when KPC cells were cultured on stiff interface (Fig. 2A). Chloroquine, a lipophilic amine that impairs lysosomal acidification, also rescued cellular YAP1 in KPC cells grown on stiff interface, which was in line with the absence of activated CtsH and accumulation of its substrate LC3-II (Fig. 2B). However, when the cancer cells were cultured in soft 3D collagen, YAP1 was degraded predominantly through the lysosomal machinery rather than the proteasome, since MG132 failed to rescue YAP1, but could restore p21, showing efficacy of the inhibitor and activation of proteasome pathway by the cells in soft stroma, whereas chloroquine and E64 were capable to rescue YAP1 (Fig. 2C), which was in agreement with the increased expression and activation of lysosomal cathepsins (Fig. 1E). Time course experiments showed that CQ treatment, but not MG132 treatment, was capable to rescue YAP1 in association with autophagy-lysosomal stress and inhibition, as indicated by LC3 accumulation (Fig. 2D-E), which further verified that stromal softness-mediated YAP1 degradation in cancer cells occurred mostly through autophagy-lysosome pathway rather than proteasome-mediated degradation pathway. To define the lysosomal-cysteine proteinases involved in YAP1 degradation, we knocked down the major lysosomal cysteine-type cathepsins, CtsB and CtsL, which were abundantly expressed by the cells in soft stroma. As shown in Fig. 2F and G, knockdown of CtsL resulted in the restoration of YAP1. In contrast, knockdown of CtsB had no effect on YAP1 levels, indicating substrate preference among the lysosomal cathepsins for YAP1 degradation by the cells in soft stroma. Cystatins, the endogenous inhibitors of cysteine cathepsins, are known to be associated with poor prognosis in PDAC patients [24]. Consistently, we found that overexpression of cystatin A (CSTA), known for its inhibition of cathepsin H/L, in KC cells restored YAP1 expression (Fig. 2H), and a similar result was confirmed by overexpression of cystatin B (Fig. 2I). Importantly, the cystatin B promoted lysosomal stress and the expression of EMT markers, as indicated by the increased expression of Twist1 and Vimentin and down-regulation of E-cadherin (Fig. 2I).

Fig. 2

Stromal softness induces YAP1 degradation in cancer cells through lysosomal flux (A) Western blot analysis of KPC cells, seeded on 2D monolayers and treated with the proteasome inhibitor MG132 at 0, 10, 20 and 50 µM for 6 h. (B) Western blot analysis of KPC cells cultured on 2D monolayers and treated with the lysosome inhibitor chloroquine at 0, 20 and 50 µM for 6 h. (C) Western blot analysis of cells embedded in 3D collagen, treated with DMSO, MG132, E64 and chloroquine for 12 h. (D-E) KPC cells were treated with MG132 (20 µM) or CQ (50 µM) for different periods of time (0, 3, 6–9 h), followed by Western blot analysis. (F) The knockdown efficiency of lenti-shRNA on CtsB and CtsL in KPC cells was determined by RT-qPCR analysis. (G) Western blot analysis of the cells transfected with the indicated shRNA and cultured on 2D monolayers or embedded in 3D collagen for 24 h. (H-I) Western blot analysis of KC cells over-expressing Flag-Cystatin A and Flag-Cystatin B in 3D collagen. (J) Western blot analysis of KPC cells embedded in 3D fibrin gel at 3 (0.6 kPa), 6, 9, 12, and 24 (2.6 kPa) mg/mL or seeded on the fibrin coated plastic for 24 h. The experiments were repeated more than twice, and typical results are presented. The values are presented as mean ± SD; ***p < 0.001 vs. sh-ctrl group; ###p < 0.001 vs. sh-ctrl group

Cancer progression is intimately related to wound healing, by which fibrin clot often occurs as a key component of the tumor micro-environment. Thus, we examined the effects of stromal softness on YAP1 degradation in KPC cells embedded in 3D fibrin lattices, through which the gel stiffness can be adjusted over a wider range. As shown in Fig. 2J, with increasing stiffness from 0.6 kPa (3 mg/ml fibrin) to 2.6 kPa (24 mg/ml), the expression levels of YAP1 were progressively increased to that of the fibrin-coated 2D surface. Strikingly, in response to the increased stiffness of 3D fibrin, expression of matured form of CtsL gradually decreased accompanying with YAP1 accumulation. These results indicated that the protein turnover of YAP1 by the cells in soft stroma is mostly mediated through lysosomal-cathepsins machinery, rather than proteosome machinery.

