2.1 PRRG1 is upregulated in pancreatic ductal adenocarcinoma and associated with patients’ survival

To discover key biological processes associated with PDAC malignancy, we utilized the TCGA Pancreatic Adenocarcinoma Database (TCGA-PAAD) for analysis. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment revealed that the top 350 genes positively related to PDAC prognosis were remarkably enriched in cell plasma membrane-associated events, including cell adhesion, endocytosis and the PI3K-AKT pathway (Figure S1A). Considering the dominant role of mutant KRAS in PDAC carcinogenesis, these prognosis-related genes were further intersected with 555 KRAS-coexpressed genes in TCGA-PAAD (Spearman's Correlation >.45, 168 samples, pancancer atlas) and 273 commonly upregulated genes in all three GEO malignant pancreatic cancer datasets with paired cancer and non-cancerous tissues (GSE15471, GSE16515, and GSE28735; FC > 1). As a result, 44 shared genes were obtained (Figure 1A) and these genes mostly encode proteins that were associated with the plasma membrane and engaged in integrin signals (e.g., ITGA2, ITGB6, LAMA3), transmembrane receptor signals (MET, EPS8, PRKCI, GPRC5A, EFNB2) or metal endopeptidase activities (ADAM9, ADAM10; Figure 1B; Table S1). The above analysis revealed that membrane-associated events and signalling played crucial roles in PDAC malignancy. Therefore, we focused on those enriched in the integral component of the plasma membrane, containing EFNB2, GPRC5A, PERP, PRRG1, SLC39A10, SGMS2, IL1RAP, MET, and SLC5A3. Amongst, the molecular function of a predicted vitamin K-dependent Gla protein, the single-pass transmembrane protein PRRG1, remains undetermined.

To investigate the expression profiles of PRRG1 in pancreatic cancer, we analyzed GEO datasets containing paired cancer and non-cancerous specimens. PRRG1 was significantly upregulated in all three GEO datasets (GSE15471, GSE16515, and GSE28735). Besides, PRRG1 was evidently upregulated in TCGA PAAD compared with normal pancreas from the Genotype-Tissue Expression (GTEx) databases (Figure S1B). PDAC samples, the main sample type in the TCGA PAAD dataset, were singled out for analysis and similar results were obtained (Figure 1C). In TCGA cohorts of PDAC and PAAD patients, increased PRRG1 mRNA expression was correlated with reduced overall survival and disease-free survival (Figure 1D, Figure S1C). Next, we further examined the protein level of PRRG1 by immunohistochemical staining, using a PDAC tissue microarray from the Renji cohort containing 84 PDAC samples. Predominant immunostaining of PRRG1 was detected and observed on the plasma membrane of cancer cells, with high PRRG1 expression positively correlated with poor survival (Figure 1E). Further analysis demonstrated a strong correlation between PRRG1 expression and lymphatic metastasis. (Figure 1E; Table S2).

Moreover, we found that PRRG1 was nearly undetectable in normal pancreatic acinar cells or ductal cells while it was gradually upregulated in precursor lesions (pancreatic intraepithelial neoplasia, PanINs) and PDAC cancer cells (Figure 1F). These findings indicated that PRRG1 is upregulated in PDAC and is associated with PDAC malignant progression.

2.2 PRRG1 knockdown suppresses the malignant behaviours of PDAC cells

Western blotting was used to detect PRRG1 expression in the immortalized human pancreatic ductal epithelial cell line and a majority of PDAC cell lines (Figure 2A). To explore the function of PRRG1 in PDAC progression, small interfering RNAs (siRNAs) or lentiviral shRNA plasmids targeting PRRG1 were introduced to PDAC cells for transient or stable knockdown of PRRG1 (Figure 2B). The results demonstrated that PRRG1 knockdown significantly suppressed the in vitro proliferation and anchorage-independent growth of PDAC cells (Figure 2C,D). Following serum-free starvation treatment, PRRG1 knockdown further increased apoptosis in PDAC cells, as evidenced by Annexin V/PI staining with flow cytometry (Figure S2A). Additionally, we found that PRRG1 knockdown also inhibited PDAC cell migration through the transwell polycarbonate membrane (Figure 2E).

To evaluate the impact of PRRG1 on PDAC growth in vivo, Patu8988 cells expressing lentiviral shRNA targeting PRRG1 or scrambled control cells were subcutaneously inoculated to the female M-NSG immunodeficient mice of 6-week-old. After 4 weeks of growth, the tumours were dissected and measured. The result demonstrated that the in vivo tumour growth of PDAC cells was evidently suppressed in PRRG1-knockdown groups compared with the control group (Figure 2F). Immunohistochemical staining and analysis showed that shPRRG1 tumours exhibited less proliferating and more apoptotic cells, as defined by Ki67- and TUNEL-positive cell ratio, respectively (Figure 2G; Figure S2B).

