A LGR5 reporter pig model closely resembles human intestine for improved study of stem cells in disease

Abstract

Intestinal epithelial stem cells (ISCs) are responsible for intestinal epithelial barrier renewal; thereby, ISCs play a critical role in intestinal pathophysiology research. While transgenic ISC reporter mice are available, advanced translational studies lack a large animal model. This study validates ISC isolation in a new porcine Leucine Rich Repeat Containing G Protein‐Coupled Receptor 5 (LGR5) reporter line and demonstrates the use of these pigs as a novel colorectal cancer (CRC) model. We applied histology, immunofluorescence, fluorescence‐activated cell sorting, flow cytometry, gene expression quantification, and 3D organoid cultures to whole tissue and single cells from the duodenum, jejunum, ileum, and colon of LGR5‐H2B‐GFP and wild‐type pigs. Ileum and colon LGR5‐H2B‐GFP, healthy human, and murine biopsies were compared by mRNA fluorescent in situ hybridization (FISH). To model CRC, adenomatous polyposis coli (APC) mutation was induced by CRISPR/Cas9 editing in porcine LGR5‐H2B‐GFP colonoids. Crypt‐base, green fluorescent protein (GFP) expressing cells co‐localized with ISC biomarkers. LGR5‐H2B‐GFP^hi cells had significantly higher LGR5 expression (p < .01) and enteroid forming efficiency (p < .0001) compared with LGR5‐H2B‐GFP^med/lo/neg cells. Using FISH, similar LGR5, OLFM4, HOPX, LYZ, and SOX9 expression was identified between human and LGR5‐H2B‐GFP pig crypt‐base cells. LGR5‐H2B‐GFP/APC ^null colonoids had cystic growth in WNT/R‐spondin‐depleted media and significantly upregulated WNT/β‐catenin target gene expression (p < .05). LGR5⁺ ISCs are reproducibly isolated in LGR5‐H2B‐GFP pigs and used to model CRC in an organoid platform. The known anatomical and physiologic similarities between pig and human, and those shown by crypt‐base FISH, underscore the significance of this novel LGR5‐H2B‐GFP pig to translational ISC research.

In this study, a new porcine transgenic model to identify ISCs is validated as a novel, translational model of gastrointestinal physiology. This model more closely resembles humans and is used here as a platform to study early CRC. This significant advancement in transgenic large animal model development will help researchers overcome the limitations of murine models and advance the study of the intestinal epithelium in health and disease. Created with BioRender.com.

Abbreviations

ANOVA

analysis of variance

APC

Adenomatous polyposis coli

Cas9

CRISPR‐associated protein 9

CRISPR

clustered regularly interspaced short palindromic repeats

CRC

colorectal cancer

FACS

fluorescence‐activated cell sorting

FISH

fluorescent in situ hybridization

GAPDH

Glyceraldehyde 3‐phosphate dehydrogenase

GFP

green fluorescent protein

H2B

histone 2 B

HOPX

homeodomain‐only protein homeobox

ISC

intestinal epithelial stem cell

LGR5

Leucine Rich Repeat Containing G Protein‐Coupled Receptor 5

LYZ

Lysozyme

OLFM4

Olfactomedin 4

qRT‐PCR

quantitative real‐time polymerase chain reaction

SD

standard deviation

SOX9

Sex‐determining region Y box 9

UEA

Ulex Europaeus Agglutinin I

1. INTRODUCTION

The intestine is a complex organ that serves to absorb nutrients while simultaneously preventing the translocation of potentially harmful luminal contents into systemic circulation. In part, these complex intestinal functions are maintained by a constant renewal of the inner‐most epithelial cell layer separating the systemic circulation and gastrointestinal contents. This constant epithelial regeneration is the sole responsibility of intestinal epithelial stem cells (ISCs), which reside deep in the intestinal crypt structures. Intestinal stem cells constantly divide asymmetrically to give rise to all intestinal epithelial cell lineages ¹ , ² ; they are also critical to epithelial repair following acute and chronic injury. ³ , ⁴ , ⁵ However, in the disease state of colorectal cancer (CRC), the second leading cause of U.S. cancer related deaths, ⁶ mutated ISCs play a more deleterious role by contributing to tumor development. ⁷ , ⁸ , ⁹ The most common initiating gene mutation that drives adenoma formation is in the tumor‐suppressor, adenomatous polyposis coli (APC) gene. This APC mutation is present in over 80% of CRCs. ¹⁰ , ¹¹ , ¹² , ¹³ Several groups have shown that mutation of Apc specifically in murine ISCs resulted in large adenoma formation. ⁷ , ⁸ , ¹⁴ Given the multifaceted role of ISCs, it is critical to be able to isolate and understand ISC dynamics in intestinal research. ⁸ , ¹⁵ , ¹⁶

To effectively study ISCs, tools to identify this cell population are necessary. Multiple biomarkers for ISCs have been identified ⁵ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² ; one of the more widely known and used ISC‐specific biomarkers is the transmembrane protein Leucine Rich Repeat Containing G Protein‐Coupled Receptor 5 (Lgr5). ² , ²³ While antibody‐based techniques have had limited success in identifying the discrete Lgr5⁺ ISC population, ²³ , ²⁴ the development of Lgr5 transgenic reporter mice in 2007 allowed for reliable in vivo identification of the Lgr5⁺ stem cells for studies including ISC biology, repair, and tumorigenesis. ² , ⁸ , ¹⁶ , ²⁵

Although transgenic reporter mice are both a beneficial scientific tool and the current standard for ISC research, mice do not fully replicate human physiology, disease, or drug metabolism. ²⁶ , ²⁷ , ²⁸ , ²⁹ Therefore, there is increasing interest in non‐rodent translational models to better replicate and study human gastrointestinal disease. For this purpose, pigs serve as a particularly useful animal model of human gastrointestinal pathophysiology. ³⁰ , ³¹ Pigs, like humans, are monogastric animals with similar intestinal anatomic features including ratio of intestinal length per kilogram of body weight and vascular structure. With an omnivorous diet similar to humans, pigs share numerous parallel physiologic characteristics including comparable metabolic processes, nutrient requirements, analogous mucosal barrier physiology, enteric microbiota, immune cell populations, and susceptibility to enteric pathogens. ³² , ³³ , ³⁴ , ³⁵ , ³⁶ These similarities between humans and pigs make the study of intestinal diseases including ischemia–reperfusion, necrotizing enterocolitis, acute and chronic inflammation, and short bowel syndrome better represented by a porcine model in basic science translational research. ⁴ , ³⁰ , ³¹ , ³⁷ , ³⁸ , ³⁹ Such porcine models better replicate human disease and overcome some of the limitations of small rodent models.

Thus far, one disadvantage of porcine models has been the lack of a transgenic model specifically developed to study ISCs in both homeostasis and disease. To remedy this gap in ISC research, our group validated a new transgenic porcine model, LGR5‐H2B‐GFP, as a novel means to identify LGR5⁺ ISCs for both in vivo and ex vivo studies. Here, we defined the LGR5‐H2B‐GFP^hi cells within the intestinal epithelium as the LGR5⁺ ISC population, thus validating the LGR5‐H2B‐GFP porcine model as the first of its kind. This will provide improved ways to specifically study ISCs in a non‐murine species with gastrointestinal physiology that more closely resembles humans. Furthermore, we demonstrated that APC disruption via CRISPR‐Cas9 editing in LGR5‐H2B‐GFP colonoids led to cystic, clonal organoid growth with constitutive WNT pathway activation. This colonoid platform demonstrates the novel and translational utility of this ISC reporter porcine model for CRC. Overall, this novel porcine model provides critical advancement to the field of translational gastrointestinal research.

2. MATERIALS AND METHODS

All authors had access to the study data and had reviewed and approved the final manuscript.

2.1. Porcine models

Healthy, crossbred LGR5‐H2B‐GFP littermate male and female pigs (10–40 weeks) were used for this study. ⁴⁰ , ⁴¹ All LGR5‐H2B‐GFP animals were bred and raised in the animal facility of NCSU CVM. A mix of healthy male and female (10–40 weeks) Yorkshire cross pigs were used as normal control animals (wild type). Both males and females were used for all experiments. The Institutional Animal Care and Use Committee of North Carolina State University approved all animal use (17‐028‐B and 19‐019‐B).

