Loss of Wasl improves pancreatic cancer outcome

Wasl is involved in PDAC outcome. Wasl, encoding for N-WASP, is upregulated in both human and mouse PDAC models (6, 7) and has primarily been associated with enhanced tumor invasion. To determine the clinical relevance of Wasl in human PDAC, we stratified human PDAC samples for low and high WASL expression, and we observed a correlation between low WASL expression and increased survival in PDAC patients (Figure 1A). To better understand the function of Wasl in PDAC, we generated a pancreatic conditional Wasl-KO mice in the Ptf1a^Cre/+ Kras^LSL-G12D/+ mouse model of PDAC (20) to generate Ptf1a^Cre/+ Kras^LSL-G12D/+ Wasl^ΔPanc mice (hereafter CK and CK-N^ΔPanc mice, respectively). At 4 weeks of age, CK-N^ΔPanc pancreata showed large areas of edema, with infiltrating cells and fatty metaplasia, while CK mice pancreata exhibited large areas with acinar tissue and focal metaplastic lesions (Figure 1, B and D). Consistent with the gross phenotype, CK-N^ΔPanc mice displayed an increase in the amount of pancreatic fat (Figure 1D) and an impressive infiltration of F4/80-positive macrophages, along with mast cells (Figure 1C and Supplemental Figure 1, A and B; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.127275DS1). These results were reminiscent of a pancreatitis-like phenotype. Notably, the infiltrating macrophages decreased with age, reaching a level below that present in the control terminal CK mice (Supplemental Figure 1B). In agreement with the inflammatory phenotype, CK-N^ΔPanc expressed REG3A (Supplemental Figure 1C), which is upregulated during pancreatitis and is involved in ADM (21); CK-N^ΔPanc also expressed high active β-catenin levels, which is activated during pancreatitis (22) (Supplemental Figure 1, D and E). However, CK-N^ΔPanc mice do not develop ADM at 3 months of age (6), which led us to conclude that lack of Wasl blocks ADM downstream of REG3A and β-catenin. To unravel the molecular effect of Wasl depletion, we performed RNA sequencing (RNAseq) of the pancreata of 4-week-old CK and CK-N^ΔPanc mice. The differentially expressed genes with an adjusted P value of < 0.02 (Supplemental Table 1) were subjected to gene set enrichment analysis using the program Enrichr (23, 24). In agreement with our histological data, CK-N^ΔPanc pancreatic tissues displayed an enrichment of genes involved in pancreatitis and the immune system (Figure 1E and Supplemental Table 2) and a downregulation of genes involved in the DNA replication pathway (Figure 1E and Supplemental Table 2). Notably, pancreatitis induces injury of acinar cells that has been described to lead to fatty degeneration (25), which may explain the observed fatty metaplasia. Gene Set Enrichment Analysis (GSEA; https://www.gsea-msigdb.org/gsea/index.jsp) also revealed an upregulation of adipogenesis and mast cell immunity, which were consistent with the observed “fatty” phenotype seen in the CK-N^ΔPanc mice, as well as an upregulation of inflammatory responses and IL-6 signaling (Figure 1F). In line with the increased amount of active β-catenin expression observed in CK-N^ΔPanc pancreatic tissues, our RNAseq data reveal increased expression of several β-catenin gene targets (Supplemental Table 3), which were validated by reverse transcription PCR (RT-PCR) (Supplemental Figure 1G). Notably, we observed downregulation of the β-catenin target Myc in CK-N^ΔPanc pancreatic tissues, along with downregulation of MYC targets, as shown by GSEA (Supplemental Figure 1F and Supplemental Table 3). Since inflammation and IL-6 signaling have been linked to oncogene-induced senescence (OIS) (26, 27) and Myc is an essential downstream effector of oncogenic Kras in the pancreas (28, 29), and since pancreatitis is a factor promoting PDAC (30, 31), we monitored a cohort of CK-N^ΔPanc mice until endpoint.

Figure 1

Low WASL correlates with improved survival, and loss of Wasl in oncogenic Kras–driven mouse model for PDAC impairs tumor development. (A) Kaplan-Meier survival curve of PDAC patients stratified for WASL expression using the UCSC Xena genomic browser. (B) Representative H&E staining of pancreata of 4-week-old mice. (C) H&E stainings; arrowheads highlight mast cells. Scale bars: 50 μM. (D) Morphometric quantification of the amount of ADM lesions, exocrine pancreas, and fat within the pancreatic tissue in 4-week-old animals. CK (n = 7) and CK-N^ΔPanc (n = 3). Data represent mean ± SEM. Student′s t test. (E) RNAseq was performed from 4 mice per group. The bar graph from the top upregulated and downregulated pathways (combined score: P value multiplied by z score) in 4-week-old CKN^ΔPanc mice. The pathways were obtained after running the differentially regulated genes (n = 155) listed in Supplemental Table 1 through the gene set enrichment program Enrichr. (F) Enrichment plots generated by the GSEA tool showing the signatures that were enriched in the CK-N^ΔPanc pancreata. (G) Kaplan-Meier tumor-free curves for CK (n = 12), CK-N^het (n = 12), and CK-N^ΔPanc (n = 12) mice. (H) Upper panels show macroscopic pictures of organs of endpoint mice. Note the tiny pancreas in the CKP-N^ΔPanc mouse. St., stomach; Panc., pancreas; Int., intestine. The lower panels show representative H&E stainings. Scale bars: 50 μM. (I) PDAC patients were stratified for WASL expression using the UCSC Xena genomic browser, and expression levels of the indicated genes were compared. In the box plots, the whiskers represent the 5th and 95th percentile, and the central line is the median of the data, with the remaining dots being outliers. Unpaired t test with Welch’s correction.

