Mechanism of trypsin-mediated differentiation of pancreatic progenitor cells into functional islet-like clusters

Abstract

BACKGROUND

Endogenous regeneration of pancreatic islet β-cells is a path to cure both type 1 and advanced type 2 diabetes. Pancreatic cancer cell line-1 (PANC-1), a human pancreatic islet progenitor cell line, can be induced by trypsin to differentiate into insulin-secreting islet-like aggregates (ILAs). However, the underlying mechanism has not been explored.

AIM

To explore the mechanism and signaling pathway of trypsin-induced differentiation of islet progenitor cells into insulin-secreting cells.

METHODS

PANC-1 cells were induced by trypsin to form ILAs and differentiate into insulin-secreting cells. Clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein 9 knockout and small interfering RNA knockdown techniques were used to investigate membrane proteins and downstream signaling pathways involved in the process.

RESULTS

The extracellular domain of membrane receptor E-cadherin hydrolyzed by trypsin induced the aggregation of PANC-1 cells and stimulated E-cadherin-recruited casein kinase-1γ3, which specifically phosphorylated the Ser655/Thr658 site of α-catenin in the cadherin-catenin complex, participating in the process of PANC-1 differentiation and affecting the maturation of differentiated ILAs.

CONCLUSION

The current study reveals the mechanism by which trypsin promotes PANC-1 cell differentiation into islet-like cells, providing a novel approach for endogenous islet β-cell regeneration.

Key Words: Endogenous islet β-cell regeneration; Trypsin; Pancreatic cancer cell line-1; Pancreatic islet progenitor cell; E-cadherin; Alpha-catenin

Core Tip: Trypsin induced pancreatic cancer cell line-1 cell differentiation into insulin-producing cells. Loss of E-cadherin, αE-catenin, and casein kinase-1γ3 (CK1γ3) blocked trypsin-induced differentiation. Hydrolysis of E-cadherin by trypsin recruited CK1γ3. CK1γ3 phosphorylated αE-catenin at Ser655/Thr658.

INTRODUCTION

Diabetes is a chronic metabolic disease with complex complications, with the number of patients estimated to reach 643 million by 2030 worldwide[1]. β-Cell dysfunction occurs in both type 1 and advanced type 2 diabetes, for which hypoglycemic drugs or insulin injection therapy are the mainstream of treatment, although they cannot cure the disease[2]. Therefore, the study of islet β-cell regeneration is important for cell replacement therapy and/or endogenous islet regeneration. However, the most studied pluripotent stem cells are faced with ethical dilemmas and have a carcinogenic risk[3]. Pancreatic progenitor cell (PPC) differentiation can potentially avoid these problems and achieve endogenous islet regeneration[3,4].

We previously found that trypsin induced the human pancreatic cancer cell line-1 (PANC-1), pancreatic ductal epithelial cells to differentiate into insulin-secreting islet-like cells, which responded to glucose stimulation in vitro and acted as islet cells in vivo after being transplanted[5]. There are cytokeratin-19-positive cells in pancreatic ductal cells, which can be induced to become pancreatic endocrine cells and acinar cells through different methods, such as pancreatic duct ligation, treatment with small molecules, such as activin A (ActA), exendin-4 and glucose, or overexpression of transcription factors such as pancreatic and duodenal homeobox 1, neurogenin 3, and MAF BZIP transcription factor A. Similarly, they can also be directed to differentiate into insulin-producing cells (IPCs). Therefore, pancreatic ductal epithelial cells are also considered a type of PPC[4]. We also found that the islet-like cell clusters derived from umbilical cord mesenchymal stem cells can be induced by trypsin, which can significantly reduce blood glucose after transplantation under the renal capsule in mice with streptozotocin-induced diabetes[6]. These results suggest that PPCs or mesenchymal stem cells can be induced to differentiate into islet-like cell clusters by trypsin.

Trypsin is a protein digestive enzyme produced by pancreatic tissue and is in direct contact with the pancreatic duct cells, which may participate in the process of islet regeneration in mice upon pancreatic injury. Trypsin plays an important role in the pancreas, such as tumor growth by hydrolyzing fibronectin to release fragments to mediate intracellular signals, inhibiting focal adhesion kinase (FAK) and fibroblast growth factor receptor (FGFR) signals[7], while the activation of FAK regulates glucose-dependent calcium responses and insulin secretion[8,9]. The expression of FGFR occurs during the aggregation and differentiation of precursor cells derived from adult islets and PANC-1 cells[10]. Trypsin can stimulate the mitogen-activated protein kinase (MAPK) signaling pathway in ovarian cancer cell lines[11], while the phosphorylation of p38 MAPK is associated with the dedifferentiation of pancreatic cells[12]. Therefore, it is important to investigate the mechanism of trypsin-mediated migration, aggregation and differentiation of PPCs.

