Swine reporter model for preclinical evaluation and characterization of gene delivery vectors

Abstract

Delivery of gene therapy vectors efficiently targeted to any somatic cell remains a key barrier for the development of genetic medicines. While rodent models provide insights into vector biodistribution and cellular tropism, their anatomical and physiological differences from humans limit their translational potential and studies in large animal models are often required. In this study, we developed a swine reporter model (SRM-1) to evaluate both viral and non-viral vector delivery in a large animal system. The SRM-1 model harbors a tdTomato reporter at the ROSA26 locus that can be activated by Cre recombinase or CRISPR-Cas reagents and allows for tracking of gene delivery vectors in vivo. To evaluate this model, we administered adeno-associated virus serotype 9 (AAV9) and lipid nanoparticles (LNPs) carrying messenger RNA (mRNA) systemically and found successful in vivo reporter activation across a variety of tissues. Intracerebroventricular (i.c.v.) administration of LNP-mRNA was also performed and demonstrated localized activation in cortical brain cells. In addition to systemic biodistribution studies, this model has utility for testing clinically relevant local administration methods, surgical and non-surgical, of delivery vectors. Our findings support the SRM-1 model as a valuable tool for advancing gene therapies from preclinical testing to clinical application.

Graphical abstract

A “turn-on” swine reporter model is developed to characterize local and systemic delivery of gene editors in vivo using viral or non-viral vectors. This adds the functionality of a reporter to preclinical gene delivery research in a large animal model that is more broadly accessible than nonhuman primates.

Introduction

Safe and effective targeting of somatic cells with viral or non-viral vectors is a continuous barrier to gene therapy development. These vectors can be used to deliver payloads of therapeutic nucleic acids or proteins to treat diseases, especially in those with genetic underpinnings.¹ Rodents have been essential in characterizing the tissue and cellular tropism of these viral and non-viral payloads, and fluorescent reporter mice that can be activated by genome-editing enzymes such as Cre or CRISPR have been used to track tissue and cellular targeting of vectors in vivo.^2,3,4,5,6,7 For example, reporter mice have aided in the characterization of lipid nanoparticle (LNP)-based messenger RNA (mRNA) delivery for in vivo gene delivery, with several formulations demonstrating both hepatic and extrahepatic tropism when delivered systemically.^4,5,6,7,8,9 However, anatomical and physiological differences between rodents and humans can limit the translation of these discoveries. This is especially true when studying the impact of alternate routes of administration, dosing in large animals, and the role of the immune system in vector targeting and expression. At advanced stages of preclinical testing, nonhuman primates (NHPs) have been used to study vector delivery and safety in vivo as they are more physiologically like humans. However, their small size, high cost, and low fecundity limit utility and access for most researchers. Alternative preclinical animal models, especially more accessible large animals, are vital for exploring vector and gene delivery to tissues and cells.

In this study, we have developed a swine reporter model (SRM-1) as part of the Somatic Cell Genome Editing Consortium to help with characterization of both viral and non-viral vectors.¹⁰ Swine are an increasingly attractive species for preclinical studies and have several advantages for use in biomedical research. First, the technology development around genome engineering and reproductive technologies has matured and custom model development is efficient. Highlighting these efforts are the recent advances in xenotransplantation of multi-edited swine organs, which demonstrates the utility of these technologies that are also being used to generate pig models for studying disease.^11,12 Pigs are available in a variety of body sizes to accommodate a wide range of gene delivery studies, from systemic infusions to local delivery with medical devices. Shared physiological and anatomical characteristics exist between pigs and humans, particularly in the nervous system, the cardiovascular system, the gastrointestinal tract, the respiratory system, the eye, and skin.^{13,14,15,16,17,18} Finally, swine husbandry and breeding practices are well-established due to the centuries-old reliance of pigs as a major food source. Pigs have large litters compared with other large animal model systems, typically rearing 5–15 piglets per litter depending on the breed. These piglets develop rapidly, reaching sexual maturity at approximately 6 months. Combined with a gestation period of about 4 months, this rapid generational turnover enables cost-efficient, translationally relevant, and statistically robust animal studies compared with NHPs.

Results

Generation of the SRM-1 model

We designed the SRM-1 model based on the Ai9/Ai4 mouse reporters.² It includes a constituently active CAG promoter that drives expression of a tdTomato fluorescent protein after removal of a strong transcriptional stop domain (3xStop, Figure 1A). The 3xStop is flanked by loxP sites, which allows for reporter activation by Cre recombinase, and by SpCas9 and AsCas12a guide RNA spacer sequences that allow for one-guide genome editing activation (Figure 1A). In addition to tdTomato, a sodium iodide symporter (NIS) protein is co-activated, allowing for the potential of real-time imaging of reporter activation using positron emission tomography (PET) or single-photon emission computerized tomography (SPECT) imaging in-life (Figure 1A).¹⁹ The SRM-1 reporter construct was integrated at the ROSA26 gene using homology directed repair in Yorkshire swine fibroblasts and cloned by somatic cell nuclear transfer (SCNT) to yield a founder breeding population and fetal fibroblasts (Figure S1). Resulting animals were screened for precise reporter integration by junction PCR of the SRM-1 construct into the ROSA26 gene at the 5′ and 3′ end of the integration site (Figure 1B). These founder animals harbored two copies of the reporter (Figure 1B), and homozygosity of the reporter allele was confirmed by outcrossing to wild type animals to yield all heterozygous offspring (data not shown). tdTomato expression is apparent in SRM-1 fetal fibroblasts after transfection of either Cre mRNA or CRISPR ribonucleoproteins (RNPs) (Figure 1C). To rapidly screen for effective reporter activation across many tissues during animal studies, we also developed a probe-based droplet digital PCR (ddPCR) assay that can be used to measure Cre-induced DNA recombination (Figure 1D).