CK control mice reproducibly develop PDAC at the age of about 1 year (20), and mice with deletion of 1 Wasl allele in this background (hereafter CK-N^het) developed tumors at a rate comparable with that of CK mice and with similar survival (Figure 1G and Supplemental Figure1H). Surprisingly, deletion of both Wasl alleles impaired tumor development (Figure 1H) and slightly improved survival (Supplemental Figure 1H) compared with CK mice. CK-N^ΔPanc mice developed pancreata with small areas of normal exocrine pancreas and adipose tissue surrounding sparse islets of Langerhans, suggesting lipomatosis of the exocrine tissue (Supplemental Figure 1, I and J). The remaining exocrine pancreas also had sparse cells showing SA–β-galactosidase activity (Supplemental Figure1K), which is the most accepted biomarker of senescence in vivo (32, 33). Consistent with these findings, low WASL expression in human patients correlates with increased expression of the senescence markers CDKN1A and CDKN2A and the checkpoint kinases CHK1 and CHK2 (Figure 1I). Thus, loss of Wasl impaired oncogenic Kras–driven tumor development by inducing an inflammatory and senescence transcriptional program coupled to a fatty metaplasia.

Concomitant Wasl and Trp53 loss improves survival of Kras-driven tumors and leads to tumors with senescence features. Since the CK-N^ΔPanc mice did not develop tumors, we next analyzed a mouse model with a more aggressive pancreatic tumor. As a model for fast-developing PDAC, we used the Ptf1a^Cre/+ Kras^LSL-G12D/+ Trp53^fl/fl mouse model (34) (hereafter CKP), which has a pancreas-specific homozygous deletion of Trp53 that results in loss of p53 protein expression (Supplemental Figure 2A) and which develops PDAC within 4–6 weeks in a setting of homozygous Trp53 deletion. We then assessed the impact of heterozygous and homozygous deletion of Wasl in the pancreas of the CKP mouse model. CKP mice lacking 1 Wasl allele (hereafter CKP-N^het), developed ductal adenocarcinomas, defined by the presence of neoplastic glandular cells in a dense fibrous stroma, and were indistinguishable from CKP mice (Supplemental Figure 2B). Additionally, CKP and CKP-N^het mice had a comparable life span (Supplemental Figure 2C). Thus, we used the CKP-N^het mice as a control cohort. Homozygous deletion of Wasl in CKP mice (hereafter CKP-N^ΔPanc) was validated by immunohistochemical staining for N-WASP (Figure 2A), as well as by RT-PCR analysis of Wasl expression (Supplemental Figure 2D). The CKP-N^ΔPanc mice had a significantly improved survival compared with CKP-N^het mice (Figure 2B), consistent with the improved survival observed in human PDAC patients with low Wasl–expressing pancreatic tumors. Similar to the CK-N^ΔPanc mice, the pancreatic tissues of the CKP-N^ΔPanc mice also exhibited high levels of lipomatosis (Figure 2D and Supplemental Figure 2, E and F). We also observed that the CKP-N^ΔPanc tumors were predominantly ductal adenocarcinomas with frequent mesenchymal features (50% of CKP-N^ΔPanc mice, compared with 20% of CKP-N^het; Supplemental Table 4), characterized by TWIST- and ZEB1-expressing fibroblasts and macrophages (Supplemental Figure 2, G–I). However, the tumors formed by the CKP-N^ΔPanc mice had a lower proliferation index, as shown by the decreased Ki67 staining (Figure 2C). This suggests that the improved survival seen in the CKP-N^ΔPanc mice is due to delayed tumor development.

Figure 2

CKP-N^ΔPanc mice have improved survival and develop tumors that express senescent markers. (A) Immunohistochemical staining showing the lack of N-WASP expression in pancreata of CKP-N^ΔPanc mice. Scale bars: 50 μM. The inserts show the islets of Langerhans (I.L.) as an internal positive control for N-WASP staining. (B) Kaplan-Meier survival curves for mice of the indicated genotypes. CKP-N^het (n = 12) and CKP-N^ΔPanc (n = 10). Log-rank test. (C) Murine tumor specimen were stained for the proliferation marker Ki67. The number of Ki67-positive tumor cells was quantified per relative area. Data represent mean ± SEM; Student′s t test. (D) Representative pictures of tumors developed by the indicated mice. Scale bars: 50 μM. (E) SA–β-galactosidase staining of cryosections of murine tumors. Scale bars: 50 μM. Inserts show magnification of the black boxes. (F) The amount of SA–β-galactosidase–positive cells was quantified in the tumor cell and stromal cell compartment. Data represent the mean of 2 mice per group ± SEM. Student′s t test. (G) The expression of the indicated genes was analyzed by RT-PCR. Data represent the mean of 4 of 5 tumors per group ± SEM. Student′s t test. (H) Enrichment plots generated by the GSEA tool. The IL-6/JAK/STAT3 and TNFA signaling via NF-κB signaling pathways, as well as the IFNG response signatures, were enriched in the CKP-N^ΔPanc tumors.