2.3 Gla modification is indispensable for PRRG1 membranous localization and protein stability

The function and activities of vitamin K-dependent proteins rely on proper γ-carboxylation of the Gla domain.¹⁵ Two adjacent Gla residues of the core element of “EXXXE” can bind to one Ca²⁺ and enhance hydrophobicity for their interaction with phospholipids in plasma membranes. To characterize Gla modification of PRRG1, PRRG1-Flag fusion constructs were generated and overexpression cells were established using the lentivirus expressing system. PANC1 cells overexpressing PRRG1-Flag fusion protein were cultured in a serum-free medium containing the insulin-transferrin-selenium (ITS) solution for 48 h. After that, the cells were treated with varying concentrations of vitamin K for 2 h. Subsequently, cells were lysed for immunoprecipitation using an anti-Flag antibody. Immunoblotting was then performed with a specific antibody recognizing the modified Gla domain. As a result, it was ascertained that Gla modification of PRRG1 was indeed induced by vitamin K in a dose-dependent manner (Figure 3A).

To investigate the critical role of the Gla domain in PRRG1, we generated truncated variants of PRRG1 lacking either the entire Gla domain (Gla^△26–65) or key components of the Gla domain (Gla^△34–46). Additionally, mutant forms of PRRG1 with altered Gla domain (Gla^E/A) and PPXY motif (PPXY^Y/A) were constructed, all fusion proteins were tagged with Flag, and their expression was achieved using the lentivirus system (Figure 3B). Gla modification was not detectable in these forms of PRRG1 with Gla domain deletion or mutation, while it remained in the wild-type PRRG1 and PPXY^Y/A mutant form of PRRG1 (Figure 3C). Warfarin, an inhibitor of vitamin K epoxide reductase VKOR that blocks Gla modification, had been used for years in clinical practice for preventing thrombosis.²³ We found that a low dose of warfarin that is reported to be insufficient to affect coagulation (2 µM),²⁴ can effectively abolish Gla modification of PRRG1 in PANC1 cells (Figure 3C). The wild-type PRRG1 exhibited predominant localization on the plasma membrane, whereas the Gla-mutant or truncated PRRG1 was barely detectable on the PDAC cell surface (Figure 3D). In contrast to the wild-type form, the Gla variants showed a propensity for rapid degradation, with their protein levels being strikingly increased after MG132 treatment (Figure 3E).

Furthermore, a plasmid encoding PRRG1 fused with a green fluorescent protein (GFP) was constructed, and an overexpression cell line was established in PANC1 cells. Fluorescence imaging of GFP demonstrated that treatment with warfarin markedly decreased the plasma membrane localization of PRRG1 and resulted in a reduction in its protein levels (Figure 3F). Additionally, the cycloheximide (CHX) assay illustrated that warfarin-induced disruption of Gla modification diminished the stability of PRRG1 (Figure 3G). These findings suggest that Gla modification is essential for the membrane integration and protein stability of PRRG1.

2.4 PRRG1 overexpression promotes PDAC malignancy, which is reversed by low-dose warfarin

To further validate that PRRG1 contributes to PDAC progression, we conducted both in vitro and in vivo experiments using cells that overexpress PRRG1. We also utilized a low dose of warfarin that disrupts Gla modification in the functional assays. PRRG1 overexpression led to increased capabilities of proliferation, anti-apoptosis and migration, while warfarin evidently reversed all these effects in PANC1 cells (Figure 4A–C; Figure S3A,B). Further, we established a subcutaneous tumour model by inoculating the control and PRRG1-overexpression PANC1 cells to the NOD-SCID mice. Treatment with warfarin or vehicle control in drinking water started 2 days after inoculation and proceeded before harvesting. The result showed that PRRG1 overexpression significantly enhanced the in vivo tumour growth of PANC1 cells, which was clearly reversed by low-dose warfarin (Figure 4D). Tumour growth was further confirmed by histological assessments with H&E staining and PCNA immunohistochemical staining (Figure 4E).

We also established a lentivirus-mediated Prrg1 overexpression cell line using a mouse PDAC cell line expressing luciferase, named KPC1199^luc. Consistently, Prrg1 overexpression also significantly promoted the proliferation of KPC1199^luc in vitro (Figure S3A,B). Furthermore, the cells were implanted into the pancreas of C57BL6 mice to generate a syngeneic orthotopic transplantation tumour model. Prrg1 overexpression significantly promoted the orthotopic tumour growth of KPC1199^luc, as indicated by the increased bioluminescent signals and the pancreas burden (Figure 4F; Figure S3C). The increased tumour burden induced by Prrg1 overexpression was significantly attenuated after low-dose warfarin treatment (Figure 4F; Figure S3C). Orthotopic tumour growth was further confirmed by histological assessments with H&E staining and Ki67 immunohistochemical staining (Figure 4G; Figure S3D). In conclusion, the results suggest that PRRG1 is a critical regulator of PDAC malignant progression which can be impeded by treatment of low-dose warfarin.

