Pancreatic acinar cells are a cell type of origin for pancreatic cancer that become progressively less sensitive to tumorigenesis induced by oncogenic Kras mutations after birth. This sensitivity is increased when Kras mutations are combined with pancreatitis. Molecular mechanisms underlying these observations are still largely unknown. To identify these mechanisms, we generated the first CRISPR-edited mouse models that enable detection of wild-type and mutant KRAS proteins in vivo. Analysis of these mouse models revealed that more than 75% of adult acinar cells are devoid of detectable KRAS protein. In the 25% of acinar cells expressing KRAS protein, transcriptomic analysis highlighted a slight upregulation of the RAS and MAPK pathways. However, at the protein level, only marginal pancreatic expression of essential KRAS effectors, including C-RAF, was observed. The expression of KRAS and its effectors gradually decreased after birth. The low sensitivity of adult acinar cells to Kras mutations resulted from low expression of KRAS and its effectors and the subsequent lack of activation of RAS/MAPK pathways. Pancreatitis triggered expression of KRAS and its effectors as well as subsequent activation of downstream signaling; this induction required the activity of EGFR. Finally, expression of C-RAF in adult pancreas was required for pancreatic tumorigenesis. In conclusion, our study reveals that control of the expression of KRAS and its effectors regulates the sensitivity of acinar cells to transformation by oncogenic Kras mutations.
This study generates new mouse models to study regulation of KRAS during pancreatic tumorigenesis and highlights a novel mechanism through which pancreatitis sensitizes acinar cells to Kras mutations.
The development of genetically engineered mouse models (GEMM) greatly helped to show that Kras mutation constitutes a prominent initiating event in pancreatic cancer (1–4). In these GEMM, tumors phenocopied the histologic features of human malignancies. More recently, lineage tracing experiments demonstrated that acinar cells are a cell type of origin of pancreatic cancer (5). Interestingly, adult acinar cells become progressively less sensitive to Kras mutations compared with early postnatal acinar cells (6). This sensitivity is increased in the presence of pancreatitis and when mutations of Kras and p53 are combined (6, 7).
During pancreatitis, Kras-mutated acinar cells undergo acinar-to-ductal metaplasia (ADM), eventually leading to the formation of pancreatic intraepithelial neoplasia (PanIN) and pancreatic ductal adenocarcinoma (PDAC; ref. 6). However, the mechanisms through which pancreatitis sensitizes acinar cells to mutant KRAS remain largely unexplored. The expression of oncogenic KRAS and of KRAS signaling effectors at the protein level could be determining, because a high KRASG12D dosage is critical for early human pancreatic tumorigenesis (8). Therefore, we hypothesized that pancreatitis could sensitize acinar cells to Kras mutations by modulating an expression program encompassing KRAS and possibly essential signaling effectors. Here, using novel mouse models, transcriptomic and biochemical analyses, we provide evidence supporting that pancreatitis induces expression of KRAS protein and its key effectors, to enable activation of KRAS signaling and initiate tumorigenesis.
Materials and Methods
For additional and detailed Materials and Methods, please see Supplementary Materials and Methods. This section contains information regarding the conditions of tissue staining/labeling (Supplementary Table S1), mouse lines and injection protocols (Supplementary Table S2), conditions of Western blot experiments (Supplementary Table S3), and primers used in this study (Supplementary Table S4).
