The intact epithelial Hippo pathway is essential to suppress fibroinflammatory response in the adult pancreas

To investigate the roles of LATS1/2 in the pancreas, we genetically ablated Lats1 and 2 genes specifically in pancreatic acinar cells at the adult stage. Through multiple steps of breeding (S1A Fig), we generated Lats1 and 2 double knockout mice (PL: Ptf1a^CreERLats1^fl/flLats2^fl/flRosa26^LSL-YFP), Lats1 and Lats2 single knockout mice (PL1KO: Ptf1a^CreERLats1^fl/flLats2^fl/+Rosa26^LSL-YFP; PL2KO: Ptf1a^CreERLats1^fl/+Lats2^fl/flRosa26^LSL-YFP), and flox-alleles–alone mice (Lats1/2: Ptf1a^+/+Lats1^fl/flLats2^fl/flRosa26^LSL-YFP). We used Cre alone (P: Ptf1a^CreERRosa26^LSL-YFP) mice as controls. We administered 180 mg/kg/d of tamoxifen (TAM) to 6- to 12-week-old mice via intraperitoneal (i.p.) injection for 5 consecutive days. We found that PL mice became lethargic and started losing weight by 7 days after the last TAM injection (S1B Fig). At day 10, PL mice had a weight loss of 19.2%. We found a significant reduction of blood glucose levels in PL mice, suggesting a starvation phenotype (S1C Fig). We observed a continuous loss of acinar tissue and replacement by fibrous tissue over 14 days (S1D Fig).

Histologically, PL mice exhibited severe damage in the pancreas, including acinar cell atrophy, inflammatory cell infiltration, and acinar-to-ductal metaplasia (ADM) at Day 5 after the last injection of TAM (Fig 1A). No such defects were observed in the pancreata of control groups (Fig 1A). Successful Lats1/2 deletions in the pancreas were determined by PCR analysis (S2A Fig) and western blot (Fig 1B). The protein levels of downstream effectors YAP1/TAZ were increased dramatically in PL mice, while phospho-YAP1 levels were lower in PL mice (Fig 1B). YAP1 and TAZ nuclear translocations were detected in PL mice by immunohistochemistry (IHC) staining (Fig 1C, S2B Fig), suggesting that loss of LATS1/2 in pancreatic acinar cells induced YAP1/TAZ activation, similarly to other tissues [17]. Compared to the control P mice, we detected an increased expression of ductal cell marker cytokeratin 19 (CK19) in PL mice (Fig 1D). Yellow fluorescent protein (YFP⁺)/CK19⁺ cells were only observed in PL mice, suggesting that acinar cells underwent ADM, a common hallmark of pancreatitis (Fig 1D, S3A Fig). The antigen identified by monoclonal antibody Ki 67 (Ki67) expression was found in YFP⁺/CK19⁺ cells, suggesting that proliferation was activated during ADM (S3B Fig). We found apoptotic cells in PL mice by staining of cleaved caspase-3 (S3C Fig). Consistent with hematoxylin–eosin (HE) staining, we identified infiltration of immune cells in PL mice by cluster of differentiation antigen 45 (CD45) staining (Fig 1E, S3A Fig). We detected strong staining of α-smooth muscle actin (αSMA) and significant deposition of collagen in PL pancreata (Fig 1F, S3A Fig), suggesting the occurrence of PSC activation and pancreatic fibrosis. Taken together, these results strongly suggest that the loss of LATS1/2 in pancreatic acinar cells rapidly leads to pancreatic fibrosis, a major clinical feature of CP.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Acinar cell-specific Lats1/2 deletions caused pancreatic inflammation and fibrosis in adult mice.

P, PL, L1/2, PL1KO, and PL2KO mice were administrated with 180 mg/kg/d of TAM for 5 consecutive days via i.p., respectively (n = 6). Pancreata were collected at Day 5 after the final injection. (A) Histological morphology of P, L1/2, PL, PL1KO, and PL2KO mice. L1/2, PL1KO, and PL2KO mice were littermate controls. (B) Western blot showing the levels of LATS1, LATS2, phospho-YAP1, YAP1, and TAZ in P and PL mice. Tubulin was used as the internal control. (C) YAP1 and TAZ IHC staining in P and PL mice. (D) Immunofluorescence of YFP (Green) and CK19 (Red) in P and PL mice. Nuclei stained with DAPI (Blue). (E) P and PL pancreata stained with anti-CD45 antibody. (F) P and PL pancreata stained with anti-αSMA antibody, and Sirius Red, respectively. αSMA, α-smooth muscle actin; CD45, ; CK19, cytokeratin 19; IHC, immunohistochemistry; i.p., intraperitoneal; L1/2, ; LATS1, large tumor suppressor 1; P, control; PL, double knockout; PL1KO, Lats1 knockout; PL2KO, Lats2 knockout; TAM, tamoxifen; TAZ, transcriptional coactivator with PDZ binding motif; YAP1, yes-associated protein 1.

https://doi.org/10.1371/journal.pbio.3000418.g001

Hippo disruption-induced pancreatic inflammation and fibrosis are YAP1/TAZ dependent

