In Vitro Validation of Transgene Expression in Gene-Edited Pigs Using CRISPR Transcriptional Activators

Abstract

The use of CRISPR-Cas and RNA-guided endonucleases has drastically changed research strategies for understanding and exploiting gene function, particularly for the generation of gene-edited animal models. This has resulted in an explosion in the number of gene-edited species, including highly biomedically relevant pig models. However, even with error-free DNA insertion or deletion, edited genes are occasionally not expressed and/or translated as expected. Therefore, there is a need to validate the expression outcomes gene modifications in vitro before investing in the costly generation of a gene-edited animal. Unfortunately, many gene targets are tissue specific and/or not expressed in cultured primary cells, making validation difficult without generating an animal. In this study, using pigs as a proof of concept, we show that CRISPR-dCas9 transcriptional activators can be used to validate functional transgene insertion in nonexpressing easily cultured cells such as fibroblasts. This is a tool that can be used across disciplines and animal species to save time and resources by verifying expected outcomes of gene edits before generating live animals.

Introduction

The development of CRISPR-based gene editing technologies enabled complex and efficient editing of the genome: it is now possible to add, edit, and/or delete DNA sequences of living cells and gametes of multiple species and phyla. In animals, this has been applied to genetically modified model organisms so as to better understand the function of genes, or to develop animals with improved agricultural production traits.^1–3 Gene editing with tools such as CRISPR, TALENs, and other RNA-guided endonucleases has unlocked the ability to perform germline gene knockouts, knock-ins, and base editing at high efficiency in almost any species.⁴

All RNA-guided endonucleases rely on a site-specific DNA break to initiate genomic repair. This break is repaired by homology-directed repair or nonhomologous end-joining. In gene-editing processes, the homology-directed repair mechanism of the cell can be hijacked by supplying the cell with an alternate repair template containing an altered DNA sequence or inserted gene, resulting in a knock-in. Alternatively, the nonhomologous end-joining approach is applied when the error-prone cellular machinery repairs the DNA with bases inserted or missing, which usually causes frameshift mutations thought to result in gene knockouts.⁵

Although in theory, insertion-deletion edits cause frameshift mutations and a premature stop codon,⁶ many laboratories have reported of alternate downstream start codons, active truncated variants, alternative splicing, or in frame exon-skipping that interfere with the complete knockout phenotype depending on the location of the indel.^7–17 This results in unwanted and complex phenotypes that are costly and time consuming to identify and resolve.

In large animals such as nonhuman primates, pigs, cattle, and sheep, this represents a large monetary and time investment due to high husbandry costs and long gestation lengths (such as almost 5.5 months in macaques, 4 months in a pig, and 9 months in a cow). There is, therefore, a need to develop a method to test the gene expression profile of a gene-edited organisms before generating offspring or performing experiments. At present, one of the most commonly used methods for generating precise gene knockouts or knock-ins in species such as pigs and cattle is by the use of somatic cell nuclear transfer (SCNT) where somatic cells are modified in vitro before being used for SCNT to generate offspring.^18–20 Having an edited cell line before SCNT provides the opportunity to examine the effect of the DNA edit on gene expression before generating a live animal. Unfortunately, most of the target genes of interest are not expressed in fibroblasts, one of the most commonly used SCNT donors. As proof of concept, we use two DNA-edited pig cell lines designed to express a cell lineage marker (Histone 2B-GFP or cytoplasmic GFP) under the control of a stem cell-specific promoter (LGR5). LGR5 is a marker of stem cells in a variety of tissues, including skin, gastrointestinal track, lung, and others.²¹ SCNT offspring from one line did not express the transgene properly for unknown reasons, requiring us to generate a second line that demonstrated the desired phenotypic change.

