Oncogenic KRAS and the EGFR loop in pancreatic carcinogenesis—A connection to licensing nodes

ABSTRACT

EGFR signaling has a critical role in oncogenic KRAS-driven tumorigenesis of the pancreas, whereas it is dispensable in other organs. The complex signaling network engaged by oncogenic KRAS and its modulation by EGFR signaling, remains incompletely understood. In order to study early signaling events activated by oncogenic KRAS in the pancreas, we recently developed a novel model system based on murine primary pancreatic epithelial cells enabling the time-specific expression of mutant Kras^G12D from its endogenous promoter. Here, we discuss our findings of a Kras^G12D-induced autocrine EGFR loop, how this loop is integrated by the MYC oncogene, and point to possible translational implications.

KRAS in pancreatic ductal adenocarcinoma (PDAC)

KRAS belongs to the RAS family of small GTPases, which are controlled by a molecular switch between an active GTP-bound and an inactive GDP-bound state, a process controlled by guanine nucleotide exchange factors (GEFs), which promote RAS activation, and GTPase activating proteins (GAPs), which stimulate the intrinsic GTPase activity and RAS inactivation.¹ Oncogenic mutations of KRAS occur in over 90% of pancreatic ductal adenocarcinomas (PDAC)² and according to recent sequencing studies.^3,4 activating mutations of KRAS found in PDAC patients affected codon 12 in 93% of all patients (G12D: 38%; G12V: 38%; G12R: 14%; G12C: 2%; G12S: 1%), codon 61 in 6% of the patients (Q61H: 3.5%; Q61R: 1.5%; Q61K: 1%), and codon 13 or codon 146 in 0.5% of all patients. PDAC is a malignancy with an uniquely poor prognosis and a 5-year-survival rate of less than 8%. Despite its low incidence compared with other solid cancers, it is currently the fourth leading cause of cancer-related death in the western world and is projected to be the second most common cause in the near future.

Mutant KRAS is refractory to GAP-induced GTP hydrolysis, favoring an active GTP-bound state.⁵ For a detailed description of the highly complex signaling induced by oncogenic KRAS in the pancreas, we like to refer the reader to recent reviews.^6-8 Compelling experimental evidence demonstrates that oncogenic KRAS initiates the development of PDAC via pre-neoplastic lesions, including pancreatic intraepithelial neoplasia (PanIN) lesions.^9-11 Furthermore, oncogenic KRAS is also required to maintain PDAC growth.^12,13 Therefore, analysis of signaling pathways controlled by mutated KRAS is of prominent importance to define opportunities to interfere with cancer initiation, progression, and maintenance.

Primary pancreatic ductal epithelial cells (PDECs): A relevant model to study KRAS signaling in the context of the pancreas