2.2. Human samples

Endoscopic biopsies from adult human ileum and colon deemed normal were collected at Duke University. These samples were blinded for the identity of the individuals. Sample use approved by IRB # 2020–10‐733.

2.3. Murine models

For orthotopic injection of LGR5‐H2B‐GFP/APC ^null colonoids, NOD.Cg‐Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) male mice, approximately 8–12 weeks old were used (Jackson Laboratory; Strain # 005557; RRID:IMSR_JAX:005557). The Institutional Animal Care and Use Committee of Duke University approved all animal use (A234‐21‐12).

2.4. Tissue collection

Pigs were sedated with Telazol/Ketamine/Xylazine (TKX, 0.04 mL/kg IM) prior to euthanasia (60 mg/kg IV pentobarbital sodium). Loops (10–12 cm) of the duodenum, jejunum, ileum, and spiral colon were collected and rinsed with 1× phosphate‐buffered saline (PBS).

2.5. Histology

Full‐thickness sections of the duodenum, jejunum, ileum, and spiral colon were fixed in 10% formalin and washed in 70% ethanol. Paraffin‐embedded samples were sectioned (5–8 μm thickness) and stained with hematoxylin and eosin by the North Carolina State University Histology Core to visualize crypt/villus morphology.

2.6. Tissue immunofluorescence

To identify specific cell types within whole tissue, sections of the intestine were fixed in 4% paraformaldehyde (PFA) (14–18 h; 4°C). Tissue was then transferred to 30% sucrose solution (24 h; 4°C), embedded in optimal cutting temperate (OCT) media, and frozen. Frozen blocks were sectioned at 5–8 μm thickness. To identify endogenous GFP fluorescence, slides were washed with PBS, stained with bisbenzimide H 33258 nuclear stain (1:1000, Sigma Aldrich), and sealed with a coverslip (Prolong Gold Antifade Mount; Thermo Fisher). For each slide evaluated for mosaic expression, only well‐oriented crypts were used to obtain counts (crypt‐base closely opposed to the muscularis mucosa layer and that extended and opened fully into the gut lumen). At least 100 total well‐oriented crypts for each section of intestine per pig were counted and the percentage GFP negative crypts were calculated.

To co‐localize ISC protein biomarkers SOX9 and KI67 with GFP expression in multiplex immunofluorescence assay, fixed‐frozen sections of the jejunum and spiral colon underwent heat‐induced epitope retrieval (HIER) by placing slides in reveal decloaker solution (Biocare Medical, Concord) in a Pascal pressure chamber (DakoCytomation). Tissue permeabilization was performed with PBS‐0.3% Triton X‐100. The α‐GFP (rabbit, 1:2000; Abcam catalogue #: ab183734) antibody was diluted in antibody diluent (SignalStain Antibody Diluent; Cell Signaling Technology) and applied. Washes between reagents were in 1 × Tris‐Buffered Saline, 0.1% Tween. Following antibody binding, HRP boost reagent [SignalStain Boost Detection Reagent (HRP, rabbit); Cell Signaling Technology] was applied, followed by a specific fluorophore‐conjugated tyramide signal amplification (TSA) reagent [green fluorescent protein (GFP)]. Slides were boiled in citrate stripping solution (SignalStain Citrate Unmasking Solution; Cell Signaling Technology) and then incubated in blocking reagent (Dako). Additional primary antibodies were applied overnight at 4°C [α‐SOX9 (rabbit, 1:500; Millipore catalogue #:AB5535), α‐KI67 (mouse, 1:250; Dako catalogue #:M7240) diluted in antibody diluent (Dako)]. The host‐specific secondary antibodies (conjugated to Cyan3 and Cy5; Sigma Aldrich) were applied at 1:500 in antibody diluent. Slides were then counterstained with bisbenzimide H 33258 nuclear staining and cover slipped (Hydromount; National Diagnostics).

To co‐localize α‐GFP and α‐UEA, the same procedure as above was followed, incubating first with α‐GFP. The α‐UEA antibody (rabbit, 1:500; Millipore Sigma‐Aldrich catalog #:U4754) diluted in antibody diluent (Dako) was applied overnight at 4°C. The secondary antibody (conjugated to Cyan3; Sigma Aldrich) was diluted 1:500 in antibody diluent and slides were counterstained with bisbenzimide H 33258 nuclear stain before coverslip application.

2.7. Crypt isolation

Intestinal crypt isolation was performed as previously described. ⁴ Briefly, sections of the intestine, with the mucosa exposed, were washed with cold PBS. The segments were then incubated in ethylenediaminetetraacetic acid [EDTA (30 mM)], Y‐27632 (10 mM), dithiothreitol [DTT (1 mM)], and 1× antibiotic‐antimycotic (anti‐anti, Life Technologies; Primocin, InvivoGen) for 30 min at 4°C. Tissue was transferred and incubated in EDTA (30 mM), Y‐27632 (10 mM), and anti‐anti for 10 min at 37°C. During incubation, tissue was intermittently shaken vigorously to mobilize the crypt epithelium. After incubation, tissues were transferred to ice‐cold PBS washes with anti‐anti. Tissue was shaken and transferred to additional washes until crypts were observed with minimal background debris. Crypts were then pelleted and snap‐frozen in liquid nitrogen for total RNA extraction and additional crypts immediately dissociated into single cell for fluorescence‐activated cell sorting (FACS).

2.8. Single‐cell isolation

Crypts from each intestinal segment were independently incubated with Dispase (36 U; Corning, Corning), Y‐27632 (10 mM), and DNase I (300 μg; Alfa Aeser) in 1× Hank's Balanced Salt Solution (HBSS) for 10 min at 37°C. Single cells were washed with 1× HBSS, resuspended in flow cytometry buffer (2% Bovine Serum Albumin in 1× PBS), and passed sequentially through 100 μm, 70 μm, and 40 μm filters. Cells were then stained with Propidium Iodide (BioLegend) for FACS sorting.

2.9. Fluorescence‐activated cell sorting and flow cytometry

FACS was conducted by the NCSU‐CVM Flow Cytometry and Cell Sorting Core using a Beckman Coulter MoFlo XDP. Dead cells were identified by Propidium Iodide. Based on GFP expression, single live cells were gated into Lgr5‐H2B‐GFP^neg and Lgr5‐H2B‐GFP^hi/med/lo. Single cells for culture were collected in cell culture media while cells for gene expression were collected in 1× PBS with 4% Fetal Bovine Serum, pelleted, and snap‐frozen in liquid nitrogen until RNA isolation. Flow cytometry analysis of GFP^pos/neg colonoid single, live, GFP‐expressing cells was performed using Beckman Coulter CytoFLEX. Data were further analyzed using FlowJo (BD Biosciences).

2.10. Real‐time PCR

Total RNA from whole crypt and LGR5‐H2B‐GFP single cells was extracted using Ambion PureLink RNA mini kit (Thermo Fisher Scientific). Yield and quality control of the extracts were determined by an Agilent 2100 BioAnalyzer performed by the North Carolina State University Genomic Sciences Laboratory Core using Agilent Eukaryote Total Pico Series II chips (Agilent Technologies). Samples were used with an RNA Integrity Score >6, and RNA concentration of >100 pg/μL. RNA (1 μg) was converted to cDNA using the iScript cDNA synthesis kit. Expression of intestinal stem cell genes was determined by quantitative real‐time PCR (QuantStudio 6 Flex, Applied Biosystem) using cDNA samples and the iTaq Universal SYBR green Supermix. Primers included GAPDH (F‐ ATCCTGGGCTACACTGAGGAC; R‐ AAGTGGTCGTTGAGGGCAATG), LGR5 (F‐ CCTTGGCCCTGAACAAAATA; R‐ ATTTCTTTCCCAGGGAGTGG), OLFM4 (F‐ GTCAGCAAACCGGCTATTGT; R‐ TGCCTTGGCCATAGGAAATA), ⁴ , ⁴² and eGFP (F‐ AAGGGCATCGACTTCAAGG; R‐ TGCTTGTCGGCCATGATATAG). The ΔΔC_t method was used to measure relative changes in gene expression, and samples were tested in triplicate.

