The Missing Link: Cre Pigs for Cancer Research

The Cre/loxP system is a powerful tool for the generation of animal models with precise spatial and temporal gene expression. It has proven indispensable in the generation of cancer models with tissue specific expression of oncogenes or the inactivation of tumor suppressor genes. Consequently, Cre-transgenic mice have become an essential prerequisite in basic cancer research. While it is unlikely that pigs will ever replace mice in basic research they are already providing powerful complementary resources for translational studies. But, although conditionally targeted onco-pigs have been generated, no Cre-driver lines exist for any of the major human cancers. To model human pancreatic cancer in pigs, Cre-driver lines were generated by CRISPR/Cas9-mediated insertion of codon-improved Cre (iCre) into the porcine PTF1A gene, thus guaranteeing tissue and cell type specific function which was proven using dual fluorescent reporter pigs. The method used can easily be adapted for the generation of other porcine Cre-driver lines, providing a missing tool for modeling human cancers in large animals.


INTRODUCTION
In recent decades, genetically modified mice have dominated basic and preclinical research into human cancer and many other diseases, largely because of the availability of murine ES cells and the technical ease of engineering interesting mutations. Mice are an invaluable source of knowledge but are not always the most suitable means of translating new findings into clinical application. The shortcomings of the exclusive use of rodents in preclinical studies, for example, drug trials, are now widely recognized (1) and regulatory agencies around the world are requiring preclinical trial data from non-rodent species (2). As mammals, mice and humans share many fundamental similarities, but dissimilar protein interactions, physiology, and anatomy can lead to different disease phenotypes from similar genetic lesions (3). There is therefore a strong need to investigate a wider variety of potential animal models. Pigs share many similarities with humans in body size, anatomy, and their physiological and pathophysiological responses and are playing an ever more important role in preclinical research. Many practical requirements necessary for the use of pigs as models of human cancer and cancer predisposition are in place. "Oncopigs" have been generated, carrying Cre inducible mutations in tumor suppressor (TP53 R167H ) or oncogenes (KRAS G12D ) either as transgenes (4) or as targeted mutations of the endogenous genes (5,6). As in mice, tumorigenesis is dependent on efficient and organ specific activation of the latent genes. But unlike for mice, no cancer specific porcine Cre-driver lines are available.
Here we report the efficient generation, tissue specificity, and in vivo functionality of two porcine Cre-driver lines for induction of pancreatic cancer, which has one of the highest unmet clinical needs (7,8). Using CRISPR/Cas9 guided insertion, codonimproved Cre (iCre) was placed either at the 5' or 3' end of the porcine pancreas transcription factor 1 subunit alpha (PTF1A, p48) locus, bringing iCre expression under the control of the endogenous PTF1A promoter. PTF1A is one of the key regulators in pancreas organogenesis (9) and murine Ptf1a-Cre lines are commonly used for the generation of pancreatic ductal adenocarcinomas (PDAC) mouse models (10). These pig Cre lines are now available for the generation of porcine models of pancreatic cancer.

Constructs for Genome Editing
For CRISPR/Cas9-mediated cleavage of the first exon or 3' end of porcine PTF1A, the following sgRNAs were selected (5'end targeting: 5'-gtcctcttactttgacgagg-3'; 3'end targeting: 5'acataaccagatcctcagga-3') and cloned into a pX462-Cas9 D10A nickase (Cas9n; D10A variant) (5'end) or pX330-Cas9 vector (3'end) carrying a puromycin resistance. For the generation of the targeting vectors, the homology arms were amplified by PCR from genomic DNA of German landrace pigs. The 5' targeting vector/donor DNA consists of 1053 bp 5' and 1246 bp 3' homology arms (HA) flanking the iCre-polyA sequence. The 3' donor DNA consists of 900 bp 5' and 3' HAs, a T2A self-cleaving peptide sequence, and the iCre-polyA sequence. To increase the efficiency of homology-directed repair (HDR) and the associated knock-in of iCre, the homology arms were flanked by the respective sgRNA target sequence in both targeting vectors, leading to the generation of double cut donors (11) ( Figure 1A).

Cell Culture
Porcine adipose-derived mesenchymal stem cells (ADMSCs) and porcine kidney fibroblast cells (KFCs) were isolated from a male German Landrace pig by the standard method (12). ADMSCs and KFCs were cultured with high-glucose DMEM supplemented with 10% FBS, 2 mM NEAA, 2 mM L-glutamine. Cells were passaged every 3-5 days and maintained at 50-90% confluency in an incubator at 37°C and 5% CO 2 . All chemicals were purchased from Sigma-Aldrich (St. Louis, MO).

Off-Target Analysis
Off-targets sites for sgRNA for 5' and 3'end targeting strategy were predicted with the use of CRISPOR (www.crispor.tefor.net) website tool. Five highest scoring potential off-target sites were analysed by PCR followed by sequncing.