Generation of CRISPR-edited KrasCit/Cit and KrasCit-G12D/+ mouse models
All procedures described previously were approved by the animal welfare committee of the University of Louvain Medical School (Brussels, Belgium; ethic number: 2017/UCL/MD/020). KrasCit/Cit and KrasCit-G12D/+ mice were generated to detect expression and localization of KRAS and KRASG12D proteins on tissue sections using anti-GFP antibody, as citrine is an improved variant of YFP. Citrine-KRAS expression plasmids have been used previously to study KRAS biology in cell culture studies (9–11). A targeting construct containing the Citrine cDNA flanked by sequences of the Kras locus located on both sides of the ATG start codon (see Fig. 1A) was obtained by PCR using HiFi KAPA (KK2602, Roche). A linker sequence (TCC GGA CTC AGA TCT CGA GCT) was added between the Citrine and Kras sequences. The Citrine-Kras (Cit-K) construct, in which the Citrine and Kras coding sequences are in the same reading frame, was cloned in pcDNA3.1 (+). Selected clones were validated by Sanger sequencing. To generate the Citrine-KrasG12D (Cit-KG12D)–targeting construct, we performed directed mutagenesis on the previously cloned Cit-K sequence according to the manufacturer's instructions (200523, Agilent). Sanger sequencing was performed on selected clones to detect the GGT-to-GAT mutation. The following primers were used for mutagenesis: forward, 5′-GTG GTG GTT GGA GCT GAT GGC GTA GGC AAG AGC-3′; and reverse, 5′-GCT CTT GCC TAC GCC ATC AGC TCC AAC CAC CAC-3′.
CRISPR/Cas9 technology was used to insert the Cit-K construct in the Kras locus by homologous recombination. Cit-K–targeting plasmid (10–20 ng/μL), recombinant Cas9 protein (100 ng/μL; 1081059, Integrated DNA technologies), and annealed specific crRNA/tracrRNA (2.4 pmol/μL, 5′-CCTGCTGAAAATGACTGAGTATA-3′; Integrated DNA technologies) were injected in IDTE buffer (11–01–02–02, Integrated DNA technologies) into the male pronucleus of B6D2F2 mouse zygotes using FemtoJet 4i Injector (Eppendorf). Zygotes were transferred into CD1 pseudopregnant females. Correct insertions were detected in the descendants by PCR and sequencing of PCR products. Breeding between heterozygous KrasCit/+ mice generated homozygous KrasCit/Cit mice. To generate the KrasCit-G12D/+ mouse line, the Cit-KG12D construct was inserted within the LSL-KrasG12D locus in frame with the KrasG12D gene, using CRISPR/Cas9 technology, as described above. Cit-KG12D–targeting construct, recombinant Cas9 protein, and annealed specific crRNA/tracrRNA (same sequence than above) were injected into the male pronucleus of LSL-KrasG12D/+ mouse zygotes. Zygotes were transferred into CD1 pseudopregnant females. Correct insertions were detected in the descendants by PCR and sequencing of PCR products.
Ex vivo culture of dissociated human pancreas
Pancreata from human heart-beating cadaveric donors were processed by the Beta Cell Bank (Brussels, Belgium) affiliated to the Euro-Transplant Foundation (Leiden, the Netherlands). The authors of this manuscript obtained a full written informed consent from donors to authorize the use of pancreata for research purposes. The study was conducted in accordance with the ICH-GCP guideline for good clinical practice. The use of donor pancreata was approved by the medical ethical committee of the Vrije Universiteit Brussel Hospital (Brussels, Belgium; ethic approval number: B.U.N. 143201732606). Culture of the exocrine cell fraction was performed as described previously (12). To induce ADM, cells were cultured for 3 days in Advanced RPMI1640 medium (12633–012, Thermo Fisher Scientific) supplemented with 5% FBS (F7524, Sigma Aldrich), 1% penicillin–streptomycin (15070–063, Life Tech) and 0.1 mg/mL soybean trypsin inhibitor (17075–029, Life Tech), at 37°C and 5% CO2. On the day of isolation (day 0), 100 μL of cells, corresponding to the normal exocrine fraction, were snap-frozen in liquid nitrogen and stored at −80°C, before protein extraction.