Current knowledge argues that KRAS engages several major pathways, including the RAF/mitogen-activated protein kinase kinase (MEK)/extracellular signal-regulated kinase (ERK)-, and the phosphoinositide-3-kinase (PI3K)/AKT-pathway. However, the signaling network regulated by oncogenic KRAS is certainly by far more complex and controlled by feed forward and backward loops, which are incompletely defined. In an attempt to establish a cell-based model to study early oncogenic KRAS signaling in the pancreas, we used genetically engineered murine primary pancreatic ductal epithelial cells (PDECs).¹⁴ Together with acinar cells, these cells contribute to the exocrine compartment of the pancreas.¹⁵ Expression of a tamoxifen-activatable Cre recombinase expressed from the ubiquitous Rosa26 or the ductal-specific Hnf1b promoter (R26^CreERT2 or Hnf1b-Cre^ER mouse lines) allows activation of the Kras^G12D oncogene via a Lox-Stop-Lox (LSL) strategy (LSL-Kras^G12D knock-in allele) in a temporally controlled fashion in PDECs (Fig. 1). Experimental evidence in mice support the notion that KRAS^G12D-induced de-differentiation/reprogramming of pancreatic acinar cells to duct-like cells, a process called acinar-to-ductal metaplasia (ADM), plays a prominent role in PDAC formation. Furthermore, murine ductal cells seem to be refractory to KRAS^G12D-mediated transformation.^16,17 To analyze the relevance of PDECs as a potential cell of PDAC origin, we conducted orthotopic transplantation experiments of PDECs with ex vivo activated KRAS^G12D expression.¹⁴ “KRAS^G12D-on” PDECs transplanted at a relative low number into the pancreas of immunodeficient mice, failed to form preneoplastic lesions. In contrast, increasing the number of cells to 7.5 × 10⁵ cells resulted in the development of ductal/PanIN-like structures, an observation consistent with data published by the Bar-Sagi group.¹⁸ Next we activated expression of KRAS^G12D ex vivo in PDECs and concomitantly deleted the tumor suppressor Cdkn2a, which is frequently inactivated in human PDAC. After orthotopic implantation of these cells into the pancreas of immunodeficient mice, we observed formation of invasive PDACs and metastasis.¹⁴ Similarly, concomitant activation of KRAS^G12D and inactivation of the tumor suppressor p53 by biallelic expression of mutated p53^R172H, which is the murine ortholog of the p53^R175H hotspot mutation found in human PDAC, resulted in invasive cancers after orthotopic transplantation (unpublished data). In a complementary approach, we could demonstrate that RNAi-mediated silencing of p16^INK4, which is encoded by the Cdkn2a locus, or p53 in “KRAS^G12D-on” PDECs results in PDAC formation in vivo.¹⁹ Consistent with these results, an important recent publication by the Leach group demonstrated PDAC formation by activating Kras^G12D and mutationally inactivating p53 (via expression of p53^R172H from both alleles) in the ductal (HNF1b positive) lineage in vivo.²⁰ In summary, murine PDECs can serve as the cell of origin of murine PDAC after inactivation of important tumor barriers. Thereby, this model mimics human PDAC formation, which depends on conquering the same roadblocks. Considering the exceptional cellular plasticity of the pancreas,¹⁶ it is not surprising that several cells of origin exist. Indeed, expressing oncogenic Kras in ductal, acinar, or even endocrine cells of the murine pancreas induces transformation under certain experimental conditions.^14,20-23 It will be important for future research, to determine whether PDACs derived from distinct cellular lineages are characterized by a specific mutational landscape or specific molecular alterations, reflecting the unique requirements of the different pancreatic compartments, and whether the cell of origin has an impact on the behavior of the tumor with respect to clinical outcomes as well as therapeutic vulnerabilities. In addition, a PanIN-independent road to cancer originating in the ductal compartment was suggested,²⁰ which needs further investigations. Beyond these considerations for future research directions, the presented data show that our new model is relevant to investigate oncogenic KRAS signaling in the context of the pancreas.

Figure 1.

Expression of KRAS^G12D signaling in PDECs. Depicted is the genetic strategy to activate KRAS^G12D-expression in primary pancreatic ductal epithelial cells (PDECs) in a temporally controlled manner. Treatment of PDECs with 4-hydroxytamoxifen activates a Cre recombinase, expressed from the Rosa26 or the Hnf1b promoter. This leads to the expression of KRAS^G12D, driven by the endogenous promoter.