Numerous studies have reported that protease activated receptor 2 (PAR2) is highly expressed in pancreatic tissue and mainly activated by trypsin[13,14], while E-cadherin can be affected by hydrolysis and cleavage by a variety of proteases to mediate intracellular signaling[15]. Therefore, we investigated if trypsin acted on membrane protein receptors PAR2 or E-cadherin to activate their intracellular signals to induce the aggregation and differentiation of PANC-1 cells.

MATERIALS AND METHODS

Differentiation of PANC-1 cells

The day before the induction of differentiation, PANC-1 cells (NSTI, Shanghai, China) were plated in a 12-well plate at 2.6 × 10⁵ cells per well. After washing with 1 × phosphate-buffered saline (PBS), cells in each well were digested with 300 μL 0.05% trypsin for 40 seconds at 37 °C, and then it was replaced by 1 mL differentiation medium (DM), containing Dulbecco’s modified Eagle’s medium/F-12 (Hyclone, Logan, UT, United States), bovine serum albumin (BSA) (ABW, Shanghai, China), and insulin transferrin selenium (Sigma-Aldrich, St. Louis, MO, United States). The edges of PANC-1 cells shrank, the cells gradually became rounded and brighter, intercellular spaces appeared, and the cells became single but still attached to the plate. Cells were observed under a microscope, and the 12-well plate was placed in a 37 °C incubator for culture.

Detection of islet-specific genes by polymerase chain reaction

The mature islet-like cell clusters were differentiated for 120 hours and the cells before differentiation were collected. Total mRNA extracted from cells by TRIzol (Sigma-Aldrich) was reverse transcribed into complementary DNA (cDNA) using 5 All-In-One RT Master Mix (G592; Applied Biological Materials, Shanghai, China). Polymerase chain reaction (PCR) was performed using 2 × Hieff^® PCR Master Mix (10102ES03; YEASEN, Shanghai, China). The primaries used for PCR are listed in Table 1, and the procedures are listed in Table 2.

Table 1 Primers of polymerase chain reaction for islet specific genes.

Name	Sequences
Insulin-F	5’-GGAACGAGGCTTCTTCTACACA-3’
Insulin-R	5’-TGTTCCACAATGCCACGCTTCT-3’
Glucose-F	5’-ACATTCACCAGTGACTACAGCA-3’
Glucose-R	5’-GGCAGCTTGGCCTTCCAAATAA-3’
PAX6-F	5’-AGGCTCAAATGCGACTTCAGCT-3’
PAX6-R	5’-TGTTGCTGGCCTGTCTTCTCTGAT-3’
SOX9-F	5’-GGCAAGCTCTGGAGACTTCTG-3’
SOX9-R	5’-CCCGTTCTTCACCGACTTCC-3’
β-actin-F	5’-ATGAAGTGTGACGTGGACATCC-3’
β-actin-R	5’-CAGGAGGAGCAATGATCTTGATC-3’

PAX6: Paired box 6; SOX9: SRY-box transcription factor 9.

Table 2 Polymerase chain reaction procedure.

Step	Temperature (°C)	Time	Number of cycles
Initial denaturation	94	5 minutes	1
Denaturation	94	30 seconds	35
Annealing	50-60	30 seconds
Extension	72	30 seconds/kb
Final extension	72	10 minutes	1
Cooling or holding	4	∞	1

Immunocytochemistry

Cells were fixed in 4% paraformaldehyde and permeabilized with 0.1% Triton at room temperature. Nonspecific sites were blocked in 5% BSA for 1 hour. Cells were incubated with primary antibody (Table 3) for 1 hour. After washing three times with PBS, the cells were stained with fluorescence-labeled secondary antibody for 1 hour in the dark. Images were observed using a fluorescent microscope (Zeiss, Oberkochen, Germany).

Table 3 Antibody for immunofluorescence.

Name	Cat and suppliers
Insulin	GB13121, Servicebio, Hubei Province, China
Glucagon	SAB4501137, Sigma-Aldrich, MO, United States
CSNK1G3	CSB-PA896549 LAOIHU, Cusabio, Hubei Province, China
E-cadherin	sc-8426, Santa Cruz, TX, United States

CSNK1G3: Casein kinase 1 gamma 3.

Gene knockout

Restriction enzyme BsmBI (FD0454; Thermo Fisher Scientific, Waltham, MA, United States) was used, and the nucleic acid sequences of single-guide RNA (sgRNA) listed in Table 4 specifically targeting genes were inserted into the lentivirus LentiCRISPRv2 vector[16]. Transfection reagent (Vigene Biosciences, Shandong, China) was used for cotransfection of the core plasmid and three lentiviral plasmids, including packaging plasmids, envelope expressing plasmid, and rev protein expressing plasmid into 293T cells to package the lentivirus. After 72 hours of transfection of PANC-1 cells with lentivirus, they were screened with 2 μg/mL puromycin to make stable cells lines. Western blotting and DNA sequencing methods were used to confirm the knockdown effects. Puromycin (1 μg/mL) was used to establish a stably transfected cell line.

Table 4 Single-guide RNA for knockout.