IHC and immunofluorescence
Dissected tissues were fixed in 4% paraformaldehyde for 4 hours at 4°C, with gentle rotation, before paraffin embedding. Six-micron–thick sections were first deparaffinized. Then antigen retrieval was performed using citrate (pH 6.0) or Tris-EDTA (pH 9.0) buffers using microwave (MW), Lab Vision PT Module (Thermo Fisher Scientific), or pressure cooker (CC; Supplementary Table S1). Sections were washed once with PBS and then permeabilized with PBS/0.3% Triton-100X for 5 minutes, at room temperature. Sections were blocked with solution 1 (3% low-fat milk, 5% BSA, 0.3% Triton-100X in PBS) for 45 minutes, at room temperature. Primary antibodies (Supplementary Table S1) diluted in solution 1 were added and incubated overnight, at 4°C. On the next day, slides were washed with PBS/0.1% Triton-100X and incubated with secondary antibodies diluted in solution 2 (10% BSA, 0.3% Triton-100X) at 37°C, for 1 hour. For IHC, additional incubation with streptavidin–β-peroxidase from horseradish (POD) conjugates at 37°C for 1 hour was performed before DAB staining. Immunofluorescence pictures were taken using Axiovert 200 (Zeiss) or Cell Observer Spinning Disk Confocal Microscope (Zeiss). IHC pictures were selected after scanning performed on Pannoramic P250 Flash III (3DHistech). Colocalization analyses and Pearson's coefficient calculations were performed using Zen Software (Zeiss).
Data were presented as means ± SEM. Normality and equal variance were checked before conduction of statistical analysis. Comparisons between two groups were performed using an unpaired Student t test. Comparison between three or more groups was performed using one-way ANOVA. Significant effects or interactions were further analyzed by Fisher LSD method. For all statistical analysis, the level of significance was set at P < 0.05. Analyses were performed using SigmaStat 3.1 and GraphPad Prism 8 softwares. For RNA-sequencing (RNA-seq) data, DESeq2 R package was used to determine significant differential gene expression using a model based on the negative binomial distribution. P values were corrected using the Benjamini and Hochberg approach for controlling the FDR. Genes with a corrected P < 0.05 were assigned as differentially expressed (*, P < 0.05; **, P < 0.01; ***, P < 0.001).
The sequencing data discussed in this study have been deposited in NCBI's Gene Expression Omnibus (GEO) database and are accessible through GEO series accession numbers: GSE163039, GSE163254, and GSE163263.
The expression of KRAS is restricted to a small percentage of pancreatic acinar cells
The lack of antibodies enabling to detect KRAS protein on tissue sections severely hampers the characterization of the mechanisms of differential tissue sensitivity to oncogenic Kras mutations (13). Therefore, we generated new mouse models in which a citrine gene (an improved version of YFP) is fused in frame in the wild-type (WT) Kras gene (Cit-K allele) or mutated KrasG12D gene in LSL-KrasG12D mice (Cit-KG12D allele; Fig. 1A and B; ref. 4). In these models, the resulting citrine-KRAS (Cit-K) and citrine-KRASG12D (Cit-KG12D) fusion proteins are expressed under the control of the regulatory elements of the endogenous Kras locus and can be detected using anti-GFP antibody. KrasCit-G12D/+ mice were crossed with ElaCER mice, in which the CreER recombinase is specifically expressed in the acinar cells, to obtain ElaCER KrasCit-G12D/+ (ElaCER KCit-G12D/+) mice. In the latter, acinar expression of Cit-KG12D protein is only achieved after tamoxifen injections. We validated these models by showing that Cit-K and Cit-KG12D are expressed, bind GTP, and localize to the cell membrane, like KRAS and KRASG12D (Supplementary Figs. S1A–S1E and S2A–S2E), and that Cit-KG12D was oncogenic (Supplementary Fig. S3A–S3E).