Next, we sought to determine the mechanism via which loss of Wasl expression in the CKP mice background contributed to improved survival. Oncogenic Kras induces cellular senescence, which is suppressed by loss of Trp53 (27, 33, 35). Accordingly, the lesions in tumors from CK mice were positive for the senescence marker SA–β-galactosidase, and those in CKP tumors were negative (Figure 2E). Notably, deletion of both Wasl alleles in the CKP mice background led to the appearance of SA–β-galactosidase–positive senescent tumor cells (Figures 2, E and F). The CKP-N^ΔPanc tumors also exhibited increased Cdkn1a(p21) and Cdkn2a expression, along with low Ccnd1 levels (Figure 2G) and high p16 protein expression (Supplemental Figure 2, J and K). Remarkably, these tumors exhibited a molecular phenotype similar to the tumors of human PDAC patients with low WASL expression (compared with Figure 1H). Together, these data indicate that lack of Wasl expression overcomes the escape of senescence phenotype observed in the tumors of CKP mice models lacking Trp53.

CKP-N^ΔPanc (n = 7) and CKP-N^het (n = 6) tumors were next subjected to RNAseq analysis, which identified several differentially expressed genes in the CKP-N^ΔPanc tumors, including Wips2/Ccn5 (Supplemental Table 5) that were validated by RT-PCR analysis (Supplemental Figure 2L). GSEA revealed that the CKP-N^ΔPanc tumors were enriched in genes involved in IL-6 and TNF-α signaling, as well as increased IFN-γ response compared with controls (Figure 2H). Notably these genes are all linked to pathways implicated in a senescence-associated secretory phenotype (SASP) (36).

To investigate if the senescence phenotype observed in CKP-N^ΔPanc tumors is cell autonomous, we used 2 approaches. First, we established cell lines from tumors with different Wasl allelic status (CKP, CKP-N^het, and CKP-N^ΔPanc). Characterization of these cell lines showed a strong SA–β-galactosidase activity in CKP-N^ΔPanc cells only (Figures 3, A and B). Similar to the primary tumors, Cdkn1a(p21) expression was increased concomitant with a slightly decreased Ccnd1 expression (Figure 3C) and low proliferation rates (Supplemental Figure 3A) in CKP-N^ΔPanc cells compared with CKP and CKP-N^het cells. Analysis of the culture supernatants using a cytokine profiler revealed increased secretion of some components of the SASP (36), including GM-CSF, KC (CXCL1), and IL-1ra (Figure 3, D and E) in the CKP-N^ΔPanc cells. This is consistent with the lower levels of the SASP-associated genes GM-CSF (CSF2), CXCL1, and CCL5 (MCP-2) (Supplemental Figure 3B) observed in PDAC patients with low WASL expression. These data, thus, support a senescent and slow proliferating phenotype of the CKP-N^ΔPanc cells due to lack of WASL expression. In the second approach, we isolated tumor organoids (37–39) from both CKP and CKP-N^ΔPanc mouse models. The CKP-N^ΔPanc organoids exhibited delayed growth in 3D-Matrigel cultures (Figure 3F), along with sparse but thicker organoids walls (Figure 3G and Supplemental Figure 3C) and impaired size gain through passaging compared with CKP organoids (Supplemental Figure 3D), indicating the presence of senescent features in the CKP-N^ΔPanc organoids. Altogether, these data demonstrate that lack of Wasl delays tumor cell growth and induces senescence in pancreatic tumor cells in a cell-autonomous manner and bypasses the effect of Trp53 loss.

Figure 3

Primary tumor cells isolated from CKP-N^ΔPanc mice have a senescent phenotype. (A and B) Cell lines from CKP (n = 3), CKP-N^het (n = 3), and CKP-N^ΔPanc (n = 4) were grown on a thin layer of collagen and stained for SA–β-galactosidase, quantified in B. Data represent mean ± SEM; Kruskal-Wallis 1-way ANOVA test (P = 0.01) followed by Dunn’s multiple comparisons test. (C) Analysis of gene expression by RT-PCR of primary tumor cell lines isolated from mice of the indicated genotypes (3 of 4 cell lines per group). Data represent mean ± SEM; Kruskal-Wallis 1-way ANOVA test (Wasl, P = 0.021; Cdkn1a, P = 0.029, Ccdn1, P = 0.06) followed by Dunn’s multiple comparisons test. (D) Conditioned media were collected from CKP-N^het and CKP-N^ΔPanc cells, which were grown on a thin layer of collagen for cytokine array analysis. The boxed regions on the representative blot indicate the differentially secreted cytokines. (E) Quantification of the boxed regions as described in D. Data represent the mean of 2 cell lines per group ± SEM. Student′s t test. (F and G) Organoids generated from CKP (n = 5) and CKP-N^ΔPanc (n = 5) pancreata were isolated. The passage numbers in function of the time are shown. Data represent mean ± SEM. (F) Organoids generated from CKP (n = 5) and CKP-N^ΔPanc (n = 5) pancreata were cultivated. The passage numbers in function of the time are shown. Data represent mean ± SEM. The slopes are substantially different indicating a slower growth of the CKP-N^ΔPanc organoids. (G) Representative bright-field images of organoids 14 days after isolation. The CKP-N^ΔPanc organoids were not passaged during this period, while the CKP organoids were passaged 3 times.