In this study, we describe a system using CRISPR-dCas9 transcriptional activators (TAs) as a method to transcriptionally activate transgene expression in a nonexpressing cell type. Our results show that TAs used in vitro can recapitulate the transgene expression profile of the gene of interest from in vivo tissue of transgenic offspring. Based on this, we propose that TAs could be a useful screening tool to confirmed desired expression profile of an intentional genomic alteration, such as a knockout or knock-in, before generation of a gene-edited animal, which is especially useful for laboratories investing time and resources on generating gene-edited large animals.

Materials and Methods

Animal welfare

This study was performed within strict accordance of the Institutional Animal Care and Use Committee at North Carolina State University (approved IACUC protocols 14-067-B and 17-082-B). The animals used in this study were generated by SCNT using surrogate gilts from the university herd. Euthanasia was performed by sedation followed by intravenous injection of sodium pentobarbital, which meets the recommended guidelines of the American Veterinary Medical Association for porcine euthanasia.

Generation of gene-edited pigs by SCNT

Two gene-edited pig models were generated. For model I, IRES-GFP, TALENs were designed to elicit a double-stranded break in the 3′ UTR of the porcine LGR5 gene (TALEN sites: 5′ TATAATTTGTTCCGCTAC and 3′ AAATCCGAATGGACTTAG), roughly 20 bp downstream from the stop codon. A homology directed repair template containing the internal ribosomal entry site and green fluorescent protein (IRES-GFP) sequence and regions of homology surrounding the cut site was used to mediate gene knock-in (Fig. 1A).

FIG. 1.

GFP detection in model I and model II. GFP is not detected in adult stem cell populations in IRES-GFP model. (A) Schematic of knock-in strategy for model I. Cryosections of intestine (B, C) or skin (D, E) were evaluated for GFP expression (B, D) or co-stained with DAPI (C, E). GFP is expressed in expected adult stem cell population in H2B-GFP model. (F) Schematic of knock-in strategy for model II. Cryosections sections of intestine (G, H) or skin (I, J) were evaluated for natural GFP expression (G, I) or co-stained with DAPI (H, J). Scale bar indicates 100 μm. H2B-GFP, histone 2b and green fluorescent protein; IRES-GFP, internal ribosomal entry site and green fluorescent protein.

For model II, H2B-GFP, CRISPR-Cas9 nuclease was used to create a double-stranded break in the genomic in exon 1 of the porcine LGR5 gene (CRISPR target: ACCATGGACACCTCCTCGGT). A homology-directed repair template plasmid containing histone 2b and green fluorescent protein (H2B-GFP) flanked by 1000 bp homology arms flanking the cut site was co-transfected with the Cas9 (Addgene #72247) and gRNA (Addgene #43860) plasmids, and cells were seeded at low density for colony outgrowth (Fig. 1F).

Pig fetal fibroblasts isolated from embryonic day 42 fetuses were used for gene editing and SCNT. After transfection and single cell cloning, colonies were genotyped by polymerase chain reaction (PCR) and sequencing to verify successful targeted transgene integration before SCNT. SCNT was completed as previously described²² and zygotes were surgically transferred into a surrogate and carried until term.

IHC and cell analysis of SCNT offspring

Skin and intestine were isolated immediately after euthanasia and fixed overnight in 4% paraformaldehyde. Tissues were dehydrated using 10%, 20%, and 30% sucrose solutions and then embedded in optimal cutting temperature compound. Cryosections were stained and mounted with Prolong Antifade Mountant with DAPI (ThermoFisher) and analyzed by fluorescent microscopy (Olympus).