Analysis of normal pancreas in KrasCit/Cit mice revealed a heterogeneous expression of Cit-K protein (Supplementary Fig. S4A); a majority of acinar cells displayed little or no Cit-K, whereas endocrine and ductal cells clearly expressed Cit-K protein (Supplementary Fig. S4B and S4C). In positive acinar cells, Cit-K labeling was located at the plasma membrane (Fig. 1C–E). Cit-KG12D was also detected at the plasma membrane (Supplementary Fig. S5A–S5D), indicating that the KrasG12D mutation did not significantly affect the dynamics of KRAS expression and subcellular localization in acinar cells. Intriguingly, we observed higher cytosolic levels of the Cit-K/Cit-KG12D fusion proteins, compared with the KRAS/KRASG12D proteins. However, this did not impact the embryonic development, postnatal growth, and fertility of citrine-tagged mice (Supplementary Figs. S1E and S2E). FACS analysis of KrasCit/Cit pancreata (14), which generate a stronger citrine signal due to the presence of two citrine alleles, revealed that only 24.6% ± 2.4% of acinar cells express Cit-K protein (Cit-KPos; Fig. 1F), indicating that the majority of acinar cells do not express KRAS protein. We confirmed, by qRT-PCR experiments, the identity of Cit-KPos and Cit-KLow/Neg acinar populations (Supplementary Fig. S6A–S6C). In ElaCER KrasCit-G12D/+ mice, we observed a lower fluorescence intensity and a higher noise-to-signal ratio in FACS quantification, which is explained by the presence of a single citrine-tagged allele, and the basal low expression of Cit-KG12D; this probably led to the inclusion of auto-fluorescent cells in our FACS quantification that overestimated the percentage of Cit-KG12DPos acinar cells (FACS data available in Supplementary Fig. S7A and S7B). To overcome this issue, we then quantified their number on immunolabeled sections of pancreas. This indicated that Cit-KG12D was expressed in 11.6% ± 2.4% of acinar cells (Supplementary Fig. S5D). Our RNA-seq data indicated that the LSL cassette of the Kras gene is deleted at a frequency of 60%, and allowed us to conclude that Cit-K (24.6%) and Cit-KG12D (19.3% after correction for LSL recombination) are expressed in a comparable number of acinar cells. Altogether, these results highlight the presence of two acinar populations expressing differential levels of KRAS protein.
The low sensitivity of pancreatic acinar cells to Kras mutation is associated with low expression of KRAS and its effectors
The lack of KRAS protein in the majority of acinar cells probably explains why these cells display low sensitivity to Kras mutations. Because KRAS protein is detectable in a minor fraction of acinar cells, we first analyzed the transcriptional landscape of Cit-KPos and Cit-KLow/Neg acinar cells (Fig. 2A; Supplementary Fig. S8A). We found that 70% of the expressed genes were detected in both populations (Fig. 2B). Cit-KLow/Neg acinar cells mainly express genes related to pancreatic secretion, ribosomes, and protein processing in endoplasmic reticulum (Supplementary Fig. S8B). Interestingly, in Cit-KPos cells, Kras mRNA expression and pathways associated with KRAS-driven tumorigenesis including the MAPK/ERK, JAK/STAT, and PI3K/AKT pathways, were upregulated in a statistically significant way (Fig. 2C; Supplementary Fig. S6C). However, this increase at the mRNA level did not translate into efficient pathway activation, as evidenced by the absence of P-ERK and P-STAT3 labeling from Cit-KPos cells (Fig. 2D). Importantly, in silico analysis of two independent studies on human pancreas confirmed the presence of two subpopulations of acinar cells sharing a molecular signature similar to that of Cit-KLow/Neg and Cit-KPos acinar cells (15, 16). These subpopulations express differential levels of KRAS: acinar cells expressing lower KRAS levels showed high expression of digestive enzymes, while acinar cells expressing higher KRAS levels showed an enrichment in genes of the MAPK pathway (Supplementary Fig. S9A–S9C).
Next, we investigated if KrasG12D mutation affects the activation of KRAS-related pathways. To do so, we compared the transcriptomes of acinar cells from ElaCER;YFP and ElaCER KrasG12D/+;YFP (ElaCER KG12D/+;YFP) mice, in which acinar cells express YFP under the control of the ROSA26 locus. We found no upregulation of the ERK/MAPK, JAK/STAT, and PI3K/AKT pathways in the presence of KrasG12D mutation (Supplementary Fig. S10A and S10B). Accordingly, P-ERK was undetectable by immunolabeling in acinar cells of KrasCit/Cit, ElaCER KG12D/+, or ElaCER KCit-G12D/+ pancreas (Supplementary Fig. S11A and S11B).