2.11. Single‐cell culture

Sorted LGR5‐H2B‐GFP cells were resuspended directly in growth factor‐reduced Matrigel (BD Bioscience) supplemented with 50 ng/mL recombinant human Epidermal Growth Factor (Life Technologies), 10 μM Y‐27632, 10 μM SB‐202190 (Sigma‐Aldrich), 500 nM LY‐2157299 (Selleck Chemicals), 10 nM Gastrin (Sigma‐Aldrich, St. Louis), 1 mM Nicotinamide (Sigma‐Aldrich), 500 nM A‐83‐01 (Tocris Bioscience), and 2.5 μM Glycogen Synthase Kinase 3 inhibitor (GSK3i, CHIR99021, Tocaris Bioscience). Approximately 4000 cells were plated in 50 μL of Matrigel. Each well was then overlaid with a total of 500 mL of conditioned media: 250 mL of recombinant human noggin, recombinant human R‐spondin, recombinant wingless/integrated 3a (Wnt‐3a) (L‐WRN) media with an additional 250 mL of culture media [advanced DMEM‐F‐12 (Thermo Fisher Scientific) supplemented with 1× N‐2 supplement (Life Technologies), 1× B‐27 supplement minus vitamin A (Life Technologies), 1× Glutamax (Life Technologies), 1× anti‐anti, and 1 mM HEPES buffer (Life Technologies)]. The entire volume of growth factors and conditioned media were changed 48 h after plating and subsequent 48 h intervals.

2.12. Whole crypt culture

The pelleted whole crypts from epithelial crypt dissociation were resuspended directly in growth factor‐reduced Matrigel at a concentration of approximately 50 crypts per 50 μL of Matrigel. The crypts were cultured in the same conditioned media and growth factors as the single cells.

2.13. Organoid isolation, RNA extraction, and cDNA conversion

Organoids were isolated from Matrigel using Corning Cell Recovery Solution as per the manufacturer's instructions. Organoids were pelleted, snap‐frozen in liquid nitrogen, and stored at −80°C. RNA was extracted using the Ambion PureLink RNA mini kit (Thermo Fisher Scientific). RNA quality control and quantification were performed with the Agilent 2100 Bioanalyzer as described above. Reverse transcription and was performed as previously described for cDNA derived from crypts. A PCR reaction for each sample was performed using the following primers: LYZ (F‐ GGTCTATGATCGGTGCGAGT; R‐ AACTGCTTTGGGTGTCTTGC), MUC2 (F‐ GGCTGCTCATTGAGAGGAGT; R‐ ATGTTCCCGAACTCCAAGG), ITF (F‐ TCGGTTCCCCAGAACCTGCCC; R‐ CGGGGATGCTGGAGTCGAAGC), CCK (F‐ CAAAAGGTAGACGGCGAGTC; R‐ GCGGGGTCTTCTAGGAGGTA), CGA (F‐ GACCTCGCTCTCCAAGGAGCCA; R‐ TGTGCGCCTGGGCGTTTCTT), SGLT1 (F‐ GCAGCTGTCTTCCTCTTGC; R‐ GCAAACTCGGTAATCATACGG), and FABP (F‐ CCGGCAAATACCAAGTACAGAGCC; R‐ CCTTCTCCCCAGTCAGGGTCTCC) ⁴² and iTaq Universal SYBR Green Supermix (BioRad) in an iCycler thermal cycler. To visualize the products of PCR reaction, a 1.5% agarose gel was prepared from heating agarose (Sigma Aldrich) and 1X TAE buffer (Thermo Fisher Scientific). SYBER safe DNA gel stain (10 μL, Invitrogen) was added to the melted agarose. Once the gel was set, PCR samples and ladder were mixed with DNA loading dye, loaded into lanes, and allowed to run until the DNA migrated ¾ of the length of the gel. The gel was imaged using ChemiDoc‐It 2 (Ultra‐Violet Products, Upland).

2.14. APC editing in colonoids

The APC guide plasmid for CRISPR was designed using Benchling (San Francisco, CA), the reference genome Sscrofa11.1, and created following the Joung Lab Cloning Protocol, and MLM3636 (Addgene plasmid # 43860; http://n2t.net/10ddgene:43860; RRID:Addgene 43 860) was a gift from Keith Joung. Once the guide plasmid was constructed, it was transfected into porcine fetal fibroblasts with the Cas9 plasmid VP12 (Addgene plasmid # 72247; http://n2t.net/10ddgene:72247; RRID:Addgene 72 247) ⁴³ using the Nucleofector 2b device (Lonza) setting U‐023. Post‐transfection, a PCR product of transfected cells was used to determine the editing efficiency by tracking of indels by decomposition (TIDE). ⁴⁴ This optimized APC guide (sequence 5′‐gtgttgttgggaaattcccg‐3′) for exon 9 was ordered as a sgRNA from Synthego.

To develop the LGR5‐H2B‐GFP/APC^null colonoids, LGR5‐H2B‐GFP colon crypts were isolated and plated in Matrigel with conditioned media as described above. The resulting colonoids were passaged using the cell recovery protocol described previously and dissociated to single cells by incubation in TrypLE (Thermo Fisher Scientific) and Y‐27632 (1:1000). Single colonoid cells were then aliquoted to 250 000 cells per nucleofection. For this process, CRISPR‐Cas9 was assembled as ribonucleoprotein (RNP) complex by combining 5 ug of sgRNA and 5 ug of Cas9 protein in 100 ul of nucleofector solution and incubated at room temperature for 20 min. Human Stem Cell Solution 1 within the Human Stem Cell Nucleofector™ Starter Kit (Lonza) was used for nucelofection. The aliquoted single LGR5‐H2B‐GFP colon cells were pelleted and resuspended with the RNP complex and nucleofector solution by gently pipetting, transferred into a cuvette, and then placed into the nucleofector device for nucleofection. Once the nucleofection was complete, cells were washed with culture media and plated at a density of 5000 cells per 50 μL Matrigel with conditioned media for 7 days to permit for Cas9‐mediated editing of APC. To select for APC edited cell population, LGR5‐H2B‐GFP/APC ^null colonoids were passaged as single cells and grown in growth factor reduced media without Wnt‐3a, R‐spondin, or CHIR. Knockout efficiency of APC immediately following nucleofection and after growth in selective media was determined through TIDE analysis. For gene expression, LGR5‐H2B‐GFP/APC ^null colonoids were passaged at least twice in selective media and then processed for RNA isolation and qRT‐PCR as described above. Primers for AXIN2 were F‐ ATTGTTGTTTCCCCGCACTC and R‐ CTGCTCATGGTGAGGGAGTT; CCND1 F‐ GCATCTACACCGACAACTCC and R‐ TGATCTGTTTGTTCTCCTCG; WNT3a GCGACTTCCTCAAGGACAAG and R‐ GGTCACGTGTACCGAAGGAT.

2.15. Murine colonoscopy and injection

Optical colonoscopy was performed as previously described. ⁴⁵ , ⁴⁶ For mucosal injections, LGR5‐H2B‐GFP/APC ^null colonoids, after two passages in selective media and grown for 72 h prior to injection, were suspended in 1:1 culture media/Matrigel at a concentration of 50 organoids/μL. Colonoids were delivered beneath the colon mucosa of 8–12 week old male recipient NOD.Cg‐Prkdc^scidIl2rg^tm1Wjl/SzJ (NSG) mice (JAX #005557) mice by optical colonoscopy as previously described using a custom injection needle (Hamilton Inc., 33 gauge, small Hub RN NDL, 16 inches long, point 4, 45‐degree bevel, part number 7803–05), a syringe (Hamilton Inc., part number 7656–01), a transfer needle (Hamilton Inc., part number 7770–02), and a colonoscope with integrated working channel (Karl Storz, 61 029 C). ⁴⁵ , ⁴⁶ For each mouse, 2–3 injections of 50–70 μL were performed as previously described. ⁴⁵ Mice underwent colonoscopy 1–2 weeks following injection to assess tumor formation and were followed weekly by colonoscopy. Mice were euthanized between 3–5 weeks post injection. After euthanasia, the distal 3 cm colon was collected and fixed in 10% formalin.

