Pancreatic Cancer Modeling using Retrograde Viral Vector Delivery and IN-Vivo CRISPR/Cas9-mediated Somatic Genome Editing

« Signaling through the T Cell Receptor (TCR) Complex and the Co-stimulatory Receptor CD28

Blast crisis in myeloid leukemia and the activation of a microRNA-editing enzyme called ADAR1 »

Pancreatic Cancer Modeling using Retrograde Viral Vector Delivery and IN-Vivo CRISPR/Cas9-mediated Somatic Genome Editing

June 10, 2016 by 2012pharmaceutical

Pancreatic Cancer Modeling using Retrograde Viral Vector Delivery and IN-Vivo CRISPR/Cas9-mediated Somatic Genome Editing

Curators: Larry H. Benstein, MD, FCAP and Aviva Lev-Ari, PhD, RN

Genes Dev. 2015 Jul 15; 29(14): 1576–1585.

doi: 10.1101/gad.264861.115

PMCID: PMC4526740

Pancreatic cancer modeling using retrograde viral vector delivery and in vivo CRISPR/Cas9-mediated somatic genome editing

Shin-Heng Chiou,^1,¹³Ian P. Winters,^1,¹³Jing Wang,^2,³Santiago Naranjo,¹Crissy Dudgeon,^4,^5,⁶Fiona B. Tamburini,¹Jennifer J. Brady,¹Dian Yang,⁷Barbara M. Grüner,¹Chen-Hua Chuang,¹Deborah R. Caswell,⁷Hong Zeng,^8,⁹Pauline Chu,¹⁰Grace E. Kim,¹¹Darren R. Carpizo,^4,^5,⁶Seung K. Kim,^2,³ and Monte M. Winslow^1,^7,^8,¹²

¹Department of Genetics, Stanford University School of Medicine, Stanford, California 94305, USA;

²Department of Developmental Biology, Stanford University School of Medicine, Stanford, California 94305, USA;

³Howard Hughes Medical Institute, Stanford University School of Medicine, Stanford, California 94305, USA;

⁴Rutgers Cancer Institute of New Jersey, New Brunswick, New Jersey 08903, USA;

⁵Department of Surgery, Rutgers Robert Wood Johnson University Medical School, New Brunswick, New Jersey 08903, USA;

⁶Department of Pharmacology, Rutgers Robert Wood Johnson University Medical School, New Brunswick, New Jersey 08903, USA;

⁷Cancer Biology Program, Stanford University School of Medicine, Stanford, California 94305, USA;

⁸Stanford Cancer Institute, Stanford University School of Medicine, Stanford, California 94305, USA;

⁹Transgenic, Knockout, and Tumor Model Center, Stanford University School of Medicine, Stanford, California 94305, USA;

¹⁰Department of Comparative Medicine, Stanford University School of Medicine, Stanford, California 94305, USA;

¹¹Department of Pathology, University of California at San Francisco, San Francisco, California 94143, USA;

¹²Department of Pathology, Stanford University School of Medicine, Stanford, California 94305, USA

¹³These authors contributed equally to this work.

Corresponding author: Email: ude.drofnats@wolsniwm

Little is known about how these alterations contribute to the development of metastatic and therapy-refractory PDAC. Given the inability to test gene function in human cancers in vivo, genetically engineered mouse models represent tractable and biologically relevant systems with which to interrogate the molecular determinants of each stage of pancreatic cancer development.

Identification of the mutations that drive the development of human pancreatic cancer combined with the ability to alter gene function in mice has enabled the development of genetically engineered murine PDAC models. Transgenic expression of Cre-recombinase in pancreatic cells of loxP-Stop-loxP (LSL) KrasG12Dknock-in mice (KrasLSL-G12D/+) results in deletion of the transcriptional/translational Stop element, expression of oncogenic KrasG12D, and development of lesions that closely resemble early stage human pancreatic intraepithelial neoplasms (PanINs) (Hingorani et al. 2003). Concomitant expression of a point mutant p53 allele, deletion of p53, deletion of Cdkn2a, and/or deletion of Smad4 allow(s) for the development of invasive and metastatic PDAC

(Aguirre et al. 2003; Hingorani et al. 2005; Bardeesy et al. 2006a,b; Gidekel Friedlander et al. 2009; Whittle et al. 2015).