In WT and ElaCER KG12D/+ pancreas, no change in Kras/KrasG12D mRNA levels was detected (Supplementary Fig. S12A). From the abovementioned results, we hypothesized that effectors downstream of KRAS in the RAF/MEK/ERK and PI3K/AKT signaling pathways are not or only marginally expressed, at the protein level, in the pancreas. In WT and ElaCER KG12D/+ pancreas, MEK1/2 and/or ERK1/2 were detectable but the expression of C-RAF was very low; the phosphorylation of C-RAF, MEK1/2, and ERK1/2 was low too (Fig. 3A; Supplementary Fig. S12B and S12C). Similarly, components of a molecular scaffold conveying the MAPK signal (17), namely KSR-1, GEF-H1, PP2Aa, and PP2Ac were very weakly expressed (Fig. 3B; Supplementary Fig. S13A–S13C). Concerning the PI3K/AKT pathway, p85 (PI3K regulatory subunit), p110β (PI3K catalytic subunit), and AKT were detected, whereas we found a weak expression of p110α (PI3K catalytic subunit), which interacts directly with KRAS and is essential for the formation of pancreatic neoplastic lesions (Fig. 3A; ref. 18). This low expression of critical KRAS effectors is probably responsible for the limited number of low-grade PanIN, which are observed when Kras mutations are induced in adult ElaCER KG12D/+ mice (Supplementary Fig. S14A and S14B).
Interestingly, the low expression of KRAS and its effectors seems to be pancreas specific, because lungs, an organ that spontaneously develops tumors in the presence of Kras mutations (19), expressed high basal levels of KRAS and its effectors in mouse and humans (Supplementary Fig. S15A–S15C). For example, C-RAF expression, which is essential for lung tumor development, was 50-fold higher in adult lungs compared with adult pancreas (Supplementary Fig. S15C; ref. 20). Cit-K labeling in lungs confirmed the wide expression pattern of KRAS in this organ (Supplementary Fig. S16A and S16B). Together, our data suggest that the low protein expression of KRAS and critical effectors of the RAF/MEK/ERK and PI3K/AKT pathways leads to a weak responsiveness to Kras mutations in the pancreas.
Pancreatitis stimulates the expression of KRAS and its downstream effectors
To investigate how pancreatitis cooperates with oncogenic Kras mutations to promote tumorigenesis, we hypothesized that pancreatitis induces the expression of KRAS and its downstream effectors. Therefore, acute pancreatitis was generated by cerulein injection. Pancreatitis resulted in a 2.5-fold increase in Kras (Kras and KrasG12D) mRNA (Supplementary Fig. S12A), and 3.4-fold increase in KRAS (KRAS and KRASG12D) protein in both WT and KrasG12D backgrounds (Fig. 3A; Supplementary Fig. S12C). This was associated with a parallel 2.7- and 2.9-fold increase in the percentage of acinar cells expressing Cit-K or Cit-KG12D, respectively, as estimated by FACS and/or tissue quantification (Fig. 3C and D; Supplementary Fig. S17A–S17C), indicating a similar increase in KRAS and KRASG12D expression in the presence of pancreatitis. These data were supported by immunodetection of membrane-located Cit-K and Cit-KG12D on pancreas sections from KrasCit/Cit and ElaCER KCit-G12D/+ mice treated with cerulein in an acute setting (Fig. 3E and F). Using the PdxCre KCit-G12D/+ model, we found that in the presence of pancreatitis, Cit-KG12D expression was significantly increased in PanIN compared with metaplastic acini, suggesting the need for a higher expression of mutated KRAS to achieve neoplastic transformation (Supplementary Fig. S18A and S18B). These experiments allow us to conclude that the increase in KRASG12D expression seen by Western blot at early stages (1 week of cerulein treatment), when only metaplasia is present, is mainly due to an increase in the number of KRASG12D-expressing cells rather than an increase in the expression level of KRASG12D by the cells.
Cerulein treatment of ElaCER KrasG12D/+ mice was associated with a drastic 46- and 4.5-fold increase in the expression of C-RAF and p110α (Fig. 3A; Supplementary Fig. S12C), although c-Raf mRNA levels remained unchanged (Supplementary Fig. S19A). Expression of C-RAF was significantly higher in KrasG12D mice, compared with WT mice; however, this observation did not apply to KRAS (Supplementary Fig. S12C). We also observed a 15-, 14-, and 18-fold increase of KSR-1, GEF-H1, and PP2Aa in the cerulein-treated ElaCER KG12D/+ pancreata and in cultures of dissociated pancreata from ElaCER KG12D/+ mice (Fig. 3B; Supplementary Fig. S13A–S13C); this ex vivo culture spontaneously recapitulates the in vivo effects of cerulein treatment on ADM (21).