Name	Sequences
sgRNA-F2RL1-F	5’-caccgGTCTGCTTCACGACATACAA-3’
sgRNA-F2RL1-R	5’-aaacCTTGTATGTCGTGAAGCAGACc-3’
sgRNA-CDH1-F	5’-caccgCTACGGTTTCATAACCCACA-3’
sgRNA-CDH1-R	5’-aaacTGTGGGTTATGAAACCGTAGc-3’

CDH1: E cadherin; F2RL1: F2R like trypsin receptor 1; SgRNA: Single-guide RNA.

Western blotting

PANC-1 cells were harvested and lysed on ice with a cell-lysis buffer. Total protein concentration was determined with a bicinchoninic acid protein assay kit (Thermo Fisher Scientific). About 30 μg total protein was separated by 10% of sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride (PVDF) membrane. After transfer, the PVDF membrane was blocked in 5% milk. Antibodies used for western blotting are listed in Table 5. Blots were developed using the western bright enhanced chemiluminescence kit (Advansta, San Jose, CA, United States).

Table 5 Antibody for western blotting.

Name	Cat and suppliers	Dilution ratios
Phospho-α-E-catenin (Ser655/Thr658)	13231, Cell Signaling Technology, Danvers, MA, United States	1:1000
Phospho-α-E-catenin (Ser641)	21330-1, Cell Signaling Technology, MA, United States	1:1000
Alpha-E-catenin	12831-1-AP, Proteintech, Hubei Province, China	1:1000
β-actin	A3854, Sigma-Aldrich, St Louis, MO, United States	1:50000
LRP5	24899-1-AP, Proteintech	1:1000
CK1α	ab206652, Abcam, Cambridge, United Kingdom	1:2000
CK1ε	ab270997, Abcam	1:2000
CK1γ3	df10135, Affinity, Jiangsu Province, China	1:1000
E-cadherin	20874-1-AP, Proteintech	1:2000
GAPDH	2118, Cell Signaling Technology	1:3000

CK: Casein kinase; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; LRP5: Low-density lipoprotein receptor-related protein 5.

Bioinformatics analysis

Potential interacting proteins of E-cadherin, αE-catenin, and casein kinase 1 (CK1) subtypes were searched from the String database. The common intersection was obtained using the Wayne diagram, and the common interaction protein was obtained.

Statistical analyses

Data were analyzed with the Student’s t-test, using the GraphPad program, and are expressed as the mean ± SD. Differences were considered significant at P < 0.05.

RESULTS

E-cadherin contributes to the aggregation and functional maturity of PANC-1 cells into islet-like clusters induced by trypsin

Our previous[5] and current study has shown that PANC-1 cells can be induced to differentiate into functional islet-like clusters (Supplementary Figure 1). E-cadherin[17,18] and PAR2[19,20] are both transmembrane proteins that can be activated by protease hydrolysis of trypsin. To identify the direct receptor of trypsin while inducing differentiation, we explored the role of E-cadherin and PAR2. Targeting sgRNA sequences (Table 4) were designed for knockdown by Clustered regularly interspaced short palindromic repeats (CRISPR)-CRISPR-associated 9 (Cas9). Western blotting confirmed that expression of E-cadherin in the stable transgenic cells with CDH1 gene KO by clustered regularly interspaced short palindromic repeats (CRISPR)- -associated 9 (Cas9) system impure (E-cad-imp) group was significantly decreased and the knockout (KO) efficiency was high, while in groups E-cad-KO and E-cad-KO (2), two monoclonal cell lines were screened out, and the E-cadherin (CDH1) gene was completely knocked out (Figure 1A). The KO cells were induced to differentiate, most of which floated without migration, and they adhered together into cloud-like groups at 24 hours (Figure 1B). The shape of the cell mass in the E-cad-KO group was irregular, and the color was yellowish, which differed significantly from that of the control islets (Figure 1C and D). Diphenylterazine (DTZ) staining in the KO group was lighter than that of the control group, and some cells had no obvious staining (Figure 1D), indicating that there were fewer IPCs while E-cadherin was deficient. PCR detection also showed that, in the E-cad-KO group, there was endocrine function and high expression of insulin as in the control group, but expression of glucagon was higher, paired box 6 (PAX6) expression was significantly decreased, and expression of SRY-box transcription factor 9 (SOX9) was lower than in the control group (Figure 1E). Immunofluorescence (IF) showed that there was high expression of insulin in the KO and control groups, while expression of glucagon in the KO group was significantly higher (Figure 1F). Among the cells expressing insulin in the KO group, colocalization of insulin and glucagon was seen in differentiated cells (Figure 1F). Some of the cells were double hormone-expressing cells, indicating the presence of immature β-cells. However, compared to normal cells, the cells with knockdown of PAR did not have any morphological and functional differences in the migration and aggregation into insulin-secreting islet-like aggregates (ILAs) (Supplementary Figure 2). These phenomena confirmed that E-cadherin affected the morphology and maturation during the differentiation of islet-like clusters.