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4526740/

Pancreatic ductal adenocarcinoma (PDAC) is a genomically diverse, prevalent, and almost invariably fatal malignancy. Although conventional genetically engineered mouse models of human PDAC have been instrumental in understanding pancreatic cancer development, these models are much too labor-intensive, expensive, and slow to perform the extensive molecular analyses needed to adequately understand this disease. Here we demonstrate that retrograde pancreatic ductal injection of either adenoviral-Cre or lentiviral-Cre vectors allows titratable initiation of pancreatic neoplasias that progress into invasive and metastatic PDAC. To enable in vivo CRISPR/Cas9-mediated gene inactivation in the pancreas, we generated a Cre-regulated Cas9 allele and lentiviral vectors that express Cre and a single-guide RNA. CRISPR-mediated targeting of Lkb1 in combination with oncogenic Kras expression led to selection for inactivating genomic alterations, absence of Lkb1 protein, and rapid tumor growth that phenocopied Cre-mediated genetic deletion of Lkb1. This method will transform our ability to rapidly interrogate gene function during the development of this recalcitrant cancer.

Keywords: CRISPR, genome editing, mouse model, pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) is an almost uniformly lethal tumor type that is projected to become the second leading cancer killer in the United States by 2030 (Rahib et al. 2014). PDAC patients have a 5-year survival rate of ∼5%, underscoring the need for novel approaches to accelerate the molecular characterization of this disease. Although high-prevalence mutations have been identified in pancreatic cancer, these tumors also incur low-frequency mutations and genomic alterations, interact with their extensive and complex stromal environment, and undergo poorly characterized changes in their gene expression programs (Biankin et al. 2012; Waddell et al. 2015). Despite the potential importance of these molecular and cellular changes, very little is known about how these alterations contribute to the development of metastatic and therapy-refractory PDAC. Given the inability to test gene function in human cancers in vivo, genetically engineered mouse models represent tractable and biologically relevant systems with which to interrogate the molecular determinants of each stage of pancreatic cancer development.

Identification of the mutations that drive the development of human pancreatic cancer combined with the ability to alter gene function in mice has enabled the development of genetically engineered murine PDAC models. Transgenic expression of Cre-recombinase in pancreatic cells of loxP-Stop-loxP (LSL)Kras^G12D knock-in mice (Kras^LSL-G12D/+) results in deletion of the transcriptional/translational Stop element, expression of oncogenic Kras^G12D, and development of lesions that closely resemble early stage human pancreatic intraepithelial neoplasms (PanINs) (Hingorani et al. 2003). Concomitant expression of a point mutant p53 allele, deletion of p53, deletion of Cdkn2a, and/or deletion of Smad4 allow(s) for the development of invasive and metastatic PDAC (Aguirre et al. 2003; Hingorani et al. 2005; Bardeesy et al. 2006a,b;Gidekel Friedlander et al. 2009; Whittle et al. 2015).

These in vivo models have been instrumental in our understanding of the genetic determinants of cancer progression as well as the functional interactions of neoplastic cells with the immune system and stromal environment. However, using conventional genetically engineered autochthonous mouse models of PDAC to interrogate gene function is complicated by inherent practical and biological limitations of these systems. Existing mouse models typically fail to model the adult onset of pancreatic cancer and induce genomic alteration in nearly every cell in the pancreas. In the decade since the first genetically engineered PDAC models were developed, few technical advances have been made, and generating the mice required to investigate a gene of interest in the established PDAC models remains a time-consuming and costly endeavor (Aguirre et al. 2003; Hingorani et al. 2003,2005; Saborowski et al. 2014).