Figure 1.

Generation of the SRM-1 Cre- or CRISPR-inducible swine reporter model

(A) Schematic illustration of the SRM-1 reporter integrated into the swine ROSA26 gene. Reporter design includes loxP sites (purple triangles) and CRISPR sites (blue and red arrows) for Cre- or genome editing-mediated reporter activation, respectively. (B) PCR of the integration junction of the SRM-1 construct in the ROSA26 gene and SRM-1 construct copy number measured by ddPCR in founder SRM-1 animals. Asterisk indicates an animal with imprecise reporter integration. (C) The percentage of tdTomato-positive SRM-1 fibroblasts measured by flow cytometry after transfection with either Cre mRNA, SpCas9 RNP, or AsCas12a RNP. Open circles are individual data points; error bars are standard error of the mean. (D) Schematic of the probe-based assay developed to measure SRM-1 DNA recombination by Cre. Without Cre-induced DNA recombination of the loxP sites, the SacI restriction enzyme will cut the DNA and not allow for PCR from the forward and reverse primers (black arrows). After DNA recombination, the SacI site is removed from the amplicon and PCR will cause degradation of the FAM-conjugated probe (gray arrow) which will release the FAM fluorescence (green arrow). (E) Comparison of SRM-1 reporter activation by tdTomato expression measured by flow cytometry and by DNA recombination measured by ddPCR across a 100-fold drop in Cre mRNA transfection. Open circles are individual measurements and bars are the average measurement for tdTomato (black) and ddPCR (purple). Brackets with numbers are the difference in percentages between the average flow cytometry measurement and the average ddPCR measurement.

In vivo activation of SRM-1 by systemic delivery of AAV9-Cre

To assess SRM-1 reporter activation in vivo, an adeno-associated virus serotype 9 (AAV9) vector expressing Cre recombinase was infused systemically into a male SRM-1 piglet at a dose of 7 × 10¹² genome copies per kilogram body weight (Figure 2A). Biodistribution and viral expression were measured across 25 tissues and compared with reporter recombination rates (Figures 2B, 2C, and S2). The liver harbored the most viral genomes, although RNA expression was lower compared with other tissues, suggesting a silencing of the viral genome (Figures 2B and S2). In addition, viral genomes were present in the brain and spinal cord with an absence of viral expression or reporter recombination (Figures 2B and S2). Four of the top nine tissues with over 5% SRM-1 recombination were muscle tissues (skeletal muscle, diaphragm, heart, and tongue; Figure 2C). While vector tropism of AAV9 in both large and small mammals is myotrophic, comparing the recombination rates with both viral genome copies and expression suggests that muscle tissues were enriched in their Cre-mediated activation (Figure 2B).^20,21,22

Figure 2.

SRM-1 is activated in vivo after intravascular delivery of AAV9-Cre

(A) Schematic illustration of systemic AAV9 in vivo study. A 23-day-old male SRM-1 piglet was injected with 7.27 × 10¹² viral genomes per kilogram of AAV9-CB-Cre, and tissues were harvested 16 days after for molecular and histological analysis. (B) Heatmap comparing viral genome biodistribution, viral expression, and Cre-mediated SRM-1 recombination across tissues collected. Each group was normalized to the highest (100, red) and lowest (0, white) values within the group. (C) The percentage of SRM-1 recombination of the loxP sites by Cre in tissues measured by ddPCR. Purple bars are measurements from an AAV-injected animal (n = 1), open circles from an uninjected SRM-1 animal (n = 1). (D) Immunohistochemistry of tissues taken from an injected and uninjected SRM-1 piglet. Tissues were co-stained for tdTomato and sodium iodide symporter (NIS) and detected with a duel-label system in which cytoplasmic tdTomato stains red and membrane-bound NIS stains brown. Scale bars, 300 μm.

In vivo protein expression of both tdTomato and NIS in tissues with Cre-mediated DNA recombination was investigated using immunohistochemistry (IHC), including a dual-stain system that was established to co-stain for each reporter protein (Figure S3). Selected tissues with both high and low reporter recombination demonstrated robust expression (Figures 2D and S4). Individual staining revealed cytoplasmic tdTomato signal, whereas NIS signal is observed in both the plasma membrane and cytoplasm (Figure S3). Interestingly, tdTomato staining shows higher background in liver tissues, so only cells strongly stained positive for both reporters should be scored as positive (Figure S3). In addition to somatic cells, it is critical to assess the potential for germ cell exposure to gene delivery vectors. Substantial reporter activation in the testicle was observed, localized to the interstitial cells of the testes, including into Leydig cells (Figure S5). These results support previous findings of effective AAV9 uptake into testicular tissue observed in mice.^23,24

Systemic delivery of LNP-mRNA into SRM-1

Non-viral delivery of nucleic acid like mRNA is a growing therapeutic modality that has applications across multiple areas of disease treatment and management.²⁵ To determine the utility of the SRM-1 model for characterizing tissue-specific LNP delivery, three piglets underwent systemic infusion with LNP-mRNA at a dose of 0.4 mg of mRNA per kilogram body weight (Figures 3A, 3B, and S6).⁷ Serum chemistries and complete blood cell counts were collected on days 1, 3, 7, and 14 to analyze potential toxicities related to systemic LNP exposure (Figure 3C). No circulating enzyme elevations were observed in animals injected with LNP, while moderate decreases in lymphocytes and increases in neutrophils and white blood cells were observed one day post injection in two LNP-injected animals compared with unexposed siblings (Figure 3C). These changes in blood counts recovered to normal values by day 3.

Figure 3.