Lysosome destinating CTSL, rather than nuclear cathepsins, digests YAP1 from the N-terminus

Hippo-kinase-mediated YAP inactivation depends on substrate phosphorylation at multiple serine residues, which collectively serves as signals for the ubiquitination and proteasome-mediated degradation [22]. However, we noticed that a five-serine-to-alanine YAP1 null variant (5SA-YAP: S61A, S109A, S127A, S164A, and S381A) was equally degraded as the wild-type YAP1 in 293T cells embedded in 3D collagen, which again implies a Hippo-independent mechanism for YAP1 degradation by the cells in soft stroma (Fig. 3A). To validate the notion based on loss-of-function analysis as shown in Fig. 2, we also performed gain-of-function analysis. Indeed, forced expression of CTSL but not CTSB, promoted YAP1 degradation in cells (Fig. 3B). Next, 293T cells were co-transfected to express a wild type CTSL or a catalytically null variant (C138S mutant) together with human YAP1 tagged with HA (Human influenza hemagglutinin) at C-terminus (Fig. 3C). As shown, forced expression of the wild-type CTSL, but not its catalytically inactive variant (C138S), led to YAP1 degradation (Fig. 3D). A previous report showed that CTSL was able to cleave the N-terminus of histone H3, which consequently impacted on embryonic stem cell differentiation [25]. Here, using antibodies against the C-terminus of histone H3, we confirmed that CTSL, but not the catalytically null variant, could cleave the N-terminus of histone H3 (Fig. 3D). In addition to lysosome destination, cysteine cathepsins can also be found in the nucleus or extracellular secretion [26]. We then investigated whether the lysosomal-sorted cathepsin L is necessary for the YAP1 degradation. As shown in Fig. 3E, CTSL variants, including M1-CTSL and SP-CTSL, both lacking an N-terminal signal peptide for ER-Golgi-lysosomal sorting, were unable to digest YAP1 and histone H3. This notion was further confirmed by immunofluorescence staining analysis of the transfected cells, revealing that most M1-CTSL and SP-CTSL variants, as well as a portion of wild type CTSL, were found in the nuclear compartment (Fig. 3F). We further captured the intermediate fragments during the course of cathepsin-mediated YAP1 degradation by adding tags at the N- or C-terminus to the substrate YAP1. We noticed that CTSL, but not CTSB, initiated YAP1 degradation starting from its N-terminus (Fig. 3G and H). The notion that the three intermediate fragments of YAP1 being recognized by antibodies against the C-terminus, but not the ones against the N-terminus, indicates that the cathepsin-mediated cleavage of YAP1 begins at the N-terminal region. The specificity of cathepsins for YAP1 degradation according to the gain-of-function analysis agreed well with the results of loss-of-function analysis (Fig. 2G).

Fig. 3

Lysosome destinating CTSL digests YAP1 from the N-terminus (A) Western blot analysis of 293T cells embedded in 3D collagen or seeded on plastic for 24 h for expression of wild-type human YAP1 or the Hippo phosphorylation null variant (5SA) of YAP1. (B) Western blot analysis of expression of CTSB, CTSL and YAP1 in 293T cells embedded in 3D collagen. (C) Schematic representation of YAP1 and CTSL variant constructs. (D) CTSL-mediated degradation of YAP1 was analyzed by Western blot analysis of 293T cells expressing wild type CTSL or catalytic inactive CTSL (C138S) together with FLAG-YAP1 (tagged at the N-terminus). The C-terminal fragments of YAP1 and N-terminus cleaved histone H3 are indicated by arrows. (E) Western blot analysis of 293T cells expressing wild-type or the signal-peptide-deletion variants of CTSL together with Flag-YAP1. (F) Confocal immunofluorescence microscopy of HA tag in 293T cells expressing HA-CTSL, HA-M1-CTSL and HA-ΔSP-CTSL. Nuclei are marked by DAPI. (G and H) Western blot analysis of 293T cells expressing CTSL or CTSB together with YAP1-FLAG. C-terminal fragments of YAP1 are indicated by arrows. (I) In vitro degradation assay of the CTSL-mediated digestion of YAP1 was carried out by incubating purified His-tagged-YAP1 with increasing doses of purified human CTSL at pH 4.0 and 37 ℃ for 30 min. The levels of C-terminal fragments of YAP1 were determined by Western blot analysis. (J and K) In vitro degradation of YAP1 by CTSL was inhibited by E64 and chloroquine. Scale bar = 10 μm. Representative images of experiments repeated 2–3 times are shown