TA gRNA design

Within the 1000 bp region directly upstream of the LGR5 transcriptional start site, 5 gRNA were designed using Benchling Biology Software, targeting 20 nucleotides on either the forward or reverse strand with the protospacer adjacent motif (PAM) sequence of NGG. CRISPR target sites include guide 1: ACAAGATTTGCTCCTCACTG, guide 2: GTCCCGCATTGTTCTACTAG, guide 3: CTCCAATGCTGTCTAACCCA, guide 4: CGGACACAAGCAGACGCACA, and guide 5: TTTCTCCACTCCGCGGCTGG. Complementary oligonucleotides were ordered from IDT, annealed, and cloned into plasmid MLM3636 (Gift from Keith Joung; Addgene #43860).²³ In brief, 1 μg of MLM3636 backbone was digested with BsmbI enzyme. Complementary oligos with 4 bp overhangs matching overhangs from digested backbone were annealed for a 10 μM concentration, combined with 20–50 ng of backbone, and incubated overnight at 16°C with T4 DNA Ligase (NEB). Product was subsequently cloned in NEB-5α competent cells, cultured, and plasmids were extracted and sequenced to verify proper sequence.

Cell culture and transfection

Pig fibroblasts were cultured at 37°C and 5% CO₂ and passaged with 0.05% Trypsin/EDTA at 80% confluence. Culture media consisted of Dulbecco's Modified Eagle Medium (Gibco) supplemented with 15% fetal calf serum (Gemini) and 1% penicillin/streptomycin. Cells were transfected by nucleofection (Lonza) using the U-023 setting. For each transfection, 5 × 10⁵ cells of each group were transfected with 1 μg total of DNA: 500 ng of 5 gRNA plasmids (100 ng each) generated as described earlier, in combination with 500 ng VP64-dCas9 (Gift from Keith Joung; Addgene #47107). As a positive control, 1 μg of pmaxGFP (Lonza) was transfected for cytoplasmic GFP expression. Detection of GFP was analyzed by epifluorescent microscopy (Olympus) followed by flow cytometry; at least 10,000 events were recorded for each treatment (Cytoflex; Beckman Coulter).

Reverse transcription quantitative polymerase chain reaction analysis

RNA was extracted from fresh cell pellets with Quick-RNA Microprep Kit (Zymo). cDNA was synthesized (AffinityScript Multiple Temperature cDNA Synthesis Kit) according to the manufacturer's instruction. For quantitative polymerase chain reaction (qPCR) amplification, cDNA and primers were added to SYBR green mastermix (Bio-Rad) and amplified according to the manufacturer's instructions using a two-step amplification, denaturing at 98°C and annealing/extending at 60°C. Listed from 5′ to 3′, primers for LGR5 were forward: CCTTGGCCCTGAACAAAATA, reverse: ATTTCTTTCCCAGGGAGTGG. For GAPDH forward: ATCCTGGGCTACACTGAGGAC, reverse: AAGTGGTCGTTGAGGGCAATG. All reactions were performed in duplicate. Amplification was quantified using a delta-delta CT analysis, normalizing to GAPDH and nontransfected wild-type cells.

Statistical analysis

All results are shown as mean with standard error. Statistical analysis was performed using JMP Pro 14 (SAS Institute). When appropriate, a one-way paired Student's t-test was used to test for significance.

Results

Model I: 3′-LGR5-IRES-GFP. GFP is not detected in cells/tissues from gene-edited pig

We first attempted to tag cells that express LGR5 with a cytoplasmic GFP marker. We began by integrating a DNA construct containing an IRES-GFP into the 3′ untranslated region of the endogenous pig LGR5 locus, directly downstream of the final exon. To do this, we designed TALE Nucleases to create a double-stranded break in the DNA at the desired locus, and simultaneously transfected the TALENs with a homology directed repair template into pig fetal fibroblasts (Fig. 1A). Colonies selected and screened by PCR and sequencing analysis showed the successful insertion of the transgene (Supplementary Fig. S1A), and cells were subsequently used for SCNT. However, GFP fluorescence was not detected in the expected locations of fluorescence, including the skin and gut of tissue from piglets harboring the transgene (Fig. 1B, E).