2.16. Multiplex RNA FISH

Fluorescent in situ hybridization to detect mRNA targets (RNAscope, Advanced Cell Diagnostics) was performed in accordance with the manufacturer's instructions. Briefly, formalin‐fixed paraffin embedded intestinal tissue from human, mouse, or pig was sectioned at 7 μm. Slides were deparaffinized with xylene, then heat‐induced epitope retrieval treated followed by protease digestion. The tissue was hybridized with the species‐specific probes targeting LGR5, HOPX, SOX9, LYZ, and OLFM4 mRNA or a positive control probe peptidylprolyl isomerase B (PPIB) (Advanced Cell Diagnostics), followed by chromogenic or Tyramide Signal Amplification technology development per user manual. Porcine tissue slides were then incubated in primary antibody α‐GFP (abcam catalogue # ab183734), diluted (1:50) followed by secondary antibody (Alexa Fluor 488). Slides were washed, then mounted with Prolong Gold Antifade Mount with DAPI (Thermo Fisher) and imaged by confocal microscopy.

2.17. Image acquisition

Images were captured on an inverted fluorescence microscope (Olympus IX83, Tokyo) fitted with a monochrome digital camera (ORCA‐flash 4.0, Hamamatsu) and color camera (DP26; Olympus). Additional images were also captured on a Fluoview FV3000 upright confocal microscope (Olympus) and on a Zeiss LSM 880 microscope.

2.18. Crypt/villus and fluorescent intensity measurements

Crypt depth and villus length were measured using the line tool in ImageJ (NIH). The intensity of GFP and bisbenzimide H 33342 fluorescence was also measured by the measure tool in ImageJ.

2.19. Statistics

All statistical analyses were performed using Prism software (GraphPad Software, La Jolla). Outliers were identified by ROUT method and a Shapiro–Wilk normality test was performed on raw data, followed by either student's t‐test or two‐way ANOVA with Tukey's multiple comparisons. Data are reported as means with standard deviations. The α‐level for statistical significance was defined as p < .05.

3. RESULTS

3.1. LGR5‐H2B‐GFP porcine model generation

The LGR5‐H2B‐GFP pig model was developed using CRISPR/Cas9 mediated gene knock‐in to insert a fused GFP and histone 2B (H2B) transgene into one allele of the LGR5 locus, under the control of the LGR5 promoter (Figure 1). ⁴⁰ , ⁴¹ Thus, any cells that express LGR5 also produce intranuclear GFP, allowing for enhanced signal resolution in contrast to the cytoplasmic GFP produced in current mouse models.

FIGURE 1.

Transgenic approach for generating the LGR5‐H2B‐GFP pig. Schematic representation of the H2B‐GFP insert into the targeted LGR5 locus via CRISPR/Cas9.

3.2. LGR5‐H2B‐GFP pigs exhibit normal intestinal architecture

To demonstrate normal intestinal architecture of LGR5‐H2B‐GFP transgenic pigs, histologic sections of the duodenum, jejunum, ileum, and ascending colon biopsies from the transgenic animals were compared with the same sections from wild‐type control pig biopsies (Figure 2A). LGR5‐H2B‐GFP pigs showed no differences in villus height or crypt depth compared with wild‐type animals (Figure 2B), suggesting there were likely no major intestinal mucosal architecture differences within the transgenic pigs. Next, we wanted to further define the specific location, phenotype, and function of LGR5‐H2B‐GFP^hi cells within the intestinal mucosal epithelium to show that they are consistent with known ISC‐identifying features in other species.

FIGURE 2.

LGR5‐H2B‐GFP pig intestinal architecture does not differ from wild‐type pigs. (A) Representative hematoxylin and eosin staining of duodenum, jejunum, ileum, and ascending colon from wild‐type and LGR5‐H2B‐GFP pigs. Scale bar 50 μm. (B) Villus length and crypt depth measurements (μm) from wild‐type and LGR5‐H2B‐GFP pigs using ImageJ line tool (NIH). Statistical analysis by student's t‐test for each intestinal segment revealed no significant differences in measurements between pigs (p > .05); data points represent 3–5 pigs per section with 10 villi/crypt axis measured per pig. Images were acquired on an IX83 inverted fluorescent microscope (Olympus) using a DP26 color camera (Olympus), 10X objective (LIC Plan FLN; Olympus), and cellSens software (Olympus).

3.3. LGR5‐H2B‐GFP cells are crypt‐base epithelial cells that co‐express known intestinal epithelial stem cell biomarkers

In the pig, small intestinal and colon crypt epithelial cell populations include ISCs, Paneth‐like cells, and transit‐amplifying cells. ⁴² , ⁴⁷ The ISCs and Paneth‐like cells are positioned in an alternating fashion within the crypt‐base. Transit‐amplifying cells, the progeny of cycling ISCs, migrate up the crypt axis as they differentiate into mature intestinal epithelial cell populations. To define the location of LGR5‐H2B‐GFP⁺ cells and determine if GFP⁺ cell location was consistent with known ISC location based on previous literature, ¹⁴ , ⁴² endogenous GFP expression in LGR5‐H2B‐GFP porcine intestine was analyzed using fluorescent microscopy (Figure 3A). Consistently, GFP‐positive cells localized to the crypt‐base epithelium within all segments of the intestine (Figure 3A). In jejunal and colon crypts, GFP fluorescent intensity normalized to DAPI was quantified for each nucleus starting at the crypt‐base cell position 0 up to cell position 30 in the crypt neck. This quantification revealed a gradient of GFP expression. The brightest cells were noted at the crypt‐base (cell position 0) and intensity decreased as cell position increased up the crypt axis (Figure 3B,C). These findings are expected given the construct of the model (Figure 1). ⁴⁸ As crypt‐base LGR5‐H2B‐GFP⁺ ISCs divide to give rise to LGR5⁻ transit‐amplifying populations, intranuclear H2B‐GFP is still present and its GFP signal intensity is diluted for each cell division of the daughter cells. Thus, loss of H2B‐GFP signal can be used as a “clock” to examine cell lineage within the intestinal epithelium. While the majority of transit‐amplifying daughter cells migrate upwards towards the villus tip during differentiation, a small subset is expected to move down towards cell position 0 as they develop into Paneth‐like cells. ⁴⁹ Therefore, quantification of fluorescence intensity from cell position 0 to 30 does not create a precise linear plot (Figure 3B,C). However, visualization of the highest intensity GFP expression only within the crypt‐base epithelium supported the conclusion that the LGR5‐H2B‐GFP‐positive cells are in the known location of ISCs.

FIGURE 3.

GFP‐expressing cells are crypt‐base epithelial cells and GFP expression decreases in a gradient along the crypt axis. (A) Representative images of endogenous GFP expression (Figure 3 cont.) (green) and bisbenzimide H 33342 (blue) within the duodenum, jejunum, ileum, and spiral colon of LGR5‐H2B‐GFP pigs. Scale bar 100 μm, insets 50 μm. Images were acquired on a Fluoview FV3000 upright confocal microscope (Olympus), 10X and 30X objectives, and cellSens software (Olympus). (B,C) Means and standard deviations of GFP intensity normalized to bisbenzimide H 33342 intensity quantifications per cell nucleus within jejunum and colon crypts, with a representative image of cell positions in jejunum and colon crypt; data points represent means and standard deviations from n = 3–8 pigs with 3–5 crypts measured per pig. Images acquired as in A. Intensity measured using ImageJ measure tool (NIH). Scale bar 50 μm.

In the process of examining endogenous GFP expression by fluorescent microscopy, it became apparent that not every intestinal crypt expressed comparable levels of GFP to its neighboring crypts (Figure 4A). As shown in the histologic image of endogenous GFP expression in the colon, clusters of adjacent crypts appeared low in GFP expression (GFP^dim). The calculated percentage of GFP^dim crypts showed that the colon had the highest rate of mosaic crypts (29.78+/−6.58%) followed by duodenum, jejunum, and ileum (15+/−8.72%; 4.2+/14.03%; 0.8+/−1.79% respectively) (Figure 4B). To verify that GFP^dim crypts still expressed LGR5, FISH using RNAscope technology was performed on sections of the colon (Figure 4C). Overlay with α‐GFP antibody demonstrated that regardless of GFP expression, all crypts expressed LGR5.

FIGURE 4.