Figure 1 E-cadherin contributes to aggregation of human pancreatic cancer cell line-1 into islet-like clusters induced by trypsin. A: Expression of E-cadherin protein was detected by western blotting, reflecting knockout (KO) efficiency of E-cadherin 1 gene (CDH1), E-cad-imp group: Stable transgenic cells with CDH1 gene KO by clustered regularly interspaced short palindromic repeats (CRISPR)- -associated 9 (Cas9) system impure, E-cad-KO and E-cad-KO (2): Monoclonal cells screening from stable KO cells; B: Representative images of KO cells at five time points after digestion with trypsin before and after CDH1 KO, magnification 10 ×; C: Morphology of islet-like cell clusters differentiated in vitro was compared with that of natural islets; D: Representative image of diphenylterazine (DTZ) staining of cell clusters formed at 96 hours before and after CDH1 KO, magnification 10 ×; E: Polymerase chain reaction detection reflected the changes in expression of the transcripts for mature islet-related genes insulin, glucagon, paired box 6 (PAX6), SRY-box transcription factor 9 (SOX9) before and after CDH1 KO; F: Representative immunofluorescence images of insulin, glucagon, and 4’,6-diamidino-2-phenylindole (DAPI) signal in differentiated cell clusters before and after CDH1 KO, magnification 20 ×, increased to 63 ×. CON: Control; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase.

αE-catenin is related to trypsin-mediated PANC-1 cell aggregation and differentiation

αE-catenin is crucial in islet development and binds to β-catenin and E-cadherin, playing an important role in cell-cell adhesion. We speculated that αE-catenin might be an important node for differentiation induced by trypsin. To test this, a KO cell line was constructed using the CRISPR-Cas9 system, with targeted sgRNA sequence (Table 4). Western blotting showed that the KO cell line did not express αE-catenin (Figure 2A). When the KO cells were induced to differentiate, more cells floated at 24 hours, and the floating cells gathered together later to form diffuse and floating clusters, without the process of slow migration (Figure 2B). The cell mass in the KO group was easy to disperse into small cell clusters, and DTZ staining was not as obvious as in the control group (Figure 2C), indicating that less mature IPCs were formed in αE-catenin KO cells. These results indicate that αE-catenin is indispensable in the process of clustering and differentiation of PANC-1 cells mediated by trypsin, similar to E-cadherin.

Figure 2 αE-catenin is related to trypsin-mediated human pancreatic cancer cell line-1 aggregation and differentiation. A: Expression of αE-catenin protein was detected by western blotting, reflecting knockout (KO) efficiency of catenin alpha 1 (CTNNA1); B: Representative images of KO cells at five time points after digestion with trypsin before and after CTNNA1 KO, magnification 10 ×; C: Representative images of diphenylterazine (DTZ) staining of cell clusters formed at 96 hours before and after KO of CTNNA1, magnification 10 ×. CON: Control.

Phosphorylation of αE-catenin at Ser655/Thr658 is necessary during differentiation of PANC-1 cells induced by trypsin

The phosphorylation of αE-catenin at Ser655/Thr658 by CK1 or Ser641 by CK2 plays a critical role in αE-catenin functioning. To further explore how αE-catenin plays a role after trypsin treatment, we explored phosphorylation at sites Ser655/Thr658 of αE-catenin. The phosphorylation of αE-catenin at Ser655/Thr658 was detected by western blotting at different differentiation time points (Figure 3A). During the whole differentiation process, the phosphorylation level of the Ser655/Thr658 site changed, showing a trend of increasing at early times, and decreasing and then increasing again. Phosphorylation peaked when PANC-1 cells were treated by trypsin and cultured in DM for 2 min (Figure 3B).

Figure 3 Phosphorylation of αE-catenin at Ser655/Thr658 during human pancreatic cancer cell line-1 differentiation and effects of casein kinase-1 inhibitor D4476 on human pancreatic cancer cell line-1. A: Representative western blot images reflecting protein expression of phospho-αE-catenin (p-αE-catenin) (Ser655/Thr658) at different time points during human pancreatic cancer cell line-1 (PANC-1) differentiation; B: Statistical analyses of the gray value of western blot images using ImageJ and GraphPad Prism, reflecting the differences in expression of p-αE-catenin (Ser655/Thr658); C: Chemical structure of D4476, inhibitor of casein kinase-1; D: Representative images of PANC-1 cells treated with 2, 5, or 10 μM D4476 for 72 hours, magnification 10 ×; E: Western blot images showed expression of p-αE-catenin (Ser655/Thr658) of PANC-1 cells treated with 2 μM D4476; F: Western blot images showed expression of p-αE-catenin (Ser655/Thr658) of PANC-1 cells treated with 2 μM D4476 at different times; G: Representative images of PANC-1 after 72 hours treatment with 2 μM D4476 at different time points after digestion with trypsin, magnification 10 ×. ^aP < 0.05. ^bP < 0.01. P calculated vs control (CON). GAPDH: Glyceraldehyde-3-phosphate dehydrogenase.