Lats1/2 deletion in adult acinar cells dramatically up-regulated YAP1/TAZ in the pancreas and induced YAP1/TAZ nuclear translocation (Fig 1B and 1C, S2B Fig). Thus, we tested whether the inflammation and fibrosis in Lats1/2 null pancreata were mediated by YAP1/TAZ. Loss of YAP1/TAZ in adult acinar cells (Ptf1a^CreERYAP1^fl/flTAZ^fl/fl, PTY) did not result in overt defect (S4A Fig). We then crossed PL mice with Yap1^fl/flTaz^fl/fl mice to generate Ptf1a^CreERRosa26^LSL-YFPLats1^fl/flLats2^fl/flYap1^fl/flTaz^fl/fl (PLTY) mice. The PLTY mice were administered with 180 mg/kg/d of TAM for 5 consecutive days to induce the quadruple deletions of Lats1/2 and YAP1/TAZ genes in adult pancreatic acinar cells. P and PL mice were used as controls. All pancreata were harvested at Day 5 after the final injection of TAM (S4B Fig). The deletions of Lats1, Lats2, Yap1, and Taz were confirmed by western blot (Fig 2A, S4C Fig). HE staining revealed that ablation of YAP1/TAZ rescued the pancreatitis phenotype of PL mice (Fig 2B). PSC activation, immune cell infiltration, and ADM were remarkably reduced in PLTY mice when compared with PL mice (Fig 2C). These data strongly suggested that Lats1/2-deletion–induced inflammation and fibrosis depend on YAP1/TAZ.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Loss of YAP1/TAZ in acinar cells rescues Lats1/2-deletion–induced pancreatic inflammation and fibrosis.

P, PL, and PLTY mice were subjected to 180 mg/kg/d of TAM for 5 consecutive days via i.p. (n = 6). Pancreata were collected 5 days later. (A) Western blot of Lats1, Lats2, YAP1, and TAZ in P, PL, and PLTY mice. Tubulin was used as the internal control. (B) Histological examination by HE staining among P, PL, and PLTY groups. (C) Immunofluorescence staining of αSMA (Green), CD45 (Red), and ADM (Green + Red) among P, PL, and PLTY groups. Nuclei stained with DAPI (blue). Quantification of αSMA and CD45 immunostaining in P, PL, and PLTY mice showed 90.1% reduction of inflammation and 95.2% reduction of stromal reaction in PLTY mice compared with PL mice. ADM was quantified by counting YFP and CK19 double positive cell numbers. There was an 84.5% reduction of ADM in PLTY mice compared with PL mice (n = 6). **P < 0.01. Underlying numerical values can be found in S1 Data. αSMA, α-small muscle actin; ADM, acinar-to-ductal metaplasia; CD45, cluster of differentiation antigen 45; CK19, cytokeratin 19; HE, hemtoxylin–eosin; i.p., intraperitoneal; LATS1, large tumor suppressor 1; P, control; PL, double knockout; PLTY, quadruple deletions of Lats1/2 and YAP1/TAZ; TAM, tamoxifen; TAZ, transcriptional coactivator with PDZ binding motif; YAP1, yes-associated protein 1.

https://doi.org/10.1371/journal.pbio.3000418.g002

Hippo disruption-induced pancreatic inflammation is not initiated by acinar cell death

Inflammation in the pancreas is frequently induced by acinar cell injury [18]. To test whether inflammation in PL mice was induced by autonomous acinar cell death following Lats1/2 deletions, we decided to knock out Lats1/2 in a small portion of acinar cells in the adult pancreas. If the inflammation was induced by acinar cell death due to Lats1/2 knockout, injured Lats1/2 null acinar cells would be rapidly removed by immune cells and the pancreas should recover with the remaining wild-type acinar cells. We gave the mice single injections of different doses of TAM. Ptf1a^CreER induced chromosome recombination in a TAM dose-dependent manner (S5A Fig). PCR assays confirmed the occurrence of Cre-mediated Lats1/2 deletions even 4 days after a single dose of 45 mg/kg TAM injection (S5A Fig). Strikingly, all mice developed inflammation 3 weeks after TAM injection, and the severity of the inflammation was correlated with TAM dose (S5C Fig). Acinar cells can recover in cerulein-induced AP mice; however, the pancreatic lesions developed in single-dose TAM-injected PL mice did not recover over time. This long-lasting phenotype suggested that acinar damage was not the major mechanism to initiate inflammation in PL mice.

We further chose mice that received a single low-dose (45 mg/kg) TAM injection for time course analysis because they showed Cre-mediated recombination in a fraction of individual acinar cells (S5B Fig), as determined by YFP expression. We co-stained YFP with cleaved-caspase-3 in the pancreata of PL mice at 4, 8, and 12 days after a single TAM injection (45 mg/kg) (Fig 3A) to test whether these cells underwent apoptosis. The results showed no YFP+ cells that were co-stained with cleaved-caspase-3 (Fig 3B). IHC assay revealed YAP1 and TAZ nuclear translocation in the PL pancreas, supporting the successful deletion of Lat1&2 (Fig 3C). We further confirmed that Lats1/2 deletions were restricted to YFP-positive cells by PCR assays after sorting by flow cytometry (S5D Fig). Collectively, these data indicate that loss of Lats1/2 did not directly cause adult acinar cellular apoptosis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Loss of Lats1/2 does not affect acinar cell proliferation, apoptosis, or ADM directly.