KRAS and the EGFR-dependent autocrine feed-forward loop

Activation of Kras^G12D expression in PDECs ex vivo induced a proliferative response.^14,24-26 To understand the requirements of “KRAS^G12D-on” PDECs to remain in the cell cycle, we profiled mRNA expression in comparison to “KRAS^G12D-off” cells. We observed molecular signatures linked to the epidermal growth factor receptor (EGFR) pathway.¹⁴ EGFR belongs to the EGFR/ErbB family of receptor tyrosine kinases and can be activated by several ligands, including epidermal growth factor (EGF), amphiregulin (AREG), epiregulin (EREG), transforming growth factor α (TGF-α), betacellulin (BTC), heparin-binding EGF-like growth factor (HB-EGF), and epigen (EGN). Indeed, several EGFR ligands, including AREG and EREG, are induced upon expression of KRAS^G12D in PDECs and EGFR becomes subsequently autophosphorylated and thus activated.¹⁴ Furthermore, we could show that the EGFR signal is needed for KRAS^G12D-dependent cell cycle entry of PDECs.¹⁴ These data argue for an EGFR-dependent feed-forward loop engaged by oncogenic KRAS. Several lines of evidence in different model systems document the contribution of an EGFR-dependent autocrine loop toward the KRAS driven transformation in the pancreas. In organoids established from duct-like cells of Pdx1-Cre;LSL-Kras^G12D/+ mice, KRAS^G12D induces expression of EGFR ligands.²⁷ Activation of EGFR signaling by induction of EGFR and EGFR ligand expression occurs early in the KRAS^G12D-driven carcinogenesis in mice as well as in human low-grade PanINs.^28,29 Mechanistically, KRAS^G12D induces the formation of reactive oxygen species (ROS), which activate the expression of EGFR and some of its ligands via the NFκB pathway.^30,31 It is clear that ectopic expression of EGFR ligands accelerates the carcinogenesis in the pancreas.^32-34 Inversely, knockout of the Egfr gene in KRAS^G12D- and KRAS^G12V-driven mouse models of the disease completely blocked pancreatic tumor formation.^28,29 Importantly, the requirement of EGFR for oncogenic KRAS-driven tumor formation seems to be tissue-specific and depends on co-existing mutations. In contrast to the pancreas, KRAS^G12V-driven tumor formation in the lung and the intestine is EGFR-independent.²⁸ Although tissue specific engagement of oncogenic effector pathways is common in cancer biology, the molecular underpinnings for the tissue-specific requirement of the EGFR remains unclear. In addition, EGFR signaling is essential for tumor formation in a KRAS^G12V-driven Cdkn2a-deficient PDAC model, whereas it is dispensable in a p53-deficient background.^28,29 Interestingly, when we compared transcriptome profiles of “KRAS^G12D-on” PDECs to profiles of PDECs with additional biallelic expression of the p53^R172H mutant, we detected enrichment of metabolic signatures in p53-inactivated PDECs by KEGG pathway analysis (unpublished data). Such observations are compatible with the well-known anti-Warburg function of wild type p53 to impair glycolysis and to promote mitochondrial respiration.³⁵ In PDAC cells, glucose fluxes to anabolic pathways, such as the hexosamine biosynthesis pathway, which is needed for protein glycosylation, and the non-oxidative branch of the pentose phosphate pathways, which is important for ribose synthesis.¹² Numerous cancer-relevant proteins, including potent oncogenic drivers, such as the transcription factor MYC,³⁶ are regulated/activated via glycosylation.³⁷ Thus, the metabolic rewiring in KRAS^G12D-expressing cells linked to an inactivation of p53 might activate potent oncogenic drivers, thereby reducing the requirement of the autocrine EGFR-loop to activate growth and to enter the cell cycle. Therefore, disabling the anti-Warburg effect of p53 might switch the augmentation of oncogenic KRAS signaling from specific extrinsic requirements to intrinsic ones. Such a process might be linked to a specific oncogenic driver, explaining discrepant outcomes of EGFR functions in different cancer types.