LGR5‐H2B‐GFP pigs exhibit mosaic expression of endogenous GFP between crypts. (A) Representative colon histology showing both GFP^pos and GFP^dim crypts, adjacent to each other. Nuclei stained with bisbenzimide H 33342 (blue). Scale bar 100 μm, inset 50 μm. Images (Figure 4 cont.) acquired on both a Fluoview FV3000 upright confocal microscope (Olympus) using, 30× objective, and cellSens software (Olympus) and an IX83 inverted fluorescent microscope (Olympus) using monochrome digital camera (ORCA‐flash 4.0, Hamamatsu), 10X objective (LIC Plan FLN; Olympus). (B) Percentage of GFP^dim crypts counted in histologic slides from each segment of the intestine; n = 5 pigs; **p < .01, two‐way ANOVA with post‐hoc Tukey's test. (C) Colon histology of LGR5 RNAscope probe (red) with GFP antibody (green) showing all crypts, GFP^pos and GFP^dim, express LGR5 mRNA. Nuclei stained with bisbenzimide H 33342 (blue). Scale bars 100 μm. Images were acquired on a Fluoview FV3000 upright confocal microscope (Olympus), 10× objective, and cellSens software (Olympus). (D) 3‐D colon crypt culture of GFP^dim crypt on day one and day five in culture, demonstrating normal colonoid growth in culture. Images were acquired on an IX83 inverted fluorescent microscope (Olympus) using a monochrome digital camera (ORCA‐flash 4.0, Hamamatsu), 20X objective (LIC Plan FLN; Olympus), and cellSens software (Olympus). Scale bar 50 μm. (E) Representative flow cytometry histogram of GFP^pos/neg populations from single live colonoid cells derived from GFP^pos (green) and GFP^dim (gray) colonoids after 7 days in culture, showing decreased GFP (FITC) expression in GFP^dim colonoids. (F) qRT‐PCR quantification of LGR5 and GFP in isolated GFP^dim colonoids compared to GFP^pos colonoids after 7 days in culture. The ΔΔC_t method was used to measure relative fold change in gene expression compared with control (GFP^pos colonoids). The dotted horizontal line indicates control sample set = 1. GAPDH was used as the housekeeping gene. Samples were tested in triplicate; n = 3 pigs, >20 colonoids per pig per GFP expression group; ****p < .0001; statistical analysis by student's t‐test.

Despite the lack of GFP expression, GFP^dim colon crypts grew as expected in 3D culture and developed into colonoids that remained GFP^dim (Figure 4D). These GFP^dim colonoids were individually isolated from culture and analyzed by flow cytometry to quantify GFP expression compared with GFP‐positive (GFP^pos) colonoids (Figure 4E). Rare GFP‐positive cells were still apparent in the GFP^dim colonoid population. The mean fluorescence intensity of GFP^dim colonoid cells was reduced compared with GFP^pos (1.36 × 10⁵+/−7.33 × 10⁴ and 1.71 × 10⁶+/−6.29 × 10⁵, respectively). The same populations of colonoids were analyzed by qRT‐PCR to further quantify LGR5 and GFP expression (Figure 4F), which revealed that the GFP^dim colonoids, while no different in LGR5 expression compared with GFP^pos colonoids, did have significantly reduced GFP expression (0.01633+/−0.0074; p < .0001). Together, these findings suggest that monoallelic silencing or alternate splicing of the LGR5‐H2B‐GFP allele may be driving the differential GFP expression, and this variation is carried along with the process of crypt fission in the growing intestine to create the adjacent populations of GFP^dim crypts. ⁵⁰ , ⁵¹

To further characterize the LGR5‐H2B‐GFP‐positive cells and support their identity as ISCs, we examined the co‐expression of GFP with previously validated porcine ISC protein biomarkers. Porcine research is somewhat limited by the availability of porcine‐specific functional antibodies, thus, the antibody biomarkers Sex‐determining region Y box 9 (SOX9) and proliferation marker Ki67 have been used in immunohistochemical analyses to identify ISCs. ⁴ , ⁴² SOX9, a well‐accepted biomarker of crypt‐base cell populations, is not unique to the ISCs as Paneth cells and transit‐amplifying cell populations also express the SOX9 protein. ⁵² , ⁵³ , ⁵⁴ Therefore, to identify the actively proliferating ISC phenotype with no porcine‐specific antibodies available, co‐localization of α‐SOX9 with proliferation marker α‐KI67 was necessary. ⁴ Immunofluorescent analysis of both the jejunum and colon revealed that LGR5‐H2B‐GFP‐positive cells, identified by α‐GFP antibody, co‐localized with distinct cell populations of both SOX9⁺ and SOX9⁺KI67⁺ cells within the crypt‐base (Figure 5A). Cells that were SOX9⁺KI67⁺ were also GFP⁺; these observations indicate that LGR5‐H2B‐GFP‐positive cells correlate with the previously identified porcine ISC pool.

FIGURE 5.

LGR5‐H2B‐GFP positive cells co‐localize with ISC biomarkers and not Paneth‐like cells. (A) Immunofluorescent images of α‐GFP⁺ cells (green), co‐localized with two antibodies used to identify ISCs: α‐Ki67 (white) and α‐SOX9 (red) in jejunum and colon crypts. Nuclei were stained with bisbenzimide H 33342 (blue). Scale bar 20 μm. Images were acquired on an IX83 inverted fluorescent microscope (Olympus) using a monochrome digital camera (ORCA‐flash 4.0, Hamamatsu), 40× objectives (LIC Plan FLN; Olympus), and cellSens software (Olympus). (B) Immunofluorescent image of α‐GFP⁺ cells (green), α‐UEA⁺ (red) and bisbenzimide H 33342 (blue) in jejunum crypt. The arrow points to α‐UEA⁺ α‐GFP⁻ cell. Scale bar 20 μm. Image acquired on a Fluoview FV3000 upright confocal microscope (Olympus), 60X objective, and cellSens software (Olympus).

The SOX9⁺Ki67⁺GFP⁺ cells within the crypt‐base were intercalated with SOX9⁺ only cells—suspected to be the porcine Paneth‐like cell population. ⁴² , ⁴⁷ To prove that these crypt‐base GFP⁻SOX9⁺ cells were of the porcine Paneth‐like cell phenotype distinct from the GFP⁺ ISCs, additional co‐localization with α‐Ulex Europaeus Agglutinin I (UEA), a marker of Paneth cells, was performed. ⁵⁵ Indeed, α‐UEA⁺ cells were discrete GFP⁻ populations intercalated with the crypt‐base GFP⁺ cells (Figure 5B). This indicates that porcine Paneth‐like cells are distinct from LGR5‐H2B‐GFP cells.

The co‐localization of α‐GFP and α‐SOX9 within cells situated above the crypt‐base stem cell zone (Figure 5A), along with the GFP intensity plot (Figure 3B,C), confirms that the transit‐amplifying cell population (SOX9⁺) also carries some degree of GFP expression. Recognizing that the LGR5‐H2B‐GFP pig is a variation on lineage tracing given the expression of GFP in daughter progenitor cells, we hypothesized that the LGR5‐H2B‐GFP^hi cells are the ISCs. Therefore, we tested our hypothesis by evaluating these cells for their expression of ISC signature genes and ability to form enteroids from single cells.

3.4. LGR5‐H2B‐GFP^hi cells are enriched in intestinal epithelial stem cell gene biomarkers and exhibit increased growth potential of complex enteroids in culture

To show that LGR5‐H2B‐GFP^hi cells have both the enteroid‐forming capacity characteristic of ISCs and enriched levels of ISC gene biomarker expression, jejunal epithelial cells were fluorescence‐activated cell (FACS) sorted into LGR5‐H2B‐GFP^{hi/med/lo/neg} populations (Figure 6A). These populations were compared by quantitative real‐time polymerase chain reaction (qRT‐PCR) to measure known ISC biomarker gene expression. There are two described subgroups of ISCs, active and reserve, with the active population marked by LGR5 and OLFM4 expression. ² , ²² , ⁵⁶ The LGR5‐H2B‐GFP^hi cells had significantly increased expression of LGR5 compared with LGR5‐H2B‐GFP^med/lo/neg populations (9.74+/−5.25; 0.88+/−0.81; 0.34+/−0.21; 0.25+/−0.18; p < .01) as well as OLFM4 (15.06+/−11.2; 1.52+/−1.07; 0.48+/−0.34; 0.16+/−0.06; p < .05) (Figure 6B).