Somatic Cell Nuclear Transfer (SCNT)
Two selected cell clones for PTF1A 5'iCre targeting and three cell clones for PTF1A 3'iCre targeting were sent for SCNT. SCNT was performed as previously described (13).

RESULTS
Generation of PTF1A 5'iCre and PTF1A 3'iCre Pigs In the Ptf1a-Cre mouse line the Cre recombinase gene had been inserted into exon 1 downstream to the translational start codon, silencing the expression of the endogenous gene. As the basic helix-loop-helix transcription factor PTF1A/p48 is critical for pancreatic cell fate it is essential that only one allele is targeted.
To replicate the murine model targeted gene placement supported by CRISPR/Cas9 was carried out in primary porcine cells isolated from a male German Landrace pig. To ensure that only one allele was targeted a Cas9 nickase (Cas9n; D10A variant) was used (14). To improve the efficiency of HDR, both 5' and 3' short regions of homology (approx. 1 kb) of the iCre (codon improved Cre)-donor plasmid were flanked by two sgRNA target sequences, which were identical to the target sequence in exon 1 of PTF1A ( Figure 1A) (11). The expression vector for both Cas9n and the exon1 sgRNA was co-transfected with the iCre double cut donor DNA plasmid. Single cell clones were isolated and screened by 5'-and 3'-junction PCR for correct gene targeting ( Figure 1B). HDR efficiency was 25%.

PTF1A 3'iCre
PTF1A/p48 is crucial for establishing the differentiation state of acinar cells. In mice, Ptf1a loss in adult acinar cells causes ER stress and activates the apoptosis pathways (15). It is epigenetically downregulated in pancreatic ductal adenocarcinoma (PDAC) and is assumed to be a tumor suppressor gene (16). As it was unknown if a heterozygous inactivation of PTF1A would have any adverse effects on porcine organogenesis or influence PDAC tumorigenesis, an alternative approach where iCre was placed at the 3'end of PTF1A, has been used. The addition of a "self-cleaving" 2A peptide (17) should allow translation of both PTF1A and iCre from a single mRNA. Again, a double cut iCre donor vector was generated ( Figure 1A), genome editing was carried out as described above using Cas9 instead of Cas9n vector giving HDR efficiency of 11%.
For both gene editing strategies the five highest scoring potential off-target sites were determined by the CRISPOR web tool (18). PCR amplification across these sites followed by sequence analyses showed no off-target events (Supplementary Figure 1).
Five correctly targeted primary cell clones for both PTF1A 5'iCre and PTF1A 3'iCre were pooled and used for SCNT. Nine male piglets were born: four PTF1A 5'iCre and five PTF1A 3'iCre ( Figure 1C). Three piglets (two PTF1A 5'iCre and one PTF1A 3'iCre ) died shortly after birth. The remaining six piglets were healthy and fertile. Precise iCre transgene placement into PTF1A allele was confirmed by 5' and 3' junction PCR followed by sequence analysis for all PTF1A 5'iCre and PTF1A 3'iCre founder animals ( Figure 1B). In each case the presence of an untargeted wildtype allele confirmed monoallelic insertion of iCre at the PTF1A locus ( Figure 1B).

Tissue Specific Expression of Porcine PTF1A-iCre
To assess PTF1A-iCre expression in the newborn founder animals, tissue samples (pancreas, stomach, kidney, spleen, heart, muscle and lung) were collected for RT-PCR and immunohistochemical analyses. At this developmental stage iCre expression was restricted to pancreatic acinar and single gastric cells in both PTF1A 5'iCre and PTF1A 3'iCre targeted pigs ( Figures 1D, 2A, B). The other organs tested were negative. These results are in accordance with PTF1A/p48 expression in humans (https://www.proteinatlas.org/ENSG00000168267-PTF1A/tissue).

PTF1A-iCre Driver Pig Lines Support Tissue Specific Cre Mediated Recombination In Vivo
To assess the location, pattern, and extent of iCre expression during fetal development and to prove iCre function, founder boars were crossed with a dual-fluorescence reporter pig line R26 mT/mG (19). This line carries a membrane-targeted tandem dimer GFP (mGFP) gene downstream of a floxed mTomato transgene at the porcine ROSA26 locus ( Figure 3A). Cre mediated excision of mTomato results in a switch from red to green fluorescence (20), marking all cells during development in which Cre recombinase was active.
Focusing on the pancreas multi-color immunofluorescence staining was performed for cell type identification. As shown in Figure 4A mGFP is expressed in glucagon-producing a-cells, ductal cells, and acinar cells. Interestingly, for b-cells co-staining of insulin with both RFP and GFP was observed ( Figure 4B). Colocalization of these pancreatic cell types with GFP confirmed the participation of PTF1A to all pancreatic lineages during development.