2.5 PRRG1 interacts with NEDD4 through the PPXY motif and promotes NEDD4 self-ubiquitination

To identify the PRRG1 binding partners in PDAC, we immunoprecipitated cell lysates of PRRG1-Flag expressing PANC1 cells using an anti-Flag antibody, then conducted a mass spectrometry analysis to determine the identities of the prey proteins. Some recognized interactors were identified including GGCX that mediates Gla modification. Several E3 ubiquitin ligases containing the WW domain that recognize PPXY motifs were also captured. As a plasma membrane-anchoring E3 ligase that contains WW domains predictably recognizing PPXY motifs, NEDD4 was reported to have tumour suppressive roles through targeting Ras proteins or degradation.⁹ It also targets other membranous substrates for ubiquitination and degradation, including the epidermal growth factor receptors and vascular endothelial growth factor receptor 2.^25-28 Therefore, the interaction between PRRG1 and NEDD4 in PANC1 cells was confirmed through co-immunoprecipitation and Western blotting assays (Figure 5A). To investigate whether the interaction between PRRG1 and NEDD4 depends on the intracellular PPXY motif, the PRRG1 mutant form PPXY^Y/A was coimmunoprecipitated with the anti-Flag antibody. The findings revealed that the PPXY^Y/A mutant PRRG1 failed to establish complexes with NEDD4 (Figure 5A). Unexpectedly, the ubiquitination and the protein level of PRRG1 were not apparently altered by NEDD4 knockdown (Figure 5B; Figure S4), implying that binding to NEDD4 is not obligatory for PRRG1 degradation.

Both PRRG1 and NEDD4 were primarily localized on the plasma membrane, with their co-localization confirmed through immunofluorescence analysis (Figure 5C; Figure S5). Intriguingly, the protein level of NEDD4 was clearly reduced by PRRG1 overexpression in both PANC1 and Mia PaCa-2 cells, while it was obviously elevated in PRRG1-knockdown Patu8988 cells (Figure 5D), indicating a clear negative regulation of NEDD4 by PRRG1. NEDD4/NEDD4L family protein was reported to undergo self-ubiquitination and subsequent degradation upon binding to its targets.²⁹ To investigate whether PRRG1 regulates NEDD4 through the ubiquitin-proteasome system, the 26S proteasome inhibitor, MG132, was utilized concurrently. The diminished expression of NEDD4 resulting from PRRG1 overexpression was restored upon inhibition of ubiquitin proteolytic activity with MG132 (Figure 5E). Likewise, treatment with MG132 abolished the elevation of NEDD4 induced by PRRG1 knockdown (Figure 5E). Moreover, the ubiquitination of NEDD4 was clearly reduced in PRRG1-knockdown cells compared with control cells, while it was increased by PRRG1 overexpression (Figure 5F). Consistently, the CHX experiment showed that NEDD4 stability was increased in PRRG1-knockdown cells (Figure 5G). These results indicate that PRRG1 interacts with NEDD4 through the PPXY motif and promotes NEDD4 self-ubiquitination and instability in PDAC cells.

To investigate the necessity of Gla modification for PRRG1 in regulating NEDD4 self-ubiquitination, Gla-mutant or deleted constructs were employed alongside the wild-type PRRG1. The findings revealed that, unlike wild-type PRRG1, PRRG1 variants lacking Gla modification did not enhance NEDD4 ubiquitination (Figure 5H) or reduce the protein level of NEDD4 (Figure 5I). Similar results were observed in low-dose warfarin-treated cells (Figure 5H,I). In conjunction with the results depicted in Figure 3, it is evident that Gla modification plays a crucial role in the membrane integration of PRRG1 to regulate the self-ubiquitination of membrane-bound NEDD4, which can be reversed by warfarin.