Model II: 5′-LGR5-H2B-GFP. H2B-GFP is detected in cells/tissues from gene-edited pig

Given the lack of desired phenotype and GFP expression from the first model, we decided to tag LGR5-expressing cells with a pig H2B-GFP fusion protein by introducing an H2B-GFP sequence, without the IRES, into one of the LGR5 alleles. To do this, we targeted CRISPR-Cas9 to the LGR5 locus immediately downstream of the start codon, and co-transfected with a plasmid containing the H2B-GFP coding sequence and a bovine polyA signal using a homology-directed repair template (Fig. 1F). Cells containing the desired gene edits were then screened by PCR (Supplementary Fig. S1B) and used for SCNT. Tissues from piglets born from the second genotype were then analyzed for transgene expression. Unlike model I, the phenotype in model II piglets accurately reflected the genotype and H2B-GFP was detected in tissues that express LGR5 (Fig. 1F, J). Furthermore, the LGR5 expression correlated with the GFP mean fluorescence intensity (Fig. 2A, B) and, therefore, we concluded that the phenotype accurately reflected the intended outcome for these pigs.

FIG. 2.

LGR5 mRNA expression in model II cells sorted by GFP expression and in all genotypes after transcriptional activation. (A) GFP protein expression correlates with LGR5 mRNA expression. Epidermal cells extracted from model II—H2B-GFP were isolated and sorted based on GFP expression. (B) RT-qPCR analysis of relative LGR5 expression from cells sorted in (A) indicates that GFP-hi cells from the model II—H2B-GFP are significantly enriched for LGR5 expression, whereas GFP-neg cells do not express LGR5. Data representative of mean of two biological replicates performed in technical duplicate. (C) Schematic of LGR5-targeting TA design. (D) RT-qPCR analysis shows that LGR5 expression is significantly upregulated in cell lines from model I, model II and unmodified upregulate after transfection with TAs (n = 3). A delta-delta CT analysis was used to normalize LGR5 expression to GAPDH relative to nontransfected wild-type cells. ***Indicates p < 0.05. Error bars indicate SEM. RT-qPCR, reverse transcription quantitative polymerase chain reaction; SEM, standard error of the mean; TAs, transcriptional activators.

Design and test of TAs

Based on this and previous experiences by us and others,^7–17 we aspired to generate a method for testing the RNA and protein expression outcomes of a DNA edit before spending the resources to generate the piglets. Because LGR5 is not expressed in cells such as fibroblasts cultured in vitro, we were not able to validate the desired phenotype until we had already invested considerable time and resources into generating and validating the pigs by SCNT. This led us to ask whether it would be possible to test the phenotype in cells before SCNT by upregulating transgene expression using a catalytically inactive dCas9 fused to the TA VP64 for targeted gene upregulation. We subsequently designed five gRNA targeted to the 1 kb region directly upstream of the transcriptional start site (Fig. 2C). All five gRNA were encoded in plasmids and co-transfected with a plasmid encoding dCas9-VP64 into fibroblasts from wild-type, model I, and model II genotypes. Cell lines showed significant upregulation of LGR5 mRNA after transfection with the TAs as detected by qPCR (Fig. 2D).

GFP expression profile using TAs

After verifying that the transcription was upregulated, we next evaluated the GFP phenotype of LGR5 expressing cells for each genotype in vitro after transfection with dCas9-VP64 TAs. First, we examined the presence of GFP by fluorescent microscopy before and after transcriptional activation (Fig. 3A). As shown, after TA nucleofection, the wild-type and the model I cells did not contain detectable GFP, whereas the model II cells expressed nuclear GFP as expected. We then quantified this upregulation by flow cytometry and showed that GFP fluorescence was not detected in any cell line without TA treatment (Fig. 3B). After transfection with TAs for LGR5, GFP was detected in an average of 35.1% ± (14.3, standard deviation) of the LGR5-H2B-GFP cell line but not in the wild-type or IRES-GFP cell lines (Fig. 3C, Supplementary Fig. S2A). From this we concluded that cells transfected with TAs targeted to the 1 kb region upstream of the LGR5 gene locus in vitro successfully predicted the in vivo GFP expression profile of LGR5-expressing cells.

FIG. 3.