(A) Schematic diagram of experimental design (n = 3). (B) PL mice were injected once with 45 mg/kg of TAM. Cell proliferation, apoptosis, and ADM were analyzed on Day 4, Day 8, and Day 12 by co-staining anti-YFP with anti-Ki67, anti-cleaved-caspase-3, anti-amylase, and anti-CK19 antibodies. Nuclei stained with DAPI (Blue). Percentages of Ki67, cleaved caspase-3 (n = 3), and YFP (n = 4) cells were quantified by IHC profiler score. Underlying numerical values can be found in S1 Data. (C) YAP1 and TAZ were detected by IHC in PL mice after injection with 45 mg/kg of TAM. ADM, acinar-to-ductal metaplasia; CK19, cytokeratin 19; IHC, immunohistochemistry; Ki67, antigen identified by monoclonal antibody Ki 67; LATS1, large tumor suppressor 1; P, control; PL, double knockout; TAM, tamoxifen; TAZ, transcriptional coactivator with PDZ binding motif; YAP1, yes-associated protein 1; YFP, yellow fluorescent protein.

https://doi.org/10.1371/journal.pbio.3000418.g003

Hippo disruption does not directly induce autonomous proliferation in adult acinar cells

Because one of the best known functions of LATS1/2 is to control cellular proliferation and differentiation, we tested whether Lats1/2 deletion induced ADM and proliferation of acinar cells. We did not observe YFP-positive cells expressing CK19 in the mice that received a single low-dose (45 mg/kg) TAM injection, indicating that no ADM occurred in Lats1/2 null acinar cells (Fig 3B). We then examined Ki67 expression in these pancreata. No YFP⁺ cells expressed Ki67 at every time point. In addition, we quantified the number of YFP⁺ cells in these mice and found no significant differences during the 12-day period (Fig 3B). The different effects of low- and high-dose TAM suggested that Hippo disruption incudes ADM by non–cell-autonomous mechanisms. The smaller number of acinar cells with Hippo disruption induced by low-dose TAM require a longer time to create an ADM-prone microenvironment, while high-dose TAM induced Hippo disruption in a larger portion of acinar cells that can rapidly activate PSCs and immune cells to promote ADM.

To further confirm that Hippo disruption does not directly induce autonomous proliferation in adult acinar cells, we isolated acinar cells from PL mice that received 5 injections of TAM and cultured them in a collagen-based 3D model. Lats1/2 deletions alone did not enhance the proliferation of acinar cells in 3D culture. However, in the presence of exogenous TGFα treatment, the most common method used to induce ADM in mouse acinar cells [19], the sizes of spheres formed by acinar cells with Lats1/2 deletions were dramatically larger than those formed by the cells from control P mice (S5E Fig), suggesting that acinar cells with Lats1/2 deletions did not induce autonomous cell proliferation but primed acinar cells to exogenous pro-proliferative stimuli from microenvironment to initiate cell proliferation.

Hippo disruption in pancreatic acinar cells stimulates PSC activation

To achieve a longer and clearer time window to analyze the temporal sequences of alterations in PL mice, we gave the mice a single TAM injection (180 mg/kg) and analyzed the pancreata at Days 0, 5, 10, 15, and 20. The pancreata of PL mice almost disappeared by Day 20. At Day 15, the phenotype was similar to what we observed at Day 5 for mice with 5 injections of TAM (Fig 4A). CK19, αSMA, and CD45 staining by IHC in consecutive sections of PL mice was positive in all lesions at Day 15, indicating the presence of ADM, PSC activation, and immune cell infiltration (Fig 4B, S6A Fig). However, the small dispersed lesions at Day 10 in PL pancreata with YAP1/TAZ nuclear accumulation were surrounded by αSMA-positive stromal cells, implying that PSCs might be activated by the direct communication with Lats1/2 null acinar cells (Fig 4C). Indeed, many of these lesions were not accompanied by CD45-positive cells (Fig 4D). This was further confirmed by co-staining αSMA with CD45 (S6B Fig), highlighting the fact that PSCs were activated before significant infiltration of immune cells. These lesions retained acinar morphology and were negative for CK19, suggesting that PSCs were activated in the absence of ADM (Fig 4E). Together, our data indicate that activation of PSCs was an early event in PL mice.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Lats1/2 deletion in acinar cells results in PSC activation.

PL mice were injected once with 180 mg/kg of TAM; pancreata were collected and embedded at Day 0, Day 5, Day 10, Day 15, and Day 20 after injection, respectively (n = 4–6). (A) Morphological changes over time (Day 0, Day 5, Day 10, Day 15, Day 20) were examined by HE staining. (B) Anti- αSMA, anti-CD45, and anti-CK19 antibodies were used to stain activated PSCs, immune cells, and ductal-like cells, respectively. (C) αSMA, YAP1, and TAZ IHC staining in consecutive sections at Day 10. (D) αSMA and CD45 IHC staining in consecutive sections at Day 10. (E) αSMA and CK19 IHC staining in consecutive sections at Day 10. αSMA, α-smooth muscle actin; CD45, cluster of differentiation antigen 45; CK19, cytokeratin 19; HE, hematoxylin–eosin; IHC, immunohistochemistry; LATS1, large tumor suppressor 1; PL, double knockout; PSC, pancreatic stellate cell; TAM, tamoxifen; TAZ, transcriptional coactivator with PDZ binding motif; YAP1, yes-associated protein 1.

https://doi.org/10.1371/journal.pbio.3000418.g004

Macrophages are polarized toward an M2 phenotype in the Lats1/2 null pancreas.

Immune cell infiltration increased dramatically in Day 4 PL pancreata after 5 doses of TAM injection (S7A Fig). Unlike the tissue damage-induced M1 polarization of macrophages in AP, alternatively activated macrophages (M2) are dominant in CP, and activated PSCs are important inducers of M2 polarization [20]. Thus, we examined which macrophages were present in PL mouse. F4/80 staining revealed an increased number of macrophages in the pancreata of PL mice (Fig 5A). We used flow cytometry to sort CD45+CD11b+F4/80+ macrophages from the pancreata of P and PL mice (S7B Fig) 4 days after final TAM injection to examine the expression of a panel of M1- and M2-accociated genes [21]. Compared with the control P mice, macrophages from PL mice displayed a dramatic up-regulation of M2 macrophage-associated gene Ym1 (Fig 5B). Other M2 macrophage-associated cytokines and chemokines, including Il10, Tgf-β1, and Ccl17, also showed significant up-regulation (Fig 5B). M1 macrophage-associated genes such as Tnf-α, Sosc1, and Nos2 showed no significant expression change (Fig 5C), and Il-1β was barely detectable by quantitative PCR (qPCR). These data revealed that macrophages were rapidly polarized toward an M2 phenotype upon YAP1/TAZ activation in adult acinar cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Loss of Lats1/2 induces M2 macrophage polarization.