Serum chemistries and blood counts of LNP-injected piglets

(A) LNP-mRNA delivery study design schematic. Three 20-day-old SRM-1 piglets were injected with 0.4 mg kg⁻¹ of LNP-mRNA carrying Cre and tissues were harvested after 14 days (B) LNP formulation schematic of mRNA using microfluidic mixing. (C) Absolute values of serum ALT, AST, creatinine, creatinine kinase, and albumin as well as blood cell counts (white blood cells, lymphocytes, neutrophils, red blood cells) in three LNP-injected SRM-1 pigs (purple), one mock-injected SRM-1 pig (gray), and 2 non-injected wild type sibling piglets (black). The red dashed lines are the high and low references ranges supplied by the diagnostic laboratory. One blood sample per piglet was taken at each time point.

Of the 18 tissue types sampled, the liver had the highest DNA recombination rate (39%, Figure 4A). Samples were collected at three locations (proximal, middle, and distal from the portal vein) on each of four pig liver lobes and uniform distribution of LNP was observed throughout the tissue (Figure 4B). Body-wide tissue distribution suggests extrahepatic targeting was also achieved, with moderate rates of targeting in the spleen (26%), lung (7%), adrenal gland (5%), and measurable uptake in the upper GI tract (1%–2%, Figures 4A, 4C, and S7). Gonad sampling from one injected male showed moderate rates of targeting to testicular cells (13%), which appear to be excluded from the seminiferous tubules (Figures 4A and 4C). Minimal ovarian tissue uptake was observed from two injected females (1%, Figures 4A, 4C, and S7).

Figure 4.

Tissue and cellular transfection by systemic LNP-mRNA can be tracked using SRM-1 swine after multiple weeks

(A) Recombination of SRM-1 genomic DNA collected from LNP-injected SRM-1 pigs (n = 3, 1 male and 2 females) and a mock-injected male SRM-1 pig (n = 1). Open circles are individual animal measurements for LNP-injected animals with purple bars at the mean, closed circles are for mock-injected. Error bars are standard error of the mean. (B) Distribution of LNP in each swine liver lobe. R = right lateral lobe, RM = right medial lobe, LM = left medial lobe, L = left lateral lobe. Open circles are individual animal measurements with purple bars at the mean. Error bars are standard error of the mean. (C) Fluorescent histological images of SRM-1 tissues with native tdTomato-producing cells (red) and stained nuclei (blue) after LNP-mRNA (Cre) injections. Scale bars, 200 μm.

I.c.v. delivery of LNP-mRNA to cortical brain cells in SRM-1

With the goal of achieving high levels of targeting and reducing potentially undesired physiological effects by delivering lower doses, localized delivery is an attractive alternative to systemic delivery of gene therapy vectors. Delivery of viral or non-viral vectors to the brain is one such example that holds great promise for the treatment of neurologic diseases.²⁶ To test reporter activity in the brain of an SRM-1 model, we performed a stereotactic injection of 2.5 mg of Cre mRNA, packaged into LNPs, per kilogram of brain mass into the intracerebroventricular (i.c.v.) space in a female piglet (Figures 5A and S6). LNP uptake was successful but limited to the outer cortex, likely due to the distribution of the vector in cerebrospinal fluid (Figure 5B). Examination of the stained sections suggests that multiple cell types in the brain were activated along the outer cortex based on differential morphology and reporter expression patterns (Figure 5B). This model holds promise for characterizing non-viral delivery of nucleic acid to the brain, and translatable surgical procedures can be deployed to optimize regional tissue targeting.

Figure 5.

Delivery of LNP-mRNA to the brain via i.c.v. injection

(A) Schematic of stereotactic i.c.v. injection of LNP-mRNA (Cre) into a 3-day-old SRM-1 piglet. 2.5 mg of mRNA per kg brain mass was injected. (B) Whole-brain coronal sectioning and immunohistochemistry for tdTomato and NIS was performed to identify mRNA uptake (left). Cell transfection was restricted to the outer cortex with minimal penetration into tissue. At least three distinct cortical cell types of the brain were identified by morphology and staining differences (black arrows, right). Tissues were co-stained for tdTomato and sodium iodide symporter (NIS) and detected with a duel-label system in which cytoplasmic tdTomato stains red and membrane-bound NIS stains brown.

Discussion

Large animal models that provide critical insights into the complexity of vector delivery are needed to develop gene therapies. Here, we highlight a swine model that can be used to track delivery of viral or non-viral vectors by activating a genomically encoded SRM-1 reporter. Activation of the reporter requires vectors to traffic to the tissue, gain entry to the cells, and deliver a functional recombinase or DNA editing payload. We first chose to test AAV9 to validate the use of an established and well-characterized vector in an SRM-1 male piglet for in vivo analysis. Similar to AAV9 tropism in mice, we observed widespread tissue tropism in the pig.^27,28 The highest viral load in the injected piglet was measured in the liver, with Cre expression lower than tissues that had lower viral load. This suggests a silencing of viral expression in the liver, which has been observed in NHPs, or a weak viral promoter system for liver cells.²⁹ Interestingly, IHC of liver tissue revealed cells with strong tdTomato and NIS co-expression and cells with a weak positive signal for tdTomato alone (Figure S3). The reporter genes share an mRNA transcript that are separated by a 2A peptide, so tdTomato and NIS proteins are co-translated from a single ribosome (Figure 1A). Cells that do not contain both proteins are an artifact of the fixing and IHC process.