As CK1 is the kinase of αE-catenin phosphorylation, we investigated the effect of CK1 inhibitor D4476 (Figure 3C). Western blotting showed that D4476 inhibited the phosphorylation of αE-catenin at Ser655/Thr658, and concentrations of 2 μM and 5 μM were appropriate (Figure 3D and E). Phospho-αE-catenin (Ser655/Thr658) was completely inhibited by treating PANC-1 cells with 2 μM D4476 for 72 hours (Figure 3F). Therefore, PANC-1 cells were treated with 2 μM and 5 μM D4476 for 72 hours in advance, and then differentiation was initiated with trypsin, adding DM containing the corresponding concentration of D4476 for culture. Compared with the control group, the cells in the D4476-treated group were not firmly attached, and the formed cell clusters were diffuse and floating in sheets, which indicated that the differentiation process was interrupted (Figure 3G). However, inhibition of phospho-αE-catenin (Ser641) by CK2 inhibitor 2-dimethylamino-4,5,6,7-tetrabromo-1H-benzimidazole (DMAT) did not affect the process of differentiation (Supplementary Figure 3). Therefore, phospho-αE-catenin (Ser655/Thr658) was significantly involved in the process of trypsin-induced PANC-1 cell differentiation.

Effects of E-cadherin on phosphorylation of αE-catenin (Ser655/T658) are mediated by trypsin

We suspected that the phosphorylation of αE-catenin (Ser655/Thr658) mediated by trypsin was related to activation of E-cadherin. Western blotting showed that the effect of E-cad-KO was significantly different from that in the control group (Figure 4A), and phosphorylation of αE-catenin (Ser655/Thr658) induced by trypsin disappeared when E-cadherin was deficient (Figure 4B), but not when PAR2 was deficient (Supplementary Figure 4). This indicated that E-cadherin participated in the process of αE-catenin (Ser655/Thr658) phosphorylation. The plasmid phospho-E-cadherin-Green fluorescent protein with E-cadherin cDNA was transfected into the E-cad-KO cell line. The phosphorylation level of this site increased significantly after the restoration of E-cadherin protein expression, but no difference was observed before and after trypsin treatment (Figure 4C). E-cad-KO reduced, whereas E-cadherin overexpression significantly enhanced, the phosphorylation of αE-catenin at the Ser655/Thr658 site, indicating that the phosphorylation of αE-catenin at this site is closely linked to E-cadherin.

Figure 4 E-cadherin effects on phosphorylation of αE-catenin (Ser655/T658) mediated by trypsin. A: Representative western blot images reflecting protein expression of E-cadherin of cells with knockout (KO) of E-cadherin gene (CDH1); B: Representative western blot images reflecting protein expression of phospho-αE-catenin (p-αE-catenin) (Ser655/Thr658) in CDH1 KO cells and difference in trypsin-mediated αE-catenin (Ser655/Thr658) phosphorylation before and after KO; C: Representative western blot images reflecting protein expression of p-αE-catenin (Ser655/Thr658) of restoration of E-cadherin in CDH1 KO cells and difference in trypsin-mediated αE-catenin (Ser655/Thr658) phosphorylation before and after restoration (n = 2). ^aP < 0.05. ^cP < 0.001. P calculated vs control (CON). GAPDH: Glyceraldehyde-3-phosphate dehydrogenase.

Trypsin promotes binding of casein kinase-1γ3 to the E-cadherin complex

We investigated whether E-cadherin promoted the interaction between CK1 and αE-catenin directly or indirectly. The possible interaction of E-cadherin, αE-catenin and CK1 subtypes was searched by String database, and the common interacting proteins of CK1, E-cadherin and αE-catenin were obtained by Wayne diagram (Figure 5A) as follows: Catenin beta 1 (CTNNB1), catenin delta 1 (CTNND1), LDL receptor-related protein 5 (LRP5), Axin 1 (AXIN1), and adenomatous polyposis coli protein. After investigating the protein interactions, we proposed a model of contact between CK1 and αE-catenin induced by activation of E-cadherin (Figure 5B). In resting cells, E-cadherin structurally bound to LRP5, β-catenin and p120-catenin, while p120-catenin bound to CK1ε in a nonphosphorylated inactivation state, and CK1α bound constitutively to AXIN and was enriched on LRP5. When E-cadherin was stimulated by trypsin hydrolysis, there were changes in the activity or quantity of a subtype of CK1 protein in the complex, such as phosphorylation and activation of CK1ε, an increase in AXIN-CK1α enrichment and recruitment of CK1γ to the complex, and phosphorylation occurred while in contact with αE-catenin. The presence of CK1α, CK1ε, CK1γ, LRP5 and αE-catenin in the E-cadherin complex was detected by immunoprecipitation, which verified the hypothetical model (Figure 5C). The amount of E-cadherin decreased significantly after the action of trypsin, and increasing the concentration of soybean flour, an inhibitor of trypsin, blocked this change (Figure 5D), indicating that the hydrolytic degradation of E-cadherin by trypsin led to the loss of extracellular fragments. To further determine the specific subtypes of CK1 involved in αE-catenin phosphorylation, the samples treated with trypsin and DM for 2 minutes were analyzed by quantitative mass spectrometry of phosphorylated proteins. There was a significant difference in the level of phosphorylation of CK1γ3 after trypsin treatment was observed (Figure 5E). The colocalization of E-cadherin and CK1γ3 was significantly enhanced in some cells after treatment with trypsin (Figure 5F), indicating that CK1γ3 are closely related to the physiological changes of E-cadherin by trypsin, and may be the direct-acting kinase of αE-catenin phosphorylation.