The NEDD4 family, comprising NEDD4 and NEDD4-2, contains 3 to 4 WW domains that bind to the PY motifs (L/PPXY) of substrate proteins. It also includes a C-terminal HECT domain, which is responsible for its ubiquitin ligase activity. Previous evidence has shown that the WW domains weakly bind to their own LPXY motif located within the HECT domain of NEDD4-2, causing a “locked” stable state where the HECT domain remains inactive. This inhibitory WW-HECT interaction is disrupted by higher-affinity binding to proteins containing PPXY motifs, activating the HECT domain and initiating self-ubiquitination.²⁹ Aligning protein sequences revealed high similarity in domains and structure between NEDD4 and NEDD4-2, with the HECT domain of NEDD4 containing the active-site cysteine residue and conserved PY (LPXY) motif identical to its homolog NEDD4-2.^{30, 31} Since PRRG1 possesses two PPXY domains (Figure 3B), it is highly probable to bind to the WW domains within NEDD4 through these motifs, triggering the “unlocking” of the ubiquitin ligase HECT domain and activating the self-ubiquitination process, leading to the instability of NEDD4. To investigate whether the regulation of NEDD4 expression and ubiquitination relies on the PPXY motifs of PRRG1, PDAC cells overexpressing the PPXY^Y/A mutant form of PRRG1 were employed in the experiments. The findings showed that, in contrast to the wild-type PRRG1 overexpression, the overexpression of PRRG1-PPXY^Y/A did not enhance the ubiquitination of NEDD4 (Figure 5H), nor did it decrease the protein level of NEDD4 (Figure 5I). The above results demonstrate that PRRG1 promotes NEDD4 self-ubiquitination and instability in a manner dependent on the C-terminal PPXY motif.

2.6 PRRG1 knockdown reduces the signalling of oncoprotein KRAS and tyrosine kinase receptor EGFR

PRRG1 co-expressed genes in the TCGA PAAD database were significantly enriched in endocytosis, Sphingolipid signalling pathway, Ras signalling, PI3K-AKT and ErbB signalling (Figure 6A). GSEA analysis also exhibited a close correlation of PRRG1 with Ras signalling (Figure S6). Transcriptome RNA-sequencing of control and PRRG1-knockdown Patu8988 cells revealed a range of differentially expressed genes enriched in the MAPK signalling, Glycosphingolipid biosynthesis and PI3K-AKT signalling (Figure 6B). In PDAC, mutant KRAS is sustained with a GTP-binding active form, which acts as a driving factor to activate the tumour-promoting signalling pathways like PI3K-AKT, MAPK/ERK, and PLC-NF-kB pathways.⁶ The oncogenic KRAS signalling also acts in a positive feedforward loop with glycosphingolipids (GSLs).³² To address whether PRRG1 has relevance with KRAS signalling, we conducted the co-immunofluorescence staining experiment and the result showed that PRRG1 was dominantly localized at the plasma membrane with considerable colocalization with KRAS both in sparse cells and confluent cells (Figure 6C). In confluent cells, concentrated colocalization of PRRG1 and KRAS was found in “cup”-like microcompartments of the plasma membrane. Oncogenic mutant KRAS is anchored to the inner cell membrane and forms protein–lipid nanoclusters, which activate downstream effectors and induce macropinocytosis under nutrient-deficient circumstances. We discovered that PRRG1 knockdown significantly reduced the expression level of KRAS (Figure 6D) as well as its aggregation in the plasma membrane of Patu8988 cells (Figure 6E). Immunofluorescence staining with an antibody specifically targeting the GTP binding form of Ras showed that active Ras-GTP was also evidently attenuated by PRRG1 knockdown (Figure 6E). Coherently, the downstream phospho-MEK1/2 (Ser217/221) and phospho-ERK1/2 (Thr202/Tyr204) were clearly inhibited in PRRG1-knockdown cells (Figure 6F). Mutant active KRAS promotes macropinocytosis, a highly conserved type of endocytosis, facilitating the non-selective uptake of extracellular fluid to support PDAC development.⁷ In line with the regulation of PRRG1 towards KRAS, the capability of macropinocytosis was evidently suppressed by PRRG1 knockdown in serum-free conditions, as determined by quantification of the internalized TMR-dextran (Figure 6G).

It has been demonstrated that the tyrosine kinase receptor EGFR/ErbB-1 formed a feed-forward loop with oncogenic KRAS and was required for KRAS-induced pancreatic tumorigenesis.^{33, 34} Intriguingly, we found that PRRG1 knockdown also suppressed the protein level and the active form of EGFR in PDAC cells, as indicated by Western blotting and immunofluorescence detection (Figure 6F; Figure S7). Collectively, these results indicate that PRRG1 knockdown decreases the protein level and signalling of both KRAS and EGFR in PDAC cells.