Transcriptome analysis indicated that acute pancreatitis induced a strong upregulation of KRAS-related pathways, including the MAPK/ERK and PI3K/AKT pathways; the corrected P values (Supplementary Fig. S10A and S10B) were statistically much higher than those obtained when comparing the acinar Cit-KLow/Neg and Cit-KPos cells (Fig. 2C). Accordingly, we found a 12-, 20-, and 6-fold increase in P-C-RAF, P-MEK1/2, and P-ERK1/2 levels, respectively, and a 3-fold increase in P-AKT level (Fig. 3A; Supplementary Figs. S12C and S19B). ERK activation was also confirmed by immunostaining in ElaCER KG12D/+ mice treated with cerulein in a chronic setting (Supplementary Fig. S19C).
Our results indicate that the two central KRAS effectors C-RAF and p110α are increased following pancreatitis. It has been shown that p110α is required for PanIN initiation when it is genetically inactivated during embryogenesis or when it is pharmacologically inhibited in adult mice (18). However, C-RAF is dispensable for pancreatic tumorigenesis when it is genetically inactivated in pancreatic progenitor cells (22, 23). To study the role of C-RAF when deleted during adulthood, we inactivated its expression in Kras-mutated acinar cells using ElaTKG12V/+c-Raflox/lox mice. In the presence of acute pancreatitis, the ablation of c-Raf reduced acinar metaplasia by 50% (Fig. 4A and B) and inhibited ERK activation (Fig. 4C). When chronic pancreatitis was induced for 3 months, followed by 8 months of recovery before sacrifice, c-Raf ablation prevented the development of advanced PanIN and PDAC (Fig. 4D and E). Although the survival of ElaTKG12V/+c-Raflox/lox p53lox/lox mice was not affected when KrasG12V mutation and c-Raf ablation took place in pancreatic progenitor cells (23), survival was extended when KrasG12V mutation and c-Raf ablation were induced in adult mice (Fig. 4F). Because PanIN development and KRAS signaling are similarly activated by KrasG12D and KrasG12V mutants (24), our results may be extended to the G12D model and highlight an important role of C-RAF in PDAC initiation from adult acinar cells. We conclude that pancreatitis sensitizes acinar cells to Kras mutations by increasing the expression of KRAS and its main effectors, such as C-RAF, at the protein level.
EGFR signaling controls the expression of KRAS effectors
EGFR is essential for the tumorigenic activity of mutated Kras in pancreas, as shown by the lack of KRAS-induced PanIN formation from acinar cells in which Egfr has been inactivated (25, 26). This prompted us to hypothesize that adult mouse pancreas expresses low levels of EGFR and upstream KRAS effectors and that EGFR is important for the expression of KRAS effectors.
We found that EGFR, P-EGFR, GRB2, SOS1, and SOS2 were almost absent from adult pancreas, but were induced by 35-, 23-, 16-, 43-, and 68-fold, respectively, in the presence of acute pancreatitis (Fig. 5A; Supplementary Fig. S20A and S20B). To address the function of EGFR signaling, we treated cultures of dissociated pancreata from ElaCER KG12D/+ mice with the EGFR inhibitor gefitinib. As expected, EGFR, SOS1, SOS2, KRASG12D, C-RAF, and p110α expression was induced after 3 days of culture, in the absence of inhibitor (Fig. 5B); Sox9 was chosen as a marker of metaplasia (27). Gefitinib inhibited the phosphorylation of EGFR and C-RAF, and significantly reduced the induction of EGFR, SOS1, SOS2, C-RAF, and p110α (Fig. 5B; Supplementary Fig. S21A). We extended our approach to human pancreas cultures, and observed an induction of KRAS and its effectors at day 3, which was significantly inhibited in the presence of gefitinib or erlotinib, two EGFR inhibitors (Fig. 5C; Supplementary Fig. S21B).