Figure 5 Trypsin promotes binding of casein kinase-1γ3 to the E-cadherin complex. A: Wayne diagram, showing intersection of E-cadherin, αE-catenin, and casein kinase-1 (CK1) subtypes interacting proteins; B: Hypothetical model of the interaction among CK1, E-cadherin, and αE-catenin; C: Immunoprecipitation (IP) reflecting the protein composition of the E-cadherin complex; D: Quantitative mass spectrometric data analysis of phosphorylated protein, control group and trypsin representing the samples before and after trypsin treatment, and the ordinate reflects the level of phosphorylation; E: Western blot images reflect the hydrolysis of E-cadherin by trypsin and the blocking effect of different concentrations of trypsin inhibitor soybean flour (SF); F: Representative confocal images show colocalization changes of E-cadherin and CK1γ3 before and after trypsin treatment, Cy3 labeling E-cadherin (red), fluorescein isothiocyanate isomer labeling CK1γ3 (green), 4’,6-diamidino-2-phenylindole (DAPI) labeling nucleus (blue), magnification 20 ×, increased to 63 ×. ^aP < 0.05. P calculated vs control (CON). GAPDH: Glyceraldehyde-3-phosphate dehydrogenase.

CK1γ3 is related to αE-catenin (Ser655/Thr658) phosphorylation mediated by trypsin

To verify whether CK1γ3 was directly involved in trypsin-mediated αE-catenin phosphorylation, small interfering RNA (siRNA) targeting CK1γ3 was designed for knockdown. Western blotting showed that the knockdown efficiency of CK1γ3 reached nearly 50%, and protein expression decreased significantly (Figure 6A). The phosphorylation level of αE-catenin (Ser655/Thr658) after trypsin treatment was lower than that of the negative control group, and the significant differences in phosphorylation at this site mediated by trypsin disappeared after knockdown (Figure 6B), indicating that CK1γ3 was directly involved in αE-catenin (Ser655/Thr658) phosphorylation. We further induced the differentiation in CK1γ3 knockdown cells by trypsin. The differentiation of CK1γ3 in knockdown cells was similar to that in the absence of E-cadherin and αE-catenin, including 24-hour cell floating, no slow cell migration, large cloud-like aggregation of cells rather than islet-like clusters, and the staining with DTZ was weaker than that in the negative control group (Figure 6C). These results suggest that CK1γ3 participates in the differentiation process mediated by trypsin by affecting the phosphorylation of αE-catenin.

Figure 6 Casein kinase-1γ3 participated in the αE-catenin (Ser655/Thr658) phosphorylation mediated by trypsin. A: Representative western blot (WB) images reflecting the efficiency of small interfering RNA knockdown of casein kinase-1γ3 (CK1γ3); B: Representative WB images reflecting protein expression of phospho-αE-catenin (Ser655/Thr658) in CK1γ3 knockdown cells and differences in trypsin-mediated αE-catenin (Ser655/Thr658) phosphorylation before and after knockdown; C: Representative images of knockout (KO) cells at five time points after digestion with trypsin and diphenylterazine (DTZ) staining of cell clusters formed at 72 hours before and after CK1γ3 knockdown, magnification 10 ×. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ¹P calculated vs control. ²P calculated vs negative control (NC). GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; CON: Control.

DISCUSSION

Differentiation of PPCs into IPCs contributes to endogenous islet regeneration[3,4], which is important in the therapy of diabetes. Our previous studies found that trypsin induced PANC-1 cells to differentiate into islet-like clusters with expression of islet-specific genes, secreting insulin after glucose stimulation, and exerting physiological activity while being transplanted, including synthesis and secretion of insulin, thus regulating the level of postprandial blood glucose in mice[5]. However, the mechanism of trypsin induction of differentiation is unknown. Our current study showed consistent results with previous published articles (Supplementary Figure 1)[5].