FIGURE 6.

LGR5‐H2B‐GFP^hi cells are enriched for aISC genes and develop enteroids in culture. (A) Representative flow sorting gating strategy to collect LGR5‐H2B‐GFP cells for gene expression and culture based on forward scatter (FSC), side scatter (SSC), live/dead marker (Propidium Iodide/AutoF), and GFP expression. (B) qRT‐PCR analysis of active ISC gene biomarker expression in sorted LGR5‐H2B‐GFP^{hi/med/lo/neg} cell populations. The ΔΔC_t method was used to measure relative fold change in gene expression compared with control (whole crypt epithelium). The dotted horizontal line indicates control sample set = 1. GAPDH was used as the housekeeping gene. Samples were tested in triplicate; data points represent 3 different pigs *p < .05 **p < .01, two‐way ANOVA with post‐hoc Tukey's test. (C) Representative time course of growth of LGR‐H2B‐GFP^hi jejunum single cell in 3‐D culture. Scale bar 20 μm. (D) Percentage of LGR5‐H2B‐GFP^{hi/med/lo/neg} single cells that developed into enteroids by day 10 in 3D culture; data points represent 3 different pigs, ***p < .0001, two‐way ANOVA with post‐hoc Tukey's test. (E) Representative image of LGR5‐H2B‐GFP^hi cells by third passage, day 5 in 3‐D culture; n = 2 pigs. Scale bar 200 μm. All figure images were acquired on an IX83 inverted fluorescent microscope (Olympus) using a monochrome digital camera (ORCA‐flash 4.0, Hamamatsu), 10X, 20X, and 40X objectives (LIC Plan FLN; Olympus), and cellSens software (Olympus). (F) Image of gel electrophoresis confirming the presence of differentiated intestinal epithelial cell biomarker genes expressed within enteroids grown from LGR5‐H2B‐GFP^hi cells after 10 days in culture; n = 3 pigs, 30–60 enteroids per pig. Image acquired using ChemiDoc‐It 2.

Additionally, each FACS‐sorted LGR5‐H2B‐GFP population was cultured to assess enteroid growth potential. Jejunal LGR5‐H2B‐GFP^hi single cells sorted for 3D culture grew into complex enteroids typical of porcine jejunum crypt culture. These enteroids exhibited classic budding of crypt‐like structures with a central lumen by day 7 in culture (Figure 6C). Additionally, LGR5‐H2B‐GFP^hi cell cultures exhibited significantly higher percentage of enteroid growth efficiency compared with LGR5‐H2B‐GFP^med/lo/neg populations (2.47+/−0.52; 0.2+/−0.1; 0.12+/−0.13; 0.16+/−0.28; p < .0001) (Figure 6D). Furthermore, passaged enteroids from LGR5‐H2B‐GFP^hi sorted cells continued to proliferate in culture (3 passages performed) (Figure 6E). To demonstrate that enteroids derived from GFP^hi cells had the capacity to produce the post‐mitotic intestinal epithelial cell types, we evaluated complex enteroids grown from LGR5‐H2B‐GFP^hi FACS sorted cells and confirmed gene biomarker expression by PCR (LYZ, MUC2, CCK, ITF, CGA, SGLT‐1, and FABP; Figure 6F). ⁴² , ⁵⁷ , ⁵⁸ , ⁵⁹

3.5. In situ comparison of crypt‐base gene biomarkers shows similarities between pigs and humans that are not present in mice

The LGR5‐H2B‐GFP pig has a substantial potential impact as an improved translational large animal model of human intestinal health and disease. To further underscore the utility of this porcine model over traditional murine models, we used RNAscope FISH to compare key, crypt‐base biomarker gene expression between LGR5‐H2B‐GFP pig, mouse, and human ileal and colonic biopsies. In the ileum and colon for all species, LGR5 expression was identified as expected within crypt‐base cells (Figure 7).

FIGURE 7.

Crypt‐base LGR5 expression is similar between humans and pigs. Representative histologic images from the ileum and colon of healthy human donors, LGR5‐H2B‐GFP pigs, and mice with LGR5 mRNA probe (magenta) and with α‐GFP antibody (green). Nuclei stained with bisbenzimide H 33342 (gray). Scale bar 25 μm. Two to three animals/patients per species were analyzed. Images were acquired on a Fluoview FV3000 upright confocal microscope (Olympus), 10X and 40X objective, and cellSens software (Olympus).

As previously noted, active ISCs have numerous biomarkers in addition to LGR5, including OLFM4. ⁵⁶ In human samples of both the small intestine and colon, OLFM4 was previously shown by van der Flier et al. to overlap with LGR5, which was also demonstrated here (Figure 8A). Similar to humans, porcine OLFM4 was expressed both in small intestinal ISCs and colonic LGR5‐H2B‐GFP ISCs (Figure 8A). This is in direct contrast to the mouse, in which Olfm4 expression is restricted to small intestinal ISCs only (Figure 8A). ⁵⁶ Separate from the actively cycling ISCs (LGR5⁺, OLFM4⁺) are the quiescent, slower‐cycling ISCs, also known as the “+4” position stem cell population. ⁶⁰ While this population has a biomarker set once thought to be unique identifiers of “+4” ISCs, several studies have demonstrated that these markers, including Hopx, are also expressed in murine Lgr5 ⁺ cells. ⁶¹ , ⁶² Here, we again demonstrated the overlap of Hopx and Lgr5 in mouse small intestinal and colonic ISCs (Figure 8B). Furthermore, both pigs and humans exhibited similar overlap in ISC expression of LGR5 and HOPX within both the small intestine and colon (Figure 8B).

FIGURE 8.

OLFM4 and HOPX are similarly expressed between pigs and humans. Representative histologic images from the ileum and colon of healthy human donors, LGR5‐H2B‐GFP pigs, and mice with LGR5 (magenta) and either (A) OLFM4 or (B) HOPX mRNA probes (green). Nuclei stained with DAPI (gray). Scale bar 25 μm. Two to three animals/patients per species were analyzed. Images were acquired on a Zeiss LSM 880 microscope with 40X objective.

The intestinal crypt‐base contains other cell types in addition to ISCs, namely progenitor cells sitting just above ISCs and Paneth cells interspersed between ISCs. Common gene biomarkers for these cells include SOX9 (Paneth, ISCs, and progenitor cells) and LYZ (Paneth cells). ⁵² , ⁵³ , ⁵⁴ , ⁶³ , ⁶⁴ Humans, mice, and pigs produce LYZ in the small intestine. ⁴⁷ , ⁶³ However, while humans are known to express LYZ in the colon, rodents do not. ⁶³ In the pig, colonic in situ LYZ expression has yet to be defined. Here, we examined LYZ expression in human, pig, and mouse ileum and colon. In direct contrast to the mouse, both human and pig colon crypt‐bases contained LYZ ⁺ cells (Figure 9). When we identified SOX9 ⁺ cells between species, as expected, SOX9 overlapped with LGR5 ⁺ and LYZ ⁺ crypt‐base cells, indicating both ISC and Paneth cell populations, respectively (Figure 9). ⁵² , ⁵³ , ⁵⁴ , ⁶⁴ In the human and pig transit‐amplifying/progenitor cell region, SOX9+ cells were also positive for LGR5 and LYZ. This phenomenon was not appreciated in the murine tissues, where LYZ was restricted to small intestinal crypt‐base cells only (Figure 9).

FIGURE 9.

Colon crypt cells co‐express LYZ and SOX9 in humans and pigs. Representative histologic images from the ileum and colon of healthy human donors, LGR5‐H2B‐GFP pigs, and mice with LGR5 (magenta), SOX9 (cyan), and LYZ (green) mRNA probes. Nuclei stained with DAPI (gray). Scale bar 25 μm. Two to three animals/patients per species were analyzed. Images were acquired on a Zeiss LSM 880 microscope with 40X objective.