We then conducted an in vitro degradation assay by incubating of purified His-tagged YAP1 as a substrate with CTSL as a proteinase. At acidic condition of pH4.0 that is closed to lysosomal pH, YAP1 was completely digested by CTSL in a dose-dependent manner (Fig. 3I). Interestingly, three C-terminus fragments of YAP1 appeared as intermediate products, which are closely match to the YAP1 fragments observed in the cells (Fig. 3D, E). We further tested the effect of the lipophilic amine inhibitors, which can cause lysosomal stress, on CTSL-mediated YAP1 degradation in vitro. Indeed, the CTSL-mediated degradation of YAP1 in vitro could be inhibited by E64 or chloroquine (Fig. 3J and K) at the similar concentrations as those observed in cells (Fig. 2). Thus, the results from in vitro degradation and in vivo transfection experiments explicitly demonstrated that YAP1 is a bona fide substrate for the lysosomal cysteine-cathepsins. These results also demonstrate the roles of lysosomal-cathepsin and autophagic flux in regulating YAP1 degradation.

Stromal softness up-regulates PTEN and promotes autophagic flux for YAP degradation

Phosphatase and tensin homologue (PTEN), a phospholipid and phosphor-protein phosphatase, is known for its role in maintaining somatic stem cells and cancer stem cells [27]. Loss of PTEN is often associated with the dissemination of PDAC [28]. PTEN is also known to promote autophagic flux via inhibiting AKT/mTORC pathway [29]. In an animal model, disruption of PTEN accelerates the KRas-induced pancreatic cancer development [30]. Our RNA expression profiling revealed that PTEN was up-regulated in KPC cells cultivated in 3D collagen lattices, which was confirmed by RT-qPCR and Western blot analysis (Fig. 4A-B). In agreement with the elevated expression of PTEN, autophagic-lysosomal flux was consequently accelerated, as indicated by increased formation of LC3-II and expression of CtsH and CtsL, which are further related to YAP1 degradation (Fig. 4B). Likewise, up-regulation of PTEN was also observed in human BxPC-3 cells cultured in 3D collagen, in agreement with the acceleration of autophagic flux, as indicated by increased turnover of p62/Sqstm1 and expression of CTSH for the degradation of YAP1, and HDAC4 (Fig. 4C), which is in line with our previous report showing CTSH mediated degradation of HDAC4 in the context of stromal softness [31]. Along with the increase in stiffness caused by embedding of KPC cells in 3D fibrin lattices (from 3 mg/mL to 9 mg/mL), PTEN was progressively down-regulated, in line with the increased levels of AKT phosphorylation for mTOR activation and attenuation of autophagy, marked by impediment of p62 turnover, ultimately resulting in YAP1 accumulation (Fig. 4D). To determine the causal role of PTEN in autophagy-mediated YAP1 degradation, we knocked down PTEN in KPC cells. Suppression of PTEN impeded autophagic flux, as indicated by the slowed turnover of p62 and LC3, resulting in YAP1 accumulation (Fig. 4E). Moreover, knockdown of PTEN impaired lysosome biogenesis and acidification, as indicated by a decreased number of Lyso-Tracker Red fluorescent puncta (Fig. 4F). Atg5, a critical component for phagophore initiation, along with others such as Atg12, was significantly up-regulated in KPC cells cultured in 3D collagen (Fig. 1F). Consequently, shRNA-mediated knockdown of ATG5 led to the restoration of YAP1, in agreement with the attenuation of autophagic flux, as indicated by accumulation of p62 and suppression of LC3 turnover (Fig. 4G). Likewise, ATG7 acts as an E-1 enzyme for the ubiquitin-like proteins, and its knockdown also impaired the autophagic flux, resulting in YAP1 accumulation (Fig. 4H).