LGR5-targeting TAs upregulate GFP expression in model II—H2B-GFP cells, but not wild-type, model I—IRES-GFP, or nontransfected cells. (A) Fluorescent-brightfield overlay showing nuclear GFP expression in model II—H2B-GFP cells treated with TA but not in wild-type or model I—IRES-GFP. (B) Flow cytometry analysis of GFP expression in each cell type without TA treatment. (C) Flow cytometry analysis of GFP expression in each cell type treated with LGR5-TAs. Scale bar indicates 100 μm.

TAs can be used to screen pooled cells for presence of transgene

Although our proof-of-concept study has shown that the in vitro phenotype of fibroblast cell lines from gene-edited pigs faithfully recapitulates the in vivo phenotype, we next asked whether this principle could best be applied before the generation of a transgenic animal. To test this, we transfected wild-type fetal fibroblasts with CRISPR nuclease, gRNA, and an HDR template as in the generation of model II. We grew out the transfected cells in a pool of mixed genotypes until recovered, and then re-transfected them with the TAs. Presence of DNA insertions was evaluated by PCR of the pooled cell DNA (Fig. 4B, C), and as shown by the flow cytometry and microscopy analysis, GFP was expressed in about 1% of the population of the cells (Fig. 4A, Supplementary Fig. S2B). This suggests that about 3–4% of the pooled cells contains the knock-in, of which only about 25–30% were upregulated. From these results, we conclude that flow cytometry and TAs can be used to screen a pool of edited cells for cells that contain the transgene.

FIG. 4.

TAs can induce transgene expression in a pool of transfected cells. The LGR5-H2B-GFP knock-in plasmid was transfected with CRISPR-Cas9 as described for the generation of the LGR5-H2B-GFP model. Cells were recovered and expanded for 1 week before pooled cells were transfected with LGR5 TAs and evaluated for H2B-GFP expression by flow cytometry. (A) Detection of a population of GFP+ cells by flow cytometry. (B) PCR analysis of genomic DNA for H2B-GFP transgene inserted into LGR5 locus or the endogenous LGR5 allele around the CRISPR cut-site. Band from primers 1 + 2 shows that population of cells within the pool of edited cells contain the H2B-GFP insert, whereas WT cells do not. (C) Diagram of primers used in (B), amplifying the 5′ junction of the insert. PCR, polymerase chain reaction; WT, wild-type.

Discussion

In this study, we employed CRISPR-based methods for evaluating outcomes of gene editing events. As a proof of concept, we used two different gene-edited genotypes in pigs, which were both generated with the goal of adding a GFP tag to the LGR5 expressing cells. Our results show that model I, IRES-GFP animals did not show any GFP signals in any tissue examined. Although the DNA sequence of the transgene was intact based on PCR (Supplementary Fig. S1A) and sequencing, it is possible that the DNA repair mechanisms that facilitated the transgene integration also elicited mutations or deletions upstream or downstream of the intended target site. It is also possible that the transgene interfered with enhancers or cis-regulatory elements that affected gene expression. Further experiments are necessary to determine why the transgene was not expressed as expected. In contrast, model II, generated from a CRISPR-mediated knock-in of the H2B-GFP gene into the 5′ of the LGR5 gene, showed the expected nuclear GFP signal in LGR5-expressing cells. Using CRISPR-TAs, we were able to distinguish in vitro which genotype yielded the desired GFP signal profile.

Using CRISPR-TAs has advantages over traditional methods of transgene testing, such as cloning the transgene into an expression vector, for several reasons. First, the design and synthesis of gRNAs using a plasmid-based method is straightforward and can be completed much more quickly than the synthesis of an expression construct. Second, by upregulating the gene in its genomic location, it is possible to account for unexpected outcomes; for example, in model I, we detected the presence of the IRES-driven GFP using an expression vector and in gene-edited cells, but, nevertheless, GFP was not detected in the animal or the cells.