Interestingly, we also observed that mouse pancreata at postnatal day 1 show a 2-fold higher expression of EGFR compared with adult pancreata; this was accompanied with higher expression of KRAS and its effectors, with, in particular, a 20-fold higher expression of C-RAF (Fig. 5D; Supplementary Fig. S22A). These results likely explain why early postnatal acinar cells from newborns are more sensitive to oncogenic Kras mutations compared with adult acinar cells (6). Altogether, these results support that in pancreas, EGFR is required for the oncogenic activity of mutated KRAS by controlling the protein expression of crucial upstream and downstream KRAS effectors.
Data from previous studies suggest that the sensitivity of pancreatic acinar cells to an oncogenic Kras mutation decreases with their differentiation stage (28–31). The sensitivity of acinar cells to Kras mutation is most important during embryonic development, and it gradually decreases after birth. At adult stage, this sensitivity is strongly increased in the presence of an inflammatory stimulus, such as that induced by pancreatitis (6, 28–31). Once the early neoplastic lesions have formed, it seems that other mechanisms may slow down tumor progression, such as senescence, which acts as a barrier that prevents progression of low-grade PanIN into more aggressive lesions (32). Here also, pancreatitis contributes to PDAC progression by eliminating this senescence barrier (32). The molecular mechanisms through which pancreatitis sensitizes acinar cells to Kras mutations are still largely unknown. We believe that the elucidation of these mechanisms depends, at least partly, on our ability to detect KRAS on tissue sections. In the absence of immunolabeling-grade KRAS antibody, we developed in this work the first mouse models that enable the detection of WT and mutant KRAS in tissues and used these models to characterize the abovementioned mechanisms.
Using our models, we found that WT and mutant KRAS proteins are only expressed in a minority of adult acinar cells. This indicates that the adult pancreas contains two populations of phenotypically indistinguishable acinar cells, one expressing KRAS and the other not expressing it (or at an undetectable level). Taking advantage of the properties of our models, we separated these populations to compare their transcriptome and found that they have distinct transcriptional signatures. Acinar cells that do not express KRAS seem to be more dedicated to the production of digestive enzymes and their secretion, whereas KRAS-expressing cells show an enrichment in genes related to KRAS signaling. An important confirmation of the existence of these two acinar cell populations was provided by two studies of RNA sequencing performed on human pancreas (15, 16). A first population seems more dedicated to digestive functions; it expresses low levels of KRAS and high levels of acinar genes, and is similar to the acinar cells expressing low KRAS protein levels described in our study. The second is enriched in genes related to KRAS and MAPK/ERK pathway but expresses low levels of acinar genes and, thus, is similar to our KRAS-expressing cells (15, 16). Future studies will be needed to determine whether KRAS-expressing acinar cells are more prone to initiate tumorigenesis.
The absence of KRAS protein in a majority of acinar cells explains why the pancreas shows a low sensitivity to Kras mutations. Capitalizing on this result, we further explored why the minority of KRAS-expressing cells do not effectively activate KRAS-dependent pathways, like the MAPK/ERK pathway, when Kras is mutated. We found that the protein expression of crucial KRAS effectors, mainly C-RAF, was missing or only expressed at marginal levels. Our results take a fresh look at data from the literature, mainly obtained using cultured cell lines, which inappropriately claim that these effectors are widely expressed. However, emerging evidences confirm our data by showing that the effectors SOS1, SOS2, and C-RAF have the lowest abundancy among all effectors of the RAS/MAPK pathway (33).