Testicular tissue and adrenal gland were among the highest transfected tissues by both AAV and LNP, suggesting these tissues may be easily transfected in juvenile pigs and agnostic to the delivery vehicle. Adrenal targeting by AAV9 has previously been demonstrated in mice and NHPs but to our knowledge adrenal gland targeting by LNPs in large animals has not been reported.^28,30,31,32 Testes transduction has also been modest in NHP and rodent studies, which may be a meaningful difference between models that can be leveraged to study the potential for harmful germ cell transduction in mammals if the vectors penetrate seminiferous vesicles instead of the interstitial tissue as targeted by AAV9 (Figure S5). Tissues with myocytes, including skeletal muscle, diaphragm, tongue, and heart, had among the highest recombination rates despite having moderate to low viral load and expression (Figures 2B and S2). Immune-mediated responses to viral vectors can dampen viral vector load and expression in cells over time but cannot reverse DNA recombination once Cre is transiently expressed.³³ This may explain how high levels of DNA recombination could be achieved despite modest vector expression. Alternatively, pig myocytes are multinucleated, and the increase in efficiency may be caused by recombination of multiple genomes per cell transduction.^34,35 Continued characterization of systemically delivered viral vectors in large animals will be integral to understand the impact gene editing therapies can have on muscle tissues, which has been explored for the treatment of Duchenne muscular dystrophy (DMD) and other genetic skeletal diseases.³⁶

AAV9 can cross the blood-brain barrier (BBB) in both juvenile NHP and mice, a discovery which translated to a gene therapy drug for spinal muscular atrophy.^31,37 Interestingly, we found vector genomes in the spinal cord and the brain but no viral expression, reporter recombination, or positive cell staining (Figures 2B and S4). AAV9 tropism in swine motor neurons has been previously described by intrathecal administration of the vector, and one study in pigs also found uptake in the brain after intravenous (i.v.) injection at nearly twice the dose and with gene expression driven by a cytomegalovirus (CMV) promoter.^38,39,40 The AAV9 vector here expressed Cre from a chicken β-actin promoter and a CMV enhancer (CB6 promoter), which has not been extensively studied in pigs but has been shown to successfully cross the BBB and express genes in the mouse brain and motor neurons along the spinal cord.⁴¹ Our data supports no reporter expression in the spinal cord or the brain at the delivered dose but additional replicates, doses, and alternative promoters should be tested to continue characterizing pigs as a preclinical model for vector studies crossing the BBB.

Unlike AAV vectors, non-viral vectors such as LNPs do not have natural extrahepatic tissue tropism and cannot localize and express their genome in the nucleus.⁴² Targeting of the LNP to different cell types depends on several factors including the chemical makeup of the LNP, the delivery route to dictate which tissues are exposed, exogenous proteins that can interact with the LNP, and mediate receptor targeting and uptake into the cell, and ultimately endosomal escape and intracellular trafficking of the cargo.⁴³ The LNP system used in the systemic study here has previously been shown to target and express Cas9 mRNA effectively in the liver for genome editing.⁷ Replacing Cas9 mRNA with Cre mRNA, we found the highest levels of targeting in the liver and the spleen, along with testicular tissue and adrenal gland. Lung tissue showed recombination levels of the reporter at similar levels to adrenal but tdTomato expression from fixed tissue appears to be lower. This is likely due to sampling differences between the fixed tissue and DNA collections, as the adrenal gland samples had a high amount of variance in tdTomato expression in the tissue after sectioning (data not shown). Surprisingly, low levels of reporter targeting were measured across many other tissues as well. These findings highlight the sensitivity of the model and are congruent with recent evidence that pigs express protein from LNP-based mRNA delivery even in extrahepatic tissue.⁴⁴ Little research exists detailing the tissue distribution of LNP-mRNA after systemic injection in large animals, and more in-depth studies are needed to determine the translatability of these results to humans.

The safety of systemic LNP exposure is important to study in large animals. Complement activation-related pseudoallergy (CARPA) has been observed in both human and swine in response to systemic nanoparticle exposure.⁴⁵ Due to the impact on overall health, animals in this study were monitored for a CARPA response and showed no symptoms through the procedure and in-life observations. Overall, complete blood count and serum chemistries were within normal limits throughout the study; however, mild variabilities in white blood cell and neutrophil counts in two of the three LNP-injected animals were noted at 24 h post injection which may be suggestive of a mild immune response to the LNP-mRNA vector or anesthetic procedure.

The ability to test both systemic and localized delivery through clinically relevant procedures is one of the advantages of swine models. The central nervous system can only be reached systemically by a subset of vectors that cross the BBB. Viral vectors have been reprogrammed to target receptors that facilitate this process, which has been used to cross the BBB by targeting the human transferrin receptor.⁴⁶ For large non-viral molecules such as LNPs, this is currently not feasible, so alternative methods like intrathecal delivery, convection-enhanced tissue injection, and focused ultrasound are being explored, all of which would greatly benefit from preclinical testing in large animal models. Demonstrating this utility by performing an intracerebroventricular (i.c.v.) injection, we found reporter activation in cortical brain cells when exposed to LNPs from the cerebral spinal fluid. The LNP formulation contained the ionizable lipid 306-O12B, which has been found to effectively target the liver after i.v. injection⁴⁷ but has not been well-studied for brain targeting. This formulation was not able to penetrate outer cortex tissue more than 200–300 μm from the cerebrospinal fluid (CSF) at a dose of 2.5 μg per kg of brain mass. LNPs with higher tissue penetration will likely be required to enable therapeutically relevant delivery using this method.

To keep the dosage of vector low, our studies relied on dosing juveniles (aged 3 days to 4 weeks). Systemic studies in older swine that are similar in body mass to humans will be informative to understand the impact age and size can have on overall biodistribution. As in vivo studies with the SRM-1 model move to gene editing-based studies, we anticipate CRISPR-based delivery will require higher doses than Cre recombinase to activate the reporter, which has been previously observed in reporter mice and through SRM-1 cell studies here.⁶ Overall, the SRM-1 enables precise tracking of viral and non-viral delivery and activity across multiple tissues, providing a large animal model to advance gene delivery discovery and translation.