2.7 PRRG1 knockdown reduces the protein stabilities of KRAS and EGFR via NEDD4

NEDD4 binds to membranes and phospholipids in a Ca²⁺-dependent fashion through its C2 domain. This interaction is critical for regulating the ubiquitin-mediated endocytosis of several membrane proteins, including EGFR.^{26, 35, 36} NEDD4 has also been evidenced to mediate KRAS degradation through polyubiquitination on lysine 5. Considering the inhibitory regulation of PRRG1 toward NEDD4 described above, PRRG1 might contribute to stabilizing KRAS and EGFR through downregulating NEDD4. Indeed, the stabilities of KRAS and EGFR were clearly decreased in PRRG1-knockdown cells as measured by the CHX assay, while they were increased by NEDD4 knockdown (Figure 7A). Blockade of proteasome-mediated degradation with MG132 or lysosomal degradation with bafilomycin respectively eliminated the reduction of KRAS or EGFR by PRRG1 knockdown (Figure 7B). In contrast to the minimal modulation of PRRG1 (Figure 5B), NEDD4 knockdown clearly elevated the expression of KRAS and EGFR in PDAC cells (Figure 7C). Furthermore, the ubiquitination of KRAS and EGFR were increased in PRRG1-knockdown cells, which were reversed by additional knockdown of NEDD4 (Figure 7D). Consistently, the reduction of the protein level of KRAS and EGFR in PRRG1-knockdown cells was also rescued by NEDD4 knockdown (Figure 7E). In addition, enhanced interaction of NEDD4 with KRAS was detected in PRRG1-knockdown cells, as determined by the semi-quantification of the NEDD4/KRAS complex using the co-IP/WB analysis (Figure 7F). These results indicate that PRRG1 exerts a protective role towards KRAS and EGFR through negatively regulating NEDD4 in PDAC cells.

2.8 Low-dose warfarin reduced the protein levels of KRAS and EGFR that were elevated in PRRG1-overexpressing cells

We further investigated whether warfarin inhibits the upregulation of KRAS and EGFR in PRRG1-overexpression cells. The CHX assay demonstrated that treatment of low-dose warfarin, the antagonist of vitamin K and Gla modification, apparently decreased the protein stabilities of KRAS and EGFR in PANC1 cells (Figure 8A). As shown in Figure 8B, PRRG1 overexpression significantly increased the protein levels of KRAS and EGFR, accompanied by elevated levels of active phospho-EGFR and the downstream MEK-ERK signalling. All these signals were remarkably reversed by treatment of low-dose warfarin, as detected by Western blotting (Figure 8B). Additionally, immunofluorescence staining showed that the active Ras-GTP and phospho-EGFR were reinforced in PRRG1-overexpressing cells compared with vector-control cells, which were clearly reversed by low-dose warfarin (Figure 8C,D). Warfarin also reversed the increased macropinocytosis in PRRG1-overexpression cells, as detected by quantitative of the internalized TMR-dextran in serum-free conditions (Figure 8E).

Together, these results indicate that PRRG1 protects the membranous signalling of KRAS and EGFR through counteracting E3 ligase NEDD4 in PDAC cells, which can be blocked by inhibitors targeting vitamin K-dependent Gla modification.

4.1 Data mining using gene expression omnibus and TCGA

The expression dataset from the TCGA genomic project (https://portal.gdc.cancer.gov) is retrieved for investigating the mRNA expression of KRAS-related genes in PDAC. To discover the upregulated genes in PDAC with poor prognosis, the expression profiles of GSE15471, GSE16515, and GSE28735 from GEO (https://www.ncbi.nlm.nih.gov/geo) were downloaded, as well as a gene list significantly linked to poor prognosis in PDAC from OncoLnc (http://www.oncolnc.org).

To better compare the expression level of PRRG1 in PDACs and normal counterparts, we retrieved the PRRG1 expression profiles of GSE15471 (with 39 paired PDAC tumours and matching normal pancreatic tissue samples), GSE16515 (with 36 tumours and 16 adjacent tissues, all derived from patients diagnosed with stage II or III pancreatic adenocarcinoma), GSE28735 (with 45 paired PDAC tumours and matching normal pancreatic tissue samples) from GEO, as well as from TCGA databases and GTEx (https://gtexportal.org). For analyzing the prognosis value of PRRG1, the clinical data were compiled from TCGA PDAC patients, correlating survival outcomes with PRRG1 mRNA levels.

Patients were grouped into two categories based on their expression levels. In short, the lowest 71 cases were classified as having “low” expression, while the highest 72 cases were designated as “high” expression. Also, we analyzed disease-free survival using the TCGA PDAC dataset, splitting patients into nonoverlapping groups of 36 cases each, representing the top and bottom quartiles of expression levels, respectively.

4.2 Patients and tissue samples

A total of 150 PDAC patients who underwent radical surgery at Ren Ji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China, between January 2017 and December 2018 were included in the study. Only patients with complete follow-up records spanning more than 2 years were enrolled, while those who died due to perioperative complications were excluded. Tumours were staged according to the TNM classification from the American Joint Committee on Cancer (AJCC), 8th edition.⁴⁸ The follow-up data encompassed the patients’ demographic information, pathology reports, surgical details, and survival durations. Resected specimens from surgery were characterized by tumour size, tumour location, tumour invasion, lymph node status, and vascular invasion. As a result, 84 patients with complete data were included in the Renji cohort. All samples were obtained from patients who had provided informed consent prior to the operation in Renji Hospital. And protocols were authorized by the ethical review committee of the World Health Organization Collaborating Center for Research in Human Production (approved by the Shanghai Municipal Government).