Fig. 4

Stromal softness up-regulates PTEN to promote autophagic-lysosomal flux for YAP1 degradation (A) RT-qPCR analysis of Pten mRNA levels in KPC cells cultured on 2D monolayers or embedded in 3D collagen. (B-C) Western blot analysis of KPC or BxPC-3 cells cultured on 2D monolayers or embedded in 3D collagen. (D) Western blot analysis of KPC cells embedded in 3D fibrin gel at different stiffnesses (0.6 kPa to 1.3 kPa). (E) Western blot analysis of control (sh-GFP) or PTEN knockdown (PTEN shRNA) KPC cells embedded in 3D collagen. (F) Confocal imaging of Lysotracker/Hoechst staining in control (sh-GFP) or PTEN knockdown (PTEN shRNA) KPC cells embedded in 3D collagen. (G and H) Western blot analysis of control (sh-GFP), ATG5 knockdown (ATG5 shRNA) or ATG7 knockdown (ATG7 shRNA) BxPC-3 cells embedded in 3D collagen. Representative images of experiments repeated 2–3 times are shown. The values are presented as mean ± SD; ***p < 0.001. Scale bar = 25 μm

In addition to the PTEN-mediated suppression of AKT for autophagic activation and increased autophagic flux in KPC cells, other cellular factors known for their ability to promote autophagy formation/flux are also up-regulated in cells cultured in soft 3D ECM. Approximately 24 factors known for autophagy promotion were increasingly expressed in cells with soft stroma. Conversely, 8 factors known for autophagy suppression were down-regulated in cells cultured in 3D gel. Taken together, these results suggest that multiple pathways for autophagy-lysosome biogenesis and flux are mobilized by the cancer cells in sense of stromal softness.

Liver fibrosis and stiffness promote PDAC tumorigenesis and metastasis

Given that the soft niche exerted cellular dormancy through YAP1 degradation, we predicted that tissue fibrotic stiffness may awaken the dormancy for tumorigenesis. To test this prediction, we generated liver fibrosis through repeated peritoneal injections of thioacetamide for 4 consecutive weeks, followed by inoculation with a small number KPC cells (Fig. 5A). Liver fibrosis was confirmed by Masson’s trichrome staining (Fig. S2D). On day 14 after tumor cell inoculation, the liver tissues injected with KPC cells were subjected to histological analysis. In this short period, tumorigenesis was barely detected in the livers of control mice, but strikingly appeared in the fibrotic liver (Fig. 5B-C). Histological examination revealed the formation of ductal tumors at the edge of liver lobules (Fig. 5D), indicating the origin from PDAC cells. Importantly, the tumorigenesis was intimately associated with type-I collagen deposition in the fibrotic liver (Fig. 5E). Immunofluorescence and IHC demonstrated the juxtaposition of tumor cells, marked by CK19 and Ki67, which was intimately associated with the fibrogenesis and stiffness, as indicated by the formation of fibrotic septa (Fig. 5E). CtsL was highly stained as lysosomal puncta in the parenchymal cells, but was reduced in the detached and less differentiated tumor cells (Fig. 5F). Intriguingly, the levels of CtsL in adjacent parenchyma of liver fibrosis was significantly lower than that in the control group (Fig. 5F), indicating the association between fibrosis/stiffness and lysosomal stress. In contrast, YAP1 was highly expressed in the tumor cells but was moderately expressed in the parenchymal cells (Fig. 5F). Again, using double immunofluorescence microscopy, we found that CtsL was mostly presented in the lysosomal puncta of the parenchyma and moderately present in the tumor epithelial cells in PanIN state, but was mostly absent in the less differentiated PDAC-like cells, beneath of the ductal cells, where YAP1 was heavily expressed (Fig. 5G). Thus, the stromal stiffness, generated by tissue fibrosis, can promote tumorigenesis by awakening cancer dormancy.