Materials and methods

SRM-1 plasmid generation

The Ai9/Ai14 reporter sequence was modified to include SpCas9-and AsCas12a-targeting sequences flanking the 3xStop domain using Geneious Prime (Dotmatics).² To mediate integration into the swine ROSA26 gene, homology arms flanking the SRM-1 reporter sequence as well as an adjacent universal SpCas9 guide sequence were included for precise integration of the reporter by homology directed repair as previously described.⁴⁸ The resulting reporter cassette was synthesized and subcloned into a plasmid backbone containing a Kanamycin resistance gene (GenScript Biotech) for integration into the pig genome. To prepare the plasmid for cell transfections, plasmid was grown in chemically competent E. coli (TOP10, Thermo Fisher Scientific) and purification was done using the plasmid DNA MidiPrep kit (QIAGEN) and eluted in water, following the manufacturer’s recommended protocols.

Cell culture

Yorkshire swine fetal fibroblasts were cultured in DMEM (Corning Life Sciences) with 10% (vol/vol) fetal bovine serum (FBS) (Atlas Biologicals) and 1× penicillin-streptomycin solution (Gibco). All cells were cultured in a humidified incubator at 37°C and 5% CO₂. A Neon transfection system (Thermo Scientific) was used deliver nucleic acid or CRISPR RNP to fibroblast cells. To produce fibroblast cells for SCNT, 6 × 10⁵ cells were co-transfected with two SpCas9-sgRNA RNPs, one targeting the swine ROSA26 gene and one targeting the SRM-1 plasmid backbone, along with the SRM-1 plasmid using the 100 μL Neon transfection kit (Thermo Scientific, electroporation conditions: 1,800 V, 20 ms, and 1 pulse). To prepare RNPs for transfections, lyophilized guide crRNAs and tracrRNA were resuspended in Tris-EDTA (TE) buffer (pH 8.0) and duplexed together by mixing 1:1 and heating to 95°C for 5 min before dropping to 22°C. SpCas9-gRNA RNPs were made by incubating Cas9 protein with duplexed CRISPR gRNAs for 20 min at a 1:1.2 molar ratio (final RNP concentration of 3 μM). After transfections, cells were plated on culture plates, and single colonies were picked for genotyping and expansion for SCNT. To measure SRM-1 reporter activity, 6 × 10⁴ SRM-1 fetal fibroblasts were electroporated in triplicate with Cre mRNA (1–0.001 ng μL⁻¹) or with CRISPR RNPs using the 10 μL transfection kit (Thermo Scientific). CRISPR RNPs were prepared as described above using sgRNAs (with no crRNA/tracrRNA complexing). All reagents used for gene editing can be found in Table S1.

Genotyping

Fibroblast clones to be used for SCNT were screened for SRM-1 integration. Cells were grown to confluency in 96-well cell culture plates and lysed to extract genomic DNA using previously described methods.⁴⁹ The resulting lysate was PCR amplified for 35 cycles using Accustart II GelTrack PCR SuperMix (Quantabio) and run on an agarose gel to identify clones that harbored a 5′ and a 3′ integration junction amplicon. Amplicons were purified using the QIAquick PCR purification kit (QIAGEN) and Sanger sequencing was performed using commercial services. Clones were then analyzed for SRM-1 copy number using ddPCR. For live animals, tissue biopsies were taken from the tail or ear of newborn piglets and genomic DNA was extracted using the DNeasy Blood and Tissue kit (QIAGEN) before running the same genotyping assays as described above. PCR primers and probes for genotyping can be found in Table S2.

SCNT and embryo transfers

Fibroblast clones were shipped to TransOva Genetics for nuclear transfer using their established protocols. The resulting embryos were implanted into the uterus via midline incision and recipients were raised until farrowing or until fetuses were collected for fibroblast isolations.

Fetal fibroblast isolations

Fetuses were collected 40 days into the pregnancy and skin tissue was transferred to a type 1 collagenase solution at 37°C with physical agitation. Resulting cells were collected by centrifugation and cultured for 2–3 days before cryopreservation with FBS and 10% DMSO.

Flow cytometry

Fibroblast cells were lifted from tissue culture plates using TrypLE Express (Gibco) and washed with PBS before resuspension in flow cytometry buffer (1× PBS +2.5% FBS + 5 mM EDTA) to a concentration around 1 × 10⁶ cells mL⁻¹. Flow cytometry was performed at the University of Minnesota flow cytometry resource core. Samples were analyzed utilizing a BD Biosciences LSRFortessa. Cells were analyzed for tdTomato expression, and cell viability was measured using Sytox Red (Thermo Fisher Scientific) according to the manufacturer’s recommended protocols. Gating was performed by running a negative control (no transfection) to identify negative cells, and a positive control (1 ng μL⁻¹ Cre mRNA) to identify the positive cell population. To gather data, 10,000 cells were analyzed for each sample. The resulting data files were analyzed using FACSDiva software v.9.3.

SRM-1 recombination, SRM-1 copy number, viral biodistribution, and viral expression measurements

To assess SRM-1 recombination rates in fibroblast cells or from animal tissues, genomic DNA extraction was performed using the DNeasy Blood and Tissue kit (QIAGEN). DNA concentrations were measured using a Qubit 4 Fluorometer (dsDNA BR Assay kit, Thermo Scientific). Reactions were set up using ddPCR Supermix for Probes (no dUTP, Bio-Rad) with 20–50 ng genomic DNA per assay according to the manufacturer’s protocol. A duplexed assay was used consisting of a primer and probe set (FAM) that detects SRM-1 recombination and a diploid genomic DNA internal control primer and probe set (GAPDH and HEX). A SacI-HF restriction enzyme (NEB) was added to the reaction mix. After reaction assembly, droplets were generated using the AutoDG Instrument (Bio-Rad) followed by PCR amplification using the recommended 3-step PCR protocol with a melting temperature of 60°C. The resulting droplets were read by the QX200 Droplet Digital PCR system and subsequent analysis was performed on the QuantaSoft Analysis Pro software (Bio-Rad). SRM-1 recombination (%) was calculated as the ratio between the SRM-1 positive DNA concentration (FAM probe) and the diploid GAPDH positive DNA concentration (HEX probe) multiplied by 2 to account for the heterozygous (single-copy) SRM-1 line, then multiplied by 100 to make it a percentage. For each pig tissue sample, recombination rates were calculated and plotted using GraphPad Prism. Multiple liver and lung samples were measured and averaged for each animal in the systemic LNP study, for every other tissue only one sample was collected.