Systems that enable in vivo functional interrogation of genes in pancreatic cancer without the financial and temporal cost of generating new mouse alleles and incorporating them into increasingly complex mouse models could have an extremely broad impact on pancreatic cancer research. To functionally investigate the molecular changes that drive each step of pancreatic cancer development, it would be desirable to have a system in which the timing of tumor initiation and the number of lesions that form in the adult pancreas can be controlled, the number of germline-encoded alleles is minimized, and genes of interest can be eliminated without having to generate a conditional allele and breed it into a complex genetically engineered mouse model.

Here we describe methods for the direct delivery of viral vectors to the pancreas and transgenic mouse lines to allow CRISPR/Cas9-mediated genomic alterations in pancreatic cells in vivo. These systems allow titratable initiation of pancreatic tumors in adult mice and functional interrogation of candidate genes in pancreatic cancer in vivo.

Figure 1.

Pancreatic retrograde ductal injection of Adeno-Cre or Lenti-Cre induces widespread recombination and initiates the development of PanINs and ductal adenocarcinoma. (A) Diagram of the pancreatic retrograde ductal injection procedure. (B) Images of retrograde …

Chronic pancreatitis is associated with an increased incidence of PDAC in humans, and experimental evidence from mouse models suggests that adult cells may be refractory to transformation in the absence of pancreatic inflammation (; ). Therefore, we assessed whether retrograde ductal viral infection induces pancreatitis. Both Ad-Cre and Lenti-Cre infection induced acute pancreatitis with morphological evidence of ADM and focal replacement of acinar cells by infiltrating mononuclear cells (Supplemental Figs. 1C,D, 4C). The induction of pancreatitis is consistent with the efficient tumor initiation observed in adult animals following retrograde ductal injection of viral-Cre.

Figure 2.

Diverse expansion potential and metastatic ability of viral-Cre-initiated PDAC. (A) DTCs can be detected in the peritoneal cavity of Ad-Cre-infected KPTmice. Viable (DAPI^negative), lineage-negative (CD31, CD45, Ter119, F4/80)^negative cells are shown. …

Figure 3.

CRISPR/Cas9 enables in vivo genetic alteration in pancreatic cancer. (A) Schematic of the Cre-regulatedCas9 allele. (CAGGS) Cytomegalovirus immediate–early enhancer/chicken β-actin promoter; (hSpCas9) human codon-optimized Streptococcus…

Somatic genome engineering enables rapid generation of genetically defined pancreatic cancer mouse models

To assess the impact of Lkb1 deletion on pancreatic cancer, we performed retrograde ductal injections of KT;H11^LSL-Cas9/+, control KT, and KT;Lkb1^flox/flox mice with Lenti-sgLkb1/Cre as well as KT;H11^LSL-Cas9/+ mice with lentiviral vectors containing negative control sgRNAs. Control KT mice infected with Lenti-sgLkb1/Cre as well as KT;H11^LSL-Cas9/+ mice infected with a Lenti-sgRNA/Cre vector containing either a nontargeting sgRNA (sgNT) or an sgRNA targeting an inert region of the genome (sgNeo) formed only rare Tomato^positive lesions (Fig. 3F–H). Both Lenti-sgLkb1/Cre-infected KT;H11^LSL-Cas9/+ and KT;Lkb1^flox/flox mice had extensive tumor growth as early as 2 mo after tumor initiation (Fig. 3I,J, respectively).

All three groups of negative control mice developed only rare PanIN lesions within almost completely normal pancreata (Fig. 4A–C; Supplemental Fig. 9A; data not shown). The tumors that formed in Lenti-sgLkb1/Cre-infectedKT;H11^LSL-Cas9/+ mice were cystic lesions comprised of Tomato^positive, CK19^positive tall cuboidal to columnar epithelial cells with otherwise bland cytological features (Fig. 4D; Supplemental Fig. 9B). Some smaller lesions also contained high levels of mucin (Supplemental Fig. 9B). Importantly, these features were histologically indistinguishable from those found in KT;Lkb1^flox/flox mice infected with Lenti-sgLkb1/Cre (Fig. 4E; Supplemental Fig. 9C). Lenti-sgLkb1/Cre-infected KT;H11^LSL-Cas9/+ mice also had a substantially higher tumor burden when compared with all three groups of negative control mice (Fig. 4F).