However, it is important to note that the autocrine EGFR loop does not stand alone to support oncogenic KRAS-mediated transformation of the pancreas. Autocrine activation of insulin like growth factor 1 receptor (IGF1R) is required for KRAS^G12D and BRAF^V600E induced transformation of Cdnk2a/p53-deficient PDECs. This might point to the possibility that the usage of specific loops is determined by the set of inactivated tumor suppressor genes.²⁵ Furthermore, inflammation is a major risk factor for PDAC and pro-inflammatory molecules may act in an autocrine fashion to augment the signaling threshold of oncogenic KRAS.³⁸ A recent important example is the control of the KRAS-driven carcinogenesis in the pancreas by the YAP and TAZ transcription factors.^39,40 In ductal cells, it was demonstrated that KRAS^G12D engages YAP to induce expression of secreted factors, including pro-inflammatory molecules like IL6 or IL1α, needed for proliferation.⁴⁰

Integration of the EGFR-loop - connection to licensing nodes

Considering that signaling of oncogenic KRAS is highly complex, with several effector pathways acting in parallel with redundant branches modulated by autocrine signaling loops, a major goal is to define non-redundant signaling nodes. An important requirement of such a node is the authority to license profound cellular decisions, for instance to grow or not to grow, to enter the cell cycle or to stay in quiescence.

The integration of the EGFR-loop and the oncogenic KRAS signaling network in the pancreas remains unclear. EGFR signaling has been connected to canonical ERK-, AKT-, STAT3-, NFAT/AP-1, p70 S6 kinase-, or GSK3-signaling.^{28,29,32,34,41} Using the EGFR inhibitors erlotinib and gefitinib, we detected no significant influence of the EGFR-loop on KRAS^G12D-induced ERK- and AKT-signaling.¹⁴ To identify licensing nodes connected to the definite pro-proliferative decision transmitted via the EGFR, we performed unbiased transcriptome profiling. Focusing on transcription factors, we detected that the majority of signatures connected to the EGFR-loop belong to MYC- or E2F-pathways.¹⁴ Both, MYC as well as E2F transcription factors can be considered as decision-making switch factors that allow cellular growth/cell cycle entry or S-phase entrance, respectively.

Due to MYC's important role in the carcinogenesis in the pancreas and our finding that E2F is downstream of MYC in a subset PDAC cells, we further investigated the link of EGFR and MYC.^42,43 Multiple layers control and fine tune MYC transcriptional activity. In PDECs, we detected N-terminal phosphorylation of MYC in response to KRAS^G12D expression.¹⁴ It is well established that threonine 58 (T58) and serine 62 (S62) residues at the N-terminal MYC homology box I control MYC activity (Fig. 2). ERK is a kinase that can phosphorylate S62 of MYC, leading to its stabilization and activation.⁴⁴ In a hierarchical fashion, phosphorylation of S62 can be followed by a glycogen synthase kinase-3β (GSK3β)-dependent phosphorylation of T58, to limit MYC stability.⁴⁴ In KRAS^G12D-expressing PDECs, the EGFR-loop has no effect on the activating phosphorylation of ERK or the inactivating phosphorylation of GSK3β. However, we detected markedly reduced expression of both MYC protein and mRNA upon EGFR inhibition.¹⁴ Since we detected no significant changes in oncogenic KRAS-activated canonical (ERK) and non-canonical (e.g. PI3K-AKT-GSK3) signaling pathways, it is likely that the EGFR-loop controls MYC mRNA abundance. To that end, the mechanism connecting EGFR to MYC mRNA expression in PDAC remains unclear and awaits further investigations. In sum, in addition to activation of a MYC stabilization signal and the inactivation of a degradation signal by oncogenic KRAS, the EGFR loop directs a further layer of control upon acute activation of the Kras oncogene in the ductal pancreatic context (Fig. 2). Hereby, the KRAS^G12D-induced autocrine EGFR-loop allows MYC to license cellular growth and cell cycle entry.¹⁴ Beyond licensing of growth and proliferation, another molecular MYC function might be noteworthy. An ongoing current debate includes how MYC is regulating gene expression programs on a global scale^45-52 In 2012, a role for MYC to increase merely the output of an active and pre-existing cellular transcriptome was postulated, called the amplifier model.^45-47 Other models include the existence of a specific set of direct MYC target genes and it was demonstrated that MYC controls (activates/represses) a specific set of target genes,^51,52 that can secondarily feed back to global mRNA expression levels.^49,52 It will be important for further work to determine whether the relevant KRAS^G12D-induced signaling branches discussed in this article synergize to induce MYC expression levels able to amplify - directly or indirectly - the transcriptome induced by oncogenic Kras signaling (Fig. 2). Such a scenario is attractive since it allows augmentation of the KRAS^G12D-signal at a very distal position, without the need to directly manipulate the activity of the canonical (KRAS-MEK-ERK) or the non-canonical (e.g. PI3K-AKT) branch of the pathway. However, such a possibility and the contribution of such a mechanism to the transformation in the pancreas clearly demand further analysis of the MYC function on a global scale.