Notably, CRISPR-Cas systems are known to have off-target effects, which occur when there are mismatches between the gRNA sequence and the genomic DNA it binds,^24,25 and always must be considered when making changes to DNA sequences. However, since TAs represent a transient state of activation,^26,27 and generally more than one binding site is needed for effective activation, this suggests that there is little to no risk of misinterpreting results due to off-target binding of CRISPR-TAs. Nevertheless, CRISPR-Cas systems will continue to be developed to reduce off-target binding,²⁸ and thus the techniques described in this study can be updated with higher fidelity enzymes as they become available.

The application of CRISPR-TAs in vitro before generating the transgenic animals in this research would have saved our laboratory significant costs in labor, animal fees, time, and resources. As shown by us and others,^7–17 even with known DNA sequences in somatic cells, it is difficult to know the exact phenotypic outcome. For example, alternate start codons, hypomorphic proteins, alternative splicing, or undetected mutations upstream or downstream of the cut site or insert could all contribute to undesired phenotype, representing wasted resources, time, increased number of experimental animals, and misinterpretation of results. The CRISPR-TAs are a tool that all laboratories performing gene editing, especially large animal gene editing, should consider for validations of gene editing outcomes before investing in the creation of the animal. In Figure 5, we have summarized the process by which gene-editing strategies and outcomes can be efficiently and effectively analyzed for knockout or knock-in gene edits, using SCNT or microinjection strategies. Although microinjection does not use somatic cells from culture, the same knockout strategy is often applied in somatic cells to optimize CRISPR cutting efficiency and targeting strategy,^29–31 which yield the same edited sequence as the target of microinjection. In this case, those cells could be used with TAs to ensure that the modified sequence elicits gene expression as desired before proceeding with microinjection and embryo transfer. In gene knock-ins, this analysis includes proper synthesis, function, and localization of a protein. For CRISPR-based knockouts, this system can be used to ensure the absence of the protein and the nonsense-mediated decay of mRNA.¹⁷ Although our study shows knock-in detection from a pool, analysis of knockouts may require clonal expansion before analysis.

FIG. 5.

CRISPR-TAs applied to gene-edited cells in vitro can validate intended phenotype of gene edits before generation of offspring. Suggested process for applying TAs: (1) Gene editing by site-specific RNA-guided endonucleases. (2) Transfect gene-edited cells with TAs and evaluate for desired outcomes. (3) Generate gene editing offspring using the cells or same editing strategy applied previously. (4) Proceed with experiments as planned.

Furthermore, our results suggest that this novel method can be used to save time by enriching for cells that contain a properly inserted transgene from a pool of edited cells. Although this proof-of-concept study used a fluorescent marker, other methods can be used to screen for proper protein expression, such as antibody-based flow cytometry. In addition, TAs can also be used with cultured cells to develop controls for Western blot or flow cytometry or for testing target-specific reverse transcription quantitative polymerase chain reaction analysis (qRT-PCR) probes and primers.

Conclusion

In conclusion, CRISPR-Cas9 and RNA-guided endonuclease-mediated gene edits often elicit unpredictable outcomes, and in this research we applied CRISPR-TAs as a tool to evaluate outcomes in vitro. As proof of concept, we demonstrated that this technique can be successfully used to validate transgene expression in nonexpressing cell types, and that it could be coupled with flow sorting to enrich a pool of cells containing the desired genotype. We suggest that this technology presents animal researchers with an opportunity for in vitro validation of the expected expression profile of site-specific DNA modifications before the generation of live gene-edited animals.

Author Confirmation Statement

K.P. and J.P. designed experiments. K.P., J.C., S.G., and K.G. performed experiments. K.P. and J.P. interpreted results and wrote the article. All co-authors have reviewed and approved this manuscript. This manuscript has been submitted solely to this journal and is not published, in press, or submitted elsewhere.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

We gratefully acknowledge funding from National Institutes of Health R21OD019738 and R01OD023138 to J.P.

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2