Loss of Wasl leads to delocalization of N-WASP protein binding partners. Nuclear N-WASP has recently been linked to regulation of senescence (10). However, no nuclear N-WASP staining was observed in CKP and CKP-N^het tumors (Figure 2A). Therefore, to understand the mechanism driving the senescence effect of N-WASP loss, we examined its binding partners. N-WASP interacts with WIP and p120ctn through its WH1 and VCA domains, respectively (40, 41). WIP is expressed in all CKP-N^het and CKP-N^ΔPanc cancers with comparable protein levels (Supplemental Figure 4A). However, WIP hardly showed a membranous staining pattern in CKP-N^ΔPanc mouse tumors compared with its CKP and CKP-N^het counterparts (Figure 4A, Table 1, and Supplemental Figure 4B). In glioblastomas (GB), WIP is involved in the control of the endocytic/endosomal system and in the sequestration of GSK3, which is a component of the β-catenin destruction complex (17). Therefore, we next analyzed the endosomal function of CKP-N^ΔPanc cells and detected increased levels of horseradish peroxidase (HRP) uptake, indicative of increased endocytosis (Figure 4B), along with a GSK3β vesicular staining (Figure 4C). Furthermore, GSK3β staining was weak in CKP-N^ΔPanc tumor tissues compared with CKP-N^het control tumor tissues (Supplemental Figure 4C). Based on these findings, we hypothesized that there is endosomal sequestration of destruction complex components in PDAC, akin to GB. We next analyzed the expression of further components of the destruction complex including of phosphorylated β-catenin and YAP1 (42). Phosphorylated β-catenin and YAP1 were enriched in CKP-N^ΔPanc tumors (Figure 4, D, F, and G), while the levels of active (nonphosphorylated) β-catenin and MYC (a β-catenin target gene) were decreased, and there was a downregulation of MYC targets (Figure 4E and Supplemental Figure 4D). We again stratified PDAC patients with high and low WASL–expressing tumors using the Xena browser and observed that low WASL expression correlated with low expression of the destruction complex components (Figure 4H), consistent with our findings in CKP-N^ΔPanc murine tumors. Thus, our data support a mechanism of endosomal sequestration of the destruction complex components in CKP-N^ΔPanc tumors.

Figure 4

Lack of Wasl leads to the accumulation of components of the destruction complex and increased expression of inactive β-catenin. (A) Immunohistochemical staining for WIP in murine tumors. Scale bars: 50 M. Inserts show magnification of the black boxes. (B) Three cell lines per genotype were grown overnight; then, an HRP-endocytic assay was performed. The relative endocytosis level is shown. Each point represents the mean from 3 experiments, and the mean of 3 cell lines ± SEM is shown. Student′s t test. (C) Cells were grown on a thin layer of collagen, and an immunofluorescence staining for GSK3β was performed. Nuclei are counterstained with DAPI. Scale bars: 10 μM. (D) Western blot analysis showing β-catenin phosphorylation at Ser33/37 and Thr41, as well as total YAP1. Actin and ERK1 + ERK2 were used as loading controls. (E) Enrichment plots generated by the GSEA tool. The signatures for MYC targets were enriched in the CKP-N^ΔPanc tumors. (F and G) Immunohistochemical stainings for active β-catenin and YAP1. Scale bars: 50 M. The right panels show magnification of the black boxes. (H) PDAC patients were stratified for WASL expression using the UCSC Xena genomic browser, and the expression levels of the indicated genes were compared. WASL high (n = 86), WASL low (n = 89). Unpaired t test with Welch’s correction.

Table 1

Morphometric quantification of WIP staining

In addition to WIP delocalization, we observed a decrease in the nuclear fraction of p120ctn in CKP-N^ΔPanc tumor cells both by immunofluorescence analysis and Western blot analysis (Figures 5, A–C). Nuclear p120ctn has been shown to interact with, and regulates the transcriptional activity of, the multifaceted transcription factor KAISO (encoded by the Zbtb33 gene) that negatively regulates Wnt/β-catenin signaling (43–46). Therefore, we examined KAISO expression in CKP-N^ΔPanc tumor cells. Notably, we detected increased KAISO protein expression in CKP-N^ΔPanc tumors (Figure 5, D and E), as well as in the matched isolated cells, compared with controls (Supplemental Figure 5, A and B). However, there was no apparent increase in Kaiso gene expression (Supplemental Figure 5C), suggesting that Kaiso is posttranslationally regulated and/or stabilized in CKP-N^ΔPanc tumors. In addition, we detected less colocalization of KAISO with the active transcription marker H3K4 in CKP-N^ΔPanc tumor cells compared with controls (Supplemental Figure 5, D and E), suggesting a displacement of KAISO from active promoter regions (47–49). We next assessed the endogenous p120ctn/KAISO interaction using immunoprecipitation assays and detected the p120ctn/KAISO complex in CKP-N^ΔPanc tumor cells (Supplemental Figure 5F).