P and PL mice were treated with 180 mg/kg/d of TAM for 5 consecutive days; cells were isolated from both groups 4 days later (n = 3). (A) F4/80 IHC staining in P and PL mice on Day 4 after TAM injection. (B) M2 macrophage-specific genes Ym-1, Ccl17, Tgf-β and IL10 were elevated in sorted macrophages of PL mice. (C) M1 macrophage-specific genes Tnf-α, Socs1, and Nos2 were not significantly changed. **P < 0.01, *P < 0.05. Underlying numerical values can be found in S1 Data. IHC, immunohistochemistry; LATS1, large tumor suppressor 1; P, control; PL, double knockout; TAM, tamoxifen.

https://doi.org/10.1371/journal.pbio.3000418.g005

Hippo disruption activates the fibroinflammatory program in acinar cells

The observation that activation of PSCs was an early event in PL mice provoked us to examine whether Hippo pathway disruption induced the activation of fibrosis and inflammation-associated genes in acinar cells to modulate the microenvironment. Thus, we performed RNA sequencing (RNA-seq) analysis of pancreata from control P and mutant PL mice. To increase the number of Lats1/2 null cells, control P mice and PL mice were subjected to180 mg/kg/d of TAM for 5 consecutive days. Tails of pancreata from P mice were harvested at Day 1 and used as the control. Tails of pancreata from PL mice were harvested at Days 1, 2, and 3 (n = 3–5 for each time point) for RNA extraction and RNA-seq library preparation. Histological examination showed no visible defects in Day 1, 2, or 3 PL pancreata (S8A Fig). There was no positive staining of CD45, αSMA, or CK19 at Day 2, but there was some at Day 3 (S8B Fig).

We compared transcriptomes from RNA-seq data from PL mice at Days 1, 2, and 3 after TAM injection to control P mice and obtained 40, 405, and 765 differentially expressed genes (DEGs), respectively, using the significance criterion (fold-change > 2; adjusted P < 0.05; and read per kilobase of transcript per million reads mapped [RPKM] > 2). A large number of genes with expression changes were found in Day 2 PL pancreata (Fig 6A), although no significant PSC activation or immune cell infiltration was observed at this time point. We performed K-means clustering analysis and identified 2 representative clusters: 300 up-regulated genes and 232 down-regulated genes (Fig 6B). Several known YAP1/TAZ targets, including Ctgf, Cyr61, and Spp1, were among the up-regulated genes (S8C Fig). Gene ontology (GO) term analysis with the 300 up-regulated genes showed that 5 out of the top 10 GO terms were related to immune response, which is consistent with the phenotype of PL mice. On the other hand, down-regulated genes were related to endoplasmic reticulum and digestion, which are associated with normal acinar cell function (Fig 6B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. RNA-seq analysis of LATS1/2 regulated genes in mouse pancreas.

(A) Heat map analysis showed differential gene expression patterns at different days (n = 3–5). (B) Left: K-means clustering analysis showed the gene expression patterns of 2 classes (up-regulated and down-regulated genes) that were changed with time. Right: GO term analysis of the function of top up-regulated and down-regulated genes. (C) Validation of profibrotic gene expressions by qPCR. (D) The small lesions were stained with anti- αSMA, anti-SPP1, anti-YAP1, and anti-TAZ antibodies in consecutive sections. White arrow indicates region in which αSMA is already present, while red arrow indicates region in which SPP1 expression has occurred but αSMA is not yet present. (E) Co-stain of αSMA (Red) with SPP1 (Green) in P and PL (Day 2 and Day 3) mice by immunofluorescence. (F) Isolated PSCs were treated with CTGF (5 nM) or SPP1 (50 nM) for 3 days (n = 3). Untreated group serves as the control. Anti-αSMA antibody was used to stain activated PSCs (Green). Cell proliferation was evaluated by EdU incorporation (White). Nuclei stained with DAPI (Blue). (G) SPP1 IHC staining in control, AP, and CP mice. Quantitative analysis for αSMA and EdU incorporation in CTGF or SPP1-stimulated cells (n = 3). *P < 0.05. Underlying numerical values can be found in S1 Data. αSMA, α-smooth muscle actin; AP, acute pancreatitis; CP, chronic pancreatitis; CTGF, connective tissue growth factor; EdU, 5-ethynyl-2'-deoxyuridine; IHC, immunohistochemistry; LATS1, large tumor suppressor 1; P, control; PL, double knockout; PSC, pancreatic stellate cell; qPCR, quantitative PCR; RNA-seq, RNA sequencing; SPP1, secreted phosphoprotein 1; TAZ, transcriptional coactivator with PDZ binding motif; YAP1, yes-associated protein 1.

https://doi.org/10.1371/journal.pbio.3000418.g006

We specifically analyzed GO terms regarding growth factor activity (GO:0008083), cytokine activity (GO:0005125), and chemokine activity (GO:0008009), and identified a total of 12 up-regulated genes (Bmp1, Ctgf, Cxcl12, Mia, Pdgfα, Tgfβ1, Tgfβ3, Timp1, Ccl12, Cxcl10, Cxcl14, and Spp1). Interestingly, several genes involved in PSC activation (Tgfβ and Pdgfα) and fibrogenesis (Spp1, Ctgf, Cxcl12, and Cxcl16) were significantly up-regulated at Day 2, before immune cell infiltration, suggesting a potential function of acinar cell YAP1/TAZ in direct regulation of PSC activation and pancreatic fibrosis. We validated the expression changes of Ctgf, Spp1, Cxcl12, Cxcl16, Pdgfα, TGFβ1, TGFβ2, and TGFβ3 by qPCR (Fig 6C). Except for TGFβ2, all of these genes showed gradual up-regulation during the time course, strongly supporting that they may be direct targets of YAP1/TAZ.