Figure 4.

Cas9-mediated targeting of Lkb1 in the pancreas promotes tumor growth. (A) Control Lenti-sgLkb1/Cre-infected KT mice (n= 5) have small clusters of Tomato^positive cells and very rare ADMs and PanINs. (B,C) KT;H11^LSL-Cas9/+ mice infected with Lenti-sgNT/Cre …

Diverse indels at the targetedLkb1 locus were specifically detected in the neoplastic cells from these tumors (Fig. 5A,B; Supplemental Fig. 9F). The indels were all frameshift mutations or large exon deletions that included the splice acceptor or donor regions of Lkb1exon 6, further supporting the strong selective advantage of Lkb1 inactivation (Fig. 5B,C). Consistent with the presence of CRISPR/Cas9-induced frameshift mutations and large deletions, Lkb1 protein was absent from most neoplastic cells in the Lenti-sgLkb1/Cre-induced tumors in KT;H11^LSL-Cas9/+ mice (Fig. 5D,E; Supplemental Fig. 9G).

Figure 5.

Cas9-mediated targeting leads to the formation of pancreatic tumors harboring deleterious Lkb1mutations and lacking Lkb1 protein. (A) Genomic cleavage detection assay detected indels in the targeted Lkb1 locus in Tomato^positive cells isolated from Lenti-sgLkb1/Cre-infected …

The development of systems that accelerate our ability to investigate pancreatic carcinogenesis at the molecular level will be a critical step toward overcoming the dismal rate of successful treatment and low survival rate of patients with this recalcitrant cancer. The ability to induce pancreatic cancer using viral vectors will be instrumental in understanding the mechanisms that sustain tumor growth, lead to metastatic spread, and drive drug resistance. Our CRISPR/Cas9-based model should allow any gene of interest to be inactivated in pancreatic cancer in vivo without the need to generate any new mouse alleles. The method developed in this study could have a profound impact on both basic and translational pancreatic cancer research. Collectively, these methods will enable a more rapid and complete understanding of the molecular regulators of all aspects of pancreatic tumorigenesis and complement the strength of existing genetically engineered models, human patient-derived xenograft models, and studies on human and murine cell lines.

Retrograde ductal injection of Adeno-Cre and Lenti-Cre vectors allows titratable pancreatic tumor initiation in the adult pancreas. This removes the requirement for the transgenic Cre(ER) alleles used in conventional genetically engineered pancreatic cancer models and enables sparse rather than widespread expression of oncogenic Kras and deletion of tumor suppressor genes, more closely recapitulating the initiating events in human PDAC. The inclusion of a fluorescent Cre reporter in transgenic Cre(ER)-induced PDAC models leads to the fluorescent labeling of not only the neoplastic cells but also most of the nontransformed pancreatic epithelial cells, making unequivocal distinction of neoplastic from normal cells difficult.

The ability to induce PDAC in Cre-lox models without having to include a transgenic Cre line will make the molecular investigation of pancreatic cancer more rapid and less expensive (Supplemental Table 2). Additionally, with the ease of generating pancreatic tumors now approaching that of lung tumors (via intranasal or intratracheal injection), comparing the impact of the same genetic alterations on each cancer type should become standard practice and will uncover the commonalities and differences between these lethal adenocarcinomas.

PDAC models often generate tumor masses of unknown clonal origin. By incorporating a multicolor fluorescent reporter, we were able to mark individual clonal lesions and identify the relationship between primary tumors and metastases. Unexpectedly, some late time point mice had only one or two large cancers, underscoring the dramatic heterogeneity in expansion potential of pancreatic lesions initiated with identical engineered genetic events.