Figure 5

Lack of Wasl correlates with p120ctn delocalization and with increased KAISO expression. (A) Cell lines of the indicated genotype were stained for p120ctn; nuclei were counterstained with DAPI and are highlighted with dashed lines. Scale bars: 10 μM. (B) Quantification of the staining described in A. Each point represents the mean fluorescence intensity from 88–198 cells. The mean of 3 cell lines ± SEM is shown. Student′s t test. (C) Three cell lines per genotype were used for the extraction of the nuclear fraction. Western blot analysis shows a decreased nuclear expression of P120ctn in the CKP-N^ΔPanc cell lines. (D) Immunohistological staining for KAISO in murine tumors. Scale bars: 50 μM. Inserts show magnification of the black boxes. (E) Quantification of the staining described in D. Each point represents the relative KAISO expression in the murine tumor tissue. CKP (n = 4), CKP-N^het (n = 8), and CKP-N^ΔPanc (n = 8). Mean ± SEM. Kruskal-Wallis 1-way ANOVA test (P = 0.012) followed by Dunn’s multiple comparisons test.

In summary, lack of N-WASP in CKP tumors correlates with an enrichment of endocytosis concomitant with an accumulation of components of the β-catenin destruction complex and the transcription factor KAISO but with nuclear displacement of Kaiso’s binding partner p120ctn.

Inhibition of endocytosis and increased nuclear p120ctn are required to rescue the phenotype of CKP-N^ΔPanc tumor cells. To confirm the involvement of the destruction complex and p120ctn in the phenotype of the CKP-N^ΔPanc tumors, we performed the following functional experiments. First, we treated CKP-N^ΔPanc tumor cells with endocytosis inhibitors. We included Cytochalasin D (which inhibits actin polymerization), the Na⁺/H⁺ exchanger ethyl-isopropyl-amiloride (EIPA), and the ionophore salinomycin. All treatments strongly induced cytoplasmic GSK3β localization concomitant with decreased total YAP1 protein expression (Figure 6, A and B). This result indicates the successful inhibition of the GSK3β endosomal sequestration. The 24-hour EIPA treatment downregulated Ccnd1 expression in the control cell lines, consistent with published studies (50) (data not shown). All treatments slightly upregulated Cdkn1a(p21) (Figure 6C) and downregulated Ccnd1 (Supplemental Figure 6A) but had no significant effect on the number of senescent cells (Figure 6D). Therefore, this treatment — although inducing a decrease of YAP1 in the CKP-N^ΔPanc tumor cells — was not able to rescue their senescence phenotype. Next, based on the observation of decreased nuclear p120ctn in CKP-N^ΔPanc tumors cells, we treated CKP-N^ΔPanc tumor cells with leptomycin B (LMB) since p120ctn nucleocytoplasmic shuttling is sensitive to LMB treatment (Figure 6E) (43, 51). However, LMB treatment alone was not sufficient to rescue the senescence phenotype of CKP-N^ΔPanc tumor cells (Supplemental Figure 6). We next treated CKP-N^ΔPanc tumor cells with a combination of salinomycin and LMB, which target both deregulated pathways in the CKP-N^ΔPanc tumor cells. This sequential treatment of CKP-N^ΔPanc tumor cells rescued the senescent phenotype of CKP-N^ΔPanc tumor cells, as shown by the significant decrease of senescence-associated β-galactosidase activity (Figures 6, F and G) and downregulation of Cdkn1a(p21) in the CKP-N^ΔPanc tumor cells (Figure 6H). Collectively, our data confirm that Wasl loss regulates pancreatic tumor cell senescence by both inducing the endosomal sequestration of the β-catenin destruction complex components GSK3β and YAP1 and by inducing mislocalization of p120ctn.

Figure 6

Combined salinomycin and LMB treatment of CKP-N^ΔPanc tumor cells rescues the senescent phenotype. (A) CKP-N^ΔPanc cells were treated with Cytochalasin D (2 μM), salinomycin (1 μM), or EIPA (25 μM) for 48 hours; then, the cells were fixed and stained for GSK3β. Scale bars: 10 μM. (B) CKP-N^ΔPanc cells were treated as described for A, harvested, and analyzed by Western blot. (C) Three CKP-N^ΔPanc cell lines were treated as described for A for 24 hours and harvested, and the relative Cdkn1a(p21) expression was analyzed by RT-PCR. Student′s t test. (D) CKP-N^ΔPanc cells were grown on a thin layer of collagen and treated with salinomycin; then, the cells fixed stained for SA–β-galactosidase and quantified. The mean of 2 cell lines ± SEM is shown. Student′s t test. (E) CKP-N^ΔPanc cells were treated with LMB (5 ng/mL) for 20 hours; then, the cells were fixed and stained for p120ctn. Scale bars: 50 μM. (F) CKP-N^ΔPanc cells were grown on a thin layer of collagen and sequentially treated with 1M salinomycin and 5 ng/mL LMB; then, the cells fixed stained for SA–β-galactosidase. A representative image is shown. Scale bars: 50 μM. (G) Quantification of the treatment described in F; each treatment was performed in triplicates. The mean of 3 cell lines ± SEM is shown. Student′s t test. (H) Cell lines were treated as described for panel F. The relative Cdkn1a(p21) expression of 2 cell lines is shown. Student′s t test.