3.6. Disruption of APC in LGR5‐H2B‐GFP colonoids leads to cystic growth in culture and constitutive WNT/β‐catenin signaling

To demonstrate the utility of the transgenic porcine model for the study of human intestinal disease, specifically CRC, we disrupted APC in LGR5‐H2B‐GFP colonoids. To achieve this goal of an APC‐edited LGR5‐H2B‐GFP model, we used CRISPR/Cas9 editing to target part of the APC sequence in exon 9 via nucleofection in LGR5‐H2B‐GFP colonoids. Colonoid editing efficiency was confirmed by tracking of indels by decomposition (TIDE) after the edited colonoids were grown in complete media for 7 days (58.73+/−19.76%). Colonoids were then passaged into selective media, depleted of Wnt, R‐spondin, and CHIR, which enriched editing efficiency to 82.20+/−17.87% (Figure 10A). These LGR5‐H2B‐GFP/APC ^null colonoids in selective media demonstrated cystic growth that paralleled similar previous murine and human APC edited colonoid models, while unedited colonoids failed to survive (Figure 10B). ¹³ , ⁶⁵

FIGURE 10.

Development and in vitro culture of LGR5‐H2B‐GFP/APC ^null colonoids. (A) Representative readout of editing efficiency as determined by tracking of indels by (Figure 10 cont.) decomposition (TIDE) in LGR5‐H2B‐GFP/APC ^null colonoids. The bar graph represents TIDE percent editing efficiency data from LGR5‐H2B‐GFP/APC ^null colonoids in complete media for 7 days (magenta) compared with after passaging in selective media for 7 days (blue). Data points represent independent transfection replicates from 2 pigs, >100 colonoids used per data point. (B) Representative images of LGR5‐H2B‐GFP/APC ^null colonoids grown in complete media compared with selective media, and unedited LGR5‐H2B‐GFP colonoids grown in selective media. Images captured at 7 days in culture. Scale bar 100 μm. Images were acquired on an IX83 inverted fluorescent microscope (Olympus) using a monochrome digital camera (ORCA‐flash 4.0, Hamamatsu), 10X objective (LIC Plan FLN; Olympus), and cellSens software (Olympus). (C) qRT‐PCR analysis of AXIN2, CCND1, LGR5, OLFM4, and HOPX expression in LGR5‐H2B‐GFP/APC ^null colonoids grown in selective media for 3–7 days. The ΔΔC_t method was used to measure relative fold change in gene expression compared with control (unedited colonoids grown in complete media). The dotted horizontal line indicates control sample set = 1. GAPDH was used as the housekeeping gene. Samples were tested in triplicate; data points represent independent transfection replicates from 2 different pigs, >100 colonoids used per data point. Statistical analysis by student's t‐test, *p < .05.

As APC serves to suppress WNT/β‐catenin signaling in unedited epithelial cells, ⁶⁶ , ⁶⁷ we confirmed that APC mutation in the LGR5‐H2B‐GFP colonoids led to constitutive activation of WNT/β‐catenin. This constitutive activation can be identified by increased AXIN2 expression, a well‐described WNT/β‐catenin pathway target gene. ¹⁵ , ⁶⁵ , ⁶⁸ Compared with non‐edited, complete media‐grown colonoids (control), selective media‐grown APC‐edited colonoids had significantly upregulated AXIN2 (fold change 7.97+/−3.98; p < .05; Figure 10C). The same APC ^null colonoids also had significantly increased expression of other WNT/β‐catenin signaling targets including CCND1, LGR5, and OLFM4 (fold changes 6.41+/−4.93, 4.20+/−2.95, 24.42+/−11.06, respectively; p < .05). Interestingly, the APC ^null colonoids were also significantly increased in the expression of the HOPX tumor‐suppressor gene (19.53+/−11.58; p < .05).

To assess potential tumorgenicity, colonoscopy‐guided submucosal injection of the LGR5‐H2B‐GFP/APC ^null organoids was performed in NOD.Cg‐Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice as previously described. ⁴⁵ , ⁴⁶ Engraftment of LGR5‐H2B‐GFP/APC ^null colonoids was confirmed by visual evidence of a raised area of dysplasia formation positive for GFP expression upon recheck colonoscopy 10 days post‐injection (Figure 11A). At euthanasia 3–5 weeks post‐injection, dysplasia formation was confirmed by the presence of architectural aberration and abnormal nuclei orientation on histology (Figure 11B). However, at the time of euthanasia, neither endogenous GFP nor α‐GFP antibody expression was detected in the histologic section.

FIGURE 11.

In vivo engraftment of LGR5‐H2B‐GFP/APC ^null colonoids. (A) Representative colonoscopy images from LGR5‐H2B‐GFP/APC ^null submucosal injection into NSG mice colons at injection and at 10 days post‐injection recheck to confirm tumor formation. Tumor GFP expression outlined by a dotted line. (B) Representative histologic image of mouse colon with dysplasia formation. Scale bar 200 μm unless otherwise indicated. Images were acquired on an IX83 inverted fluorescent microscope (Olympus) using DP26 color camera (Olympus), 4X and 40X objective (LIC Plan FLN; Olympus), and cellSens software (Olympus).

4. DISCUSSION

To date, ISC research has relied nearly completely on murine ISC transgenic models as the gold standard for advanced analysis of ISC biology, repair, and tumorigenesis. ² , ⁸ , ¹⁶ , ²⁵ However, porcine models are increasing in popularity for translational gastrointestinal research because human and porcine intestinal anatomy and physiology are closely paralleled in size, diet, microbiome, and immune cell biology. ³⁰ , ³¹ , ³⁹ Further advancement of these porcine intestinal models for more mechanistic studies of human disease has been limited by a paucity of transgenic models.

In this study, we characterized a transgenic LGR5 reporter pig, the first of its kind, to validate its use as a large animal model to identify and isolate ISCs. Collectively, our findings reveal that the LGR5‐H2B‐GFP^hi cells, expressing nuclear GFP, are the LGR5⁺ ISCs. These findings were verified by the histologic location of GFP⁺ cells in the crypt‐base, ISC signature protein biomarker co‐localization with GFP⁺ cells, significantly increased LGR5 gene expression within the LGR5‐H2B‐GFP^hi population, and enteroid formation in culture from LGR5‐H2B‐GFP^hi single cells. Each of these findings underscores the utility of the LGR5‐H2B‐GFP pig as a translational model to identify and interrogate ISCs.

In addition, ileal and colonic biopsies of humans, mice, and pigs were compared by in situ hybridization to underscore the utility of the pig as a model for crypt‐base cell studies. While all three species demonstrated similar expression of LGR5, further in situ examination of other crypt‐base cell markers OLFM4, HOPX, LYZ, and SOX9 highlighted important similarities between pig and human that were not present in the mouse. Notably, previous literature detected OLFM4 expression in porcine colonic epithelium by qRT‐PCR ⁶⁹ ; however, to the authors' knowledge, this is the first in situ identification of porcine OLFM4 in both the small intestine and colon, which also mirrors human OLFM4 expression. This is in direct contrast to the mouse, which does not express OLFM4 in the colon. Similar OLFM4 expression in both human and porcine colon could be important for future improved modeling of intestinal diseases such as CRC where OLFM4 is a known marker for a subset of cancer cells and potential metastasis. ⁵⁶ , ⁷⁰ , ⁷¹

Moreover, both porcine and human tissue also exhibit similar in situ co‐expression of LGR5 and HOPX, which further validates several recent studies showing overlap of LGR5 and the proposed unique quiescent/ “+4” stem cell markers. ⁶¹ , ⁶² , ⁷² Comparison of in situ LYZ expression between species revealed that pig colon crypt‐base cells are positive for LYZ, similar to humans, while mice are not. This finding may be important for potential future studies of intestinal diseases such as inflammatory bowel disease (IBD). In human patients suffering from IBD, LYZ‐producing Paneth cells are over‐expressed in the colon. ⁷³ Previous studies have attempted to further investigate this disease pathophysiology by creating transgenic mice that overexpress Lyz in the colon ⁶³ ; however, given the naturally occurring similarity in LYZ expression between humans and pigs, the pig has the potential to be an improved translational large animal model for IBD research. Furthermore, similar co‐expression of LYZ, LGR5, and SOX9 located in the progenitor cell zone within both human and pig cells, but not in mouse, highlights the need for additional comparative transcriptomic and proteomic analysis between pig and human crypt‐base cells. Such analysis in the LGR5‐H2B‐GFP model could potentially use the crypt‐base cell GFP intensity gradient to identify different progeny of LGR5⁺ ISCs, thereby expanding the possible model utility.