Among these genes, the up-regulation of secreted phosphoprotein 1 (Spp1) was the most dramatic. Even at Day 1 after TAM injection, we detected an about 2-fold increase of Spp1 compared to the control P mice. Spp1 has been suggested as a potential target of YAP1/TAZ [22] and is also strongly associated with TGFβ function and hepatic stellate cell (HSC) activation [23]. Thus, we performed immunostaining of SPP1 on consecutive sections of Day 10 PL pancreata with one dose of TAM (180 mg/kg). We found that SPP1 staining correlated well with αSMA, YAP1, and TAZ staining in small lesions (Fig 6D, S9B Fig). SPP1 was strongly positive, while no apparent αSMA staining surrounded acinar cells at Day 2 and Day 3 after 5 doses of TAM (Fig 6E). Immunofluroescent staining and reverse transcription (RT)-PCR for SPP1 in P, PL, and PLYT mice pancreata after TAM treatment showed that SPP1 up-regulation due to Hippo pathway inactivation was dependent on YAP1/TAZ (S9C and S9D Fig). These findings suggest that up-regulation of SPP1 by YAP1/TAZ in pancreatic acinar cells may precede PSC activation. To further test whether SPP1 can activate PSCs, we isolated the primary PSCs by density gradient centrifugation to directly test the effect of SPP1 on PSC activation [24]. The purity of the PSCs was confirmed by Glial fibrillary acidic protein (GFAP) staining (S9A Fig). The primary PSCs were treated with SSP1 recombinant proteins (50 nM) for 3 days. CTGF is a well-known YAP1/TAZ target that can activate PSCs and is up-regulated in PL mice. We used CTGF recombinant protein as the positive control. SPP1 treatment significantly increased αSMA-positive cell numbers, similarly to what we observed in the CTGF treated group, suggesting that SPP1 can trigger PSC activation (Fig 6F). SPP1 treatment did not significantly enhance the proliferation of PSCs as determined by EdU staining (Fig 6G). We noticed that αSMA-positive cells and EdU-positive cells did not always overlap (Fig 6F), suggesting that they might be two uncoupled events in PSC activation. Together, these data supported the notion that SPP1 might be an important soluble factor secreted by acinar cells with Lats1/2 deletions to activate PSCs.

Targeting the fibroinflammatory program in acinar cells attenuates Hippo disruption-induced pancreatic inflammation and fibrosis

To test whether the activation of the intrinsic fibroinflammatory program in acinar cells contributed to Hippo disruption-induced pancreatic inflammation and fibrosis, we performed a rescue experiment using CTGF-neutralizing antibody FG-3019 (FibroGen, Inc.) [25]. CTGF is the best known YAP1/TAZ target gene and has been documented to play a central role in PSC activation [26] and fibrosis [27]. We found that Ctgf mRNA expression was up-regulated at an early time point (Day 2) after Lats1/2 deletions (Fig 7A). In addition, CTGF protein level was significantly decreased in PLTY mice compared to PL mice (Fig 7B, S10A Fig). We injected PL mice with FG-3019 or FG-hulgG (control, human Immunoglobulin G) for 2 days (i.p. 30 mg/kg/d), followed by a single TAM (180 mg/kg) injection. After this, the mice received FG-3019 or FG-hulgG injection every other day and were euthanized at Day 15 (Fig 7C). The HE staining revealed that CTGF depletion partially attenuated inflammation and fibrosis in PL mice (Fig 7D). FG-3019-treated mice achieved 61.7%, 53.5%, and 56.7% reduction of immune cell infiltration, stromal reaction, and ADM, respectively, compared with the FG-hulgG-treated group (Fig 7E), suggesting that neutralization of CTGF partially rescued the pancreatitis induced by Lats1/2 deletions in adult acinar cells. To examine whether neutralizing CTGF affected the health of the mice, we treated another group of mice (n = 6) for 30 days. Half of the FG-3019-treated mice maintained their body weight, while all FG-hulgG-treated mice lost weight by Day 12 (S10B Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Depletion of CTGF partially abolishes pancreatic inflammation and fibrosis in Lats1/2 double knockout mice.