This study shows that N-WASP acts as an oncogene in the development of PDAC by regulating protein stability and subcellular localization of components of the p120ctn/β-catenin signaling pathway.

Kras^G12D is the driver oncogenic mutation of PDAC (1). However, despite extensive research, the precise molecular mechanism involved in PDAC development and progression is not well understood. The effectors of oncogenic Kras include CDC42 and PIP2, both binding partners of N-WASP (encoded by Wasl) (12–14, 52), which is associated with poor PDAC prognosis (7). Originally, N-WASP was investigated in the pathogenesis of the human Wiskott-Aldrich syndrome, where it regulates immunity by interacting with histone methyltransferases (53). At a molecular level, N-WASP acts as an actin nucleation-factor (54); thus, it is not surprising that it is implicated in the regulation of cell migration and tumor metastasis (7–9, 11). More recently, nuclear N-WASP was reported as an inhibitor of p53-induced senescence in a model for skin tumor formation, further implicating an oncogenic role for N-WASP (10). However, N-WASP has also been shown to act as a tumor suppressor, for example, in early intestinal carcinogenesis, suggesting a tissue-specific function of this protein (55). Since N-WASP binds KRAS effectors and because actin polymerization is one of the pathways involved in PDAC development (15), we hypothesized that Wasl expression plays a tumor-promoting role in PDAC development.

Using established oncogenic Kras–based mouse models for PDAC, we show that the conditional deletion of Wasl delays PDAC development, leading to senescent tumors in the absence of the tumor suppressor Trp53, and abrogates tumor development in the CK mouse model, highlighting the oncogenic properties of Wasl in PDAC. In both mouse models, we observed increased lipomatosis, and RNAseq showed that Wisp2/Ccn5 was among the significantly upregulated genes upon loss of Wasl. Wisp2 encodes a cytoplasmic protein downregulated in PDAC, which is involved in the suppression of cell proliferation, reviewed in ref. 56; Wisp2 also regulates adipogenic commitment (57). Both known functions of Wisp2 are consistent with the low proliferation rate and the increased fatty metaplasia observed after deletion of Wasl in the CKP mouse model. Notably, Wisp2 was slightly upregulated in 4-week-old CK-N^ΔPanc mice, supporting an early function of Wisp2 upon loss of Wasl.

PDAC patients stratified for low WASL expression have improved survival and exhibit low expression of senescence markers, as well as downregulation of the Wnt/β-catenin pathway; thus, mechanistically, we focused on the tumors developed by the CKP-N^ΔPanc mice, which showed similar molecular features to low WASL–expressing human PDAC tumor tissues rather than investigating the CK-N^ΔPanc mice, which do not develop tumors.

CKP-N^ΔPanc mice developed slightly more mesenchymal tumors than control mice. This phenotype can be explained by the combined action of several factors, as described below. First, we observed high Kaiso expression in CKP-N^ΔPanc tumors. Since Kaiso is implicated in regulation of the mesenchymal phenotype and high Kaiso expression correlates with poor survival of many different tumor types (prostate, breast, pancreas; reviewed in ref. 48), this raises the possibility that Kaiso plays a role in Wasl-driven PDAC. Second, the CKP-N^ΔPanc tumors have increased YAP1 expression, which is also involved in regulating the mesenchymal phenotype (58). Last but not least, our detection of increased inflammation (which is intimately connected to induction of epithelial-mesenchymal transition EMT; ref. 59) in CKP-N^ΔPanc and CK-N^ΔPanc mice further supports a Wasl mechanism involving EMT. Interestingly, although EMT is linked to worse PDAC survival outcomes (60), our CKP-N^ΔPanc mice did not exhibit poor survival, most likely because there was no tumor metastasis. However, the improved survival of CKP-N^ΔPanc mice may also be explained by the low proliferating senescent tumors. We showed that loss of Wasl led to a cell-autonomous senescent phenotype in the tumors, consistent with strong upregulation of Cdkn1a(p21) and increased SA–β-galactosidase activity that is currently the most accepted biomarker of senescence in vivo (32, 33). Cellular senescence has been described as a tumor-suppressive mechanism (36). However, senescent cells can modify the tumor microenvironment by acquisition of SASP, which can increase the number of inflammatory cells (36, 61), as evidenced by the GSEA signature for an inflammatory response in the CKP-N^ΔPanc tumors.

Since no nuclear N-WASP was detected in CKP-N^ΔPanc tumors, we focused on N-WASP interacting proteins (WIP and p120ctn) rather than on the deregulated genes (12, 13). N-WASP interacts with the actin cytoskeleton, which in turn regulates cell motility and is not linked to senescence (12, 13). N-WASP additionally binds WIP and p120ctn. WIP is required for N-WASP and YAP1 stabilization by sequestration of the β-catenin destruction complex (12, 17, 40). YAP1 is involved in several signaling pathways, where it plays a different role depending on the upstream members (62). WIP is also involved in the regulation of endocytosis (63) and participates in several cellular processes, some of which are relevant in cancer and which may be dependent on different oncogenic stimuli (17, 64, 65). We showed that CKP-N^ΔPanc tumors displayed a small amount of membranous WIP, which probably resulted in the increased rate of endocytosis that we observed in these tumors. In line with the role of WIP in the sequestration of destruction complex components, we observed vesicular staining of GSK3β, as well as high YAP1 and phosphorylated β-catenin levels. Since inhibition of endocytosis induced cytosolic accumulation of GSK3β and decrease of YAP1 expression, we conclude that loss of N-WASP in our mouse model leads to sequestration of the destruction complex into vesicles. We recently showed that Wasl deletion impairs ADM by inhibiting fluid-phase endocytosis in pancreatic acinar cells (6). Juin and colleagues created a conditional mouse model with a mosaic deletion of Wasl and concomitant activation of oncogenic Kras and mutated p53 (Trp53^R172H) (KPCN mice) (9). In this KPCN model, Wasl loss deregulated vesicular transport, reduced metastasis, and improved survival. These 2 studies in conjunction with our current study reveal a crucial role for N-WASP in PDAC development through regulation of vesicular trafficking.