4.3 Cell lines and cell culture

We purchased the human pancreatic cancer cell lines PANC1, Patu8988. and MIA PaCa-2 from the Cell Bank of the Chinese Academy of Sciences. Additionally, the primary mouse tumour cell line KPC1199 was obtained from KPC (LSL-KrasG12D/+; LSL-Trp53R172H/+; Pdx-1-Cre, C57BL/6 background) pancreatic ductal adenocarcinoma mouse model which was generously provided by Prof. Jing Xue (Renji Hospital). We cultured all cells in Dulbecco's modified Eagle's medium (DMEM) added with 10% (v/v) fetal bovine serum. The cells were cultured with 1% antibiotics and incubated at 37°C in a humidified chamber containing 5% CO2.

4.4 Small interference RNA-mediated knockdown of gene expression

siRNAs were transfected into PDAC cells to specifically knock down gene expression. The jetPRIME transfection reagent (Polyplus) was used following the manufacturer's guidelines. After 72 h, the cells were harvested, and Western blotting was performed to evaluate the efficiency of the knockdown. The effective siRNA sequences can be found in Table S3.

4.5 Lentivirus-mediated overexpression and knockdown of PRRG1

Full-length cDNAs encoding human PRRG1 or mutant forms of PRRG1 including Gla domain truncated (Gla^△26–65 and Gla^△34–46), Gla-domain mutant (Gla^E/A) and PPXY mutant (PPXY^Y/A) were synthesized and cloned to the pSLenti-CMV-MCS-3xFlag-PGK-puro-WPRE plasmid. For stable knockdown of PRRG1, the PLKO.1 vector (Invitrogen) was used to construct the plasmid expressing short hairpin RNAs targeting PRRG1 (shPRRG1). The shRNA-encoding sequences targeting PRRG1 are listed in Table S4. We generated the lentiviral particles by co-transfection of PRRG1 or mock vectors in 293T cells with the three-plasmid packaging system (pPACKH1-REV, pPACKH1-GAG, and pVSV-G), with the help of Lipofectamine 2000 (Invitrogen, 11668-019). After transfection, we collected viruses and then measured them at both 48 and 72 h. Cells were infected with recombinant lentivirus-transducing units accompanied by Polybrene for 5 µg/mL, (Sigma-Aldrich, H9268). After 48 h of infection, we cultured cells for 14 days with 5 µg/mL puromycin (Gibco, A1113802) to eradicate uninfected cells, then further in a medium containing 1 µg/mL puromycin.

4.6 Animal studies

For constructing the subcutaneous xenograft model, we subcutaneously injected 6 × 10⁶ Patu8988 cells in the flank of 6-week-old female M-NSG immunodeficient mice. The cells were expressing stably shNC or sh-PRRG1-2 or sh-PRRG1-3. Tumours were measured with a calliper every week and the tumour volumes (V) were calculated with the formula: V = .52 × L (tumour length) × W² (tumour width). The mice were sacrificed to collect and dissect tumours for further measurements after growing for 4 weeks.

The mice were housed in isolated, ventilated cages at the Shanghai Jiaotong University Laboratory Animal Center. They were kept under a 12 h light/dark cycle, with temperatures between 22°C and 26°C. Water and sterile pellet food were provided ad libitum. This study followed the guidelines outlined in the Guide for the Care and Use of Laboratory Animals and applicable Chinese regulations. The protocol was approved by the Institutional Animal Care and Use Committee (IACUC) at Shanghai Jiaotong University under Animal Protocol number A2023083.

4.7 Histological analysis

Phosphate-buffered formalin was used to fix tissues and then embedded in paraffin for sectioning. The slides underwent H&E staining or immunohistochemical staining using antibodies with cell proliferating marker Ki-67 (GB111141, rabbit, Servicebio), PCNA (GB11010, rabbit, Servicebio), or TUNEL labelling of apoptotic cells. For immunohistochemical staining of PRRG1 in the PDAC tissue microarray, the antibody of PRRG1 (PA5-67490, rabbit, Invitrogen) was used. As previously described, PRRG1 expression was scored according to the intensity and proportion of positive cells. The scoring system was as follows: 0 for 0–5% positivity, 1 for 6–35% positivity, 2 for 36–70% positivity, and 3 for more than 70% positivity.^{49, 50}

4.8 Cell proliferation, apoptosis, and transwell-based migration assays

For the cell proliferation assay, all experiments were conducted as previously outlined.⁵¹ In Brief, cells were plated in a 96-well plate for later treatment with cell counting kit-8 (Share-bio). One hour after CCK8 treatment, the absorbance of cells at 450 nm was recorded with a BioTek Synergy HTX Multimode Reader (Agilent). Measurements were taken 24, 48, 72, and 96 h after cell plating.