(A) Ctgf mRNA expression in P and PL mice (n = 3). **P < 0.01. (B) Western blot of CTGF protein levels in P, PL, and PLTY mice. Tubulin served as the internal control. (C) Schematic diagram of experimental design (n = 6 per group). Mice without TAM injection served as control group. FG-3019: CTGF-neutralizing antibody. (D) Top: histological examination by HE staining; center: immunofluorescent staining of αSMA (Green); bottom: immunofluorescent staining of CD45 (Red). Nuclei stained with DAPI (blue). (E) Quantitative analysis of FG-3019 treatment effects on Lats1/2 deletion-induced ADM, PSC activation, and immune cell infiltration in the mouse pancreas (n = 6); *P < 0.05. Underlying numerical values can be found in S1 Data. αSMA, α-smooth muscle actin; ADM, acinar-to-ductal metaplasia; CD45, ; CTGF, connective tissue growth factor; HE, hematoxylin–eosin; IgG, human Immunoglobulin G control; LATS1, large tumor suppressor 1; P, control; PL, double knockout; PLTY, quadruple deletions of Lats1/2 and YAP1/TAZ; PSC, pancreatic stellate cell; TAM, tamoxifen; TAZ, transcriptional coactivator with PDZ binding motif; YAP1, yes-associated protein 1.

https://doi.org/10.1371/journal.pbio.3000418.g007

Pancreatitis mouse models exhibit Hippo disruption-induced molecular features

To test potential involvement of the Hippo pathway in pancreatitis, we examined the histological alterations and the expression patterns of Hippo pathway core components in time course analysis of caerulein-induced AP (Fig 8A). Pancreatic inflammation was observed from Day 1 to 3, and recovery started at Day 4. By Day 6, the pancreas appeared normal (Fig 8B). Interestingly, western blot results revealed that the protein levels of Lats1/2 were significantly down-regulated from Day 1. By Day 6, the expression of both LATS1 and LATS2 recovered. In addition, the level of phospho-YAP1 (P-YAP1) showed a similar trend as LATS kinase. Inversely, YAP1/TAZ were increased from Day 1, and when the pancreas recovered, YAP1/TAZ went down back to normal levels (Fig 8C, S11A Fig). YAP nuclei staining was observed at Day 1 in the AP pancreas (S11B Fig). These data indicated a strong relationship between the Hippo signaling pathway and inflammatory response. CP was induced by repetitive episodes of AP for 4 weeks (Fig 8D), resulting in pancreatic fibrosis and increased collagen deposition, which was stained by Sirius Red (Fig 8E). Western blot data clearly showed significant reduction of LATS1/2 levels and elevated levels of YAP1/TAZ in the CP group compared to control mice (Fig 8F, S11C Fig), suggesting that Hippo signaling is also involved in pancreatic fibrosis. SPP1 is expressed in pancreatic ductal tissues and undifferentiated pancreatic precursors [28], as well as in pancreatic cancer [29]. However, its role in pancreatitis is unknown. We found SPP1-positive staining in ductal cells in control pancreata, but also in acinar cells in caerulein-induced AP and CP, with much stronger staining in CP pancreata (Fig 8G). Taken together, we found that the Hippo signaling pathway is dynamically regulated during caerulein-induced pancreatitis, indicating a potential function of Hippo signaling in modulating pancreatic inflammation and fibrosis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. The Hippo signaling pathway is involved in inflammation and fibrosis in caerulein-induced pancreatitis mouse model.

(A) Schematic diagram of induction of AP. n = 3. (B) The AP was analyzed by HE examination on Day 1, Day 2, Day 3, Day 4, and Day 6 after the induction. (C) Western blot of LATS1, LATS2, P-YAP1, YAP1, and TAZ in untreated and AP mice. Untreated mice served as the control group. Tubulin was used as the internal control. (D) Schematic diagram of induction of CP (n = 4). (E) HE examination of CP mice. Untreated mice served as the control group. Sirius Red was used to detect collagen deposition in CP mice. (F) Western blot of LATS1, LATS2, YAP1, and TAZ in untreated and CP mice. AP, acute pancreatitis; CP, chronic pancreatitis; HE, hematoxylin–eosin; LATS1, large tumor suppressor 1; P-YAP1, phospho-YAP1; TAZ, transcriptional coactivator with PDZ binding motif; YAP1, yes-associated protein 1.

https://doi.org/10.1371/journal.pbio.3000418.g008

Ethics statement

All procedures have been approved by Institutional Animal Care and Use Committee at the University of Texas Health San Antonio.

Generation of conditional knockout mice

All animal study protocols were approved by the University of Texas Health San Antonio Animal Care and Use committee. Ptf1a^Cre-ERTM mice (The Jackson Laboratory, Bar Harbor, ME; stock number: 019378) and R26R-EYFP mice (The Jackson Laboratory; stock number: 006148) were obtained from Hebrok Lab [41]. Lats1^fl/fl and Lats2^fl/fl mice were kindly provided by Dr. Randy L. Johnson. Yap1^fl/fl and Taz^fl/fl mice were kindly provided by Dr. Eric N. Olson. We generated (1) Ptf1a^CRE-ERRosa26^LSL-YFP mice (P mice, control), (2) Ptf1a^CRE-ERRosa26^LSL-YFPLats1^fl/flLats2^fl/fl mice (PL mice), (3) Ptf1a^CRE-ERYap1^fl/flTaz^fl/fl mice (PTY mice), and (4) Ptf1a^CRE-ERRosa26^LSL-YFPLats1^fl/flLats2^fl/flYap1^fl/flTaz^fl/fl mice (PLTY mice). All offspring were genotyped by PCR of genomic DNA from the toe with primers specific for the Ptf1a^Cre-ER, Rosa26^LSL-YFP, Lats1, Lats2, Yap1, and Taz transgenes. To induce the conditional knockout, 6- to 12-week-old mice were treated with 45 mg/kg to 180 mg/kg of TAM (Sigma-Aldrich, St. Louis, MO, T5648-5G), which was dissolved in corn oil (Sigma-Aldrich, St. Louis, MO, C8267) via i.p. injection. PCR was used for validation of knockout alleles (S1 Table).