The second N-WASP binding partner that we focused on is p120ctn. CKP-N^ΔPanc tumors had less nuclear p120ctn, which is consistent with better prognosis of PDAC patients (18). p120ctn is also involved in the internalization of the GSK3β-containing destruction complex into multivesicular bodies (66); thus, the delocalization p120ctn upon loss of Wasl likely affected the vesicular sequestration of destruction complex components. P120ctn shuttles to and from the nucleus, where it interacts with the transcription factor Kaiso, which is upregulated in CKP-N^ΔPanc tumors (43–45, 48, 67). Since we detected low p120ctn nuclear levels in the CKP-N^ΔPanc tumor cells, we reasoned that p120ctn undergoes increased nucleocytoplasmic shuttling in our model. We were, however, only able to rescue the senescence-associated phenotype in CKP-N^ΔPanc tumor cells by inducing nuclear p120ctn translocation along with inhibition of endocytosis. In summary, our study implicates p120ctn nucleocytoplasmic shuttling and accumulation of destruction complex components in the senescence phenotype of CKP-N^ΔPanc tumor cells and, by extension, suggest a role for N-WASP in these processes in PDAC tumor development.

Notably, lack of Wasl — in the context of oncogenic Kras only — led to an inflammatory phenotype at an early time point. Pancreatitis is considered an important factor that promotes both human and murine PDAC (2, 30, 31). However, despite the inflammatory phenotype at 4 weeks of age, which completely regressed within 6 months, the CK-N^ΔPanc mice did not develop tumors. Since injury of acinar cells can lead to fatty degeneration in pancreatitis (60), this is consistent with the CK-N^ΔPanc phenotype at endpoint. Similar to the CKP-N^ΔPanc mouse model, GSEA in CK-N^ΔPanc mice showed an increased IL-6 signaling signature, along with an INFG response. Thus, the CK-N^ΔPanc phenotype resembles some senescence features of the CKP-N^ΔPanc mouse model. These findings are of potential interest to elucidate the role of inflammation in PDAC development and may be investigated in further studies.

In conclusion, Wasl is required for the suppression of Kras-induced senescence upon Trp53 loss (27, 35). Given the importance of senescence in PDAC (27), unraveling a connection between N-WASP, senescence, and the p120ctn/β-catenin pathway may provide strategies for the treatment of pancreatic cancer.

Mice. Ptf1a^Cre/wt, Kras^LSL-G12D/wt, Trp53^fl/fl, and Wasl^fl/fl strains have previously been described (6, 9, 20, 34, 68, 69).

IHC and immunofluorescence staining. The primary antibodies used are listed in Supplemental Table 6. Immunoperoxidase staining was performed on 4% PFA/PBS fixed paraffin-embedded tissue slides using the Vectastain-ABC-Elite and/or the ImmPRESS kit (Vector Laboratories) following manufacturer’s instructions. For immunofluorescence staining, samples were fixed with 2% PFA for 10 minutes and permeabilized with 0.1% Triton X-100 for 5 minutes. Slides were incubated overnight with primary antibodies (listed in Supplemental Table 6) at 4°C and were then incubated for 1 hour with secondary antibodies (Alexa dyes, 1:500, Invitrogen) at room temperature. Coverslips were mounted on tissue sections with VECTASHIELD HardSet Mounting Medium containing DAPI (Vector Laboratories) and examined using a Leica TCS_SP5 confocal microscope. In all experiments, samples were stained sequentially. To avoid bleed-through, single channels were scanned sequentially. Primary antibodies used are listed in the Supplemental Table 6.

Morphometric analysis. Slides from murine tissue were digitalized, and the regions of interest were quantified blindly using Aperio-ImageScope from Leica. Alternatively, several randomly chosen pictures were taken using a 10× objective from the stained slides, and positive staining was quantified blindly by different researchers.

Cell culture. Primary pancreatic tumor cells were derived from transgenic mouse tumors described above and maintained in DMEM media (Sigma-Aldrich, catalog D0819) supplemented with 10% FCS (Thermo Fisher Scientific). Cells were incubated in a humidified incubator at 37°C and 5% CO₂. Where specified, the cell culture dishes were coated with a thin layer of collagen using rat tail collagen type I (Corning) according the manufacturer′s instructions. For inhibition of endocytosis, the following inhibitors were used: salinomycin (MilliporeSigma, S6201), Cytochalasin D (Tocris, 1233/1), and EIPA (Tocris, 3378/10). LMB was purchased from Sigma-Aldrich.