To measure the SRM-1 copy number in fibroblasts for use in SCNT, the same protocol is used substituting the primer and probe (FAM) set to target the tdTomato sequence of the reporter construct. SRM-1 copy number was calculated by the QuantaSoft software, with the GAPDH reference copies at 2.

To measure the AAV9 copy number in pig tissues, the same protocol is used, substituting the primer and probe (FAM) set to target the AAV-Cre viral genome sequence. Copies of AAV in each well measured by ddPCR were divided by the input mass of DNA to determine copies of AAV per nanogram of starting material.

AAV viral expression was measured using the 1-Step RT-ddPCR Advanced kit for probes (RT-ddPCR, Bio-Rad). RNA was extracted using the TRIzol reagent method according to the manufactures protocol (Thermo Fisher Scientific). In short, 50–100 mg of snap frozen tissue was submerged in 1 mL of TRIzol and homogenized in by handheld homogenizer until there were no visible tissue pieces. Homogenized tissues were allowed to incubate in TRIzol for 5 min then 200 μL of chloroform was added and tubes were briefly inverted and incubated for 3 min. Samples were next centrifuged at 12,000× g at 4°C, and the aqueous layer was removed, transferred to a new tube, and 500 μL isopropanol was added to precipitate the RNA. The resulting solution was allowed to incubate for 10 min before being centrifuged for 10 min at 12,000× g at 4°C. The resulting RNA pellet was washed with 75% ethanol and briefly centrifuged before the supernatant was removed and the RNA was allowed to air dry. RNA was resuspended with nuclease-free water, and the RNA concentration was measured using the Qubit RNA BR Assay kit (Thermo Fisher Scientific).

To measure viral expression, RT-ddPCR was run with the AAV9-Cre Copy Number primers and probes (FAM fluorophore) with input RNA between 20 and 100 ng. Samples were then diluted 1:100 with nuclease-free water and measured using the Qubit RNA HS Assay kit (Thermo Fisher Scientific) before running the with GAPDH RNA primers and probes (HEX fluorophore). For RT-ddPCR, droplet generation and PCR cycling was performed as described above with an additional 1 h reverse transcription step at 42°C before PCR initiation. Viral Cre expression and GAPDH expression were determined by dividing copy number per by the total input RNA to calculate copies per nanogram. A heatmap was made using GraphPad Prism to compare viral biodistribution, expression, and Cre recombination in all collected tissues after systemic AAV infusion by normalizing each measurement by assigning 100% to the highest value and 0% to the lowest value and proportionally measuring each data point. All primers and probes used for ddPCR in this study can be found in Table S2.

Recombinant AAV production

Recombinant AAV (rAAV) was produced by transient HEK 293 cell transfection and CsCl sedimentation by the University of Massachusetts Medical School, Viral Vector Core, as previously described.⁵⁰ Vector preparations were determined by ddPCR, and purity was assessed by 4%–12% SDS-acrylamide gel electrophoresis and silver staining (Invitrogen). The titer of the injected rAAV was 2.2 × 10¹³, and the ITR-to-ITR size of CB-Cre was 2,388 basepairs.

IVT

For the i.c.v. injection, Cre mRNA was produced by in vitro transcription (IVT), from a linearized plasmid template containing a CleanCap-AG-compatible T7 promoter, 5′-UTR sequence, Kozak sequence, coding sequence, 3′-UTR sequence, poly(A) tract, and a restriction site (Esp3I/BsmBI). The 5′-UTR and 3′-UTR sequences were derived from those of Pfizer/BioNTech severe acute respiratory syndrome (SARS)-CoV-2 mRNA vaccine (BNT-162b2),⁵¹ and the Cre recombinase coding sequence was derived from the TriLink Cre mRNA open reading frame sequence (TriLink Product L-7211, 1351 nucleotides in length). Full plasmid sequence is provided in the supporting information. The IVT was performed using HiScribe T7 High Yield RNA Synthesis kit (NEB). CleanCap Reagent AG (3′-OMe) (TriLink Biotechnologies) was added to the reaction mixture, and UTP was replaced with N¹-methylpseudouridine-5′-triphosphate (TriLink Biotechnologies). Following a 2 h incubation at 37°C, DNase I (NEB) was added to digest DNA templates. Cre mRNA was purified using the Monarch RNA Cleanup kit (NEB) and high-performance liquid chromatography.

LNP production and characterization

For swine systemic injection studies, LNPs were formulated with an amine-to-RNA-phosphate (N:P) ratio of 4.5. The LNP components were dissolved in 100% ethanol with the following molar ratios: 45 mol% LP01 lipid, 44 mol% cholesterol, 9 mol% distearoylphosphatidylcholine (DSPC), and 2 mol% 1,2-Dimyristoyl-sn-glycero-3-methoxypolyethylene glycol (DMG-PEG 2000). The mRNA cargo was dissolved in 25 mM citrate buffer (pH 4.5), resulting in a concentration of mRNA of approximately 0.45 mg mL⁻¹. LNPs were formed by microfluidic mixing of the lipid and RNA solutions using a NanoAssemblr Ignite (Precision NanoSystems, Cytiva) at a flow rate of 2 mL min⁻¹, following the manufacturer’s protocol. The remaining buffer was exchanged into PBS (100-fold excess of sample volume) overnight at 4°C under gentle stirring using a 10 kDa dialysis cassette. The resultant mixture was then filtered using a 0.2 μm sterile filter. The filtrate was stored at 2°C–8°C. Encapsulation efficiencies were determined by the RiboGreen assay.⁵² Particle size and polydispersity were measured by dynamic light scattering (DLS) using a Malvern Zetasizer DLS instrument.