Figure 2.

Integration of oncogenic KRAS signaling by MYC in PDECs. GTP-bound oncogenic RAS induces several effector pathways. Depicted are the pathways with clear and definite impact on the carcinogenesis in the pancreas. Canonical signaling leads to activation of the RAF - (mitogen-activated kinase/ERK kinase) MEK - extracellular regulated kinase (ERK) protein kinase signaling pathway. Non-canonical signaling initiates the phosphoinositide 3-kinase (PI3Ks)-AKT-glycogen synthase kinase 3 (GSK3) cascade. ERK can phosphorylate MYC at serine 62 (S62), contributing to its stabilization and activity. AKT-mediated inhibitory phosphorylation of GSK3 impairs MYC threonine 58 (T58) phosphorylation and degradation. The KRAS^mut induced EGFR-loop increases MYC mRNA expression. The function of MYC to license growth and cell cycle entry is depicted.

Conclusion

Considering I) the complexity and the difficulties to analyze endogenous mouse models of PDAC, II) the need for in vitro systems to analyze signaling pathways mechanistically, III) the cellular plasticity in the pancreas, and IV) several different cells of PDAC origin, we aimed at developing a novel cellular PDEC-based system that addresses these issues allowing functional studies of early events of oncogenic KRAS signaling in the ductal compartment of the pancreas. We validated the new system, which is highly flexible, versatile and fast in vitro as well as in vivo and provide functional evidence for an EGFR signaling loop that controls MYC activity in pancreatic carcinogenesis.

The importance of cooperation between RAS and MYC for transformation has been demonstrated more than 3 decades ago.⁵³ Our findings connect these 2 pathways in an in vitro model of PDAC and illustrate that EGFR is needed for this link, thus contributing to explain the critical role of EGFR for the tumorigenesis in the pancreas.

Despite intensive efforts and recent progress, oncogenic KRAS clinically still remains a challenging target.⁵⁴ Therefore, it will be important to see whether molecular circuits discussed in this article are relevant at the stage of tumor maintenance and whether such molecular knowledge can be exploited to develop novel therapeutic concepts. Only a subpopulation of PDAC patients benefit from EGFR inhibition.⁵⁵ The pre-clinical data discussed in this commentary may argue that especially p53 wild type PDACs depend on an extrinsic augmentation of oncogenic KRAS signaling, such as the activation of the EGFR-loop. Indeed in a recent study, regular p53 expression, indicative for wild type p53, was linked to an improved progression free survival of EGFR inhibitor treated PDAC patients.⁵⁶ Although such observations point to the power of a detailed molecular understanding of the disease for clinical study design and patient stratification, the study needs definite further prospective validation. Considering MYC as an important integrator of the KRAS signal may pave the way to novel therapeutic efforts to target MYC in PDAC.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We would like to apologize for not citing any relevant reports due to the need to selectively choose examples, lack of space, or an oversight on our part.

Funding

This work was supported by: Deutsche Krebshilfe [110908 to G.S. and 111273 to M.R.], Wilhelm-Sander Stiftung [2016.004.1 to G.S.], Else Kröner-Fresenius-Stiftung (2016_A43 to M.W.), Deutsche Forschungsgemeinschaft (DFG) [SFB824/C9 to G.S. and D.S. / SCHN 959/3–1 to G.S.], and DKTK Joint Funding [D.S., and G.S.].