For the apoptosis assay, cells were cultured in a six-well plate and deprived of serum to promote apoptosis. After 48 h, cells were stained with propidium iodide and annexin V-FITC (Share-bio). We determined apoptosis in cells with flow cytometry using a BD FACSCalibur instrument (BD Biosciences).

For the transwell-based migration assay, we cultured 2 × 10⁴ cells on the upper side of the chamber in the serum-free DMEM medium. As a normal chemoattractant, the lower chamber was added with DMEM medium containing 10% FBS. After a 24 h incubation, cells that did not migrate were carefully removed from the upper surface of the membrane using a cotton swab. The cells that migrated to the lower side of the filter were then fixed with 4% paraformaldehyde for crystal violet staining. Finally, migration was evaluated under a microscope, and the number of migrated cells was quantified from three randomly selected fields in each of the three replicate experiments.

4.9 The soft agar colony formation assay

First, we applied a 1.2% agarose (Share-bio) layer to the bottom of 6-well plates in DMEM medium containing 10% FBS. Next, 10 000 cells were combined with .6% agarose and placed as the top layer. Two-hundred microlitre medium was placed on the top agarose layer and replaced every 3 days. We cultured the plate at 37°C with 5% CO2 for 4 weeks, allowing the colonies to form. Afterwards, colonies were stained with 200 µL of MTT (0.5 mg/mL) for 30 min in an incubator. Colonies were quantified with three replicates.

4.10 Detection and quantification of macropinosome

Macropinosome visualization experiments were carried out as previously noted.⁵² Briefly, we incubated cells on an Ibidi 8-well culture slide in starvation with serum-free medium for 18 h. To visualize macropinosomes, a 70 kDa high-molecular tetramethylrhodamine–dextran (TMR-dextran, D1818, Invitrogen) was added to the serum-free medium at 1 mg/mL for 40 min in the incubator. After that, we rinsed the cells with cold PBS five times (5 min for each rinse) and then fixed them with 4% paraformaldehyde. Cellular internalization of TMR-dextran was captured with a confocal microscope (Leica). We randomly counted at least 200 cells for quantitative analysis. We analyzed the images using the ImageJ software (ImageJ). The macropinocytic index is calculated as the total fluorescent area of TMR-dextran divided by the total cell area, multiplied by 100.

4.11 Transcriptome RNA-sequence

Total RNA from control and PRRG1 knockdown Patu8988 cells (three biological replicates) was extracted. The mRNA was harvested by RNeasy Mini Kit (Qiagen). Libraries were constructed with a TIANSeq Fast RNA Library Prep Kit (TIANGEN). Illumina NovaSeq 6000 platform was used to sequence the library preparations. We use the DESeq2 R package to analyze differentially expressed genes in the transcriptome. Genes with fold change greater than 1.5 and p-values below .05 were considered significantly different. We used the DAVID functional annotation tool to analyze KEGG pathway enrichment results. KEGG terms with adjusted p-values below .05 were regarded as significantly enriched. The accession number for the RNA-seq data is GSE273196.

4.12 Reagents and inhibitors

To determine protein stability, we treated cells with cycloheximide (C7698, Sigma-Aldrich) at a concentration of 10 µg/mL for 0, 1, 2, and 3 h; or for 0, 1.5, 3, and 4.5 h prior to harvesting. Cell lysates were subjected to Western blotting. We assessed the relative protein levels by quantification of the band density of the specific protein in comparison with the internal control. To investigate whether the alteration of the protein level was attributed to the ubiquitin-proteasome system, we added MG132 (HY-13259, MedChemExpress) or vehicle control to the medium at a final concentration of 10 µM for 2 h. To detect whether the lysosomal protein degradation system was involved, Bafilomycin A1 (HY-100558, MedChemExpress) was applied. To block vitamin K-dependent Gla modification, warfarin (S4545, Selleck) was utilized. Ethylisopropylamiloride (EIPA, HY-101840, MedChemExpress) was used as an inhibitor of micropinocytosis.