To specifically demonstrate the use of the LGR5‐H2B‐GFP porcine model for intestinal disease research, we used CRISPR/Cas9 editing to create APC‐disrupted colonoids. These LGR5‐H2B‐GFP/APC ^null colonoids showed APC editing efficiency of over 80%, developed cystic organoid structures in growth factor depleted selective media, and had increased expression of WNT/β‐catenin pathway target genes AXIN2, CCND1, LGR5, and OLFM4. Similar colonoid growth characteristics and WNT/β‐catenin pathway target gene expression changes have been previously described in both human and murine APC knock‐out colonoid experiments, ¹³ , ⁶⁵ although the over‐expression of OLFM4 in this new porcine model is of particular interest. Given both the known absence of Olfm4 in the murine colon, and the known over‐expression of OLFM4 in early‐stage CRC formation in human patients, the increase in OLFM4 in this transgenic APC ^null porcine colonoid model further underscores its translational potential for human disease study. ⁵⁶ Interestingly, the LGR5‐H2B‐GFP/APC ^null colonoids also showed increased expression of known tumor‐suppressor gene HOPX. ⁷⁴ , ⁷⁵ In a recent lung cancer study, HOPX was shown to inhibit WNT/β‐catenin signaling. ⁷⁶ In the colonoid model presented here, APC ^null colonoid cells may be over‐expressing HOPX in an attempt to control the upregulated cellular proliferation. Further investigation is warranted to understand this relationship between HOPX and WNT/β‐catenin signaling in this porcine colonoid CRC platform.

While orthotopic, submucosal injection of these porcine colonoids into immunodeficient mice led to dysplastic growth that expressed GFP on 10‐day post‐injection colonoscopy, neither endogenous GFP nor α‐GFP antibody was appreciated within the dysplastic region on histology at 4 weeks post‐injection, indicating a likely loss of the porcine GFP⁺ cells over time. The lack of more developed adenoma formation in this model suggests the need to induce additional CRC mutations including KRAS, SMAD4, and TP53 in the LGR5‐H2B‐GFP colonoids to increase in vivo tumorgenicity. ⁴⁵ , ⁴⁶ , ⁶⁵ , ⁷⁷ Despite these current shortcomings in the in vivo application, this novel combination of ISC reporter and CRC gene mutation in a translational porcine model is important to progress the field of CRC research. Previously, Flisikowska et al. created an APC mutated pig line, which develops dysplastic adenomas in the large intestine, similar to the precancerous lesions in human patients. ⁷⁸ These pigs have contributed greatly to understanding CRC cellular mechanisms and improving diagnostics/therapeutics for human CRC patients in ways that previous mouse models could not. ⁷⁹ , ⁸⁰ , ⁸¹ , ⁸² Historically, colon cancer has been difficult to model completely in genetically modified mice due to the tendency of tumors to form in the small intestine instead of the colon. ⁷⁷ Furthermore, the small size of mice makes it difficult for clinicians and researchers to improve endoscopic and surgical techniques to detect and remove CRC. Additional porcine CRC models would significantly advance the field of translational CRC research in these areas. In particular, given the previously described role of LGR5⁺ cancer stem cells in CRC tumor development, ⁷ , ⁸ , ⁹ , ¹⁴ the LGR5‐H2B‐GFP pig will allow critical in vivo ISC tracking and isolation from tumors. Inducing the other common sequential CRC gene mutations in the LGR5‐H2B‐GFP colonoids beyond APC will likely allow modeling of the adenoma‐carcinoma sequence with potential to form invasive carcinoma upon orthotopic auto‐transplantation. ⁴⁵ , ⁴⁶ , ⁶⁵ , ⁷⁷

The GFP^dim crypts observed in this study have also been demonstrated in ISC transgenic mice. Mosaicism has been long known to occur in the crypts of LGR5‐eGFP‐IRES‐CreERT2 mice. ² , ⁸³ It is unknown, in that model, what specific mechanisms drive the process, although it has been hypothesized to be due to the inserted DNA sequence itself or the location of the insertion. ⁸⁴ Research on other transgenes supports that during development, alterations to the DNA sequence can result in variegation that leads to stochastic levels of expression of transgenes from cell to cell, and this is especially true in essential genes, termed “spatial effect variegation” (different than position effect variegation due to random insertion). ⁸⁵ , ⁸⁶ Our data confirm that this pattern is also observed within a different species (porcine), with a different transgene organization, and insertion in the 5′ end, and further that the silencing alters H2B‐GFP expression but not LGR5 expression, as confirmed by RNAscope in situ hybridization and qRT‐PCR (Figure 3C–F). Because GFP^dim crypts were observed in adjacent clusters (Figure 3A), our data support the idea that the mechanism driving differential GFP expression is likely occurring early in intestinal development. Subsequent crypt fission of a GFP^dim crypt as the intestine grows leads to adjacent populations of GFP^dim crypts. Further research needs to be done to determine if this observation is reflective of a previously unreported LGR5 gene transcriptional regulatory system (i.e., splicing variants, alternate promoters, or other epigenetic modifications) in the colon.

The variable percentage of mosaic crypts noted between sections of the intestine can likely be attributed to differential rates of crypt fission. In both porcine and murine studies, compared with the small intestine and cecum, the rate of crypt fission in the colon is markedly increased. ⁵⁰ , ⁸⁷ , ⁸⁸ Despite the mosaic pattern in the crypts, it is still clear that within the crypts that express GFP, we can reliably define the LGR5‐H2B‐GFP^hi cells as LGR5+ ISCs. Furthermore, this mosaic expression has not been observed in other tissues in this pig line. This suggests that this mosaicism is unique to certain tissues, in this case, the intestinal tract. ⁴⁰

The LGR5‐H2B‐GFP transgenic pig serves as a milestone large animal model to reliably identify and isolate ISCs using the LGR5 gene. Such advancement in large animal transgenic models allows for critical improvement in translational basic science studies involving intestinal pathophysiology. In combination with previous ISC knowledge gained from murine studies and this new development in ISC identification in a translational large animal model, we can better model pathophysiology and discover treatments for numerous clinical intestinal disease processes including CRC, ischemia–reperfusion injury, IBDs, neonatal necrotizing enterocolitis, allograft failure, and short bowel syndrome.

AUTHOR CONTRIBUTIONS

Cecilia R. Schaaf, DVM, PhD, and Liara M. Gonzalez, DVM, PhD, contributed to the conceptualizing, designing, analyzing, and interpreting of all study data. Cecilia R. Schaaf, DVM, PhD, performed and acquired the data on all components of this study and led the drafting of the manuscript. Jorge A. Piedrahita, PhD, and Kathryn M. Polkoff, PhD, conceptualized, designed, and created the transgenic pig. Amy S. Stewart, DVM, PhD, and John Freund acquired data, supported analyzing and interpreting data related to intestinal stem cells. Jatin Roper, MD, Jorge A. Piedrahita, PhD, and Breanna Sheahan, DVM, PhD, conceptualized the LGR5‐H2B‐GFP/APC^null colonoids and mouse engraftment studies. Amber Carter, with the help of Kathryn Polkoff, performed the research, acquired, and analyzed the data related to colon cancer studies. Joshua C. Snyder, PhD, and Joshua Ginzel conceptualized and assisted in acquiring and analyzing the data obtained from RNA scope assays. All authors were involved in editing and revising the manuscript.

FUNDING INFORMATION

HHS | National Institutes of Health (NIH) (R21 OD019738, R21CA256001, P30 DK034987), HHS | NIH | Office of Research Infrastructure Programs, National Institutes of Health (ORIP) (K01 OD019911‐01A1).

DISCLOSURES

Jorge Piedrahita has a non‐exclusive license to distribute pigs through RenOVAte Biosciences. All other authors declare no conflicts of interest.

Schaaf CR, Polkoff KM, Carter A, et al. A LGR5 reporter pig model closely resembles human intestine for improved study of stem cells in disease. The FASEB Journal. 2023;37:e22975. doi: 10.1096/fj.202300223R

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available in the Methods section of this article. The LGR5‐H2B‐GFP pigs will be made available for distribution through Renovate Biosciences and the NIH Swine Resource and Research Center.