Organoid isolation and maintenance. Tumor cells were isolated and cultured as previously described (19, 37, 38, 70, 71). Briefly, after euthanizing the mice, the pancreas was quickly harvested and washed in PBS plus penicillin-streptomycin. The pancreas was then minced into small pieces and further digested with collagenase type IV (Sigma-Aldrich) for 1 hour. Additional digestion with Trypsin for 5 to 10 minutes was performed. Pancreatic cells were further washed and centrifuged at 200 × g for 5 minutes at 4 °C. The cells were mixed with Matrigel, and the mixture was plated (50 μL per well of a 24-well plate) and allowed to solidify. After 20 minutes, when the Matrigel had solidified, complete culture media (37) was added to cells.

Immunoprecipitation and Western blots. Immunoprecipitation and Western blot analyses were performed as previously described (6, 44). Primary antibodies used are listed in Supplemental Table 6. Antibody binding was visualized using HRP–labeled secondary antibodies (1:10,000) (GE Healthcare, catalog NA935V, NA394V, and NA931V) and ECL reagent (GE Healthcare).

Quantitative PCR. Total RNA extraction was performed according to manufacturer’s instructions using QIAGEN RNeasy Plus Micro Kit (catalog 74034) for organoids and the NucleoSpin RNA kit (Machery-Nagel) for all other samples. RNA was reverse transcribed to cDNA using Super ScriptIII reverse transcriptase and oligo(dT) primers (Thermo Fisher Scientific) following manufacturer’s instructions. The resultant cDNA was used as template for quantitative PCR (qPCR). RT-PCR was performed in duplicates, using the iTaq universal SYBR Green (Bio-Rad) according to manufacturer’s instructions. Primers used are listed in Supplemental Table 7. The relative amount of each gene of interest was normalized to the housekeeping gene XS13 (an acidic ribosomal protein, which is constitutively expressed at the same level in normal, cancerous, and inflamed pancreas; ref. 24) using the ΔΔCT method.

RNAseq and GSEA by Enrichr. RNA extraction was performed as described for qPCR. RNA concentration was fluorimetrically determined using the QuBit 3.0 system (Thermo Fisher Scientific). Bulk 3′-prime transcript end RNAseq (SCRB-Seq) libraries were prepared as described below. RNA was reverse transcribed using oligo(dT) primers decorated with sample barcodes, unique molecular identifiers (UMIs), and adapters (Integrated DNA Technologies). cDNA from all samples was pooled, and unincorporated primers were digested using Exonuclease I (New England Biolabs). The cDNA pool was then amplified with KAPA HiFi ReadyMix (KAPA Biosystems). To obtain sequencing libraries, 0.8 ng of cDNA was fragmented, and 3′ ends were amplified with the Nextera XT Kit (Illumina) using a specific primer for the adapter on the 3′-end. The library was paired-end sequenced on a HiSeq1500 with 16 cycles for read 1 to decode sample barcodes and UMIs, and 51 cycles on read 2 into the cDNA fragment. Transcript and gene definitions were used according to the ENSEMBL annotation release 75. GSEA was performed using the web tool Enrichr, available at http://amp.pharm.mssm.edu/Enrichr (23, 24). In addition, GSEA Software v4.0.3 (72) was used to calculate normalized enrichment scores (NES), false discovery rate (FDR) values, and P values for gene sets in the Molecular signatures database (MSigDB) (73). The genes were preranked according to the fold-change and P values

Endocytosis assay. A fluid-phase HRP uptake assay was used as described previously (6). Briefly, samples were incubated with 1 mg/mL HRP (MilliporeSigma) for 5 minutes. After pulse, chase, and extensive wash, cells were lysed (10 mM TRIS, 0.05% Triton X-100). The amount of internalized HRP was quantified by measuring the peroxidation of o-Dianisidine (MilliporeSigma) after addition of H₂O₂ at 405 nm. The amount of internalized HRP was normalized to the total protein content measured by BCA protein assay (Pierce).

SA–β-galactosidase staining. SA–β-galactosidase staining was performed as previously described (33). Briefly, the fixed samples were incubated with a staining solution containing 20 mg/mL of X-Galactosidase (Roth) for 1–3 days at 37°C and then mounted. Stained cells were examined with a Zeiss inverse microscope, and images were quantified blindly with ImageJ (NIH).

In silico patient survival. For stratification of PDAC patients, we used the UCSC Xena Platform (http://xena.ucsc.edu) (74) with the TCGA Pancreatic Cancer (PAAD) data set filtered for primary tumors (n = 185).

Statistics. Statistical analyses were performed using the GraphPad Prism version 5.0a for Mac OS X. We applied unpaired, 2-tailed Student′s t test, Mann-Whitney U test, the 2-way ANOVA, and the log-rank test; for multiple comparisons, we used the Kruskal-Wallis test followed by a Dunn’s multiple comparisons test, as specified in the legends. A P value less than 0.05 was considered significant.

Study approval. The present study in animals was reviewed and approved by the Government of Bavaria, Munich, Germany (AZ 5.2-1-54-2532-46-2014) and the IACUC of the Technical University of Munich. Experiments were conducted in accordance with the German Federal Animal Protection Laws.