During the process of differentiation induced by trypsin, cell migration and aggregation were observed (Supplementary Figure 1A)[5], with formation of nearly spherical islet-like cell clusters, which is similar to the results of Wei et al[20] from a morphological perspective[21]. Many cells in ILAs differentiated into IPCs, which were further confirmed by scarlet staining of DTZ, upregulated expression of islet-related genes like insulin and PAX6 tested by PCR and obvious expression of insulin protein detected by IF (Supplementary Figure 1B and C). The migration of ductal cells from the epithelium can also be observed in islet regeneration in vivo[22]. Pancreatic ductal epithelial cells are considered PPCs, for cytokeratin 19-positive cells that can be induced differentiation into IPCs in different ways, such as small molecule (like ActA, EX-4 and glucose) treatment[4]. PANC-1 cells are derived from human pancreatic ductal cancer cells, and express progenitor cell markers such as C-kit and stem cell markers[23]. Human pancreatic tumor cell lines AsPC-1 and PANC-1 can form islet-like cell clusters and express pancreatic-development-related genes after short-term treatment with trypsin, but AsPC-1 cells only produce glucagon. In contrast, PANC-1 cells produce insulin[24]. These findings confirm PANC-1 cells as an ideal in vitro model and trypsin as a differentiation inducer for studying endogenous β-cell regeneration. In vivo, trypsin is a protein digestive enzyme produced by pancreatic tissue, in contact with the pancreatic duct, which may participate in the process of islet regeneration. Therefore, to explore the mechanism of trypsin-mediated migration, aggregation and differentiation of PANC-1 is important.

When looking for trypsin-acting receptors, we found that the extracellular part of the transmembrane protein E-cadherin could be affected by protease hydrolysis, including matrix metalloproteinases[17], metalloproteinases[18], secretase[25], and fibrinolytic enzymes[26], while the intracellular part bound a large number of proteins, including membrane receptors (epidermal growth factor receptor, insulin-like growth factor 1, and LRP5), adhesion-related proteins (p120-catenin, α-catenin and β-catenin), and tyrosine kinases, such as Ras-related C3 botulinum toxin substrate, and phosphatidylinositol 3-kinase[27]. E-cadherin is related to a variety of signaling pathways, including Hippo, transforming growth factor 1 and Wnt[15], which are related to the development and maturation of islets. As a cell adhesion molecule, E-cadherin is closely related to cell migration, which is intuitively observed during differentiation induction. Our results showed that trypsin hydrolyzed the extracellular domain of E-cadherin, changed cell adhesion, promoted cell migration, and affected the aggregation and differentiation of PANC-1 cells. After the CDH1 KO cell line was successfully obtained using the CRISPR-Cas9 system (Figure 1A), the cells in the KO group showed obvious morphological changes during differentiation, with floating and disappearance of interaction with the culture plate substrate. There was no slow cell migration, and the cells finally formed a loose flaky cluster rather than a tight islet-like cluster (Figure 1B-D). This was confirmed by immunofluorescence, showing that the cell mass was completely dispersed (Figure 1F). There were some phenomena such as upregulation of islet endocrine development regulatory factor PAX6, no significant downregulation of SOX9, increased expression of glucagon, and co-expression of double hormones (insulin and glucagon) (Figure 1E and F). Some studies have shown that endocrine cells co-expressing double hormones (insulin and glucagon, insulin and somatostatin)[28,29], or multiple hormones appeared when inducing human pluripotent stem cell differentiation in vitro, are closely related to immature fetal islet cells. Although they secrete insulin, they do not have the function to respond to glucose stimulation, and complete endocrine function can be exerted only when they further mature to single-hormone-expressing cells[30]. This suggests that the endocrine cells are not mature enough while differentiating in the absence of E-cadherin. The positive expression of SOX9 is considered a marker of PPCs, and the downregulation of SOX9 expression during pancreatic development is regarded as one of the markers of islet neogenesis[31]. In our results, there were a large number of SOX9-positive cells after differentiation in the KO group, indicating that many cells were not completely differentiated compared to control cells.

PAR2 is a membrane receptor of trypsin, which is hydrolyzed specifically by trypsin and can mediate a variety of intracellular signals by coupling G protein[19,20]. Some studies have proved that PAR2 is indispensable in islet proliferation in mouse models of pancreatitis[32] and can promote tumor migration[33], which is speculated to be a receptor of trypsin mediating aggregation and differentiation. However, we found that PAR2 was not a critical receptor in this process, with no obvious morphological and functional differences in the cell migration, aggregation into ILAs, and DTZ staining, while PAR2 had a low expression level (Supplementary Figure 2).

Cadherin-catenin complex is formed by α-catenin binding with β-catenin and E-cadherin, which plays a vital role in cell-cell adhesion[34]. αE-catenin significantly participates in intercellular junctions[35], cell proliferation[36], Hippo signaling pathway[37] and hedgehog signaling pathway[38], affecting cell migration and polarity[39], and interfering with the Wnt/β-catenin pathway[35]. We confirmed that in the developed pancreas of mice without αE-catenin, there was no normal islet morphology, but more SOX9 positive progenitor cells, fewer endocrine cells, and an increase in the proportion of α-cells[31], which indicated that αE-catenin was a key regulatory protein in endocrine cell differentiation. Our study confirmed that the cells floated quickly and formed loose cell groups with light DTZ staining, while αE-catenin was deficient (Figure 2B and C), which was similar to the situation in the absence of E-cadherin (Figure 1B and C). αE-catenin is phosphorylated during cell migration and aggregation, and its phosphorylation sites mainly involve the three sites of Ser641, Ser655 and Thr658[40]. We found that the CK1 inhibitor D4476 inhibited p-αE-catenin (Ser655/Thr658) and affected the differentiation of PANC-1 cells induced by trypsin (Figure 3). However, CK2 inhibitor DMAT inhibited p-αE-catenin (Ser641) but did not affect differentiation (Supplementary Figure 3), indicating that trypsin-induced differentiation of PANC-1 cells is related to the phosphorylation of Ser655/Thr658 of αE-catenin but not Ser641.

