Twists and turns in Kras-driven tumor initiation

Accumulation of genetic mutations progressively drives carcinogenesis, which is affected by the cell lineage and organ-specific microenvironment. Reflecting the diversity in such cooperation, frequently mutated genes in cancer vary significantly according to the organ wherein carcinogenesis is initiated. Notably, a subset of cancers is almost exclusively associated with a particular genetic alteration, such as KRAS activating mutations in pancreatic ductal adenocarcinoma (PDA). Genetically engineered mice (GEM) have been considered the gold standard as a disease model of cancer. Pancreas-specific reconstitution of mutant Kras in mice invariably leads to the development of pre-neoplastic lesions, pancreatic intraepithelial neoplasia (PanIN), which frequently progresses to PDA in a long latency, recapitulating the multi-step pancreatic carcinogenesis in humans.

Alternatively, we have previously established a cell-based streamlined assay to probe pro-tumorigenic genetic interactions [1]. We adopted the Matrigel-based organoid culture technique because it enables long-term and physiological propagation of tissue stem cells. By integrating lentiviral gene transduction of murine organoids and their subsequent allograft implantation in immunodeficient mice, we assessed the tumorigenicity of each oragnoid [2–4]. In most cases, the outcomes were essentially similar to those observed in GEM models, even in the absence of an organ-specific microenvironment or cellular immunity, highlighting the validity and robustness of this ex vivo model [5,6]. Although at least two genetic alterations are usually required for developing neoplasms, mutant Kras alone in the pancreatic organoid exceptionally gave rise to PanIN-like lesions in 50% of tested cases [3], confirming the critical roles of mutant Kras in pancreatic tumorigenesis using this model. Together with the notion that Kras is a bona fide oncogene, we had hypothesized that Kras, regardless of wild-type (WT) or mutant, would simply confer pancreatic organoids with advantages in both in vitro and in vivo setting. However, thorough examination of organoids at each step of this model revealed that tumor initiation in the pancreas is more complicated than initially thought.

In Kras^LSL-G12D/+ mice, transcription of Kras^G12D is induced by Cre-mediated excision of the LoxP-Stop-LoxP (LSL) cassette. Successful in vitro recombination was verified by the emergence of the G12D amplicon in genomic PCR, while the retention of the LSL amplicon indicated the presence of residual cells without recombination. Given that our infection efficiency in organoids was approximately 90% [1], organoids before inoculation normally retained the LSL, although its precise quantification by regular PCR was technically challenging. In contrast, LSL was absent in tumor-derived organoids, strongly suggesting the definite requirement of Kras^G12D for tumor development. To evaluate the proportion of Kras^G12D-expressing cells before inoculation, we conducted puromycin selection as these cells are LSL-negative and puro-sensitive. Surprisingly, puro-sensitive cells decreased from 90% to 50% at 4 weeks after Cre induction (Figure 1), raising the possibility that Kras^G12D could be inferior to Kras^WT in the propagation of pancreatic organoids under standard culture conditions [3]. As Kras-activated cells are known to become less dependent on the EGF signaling pathway, we depleted EGF along with other stem cell niche factors from the medium for two weeks. As expected, Kras^WT cells did not survive this harsh culture condition, leading to 100% enrichment of Kras^G12D-expressing cells. In addition, the protein level of Kras^G12D was significantly increased after puromycin selection than anticipated by enrichment of Kras^G12D-expressing cells, suggesting positive in vitro selection for cells with higher magnitude of Kras activation. PanIN-like lesions were induced with complete penetrance with a pure cell population with a hyperactive Kras pathway [3]. Based on these findings, we inferred that the initially observed lower tumorigenesis rate in this model than that in the GEM model might be attributable not only to the lack of an organ-specific microenvironment, but also to the underrepresentation of Kras hyper-activated cells before inoculation, which is facilitated by negative selection under standard culture conditions.

Intriguingly, the situation was observed to be different under a p53-null setting. After Cre introduction in pancreatic organoids from Kras^LSL-G12D/+; Trp53^fl/fl mice, the recombination rate quickly reached 100% for both genes, and PDA development was invariably observed, underscoring the synergy between mutant Kras and p53 loss. We noted occasional PDA development from pancreatic organoids caused by Kras^G12D alone, suggesting the acquisition of pro-tumorigenic alterations in the subcutis. We identified the deletion of the Kras^WT allele in the PDA-derived organoid, which is also recurrently detected in human cancers carrying KRAS mutations. This finding was in line with the notion that Kras^WT competes with Kras^G12D, thereby acting as a relative tumor suppressor. Deletion of the Kras^WT allele was frequently observed in Kras-driven tumorigenesis in the oviduct [7] and endometrial organoids [8]. These findings suggest that cells with hyperactive Kras pathway are selected in the subcutaneous tissue, in which synergistic effects of Kras^G12D with the loss of Trp53 and Kras^WT provides a prominent growth advantage. It depended on organs whether Kras^G12D-expressing cells were selected or declined, pointing toward the notion that carcinogenesis-related research must be conducted on the basis of organs and genes.

Although PDA development is highly dependent on activating mutations in KRAS, we showed that this is not a straightforward process. We uncovered anti-tumorigenic effects of both Kras^WT and Kras^G12D, while demonstrating that Kras^G12D-expressing cells are advantageous under only some stringent conditions associated with tumor initiation, or in the context of the loss of p53 or Kras^WT, which mirrors tumor progression. All these findings were obtained using the ex vivo carcinogenesis model, thereby warranting the use of this approach and comparison with corresponding GEM, to dissect the molecular events during tumor initiation in any organ.