For i.c.v. injection studies, LNPs were formulated following a previously published protocol.⁵³ A total lipid solution at a concentration of 3 mM in 100% ethanol was prepared by combining ionizable lipid 306-O12B (synthesized in house according to published protocol), DSPC (Avanti Polar Lipids), cholesterol (Sigma-Aldrich), and DMG-PEG2000 (Avanti Polar Lipids) at a molar ratio of 60:10:38.5:1.5.^47,54 Cre mRNA was prepared in 10 mM sodium citrate, pH 4. The molar ratio of ionizable lipid to mRNA was 5:1. Total lipid and mRNA solutions at 3:1 volumetric ratio were combined by microfluidic mixing using the NanoAssemblr Ignite (Precision NanoSystems, Cytiva) at a 12 mL min⁻¹ total flow rate. The collected mRNA-LNP sample was immediately diluted in an equal volume of 1× PBS and transferred into a Slide-a-lyzer dialysis cassette 20 kDa MWCO (Thermo Fisher Scientific) for overnight dialysis in 1× PBS at 4°C. LNP solution was filtered using a 0.22 μm syringe filter and concentrated with an Amicon 100 kDa MWCO centrifugal filter. Particle size and polydispersity index were characterized by DLS (Zetasizer, Malvern Panalytical). Encapsulated mRNA concentration was quantified using Quant-it RiboGreen RNA Assay kit (Invitrogen). LNP samples were kept at 4°C until injection within 72 h of formulation. Quality assessment data for all LNPs can be found in Figure S6.

Animal experiments

All swine procedures involving animals were conducted in compliance with Recombinetics institutional animal care and use committee (IACUC) and all animals euthanized were done so in accordance with American Veterinary Medical Association-approved methods. For systemic vector delivery experiments, piglets were anesthetically induced with intramuscular (i.m.) injection of Telazol/Ketamine/Xylazine (TKX). Anesthetic depth was verified and vitals (electrocardiogram, heart rate, temperature, and respiratory rate) monitored every 10–15 min during the procedure. Piglets were placed in dorsal recumbency and the injection site sterilely prepped. A 1–2 cm incision was made to isolate the target vein, where a 1.5 inch i.v. catheter was placed. The test article was administered over 30 s. The site of surgical cut-down was closed, and the animals were taken to post-operative care for observation. Upon full recovery, animals were returned to study-specific group housing for in-life phases of the project.

For the AAV study, AAV9 harboring a CMV enhancer and a chicken β-actin promoter expressing Cre recombinase was purchased from the University of Massachusetts, Vector Core (Construct AAV9.CB-PI-Cre). A litter of SRM-1 piglets was screened for neutralizing antibodies (Nab) against AAV9 in serum by the University of Pennsylvania, Immunology Core. A 23-day-old 8.8 kg male SRM-1 piglet with no detectable anti-AAV9 Nabs was injected with the AAV9 vector at a dose of 7.27 × 10¹² genome copies, per kilogram body weight. A full necropsy was performed 16 days post injection and tissues were both snap frozen and fixed in 10% neutral-buffered formalin overnight. Fixed tissues were washed the following day in 1× PBS and transferred to 70% ethanol until further processing.

For the LNP systemic study, three 20-day-old SRM-1 piglets were injected with 6 mL LNP carrying Cre mRNA at a dose of 0.4 mg mRNA per kilogram body weight. One age-matched SRM-1 piglet underwent the same surgical procedure but was injected with 6 mL of sterile saline. Blood draws (serum and EDTA-whole blood) were performed on days 0, 1, 3, 7, and 14 on these animals as well as two sibling wild type piglets for a total of 6 samples. Samples were shipped to IDEXX Laboratories for serum chemistry and blood counts, and assay reference ranges were provided by the company. A full necropsy was performed 14 days post-injection and tissues were snap frozen and fixed in 4% paraformaldehyde (PFA) for further processing.

For the LNP i.c.v. delivery, a 3-day-old piglet was placed under anesthesia and into a stereotactic injector frame. The injection site was identified through established landmarks and a Hamilton needle attached to an automated injector were used to infuse a total of dose of 80 μg (2.5 μg kg⁻¹ brain mass) of Cre-mRNA packaged in 100 μL LNP at a rate of 25 μL min⁻¹. After completion, the needle was slowly extracted, and the piglet was recovered and returned to the group housing for in-life phases of the study. A necropsy was performed 19 days after injection to remove the whole brain and fix in 10% NBF for further processing.

Mouse animal procedures for LNP validation were approved by the University of California, Berkeley (UCB) IACUC. All facilities used during the study period were accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International (AAALAC). To determine the in vivo stability and delivery efficiency of LNPs with the same formulation used for the systemic pig study, wild-type BALB/cJ mice (male and female, Jackson Laboratory, stock #000651), aged 10–16 weeks and weighing 18–25 g were used for firefly luciferase mRNA delivery experiments. LNPs were injected into mice at a dose of 0.25 mg mRNA per kg body weight via the retro-orbital route (i.v.), and LNP/mRNA complexes were mixed with PBS pH 7.4 right before injection, resulting in a total volume of 100 μL per mouse. 4–6 h later, a D-luciferin solution in PBS pH 7.4 solution (150 mg kg⁻¹) was injected into the IP cavity of mice. After 10 min, the luciferase luminescence intensity of the whole body and organs (lungs, liver, kidneys, heart, and spleen) was imaged using an IVIS spectrum imager Figure S6. (PerkinElmer).