We also discovered that pancreatitis leads to a significant increase in the protein expression of KRAS and its effectors. This is in line with previous studies, which described the existence of a biological threshold for the expression and activity of KRAS that needs to be reached to initiate tumorigenesis (34–36). Our study reveals that this observation also extends to the expression levels of KRAS effectors. Accordingly, we confirmed that pancreatitis induces pancreatic neoplasia through upregulating the expression of C-RAF, as the ablation of the latter hampered tumorigenesis from adult acinar cells. The fact that protein expression of KRAS and its effectors is correlated with the sensitivity of acinar cells to Kras mutations, is supported by our observation that pancreata from newborns, which develop more lesions in response to Kras mutations than adult pancreata (6), express higher levels of KRAS and its effectors, as compared with adult pancreata. Another supporting independent observation comes from Collins and colleagues showing that increased protein expression of oncogenic KRAS in an adult pancreas, from an ectopic promoter, is associated with a significant and rapid development of PanIN and PDAC, without requiring experimental pancreatitis (37). Contrariwise, the models in which oncogenic KRAS is expressed from its own promoter require experimental pancreatitis to observe a significant and rapid development of PanIN and PDAC (6).
Our results also provide an explanation for the role played by EGFR in pancreatic tumorigenesis (25, 26). We found that the weak protein expression of upstream and downstream effectors, in pancreas, renders ineffective the initiation of Kras-related tumorigenesis. In this context, pancreatitis increases the expression and activation of EGFR, which in turn promotes the expression of KRAS effectors, including C-RAF. Indeed, the expression and activation of EGFR can be maintained through a positive feed-forward loop activated by inflammation. This loop seems to include EGFR, KRAS effectors, and the transcription factors Sox9 and NFATc1 (35, 38, 39). Our data connect EGFR activation with the expression levels of KRAS effectors in mice but also in humans, indicating a similar regulation in both species. The ability of EGFR to increase the levels of KRAS effectors explains its essential role in mutant KRAS-driven pancreatic tumorigenesis (25, 26).
According to our results, it appears that the mechanisms of tolerance to oncogenic stress are tissue specific. In the skin, the epidermis resists oncogenic Pik3ca mutations by inducing the differentiation of progenitor cells, leading to the suppression of abnormal cell growth (40); it also tolerates oncogenic Hras mutations via a selective translational mechanism that inhibits cell renewal (41). In the pancreas, the mechanism found here is passive as it depends on the low abundancy of multiples key KRAS effectors, resulting in an ineffective KRAS signaling. We do not exclude that this passive mechanism coexists with a more active mechanism, which would aim, for example, to curb the activity of KRAS signaling.
In conclusion, our study reveals that the expression of KRAS protein and its effectors is heterogeneous in adult pancreas, and that it is induced by pancreatitis. Low concentration of several constituents of the KRAS pathway is associated with an ineffective signaling, and consequently, the occurrence of a Kras mutation does not translate into increased signaling activity. Comparable regulations could affect other genes frequently mutated in cancer, such as other members of the RAS family, namely HRAS and NRAS, and other effectors, such as B-RAF. Together, our observations underline the need to consider oncogene expression levels when investigating tumorigenic mechanisms, and pave the way for future anticancer strategies targeting the control of oncogene expression.
No disclosures were reported.
M. Assi: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. Y. Achouri: Resources, data curation, investigation, methodology. A. Loriot: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. N. Dauguet: Data curation, software, formal analysis, validation, visualization, methodology. H. Dahou: Data curation, investigation, methodology. J. Baldan: Resources, data curation. M. Libert: Resources, data curation. J.S. Fain: Software, methodology. C. Guerra: Resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. L. Bouwens: Resources, data curation, methodology. M. Barbacid: Resources, data curation, software, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. F.P. Lemaigre: Formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. P. Jacquemin: Conceptualization, resources, formal analysis, supervision, funding acquisition, validation, investigation, visualization, writing–original draft, project administration, writing–review and editing.
The authors thank Mourad El Kaddouri and Jean-Nicolas Lodewyckx for expert technical assistance, and Doris Stoffers and Tyler Jacks for providing mouse strains. This work was supported by grants from the Foundation for Cancer Research (#2016-089, #2018-078, and #2018-076 to F.P. Lemaigre and P. Jacquemin) and Télévie (#7.4595.15 and #7.4502.16 to F.P. Lemaigre and P. Jacquemin). M. Assi is Research Assistant at the National Fund for Scientific Research (FRS-FNRS, Belgium). P. Jacquemin is Senior Research Associate at FRS-FNRS, Belgium.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.