Tissue preparation

Mice were euthanized, and pancreata were dissected into ice-cold phosphate-buffered saline (PBS). For histological examination, mice pancreata were fixed in 4% paraformaldehyde (PFA) overnight at 4°C and then processed for paraffin or O.C.T embedding. Sections were cut at 5 μm and then used for HE staining or immunostaining.

Immunostaining

Primary and secondary antibodies used in this study are listed in S2 Table. For IHC staining, tissues were deparaffinized, rehydrated, and immersed and heated in a citrate-based antigen unmasking solution (Vector, Burlingame, CA, H300) for 30 minutes. Endogenous peroxidase activity was quenched with 3% hydrogen peroxide in ddH₂O. Sections were blocked with 5% donkey serum in 0.1% TBST (Tween20, Sigma-Aldrich, St. Louis, MO, P1379-500 ml) for 1 hour at room temperature and then incubated with primary antibody at 4°C overnight. Biotin-labeled secondary antibodies (1:250, BD Pharmingen) and Streptavidin-Horseradish peroxidase (SAv-HRP, BD Pharmingen, San Jose, CA; 550946) were applied to amplify the target antigen signal, and then sections were developed with DAB substrate (Dako, Santa Clara, CA; K3468), followed by counterstaining with hematoxylin (Biocare Medical, Pacheco, CA). Finally, sections were covered with poly-Mount medium (Polysciences, Warrington, PA; Category number 08381). For immunofluorescence staining, frozen sections were blocked with 10% donkey serum plus 1% bovine serum albumin (BSA) in PBST (0.025% Triton X-100, Fisher Scientific; Waltham, MA, CAS9002-93-1) for 1 hour at room temperature. Sections were incubated with primary antibodies at 4°C overnight followed by incubation with the Alexa Fluor secondary antibodies (1:250, Jackson ImmunoResearch, West Grove, PA) for 1 hour at room temperature. Sections were covered with a drop of ProLong Gold anti-fade reagent with DAPI (Invitrogen, Carlsbad, CA, P36935) for observation. All images were obtained with a Microsystems DMI6000 B microscope and microscope software (Leica Microsystems, Buffalo Grove, IL).

Western blot analysis

Proteins were isolated from the tail of the pancreas. Briefly, pancreata were homogenized in RIPA lysis buffer containing phosphatase and protease Inhibitors (GenDEPOT, Katy, TX; P3100-001 and P3200-001) on ice. After 10-cycle of sonication (Bioruptor Pico Sonication System, Diagenode, Denville, NJ, Category Number B01060001), proteins were extracted, and the concentration was determined by a BCA kit (Pierce, Rockford, IL; 23228). Equal amounts of 50 μg protein were resolved in SDS-PAGE, then transferred to PVDF membrane (ThermoFisher Scientific, Waltham, MA, Product number 88520) and blocked with 5% milk in PBST (0.1% Tween 20, Sigma-Aldrich, St. Louis, MO, P1379-500ml) for 1 hour at room temperature. Membranes were incubated with primary antibodies at 4°C overnight and then exposed to secondary antibodies for 1 hour. Finally, protein expression was determined by treating the membrane with Clarity Western ECL substrate (Bio-Rad, Hercules, CA, Category number 1705060) and measuring luminescence with an Amersham Imager 600 system (GE Healthcare, Waukesha, WI). (All antibody information and working concentrations are shown in S2 Table).

RT and real-time PCR analysis

Total RNA was extracted from the tail of the pancreas using a lysis binding buffer (Ambion, Austin, TX; Category AM1560) and Direct-zol RNA miniprep kit (Zymo Research, Irvine, CA, R2052). The RNA concentration was measured using Nanodrop2000 (ThermoFisher Scientific, Waltham, MA). The amount of 400 ng of RNA was subjected to RT by using a high-capacity cDNA RT kit (Applied Biosystems; 4368814), following the manufacturer’s protocol. The cDNA samples were amplified with iTaq Universal SYBR Green Supermix (Bio-Rad, Hercules, CA; 172–5271) in a CFX96 Real-Time system (Bio-Rad, Hercules, CA). GAPDH was used as endogenous control for normalization, and the relative expression of mRNA was calculated based on the ^ΔΔCt method. The information of the primers is available in S3 Table.

Flow cytometry analysis

Single-cell solutions from P and PL mice were prepared according to previous reports. Briefly, 6-week–aged mice P and PL were treated with 180 mg/kg of TAM for 5 consecutive days. Four days later, mice were anesthetized and then perfused with cold PBS. Pancreas was digested by collagenase (Sigma-Aldrich, St. Louis, MO; C9891-1G); then, cells eventually filtrated through a 5 ml polystyrene round-bottom tube with cell-strainer cap (Corning, Corning, NY; number 352235). All antibodies used for flow cytometry were purchased from Biolegend (San Diego, CA): APC-CD45 (103111; 1:100), Percp/Cy5.5-CD11b (101228; 1:100), and BV421-F4/80 (123137; 1:100). Cells were incubated with the antibodies in FACS buffer at 4°C for 15 minutes. After incubation, cells were washed once and resuspended into FACS buffer. Sorting experiments were conducted on BD FACSAria (BD Biosciences, Franklin Lakes, NJ) with BD FACSDiva software (BD Biosciences, http://www.bdbiosciences.com/en-us/instruments/research-instruments/research-software/flow-cytometry-acquisition/facsdiva-software).