Fig. 5

PDAC tumorigenesis and metastasis are promoted in the fibrotic liver (A) Schematic representation of the experimental design used to determine the impact of liver fibrosis/stiffness on tumorigenicity. The mice were treated with or without TAA for 4 weeks, followed by inoculation of UN-KPC961 cells for 2 weeks (n = 6). (B) Gross examination (upper row) and H&E staining (lower row) of control and tumor-bearing livers. Scale bar = 500 μm. (C) Quantitation of tumor numbers in normal or fibrotic liver sections. (D) H&E staining of normal or fibrotic liver sections, scale bar = 200 μm. (E) Immunohistochemical staining of type-I collagen, Ki67 and CK19 in normal or fibrotic liver sections, scale bar = 100 μm. (F) Immunohistochemical staining of YAP1 and CtsL in normal or fibrotic liver sections, scale bar = 100 μm. (G) Immunofluorescence staining of CtsL and YAP1 in PanIN and adjacent hepatic parenchyma. The PanIN ducts are indicated by brown dashed lines, and malignant cells are indicated by pink arrows, scale bar = 25 μm. The values are presented as mean ± SD; **p < 0.01

Lysosomal cathepsins are down-regulated in the human PDAC tissues, in association with fibrotic stiffness, YAP1 accumulation and poor prognosis in PDAC patients

Finally, we examined the potential inverse association between the cysteine cathepsins and YAP1 in the context of fibrotic stiffness in human PDAC tissues. We determined the protein expression levels of YAP1 and CTSL in human PDAC through tissue microarray analysis (n = 30 for PDAC, n = 29 as control). Immunohistochemical analysis showed that YAP1 was overwhelmingly stained in the cells of pancreatic ducts, and cancerous stromal cells as well (Fig. 6A-B). Conversely, CTSL was mostly absent in the corresponding cancer cells. Fibrotic septa and tensile stiffness in the PDAC tissues were indicated by the expression of tissue transglutaminase (TGM2) (Fig. 6C), which can crosslink ECM proteins to generate tensile force and stromal stiffness. We also retrieved the data from the Human Protein Atlas and examined the expression of CTSL and survival time of PDAC patients. We found that malignancy was negatively related to CTSL expression in adenocarcinomas. A typical CTSL-stained image of human PDAC tissues derived from the Human Protein Atlas was presented in Fig. 6D, which shows negligible staining of cathepsin L in the fibrotic ductal zooms. Cox proportional hazards regression was used to assess the relationship between CTSL expression levels and survival outcomes of PDAC patients. The hazard ratio (HR) was calculated to be 1.47, indicating that high expression levels of CTSL are associated with the favorable prognosis of PDAC (Fig. 6E). In contrast, high expression levels of cystatin A and cystatin B, the endogenous CTS inhibitors as an indication of lysosomal stress, are related to the unfavorable prognosis in PDAC patients (Fig. S2A, B). Taken together, these lines of clinical data, in agreement with our in vitro and in vivo results, support the hypothesis that stromal softness and its exerted autophagic-lysosomal flux can confine the oncogeneically transformed cancer cells to cellular dormancy.

Fig. 6

High expression of lysosomal-cathepsins correlates with YAP1 down-regulation, less fibrosis and survival of PDAC patients (A) Representative immunohistochemistry images of YAP1 and CTSL staining in pancreatic tissues from PDAC patients (n = 30 for PDAC, n = 29 for control). (B) Semi quantitation of CTSL staining in PDAC cells vs. the control cells was determined by densitometry analysis. (C) Fibrotic stiffness in human PDAC tissues was determined as expression of tissue transglutaminase 2 (TGM2) in adjacent to the tumor ducts by immunohistochemistry. (D) A typical image of human pancreatic cancer slide stained for CTSL, from the Human Protein Atlas database. Fibrotic stiffness is indicated by dashed green circles. (E) Decreased expression of CTSL is associated with poor prognosis in PDAC patients (from the Human Protein Atlas database). The values are presented as mean ± SD; **p < 0.01. Scale bar = 50 μm