We speculated that trypsin promoted the phosphorylation of αE-catenin (Ser655/Thr658) by hydrolyzing E-cadherin. We observed that the phosphorylation of αE-catenin (Ser655/Thr658) induced by trypsin disappeared significantly in E-cadherin KO cells (Figure 4A and B), and phosphorylation of this site increased dramatically while restoring expression of E-cadherin (Figure 4C). However, the low expression of PAR2 did not affect phosphorylation of this site (Supplementary Figure 4). These results inferred that hydrolysis of E-cadherin by trypsin changed the inhibitory function of intracellular and extracellular binding and complex stability containing E-cadherin, or recruited some phosphorylated kinases to interact with proteins in the complex (including αE-catenin), mediating intracellular changes. Therefore, we searched for interactive proteins through the String database and Wayne analysis (Figure 5A), combined with literature investigation. CK1γ can be recruited into large complexes to play the role of phosphorylating LRP5 when Wnt ligands stimulate LRP5[41], and the protein p120-catenin expressed by CTNND1 binds to the intracellular domain of E-cadherin to regulate its stability[42]. Otherwise, the constitutive binding of p120-catenin with CK1ε can regulate the signal transduction process of the Wnt pathway[43]. N-cadherin-LRP5-AXIN has been confirmed to have direct interaction and can negatively regulate the signal transduction of Wnt/β-catenin[44]. AXIN is also bound constitutively to active CK1α, which affects their affinity for LRP5 by regulating the level of AXIN phosphorylation[45]. Based on these findings, a hypothetical action model was constructed (Figure 5B). E-cadherin bound structurally to LRP5, p120-catenin and β-catenin, while αE-catenin bound to β-catenin, and CK1ε to p120-catenin. After stimulation by trypsin, CK1ε may be phosphorylated and activated, and CK1α-AXIN and CK1γ are recruited to interact with αE-catenin directly. We verified the complex of CK1 subtypes, E-cadherin and αE-catenin in the model by immunoprecipitation (Figure 5C). Through quantitative mass spectrometric analysis of phosphorylated proteins, we found that CK1γ3 changed significantly after treatment with trypsin, while other CK1 subtypes were not affected (Figure 5D). This was confirmed by colocalization with E-cadherin induced by trypsin (Figure 5F), reflecting the recruitment of CK1γ3 after E-cadherin hydrolysis, which further proved that trypsin-mediated αE-catenin (Ser655/Thr658) phosphorylation was effected through CK1γ3. Previous studies have shown that when LRP5/6-E-cadherin is stimulated, it recruits AXIN-CK1α and CK1γ to bind to a large complex to induce phosphorylation[41], which is similar to our results. When there was 50% knockdown of CK1γ3 by siRNA, the significant change in phosphorylation of E-catenin (Ser655/Thr658) induced by trypsin disappeared (Figure 6A and B) and blocked aggregation and differentiation induced by trypsin, evidenced by the phenomena including cells floating, no migration, loose aggregates, flake cell group morphology and relatively light DTZ staining (Figure 5C), which were similar to that in absence of E-cadherin (Figure 1) and αE-catenin (Figure 2).

This study had some limitations. We were not able to directly prove that CK1γ3 binds more to the cadherin–catenin complex after stimulation by trypsin through protein quantification, and the changes in intracellular signals mediated by αE-catenin (Ser655/Thr658) phosphorylation were not evident. The specific mechanism and downstream signals warrant further exploration for the effect of αE-catenin (Ser655/Thr658) phosphorylation.

Overall, we determined that trypsin directly acts on the membrane receptor protein E-cadherin, hydrolyzed by trypsin to recruit CK1γ3 to bind to the intracellular cadherin–catenin complex. CK1γ3 affected the aggregation and differentiation of PANC-1 cells by phosphorylating αE-catenin (Ser655/Thr658). The deletion or decrease of proteins involved in this process, that is, E-cadherin, αE-catenin and CK1γ3, could block the formation of ILAs and the maturation of endocrine function of IPCs.

CONCLUSION

Our study confirmed the mechanism by which trypsin induces differentiation at the membrane protein level, providing a theoretical basis for endogenous islet β-cell regeneration, and helping to find alternative endogenous β-cell regeneration inducers.

ACKNOWLEDGEMENTS

The authors appreciate the technical assistance by Wu XN, Chen JJ, Yun Y, from the School of Pharmaceutical Sciences, Xiamen University, and Wu YY, from the School of Life Sciences, Xiamen University. We also thank Kong CC, from the School of Medicine, Xiamen University, for supplementary of data about the islets of mice.