Isolation and culture of mouse PSCs

Primary mouse PSCs were isolated using a previously reported gradient centrifugation method with a few modifications [24]. Briefly, pancreata were dissected from wild-type mice at the ages of 4 to 6 weeks. After all adipose tissues were carefully trimmed off, pancreata were cut into small pieces in cold HBSS buffer. Tissues from 2 mice were transferred into a 50 ml tube containing 10 ml HBSS buffer (0.05% pronase, 0.035% collagenase P, 0.1% DNAse, 10 mM HEPES, and 0.01 Trypsin inhibitor, Sigma-Aldrich). Tissues were incubated in a 37°C water bath for 10 minutes and then dissociated by pipetting up and down with a 10 ml tip 10 times. The tissues were incubated in a 37°C water bath for an additional 5 to 10 minutes and dispersed by pipetting gently up and down with a 10 ml pipette 10 times. The cell suspension was filtered through a 70 μm cell strainer. Cells were washed twice and suspended in 2 ml HBSS buffer containing 0.3% BSA. The amount of 4 ml 60% Optiprep was added to the cell suspension and mixed well in a 15 ml tube; 2 ml HBSS buffer was carefully added to the top of the Optiprep gradient. PSCs were separated by centrifugation at 1400g for 20 minutes. PSCs were suspended in DMEM:F-12 medium containing 10% FBS, 15 mM HEPES, 4 Mm glutamine, and antibiotics (penicillin 100 U/ml and streptomycin 100 μg/ml). Cells were seeded on a 10-well microscope slide (6,000 cells/well) and treated with CTGF and SPP1 for 3 days.

Acini preparation and treatment

Immediately after euthanasia, pancreata were dissected out, rinsed in cold 1X Hank’s balanced salt solution, and cut into small pieces. The small tissue pieces were digested with collagenase P (200 μg/ml in 1X HBSS) for 20 minutes at 37 °C. Released cells were washed 3 times in cold 1X HBSS with 5% fetal bovine serum. Large clumps of connective tissues were removed by filtering with 100-μm nylon meshes. Cell suspensions were carefully layered on top of 30% fetal bovine serum for centrifuge (300g, 2 minutes at 4 °C) and then were resuspended in 3D culture media (RPMI, 0.5% fetal bovine serum, 0.1 mg/ml soybean trypsin inhibitor and antibiotics). Forty-eight-well tissue culture plates were coated with a 150 μl/well collagen layer (1 part 10X RPMI, 9 parts 3 mg/ml collagen, neutralized with 4.2% NaHCO₃) and incubated for 1 hour at 37 °C. The acini suspension was mixed with collagen 1:1 and plated (0.3 ml/well). The cell–collagen mix was allowed to solidify for 1 hour at 37 °C before adding 0.5 ml 3D culture media with TGFα.

RNA-seq and data analysis

Total RNA was isolated using RNA lysis binding buffer (Ambion, Austin, TX; Category number AM1560) at Days 1, 2, and 3, as well as control condition. RNA quality was assessed by Bioanalyzer, and poly A(+) RNA was isolated by oligo-dT purification and fragmented using divalent cations under elevated temperature. cDNA fragment libraries were synthesized following the TruSeq mRNA-seq Library Preparation protocol (Illumina). For a total of 16 samples, we obtained about 31.4 million sequence reads per sample, using Illumina HiSeq system at the Greehey Children’s Cancer Research Institute Genome Sequencing Facility, utilizing a 50 bp single-read sequencing protocol.

Short sequence reads were then aligned with TopHat [42] to mouse genome (UCSC mm⁹ https://genome.ucsc.edu/cgi-bin/hgGateway?db=mm9), allowing no more than 2 mismatches in the alignment. Expression abundance of each gene was evaluated with read count and RPKM. Differential gene expression was calculated using DESeq [43] to obtain fold-change, P value, and adjusted P value by Benjamini-Hochberg correction for multiple tests. DEGs were selected based on the following criteria: (1) fold-change > 2, (2) adjusted P < 0.05, and (3) RPKM > 1. Using the significance criterion and for comparison between Days 1, 2, and 3 to control, we obtained 40, 405, and 765 DEGs, respectively. To access the overall expression change across time, a heatmap was created using all DEGs from 3 comparisons. We also performed K-means algorithm (MATLAB, Mathworks) with all DEGs into 9 clusters and then merged up-regulated and down-regulated genes into 2 representative clusters (300 and 232 genes, respectively). Biological functional assessment of these DEGs was performed by using Database for Annotation, Visualization and Integrated Discovery (DAVID; http://david.abcc.ncifcrf.gov/), with the top 10 most enriched biological processes (BPs) provided in the figure.

CTGF inhibition experiment

PL mice were injected once with neutralizing mAb FG-3019 (30 mg/kg, FibroGen) by i.p. administration. Twenty-four hours later, mice were subjected to 180 mg/kg of TAM once, and then continued to be treated with FG-3019 every other day. Human IgG (equal amount, Jackson ImmunoResearch) was used as the control. Mice were euthanized at Day 15 after TAM injection. Pancreatic tissue sections were stained with HE to quantify the severity of inflammation.

Caerulein-induced acute and CP

Caerulein (American Peptide Company, 46-1-50) was used to induce both acute and CP mouse models. For AP, caerulein was administered through 8 i.p. injections (50 μg/kg) at hourly intervals per day for 2 days [44]. Pancreatic inflammation was assessed by HE staining on Days 1, 2, 3, 4, and 6 after induction of AP. For CP, mice were subjected to repetitive episodes of AP (50 μg/kg, 8 injections at hourly intervals) every other day for 4 weeks. Mice were euthanized 4 weeks after the initial caerulein injection.

Statistical analysis

All data are expressed as mean ± standard error of the mean. GraphPad Prism 5 Software was used to analyze the data with ANOVA and Student t test.

Transcript profiling

All RNA-seq files are available from GEO database (accession code: GSE111640).