4.13 Co-immunoprecipitation and mass spectrometry

To identify proteins that bind to PRRG1, we performed immunoprecipitation (IP) with the anti-DDDDK tag (FLAG tag) antibody (66008-4-Ig, mouse, Proteintech) with Protein A/G Magnetic Beads (B23202, Selleck). Samples from PANC1 and Patu8988 cells were processed for SDS-PAGE electrophoresis, followed by staining with Coomassie blue as well as mass spectrometry (MS) analysis using an Easy nLC1200 system (Thermo Scientific). For verifying the specific interacting proteins, we incubated cell lysates with magnetic beads conjugated with DDDDag antibody or IgG for 2 h at 25°C. After that, we washed the beads for 5 min with PBS/Tween-20 three times before releasing the immunoprecipitation mixtures with 1*loading buffer. Finally, we performed Western blotting on the protein lysates to detect specific antibodies.

4.14 Protein ubiquitination

To assess the ubiquitination levels in specific proteins, we pretreated cells with MG132 for 2 h and harvested using lysis buffer (P70100, NCM Biotech) containing protease and phosphatase inhibitor cocktail (P002, NCM Biotech). We performed immunoprecipitation (IP) on the protein lysates using either an anti-ubiquitin antibody (10201-2-AP, rabbit, Proteintech) or an anti-DDDDK (FLAG tag) antibody (66008-4-Ig, mouse, Proteintech), followed by capture with protein A/G MAGNETIC BEAds (B23202, Selleck). After incubation, we analyzed the immunoprecipitated samples using SDS-PAGE electrophoresis followed by western blotting. The samples immunoprecipitated with the anti-ubiquitin antibody were probed using anti-KRAS, anti-EGFR, or anti-PRRG1 antibodies. For the samples immunoprecipitated with the anti-Flag antibody, an anti-ubiquitin antibody (10201-2-AP, rabbit, Proteintech) was used for probing.

4.15 Immunofluorescence staining

For immunofluorescence staining, cells cultured on Ibidi 8-well culture slides were fixed for 10 min with 4% paraformaldehyde. Next, cells were permeabilized with .1% Saponin (Abmole Bioscience, M9618) and then blocked with 1% BSA at 25°C. Primary antibodies against PRRG1 (PA5-67490, rabbit, Invitrogen), Flag tag (66008-4-Ig, mouse, Proteintech), NEDD4 (21698-1-AP, rabbit, Proteintech), Ras-GTP (26909, mouse, NewEast Bioscience), KRAS (12063-1-AP, rabbit, Proteintech), pan Ras (60309-1-Ig, mouse, Proteintech), and phospho-EGFR Tyr1068 (3777s, rabbit, CST) were used for incubation at 4°C overnight. Subsequently, species-matched secondary antibodies conjugated to Alexa Fluor-488 or -594 were applied. The nucleus was stained with Hoechst and then sealed by an anti-fade mounting medium (Beyotime, P0126). We captured immunofluorescence images using a confocal microscope (Leica).

4.16 Western blotting

Cells were lysed with lysis buffer (P70100, NCM Biotech) containing Phosphatase and Protease Inhibitor Cocktail (P002, NCM Biotech). We performed protein extraction and separation using SDS-PAGE, followed by transfer onto a .22 µm pore-size nitrocellulose membrane. The membrane was incubated with 5% non-fat milk for 1 h to block nonspecific binding sites. Then, we performed the incubation of primary antibodies targeting PRRG1 (14103-1-AP, rabbit, Proteintech), Gla (3570, mouse, BioMedica Diagnostics), NEDD4 (21698-1-AP, rabbit, Proteintech), KRAS (12063-1-AP, rabbit, Proteintech), beta Actin (GB15001, mouse, Servicebio), MEK1/2 (11049-1-AP, rabbit, Proteintech), phospho-MEK1/2 (Ser217/221) (9154T, rabbit, CST), ERK1/2 (11257-1-AP, rabbit, Proteintech), phospho-ERK1/2 (Thr202/Tyr204) (80031-1-RR, rabbit, Proteintech), EGFR (sc-373746, mouse, Santa Cruz), and phospho-EGFR (Tyr1068) (3777s, rabbit, CST) at 4°C overnight, and then exposed to species-specific secondary antibodies (7074P2, rabbit, and 7076P2, mouse, CST).

4.17 Statistical analysis

All statistical analyses were conducted with a minimum of three independent experiments, and the data are presented as mean ± SD. Related graphs were drawn in GraphPad Prism 10 software (GraphPad Software, Inc). To determine statistical significance across multiple experimental groups, we conducted one-way analysis with Dunnett's post hoc test or two-way analysis with Bonferroni's post hoc test. Clinical data were analyzed using SPSS 16.0 for Mac (IBM). Cumulative survival times were estimated through the Kaplan–Meier method, and comparisons were made using the log-rank test. The relationship between PRRG1 expression and clinical features in PDAC patients was assessed via Fisher's exact test or the χ2 test. Statistical significance was set at a p-value of less than .05.