IHC of tissues

For the AAV systemic study, small sections of tissue (1–3 cm³) were allowed to fix in 10% NBF over a 24 h period before being washed 3 times in PBS, with larger sections (>3 cm³) allowed to fix for 72 h, and the whole intact brain allowed to fix for 7 days. Tissues were sent to Scientific Solutions Inc. pathology service for paraffin embedding, sectioning, and staining using their established methods. A rabbit anti-red fluorescent protein (RFP) antibody (Rockland 600-401-379) and a mouse anti-NIS (Abcam ab242007) were used to detect tdTomato and NIS, and detection of positive staining was done using a dual-stain system (ImmPRESS Duet Double Staining Polymer Kit, Vector Laboratories) with alkaline phosphatase enzyme converting ImmPACT Vector Red substrate (red in color) for tdTomato staining and horseradish peroxidase converting ImmPACT DAB (3,3′-diaminobenzidine) EqV substrate (brown in color) for NIS staining, or by single staining of ImmPACT Vector Red substrate or ImmPACT DAB. The resulting stained tissues were scanned with a Leica Aperio CS2 whole slide scanner. Aperio ImageScope software was used to visualize and analyze the resulting digital images.

Immunofluorescence imaging of testis tissues

Testis tissues collected from the AAV9-injected animal was analyzed by immunofluorescence using previously published methods.⁵⁵ In short, formalin-fixed paraffin-embedded samples were sectioned into 5 μm slices and placed on positively charged glass slides. Sections were deparaffinized with xylene and rehydrated, followed by antigen retrieval by boiling in citrate buffer and permeabilized in cold methanol followed by a Triton X- solution. Blocking was performed before primary antibody administration. tdTomato was stained with RFP antibody (1:200 dilution in PBS, Rockland 600-401-379S) and aromatase/CYP19A1 was stained with an aromatase antibody (1:100 dilution in PBS, Novus Biologicals NBP3-07826-20 μg). Following overnight incubation of the primary antibody, slides were washed and incubated with secondary antibodies. For tdTomato, an Alexa Fluor 596 donkey anti-mouse secondary was used (1:400 dilution in PBS, Jackson ImmunoResearch Laboratories 711-585-152) and for aromatase, an Alexa Fluor 488 donkey anti-rabbit secondary was used (1:400 dilution in PBS, Jackson ImmunoResearch Laboratories 714-547-003). Following incubation in the secondary antibody, slides were washed and co-stained with 4′,6-diamidino-2-phenylindole (DAPI) while mounting a coverslip (ProLong Gold antifade reagent with DAPI, Invitrogen). Slides were allowed to dry and images were captured on a Leica DM6000B microscope with an attached Leica DFC7000T microscope camera with GFP, TxRed, and DAPI filters.

Tissue preparation and fluorescent imaging of LNP-injected pigs

After necropsies, small sections of tissue (1–3 cm³) were allowed to fix over a 24 h period in 4% PFA before being washed with PBS, followed by a transition to a 15% then 30% sucrose solution in PBS. Tissues were transitioned between sucrose concentrations once they sunk to the bottom of the tube. Following sucrose treatment, tissues were embedded in Tissue-Tek optimal cutting temperature compound (O.C.T., Sakura Finetek USA) and snap frozen in embedding cassettes with liquid nitrogen-cooled aluminum blocks. Cryosectioning of the OTC-tissue blocks was performed on a Leica CM1860 cryostat, and 5–10 μM sections were placed on positively charged microscope slides. Slides were co-stained with DAPI while mounting the coverslip (ProLong Gold antifade reagent with DAPI, Invitrogen). Images were captured on a Leica DM6000B microscope with an attached Leica DFC7000T microscope camera with TxRed and DAPI filters.

Data and code availability

All data provided in this manuscript are available for sharing by request to the corresponding author.

Acknowledgments

We thank the entire Recombinetics laboratory and animal production staff for their support and advice. We also thank Dr. Ina Dobrinski and Dr. Nathalia de Lima e Martins Lara for their helpful advice and protocols. We thank the core facility staff at the University of Minnesota Flow Cytometry Resource, the University of Massachusetts Viral Vector Core, and the University of Pennsylvania Immunology Core as well as the staff at Scientific Solutions. Work performed at Recombinetics was supported by the Somatic Cell Genome Engineering Consortium grant from the NIH Office of the Director (U24OD026641). We thank Oleg Mirochnitchenko for his advice and guidance. Work performed at the University of Massachusetts was supported by the Rett Syndrome Research Trust. We also thank Adam Hedger for synthesizing the 306-O12B lipid, and Li Li for access to the Zeta Malvern instrument to measure LNP physical characteristics. Figures were designed with BioRender.

Author contributions

J.M.C. and D.M.K. designed and oversaw in vivo studies with the help of D.F.C.; J.M.C., D.M.K., H.D., and R.M. ran and analyzed samples from pig studies; J.M.C. and D.A.W. designed and oversaw cell manufacturing and cell studies with the help of R.O.A.; H.H., S.Z., and N.M. produced and characterized LNP-mRNA for the systemic study; Y.A.H.N. and J.K.W. produced and generated LNP-mRNA for the i.c.v. injection study; N.K. performed testis staining with the help of J.M.C.; J.M.C., D.M.K., and D.F.C. wrote the paper with input and edits from all authors.

Declaration of interests

J.M.C., D.M.K., H.D., D.A.W., R.O.A., and D.M.C. are current employees of Recombinetics, which seeks to commercialize the SRM-1 model. N.M. and H.H. are founders of Opus Biosciences. J.K.W. is a co-founder of Nucyrna Therapeutics and an advisory board member of EnPlusOne, PepGen, and 6-fold.