DISCUSSION
Cre/loxP, Flp/FRT, and Dre/Rox are site-specific recombinase systems, which can be employed to activate or inactivate conditionally targeted genes in a tissue specific manner (21). The Cre/loxP system has been extensively used in mice for lineage tracing, to study gene and cell function, induction, and development of cancer and a large number of Cre-driver lines are available (22). This is in strong contrast to livestock species where very few Cre transgenic animals have been generated, based solely on random integration into the genome, showing expression but not in vivo functionality (23,24). In the two pig lines reported here, the Cre recombinase is under the control of an endogenous promoter ensuring a correct spatial and temporal expression pattern, and in vivo Cre mediated recombination was confirmed.
Until recently targeted gene placement has been technically difficult in livestock species due to the lack of functional pluripotent stem cells. Gene targeting experiments were restricted to somatic cells, where homologous recombination is 100-fold less efficient. Targeting of non-expressed genes, such as pancreas specific genes, in porcine fibroblasts only became realistic with the advent of genome editing. Here we show that a method previously employed for precise knock-ins in induced pluripotent stem cells (11) allows highly efficient and precise gene placement (up to 25%) in somatic cells, which remain nuclear transfer competent. This approach is generally applicable and will enable the efficient generation of other Cre driver lines.
The functionality of PTF1A-iCre was confirmed by crossbreeding founder animals with our dual fluorescent reporter pig line (19). This not only proved the efficient removal of the floxed mTomato gene in the pancreas but also indicated which pancreatic cell types were derived from early PTF1A expressing progenitor cells, including acinar and ductal cells, a-cells but not all b-cells. After birth PTF1A-iCre expression became restricted to acinar cells. This is consistent with findings in mice, where Ptf1a was shown to be essential for commitment to pancreatic fate (25) and the formation of pancreatic multipotent progenitor cells (26). Later in development, it is expressed in acinar cells where it maintains acinar cell identity (27). Very localized PTF1A expression was found in the stomach both in the embryo and after birth. Gastric and pancreatic developmental programs are related, and low-level expression has also been observed in human stomach samples. During fetal development, PTF1A/p48 is also expressed in the cerebellar ventricular zone where it is required for the development of cerebellar GABAergic neurons (28). Accordingly, widespread expression of green fluorescence protein was observed in the cerebellum of R26 mT/mG+/-PTF1A-iCre pigs.
Based on its importance during pancreas development and cerebellar neurogenesis, mutations in PTF1A can cause cerebellar ataxia and permanent neonatal diabetes mellitus in humans (29). With regard to pancreatic cancer, deletion of Ptf1a causes acinar to ductal metaplasia and dramatically enhances KRAS-driven acinar cell transformation in mice and downregulation is observed in human PDAC (16), leading to the postulation that PTF1A/p48 functions as a tumor suppressor gene. By placing iCre at the 5' end of the porcine PTF1A gene we replicated existing mouse models and have inactivated one PTF1A allele. By introducing iCre at the 3' end, PTF1A/p48 protein expression should not be affected. It will be interesting to see if this alters tumorigenesis. Breeding is underway to obtain triple modified pigs (KPC pigs) in which KRAS G12D and TP53 R167H are activated in a tissue specific manner by either PTF1A 5'iCre or PTF1A 3'iCre and -as in KPC mice-initiates PDAC as the main pathophenotype. The KPC mice showed no phenotype outside the pancreas, no gastric or neuronal cancers during their lifetime despite high PTF1A expression in the cerebellum, suggesting a tissue-specific and context-dependent response (30). While unlikely, it remains to be seen if this will be different for the pig KPC model.
Pancreatic cancer is one of the most lethal forms of cancer, with fewer than 20% of people surviving for longer than twelve months after diagnosis. The usefulness of large animal models for translational research and as pre-clinical models has been documented (31)(32)(33)(34)(35). Modelling PDAC in large animals was hampered by the lack of Cre-driver lines. Besides fulfilling this essential requirement for a porcine PDAC model, the technology employed enables the efficient generation of other Cre transgenic lines for cancer research. In combination with reporter pigs, these Cre-driver lines can also support genetic fate mapping.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
All animal experiments were approved by the Government of Upper Bavaria (permit number ROB-55.2-2532.Vet_02-18-33).

AUTHOR CONTRIBUTIONS
DK, TF, HL, DS, and AS conceived the study. DK generated DNA constructs, performed transfections, and analyzed the cell clones. MK, BK, VZ, and EW performed SCNT and embryo transfer. KF, TF, and DK collected samples. DK performed molecular analysis of the PTF1A-iCre founder animals. DK, TF, CY, and LB performed histological analysis. DK wrote the manuscript. KF, TF, and AS edited the manuscript. All authors critically reviewed and approved the final manuscript.